Event attribution studies, which assess whether and to what extent human-induced climate change has made extreme weather events more likely or severe, have become routine in recent years. For many regions, multiple studies now exist for the same type of extreme event, with research on heatwaves dominating in Europe and globally, while studies on heavy rainfall are the most represented ones in Asia and North America. However, significant gaps remain, particularly for small island states, which have been largely neglected by attribution research. The growing abundance of studies in certain regions and for certain hazards raises questions about the added value of additional attribution analyses, for example, extreme heat in Europe or India, or heavy rainfall in Ireland or China, where sufficient evidence already exists. While the precise definition of an extreme event can influence quantitative attribution results, recent findings indicate that the effect of different datasets often explains more variance than event definition, particularly for temperature extremes. This lecture will present these new insights, drawing on a decade of experience from World Weather Attribution, and discuss their implications for the broader field of event attribution and for proposed operational services, including when new studies are necessary and how methodological choices affect the interpretation of results.

How to cite: Otto, F., Barnes, C., Keeping, T., Philip, S., Pinto, I., Clarke, B., Zachariah, M., and Bergin, C.: The added value of yet another attribution study , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3780, https://doi.org/10.5194/egusphere-egu26-3780, 2026.

Mesozoic Oceanic Anoxic Events (OAEs) are critical geological episodes linked to global carbon cycle perturbations, climate warming, and ecosystem restructuring. However, the regional expression of OAEs in the eastern Tethys remains insufficiently constrained. This study focuses on the Early Jurassic Toarcian OAE (T-OAE)—integrating petrological, mineralogical, and geochemical analyses of two key sections to reconstruct Early Jurassic sedimentary evolution, paleoclimate-paleoenvironment dynamics, and their responses to the T-OAE. Pronounced negative carbon isotope excursions (CIEs) are recorded in both marine strata, correlatable with global T-OAE records. Intensified continental chemical weathering  and enhanced terrigenous detrital input are common responses of the eastern Tethys to T-OAE, driven by global warming. Redox proxies reveal oxic-suboxic conditions in open marine settings of the eastern Tethys during OAEs, regulated by regional factors (water depth, basin restriction, freshwater input), contrasting with the anoxic-euxinic environments in the western Tethys. Bioproductivity showed spatial heterogeneity: organic matter accumulation was controlled by redox conditions and productivity, with high accumulation in restricted lagoons versus low-moderate in open shelves.

This study reveals the regional response patterns of the eastern Tethys to Mesozoic OAEs, highlighting the spatial heterogeneity of redox and productivity dynamics, and provides new insights into the Mesozoic climate-ocean-biosphere system.

How to cite: Yi, J. and Fu, X.: Sedimentary Environment Evolution and Response to Mesozoic Toarcian Oceanic Anoxic Event (T-OAE) in the Eastern Tethys, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-279, https://doi.org/10.5194/egusphere-egu26-279, 2026.

Microbially Induced Sedimentary Structures (MISS) are syndepositional primary structures that occur both in some of the earliest forms of life and in modern environments. Throughout geological time, microorganisms developed metabolic strategies that enabled their establishment and proliferation in a wide range of settings, including arid environments such as deserts. However, MISS are widely recognized in tidal flats and other shallow-marine environments, whereas examples preserved in continental deposits remain comparatively scarce. A comprehensive review of Precambrian MISS occurrences indicates a notable expansion of documented records during the Mesoproterozoic, coincident with the assembly of the Columbia supercontinent and a concurrent rise in atmospheric oxygenation. These global transitions may have promoted the ecological diversification of microbial communities and facilitated their dispersal into progressively drier continental interiors. Under favorable conditions, microorganisms proliferate and form microbial mats that interact with external factors such as sedimentation, currents, erosional processes, and other physical drivers. Their presence in arid terrestrial deposits is thus of considerable importance, as it underscores how microbial communities evolved and developed adaptive capabilities that enabled them to colonize and persist within intermittently wet landscapes subjected to elevated environmental stress. This study documents the occurrence of MISS within continental desert depositional systems of the Mangabeira Formation, São Francisco Craton, Brazil (1.6 Ga), preserved in wet sandsheet deposits. These occurrences broaden the sparse record of MISS in Proterozoic desert environments and offer new constraints on the capacity of early microbial communities to endure highly stressful and intermittently wet conditions. To investigate the conditions that enabled microbial establishment in this ancient desert, this study applies a multi-method approach integrating sedimentological, stratigraphic, petrographic, and microtextural datasets. The vertical succession reveals six distinct drying-upward cycles, each associated with fluctuations in groundwater level that periodically generated stable, moisture-rich surfaces suitable for microbial mat development. Within these intervals, MISS occur in millimetric heterolithic laminites displaying wavy–crinkly lamination, wrinkle marks, roll-up structures, deformational features, authigenic minerals with convolute morphologies, trapped grains, and organic carbon remnants. Complementary Raman analyses reveal characteristic carbonaceous peaks (at ~1370, 1590, and 1610 cm⁻¹), confirming the presence of organic carbon and kerogen. Collectively, the integrated dataset indicates that microbial colonization in the Mangabeira Formation was episodically favored by groundwater-controlled moisture stability, which enhanced substrate cohesion and enabled the formation of distinctive biosedimentary fabrics. These findings, contextualized within the broader Mesoproterozoic expansion of MISS, highlight the capacity of early microbial communities to establish themselves in hydrologically stressed desert landscapes and refine the sedimentological and geochemical criteria necessary for recognizing MISS in deep-time continental systems.

How to cite: Feitosa, A., Bállico, M., Souza, E., Scherer, C., Callefo, F., Balbinot, V., Tatsch, G., Yokoyama, E., Leite, A., Reis, A., Silva, S., and Santos, A.: Microbially Induced Sedimentary Structures in a Mesoproterozoic Erg System: A Case Study from the Mangabeira Formation, Brazil, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-633, https://doi.org/10.5194/egusphere-egu26-633, 2026.

How to cite: Gkatzogias, A., Bitas, D., Georgiou, K., Amditis, A., and Michalis, P.: From Divers to Communities: An IoT-Based Crowdsourcing Sensing Approach to Protect Underwater Heritage Sites, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1934, https://doi.org/10.5194/egusphere-egu26-1934, 2026.

Our cultural memory is permanently endangered and, often, damaged irretrievably, given that cultural disasters, are quite repetitive, whether in cases of man-made or of natural disasters, armed conflicts and climate change combined with earth’s tectonic activity constituting, potentially, the primary causes.

The southwestern Peloponnese in Greece, precisely the region of Pylia, due to its proximity to the Hellenic trench, is considered to be tectonically as one of the most active areas in Greece, representing a major subduction zone. At the same time, it constitutes also a broad area with a very long history and a wealth of archaeological sites and relics. Focusing on the western coastline of the said region, directly exposed to variations of sea level height, as well as to conditions of rapid erosion due to the aggressive components of seawater, it should be considered as an area of ​​urgent priority to be monitored and protected.

Specifically, Voidokilia bay at the western coast of Messenia Prefecture, to the north of Navarino bay, a highly fragile ecosystem, actually under a NATURA Network protection status, constitutes the related case-study. Nevertheless, Voidokilia bay, is also one of the most attractive landscapes worldwide, thus facing rapid tourism challenges, ending up in increased disaster risks, both for the cultural as well as for the environmental assets. Accordingly, an interdisciplinary methodology within the scientific field of digital humanities has been applied for digitizing archaeological sites, excavated or not, underwater or not, as well as for highlighting their interdependence with the wider Messenia region’s archaeological sites network, further combined with trade routes connecting southwestern Peloponnese and the Aegean islands, via southeastern Peloponnese and Attica, thus fulfilling the notion of applied archaeology.

In this context, applying geoinformatics proved to be the most effective methodology for the related holistic cultural heritage management, in the perspective of an effective strategic planning on the part of the State apparatus, whether for private or public works, taking also into consideration charters, European directives and good practices, already applied worldwide, such as people’s community inclusion, in the direction of a public archaeology model.

The cornerstone of the specific research procedure has been extensive documentation, integration of different data types, such as archaeological, bibliographic, (palaeo)environmental, geospatial, remotely sensed imagery, for building up the sites’ multidimensional profile and revealing spatial relations and settlements’ interdependence, further highlighting the related buffer zones, in the perspective of delineating wider areas of archaeological profile, for anticipating the long-standing threats to archaeological assets such as rapidly increasing tourism, mismanaged development, poor excavation and looting, lack of conservation, climate change posing further significant threats to cultural heritage assets.

Conclusively, further constituting a potential contribution to the archaeological cadastre, already established by the Hellenic Republic, as well as proposing mild tourism development for keeping the balance between urban regeneration and environmental protection, in accordance with the Sustainable Development Goal 11-Sustainable cities and communities, one of the 17 SDGs established by the United Nations General Assembly in 2015, with the official mission to “Make cities inclusive, safe, resilient and sustainable”.

How to cite: Chroni, A. and Karathanassi, V.: The western Peloponnese coastline cultural landscape: cultural heritage management policy tools for making cities inclusive, safe, resilient and sustainable, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-2191, https://doi.org/10.5194/egusphere-egu26-2191, 2026.

Instrumental long-term climate records are scarce worldwide, especially in tropical regions such as the Brazilian Amazon. The lack of systematic data prior to the 20th century limits understanding of long-term climate variability and extreme events. This study is situated within the field of Historical climatology. It aims to contribute to knowledge of climate extreme events that occurred during periods of limited climatological data through analysis, focusing on the flooding events of 1859/60 and 1892 in the Amazon River basin. The research aligns 19th-century historical documents, such as newspapers, periodicals, official correspondence, and travel logs, with climate information from dendrochronological studies to reconstruct the magnitude, duration, and impacts of these flood events. Methodologically, it is constructed through an interdisciplinary approach combining environmental history, historical climatology, and hydrology, using written records as climate proxies that provide crucial information on river levels, rainfall seasonality, and flood persistence. Analysis of tree-rings from the Amazonian trees known as “Cedro-Vermelho” (Cedrela Odorata) indicates that the floods of 1859/60 and 1892 were among, if not the, most severe flooding extremes of the 19th century on the Amazon River basin. Historical descriptions of damage to livestock, farming, agriculture, urban infrastructure, and the living conditions of the population, which mainly consisted of “ribeirinhos”, a traditional culture and way of life near the rivers, back this up. The results demonstrate that rescuing and systematizing historical climate information from a region with a traditional lack of instrumental records helps fill a gap in tropical data. Regions such as those analyzed in this study have often been overlooked in historical climate research. The potential of historical records combined with dendrochronological analysis has proven extremely promising. It allows not only cross-validation of information but also the recovery of climate data through a non-conventional method of analyzing climate before the 20th century, helping to build a more comprehensive understanding of past climate in the vast Amazonian territory.

How to cite: Sousa, K., Granato, D., Neto, A., and Nunes, E.: Historical floods of the 19th century in the amazon, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3138, https://doi.org/10.5194/egusphere-egu26-3138, 2026.

The preservation of underwater and coastal cultural heritage is challenged by climate change, including sea-level rise, coastal erosion, extreme events and long-term environmental degradation. These threats require not only scientific monitoring and risk assessment, but also active engagement of local communities and stakeholders to foster awareness and citizen centered resilience.

This contribution presents the Augmented Reality (AR) crowdsourcing mobile application, to engage citizens, divers and local communities in exploring heritage sites, understanding climate-related risks and contributing to resilience strategies with collection of ground-truth data. The AR mobile application focuses on providing complex scientific knowledge into intuitive, place-based experiences accessible to non-expert audiences through interactive 3D reconstructions, contextualized information supported by visualizations of environmental and site-specific data over time. The AR mobile app supports enhanced learning and strengthens connections between citizens, heritage sites and scientific evidence, by allowing users to visualize digital content within their physical surroundings.

The application has been deployed across seven underwater and coastal pilot sites: the Equa Shipwreck (La Spezia, Italy), the Albenga A Shipwreck at Gallinara Island (Italy), the Hiorthhamn Arctic mining station (Svalbard, Norway), Lake IJssel (The Netherlands), the B-24 Liberator aircraft wreck (Algarve, Portugal), the Castle of Mykonos (Greece) and the Nissia Shipwreck (Cyprus). Each pilot features a tailored AR campaign reflecting its specific heritage value, ranging from Roman cargo vessels and WWII wrecks to Arctic industrial remains and coastal fortifications. Site-specific content visualises relevant climate hazards such as erosion, sea-level rise, storm impacts and material decay, while enabling users to explore excavation layers, alternative site states and historical reconstructions. The AR experiences are built using optimised 3D scans, reconstruction models, archival imagery, curated scientific content with interactive Points of Interest. Dynamic visualisations illustrate processes such as habitat formation, sediment movement, and structural transformation, supporting a deeper understanding of how environmental change affects heritage over time. All content has been developed in close collaboration with domain experts to ensure scientific accuracy and educational value.

A citizen-engagement study has been conducted to assess usability, user motivation, and the application’s effectiveness in raising awareness of climate risks to cultural heritage. Full validation across all pilot sites is taking place, ensuring that results reflect the cultural, geographic and environmental diversity of the seven pilot sites.

Acknowledgement:

This research has been funded by European Union’s Horizon Europe research and innovation programme under THETIDA project (Grant Agreement No. 101095253) (Technologies and methods for improved resilience and sustainable preservation of underwater and coastal cultural heritage to cope with climate change, natural hazards and environmental pollution).

How to cite: Koukoudis, K., Katika, T., Touramanis, A., Amditis, A., and Michalis, P.: Connecting Citizens, Science, and Vulnerable Heritage: An AR-Based Approach to Climate Resilience , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3862, https://doi.org/10.5194/egusphere-egu26-3862, 2026.

Tide-gauge observations are among the most reliable sources for assessing sea-level variability and its seasonal and temporal changes in coastal regions. This study analyzes sea-level records obtained from tide gauges along the Angolan coast for the period 2015–2020, with the objective of characterizing the seasonal and interannual variability of tides. The methodology included quality control and pre-processing of hourly and monthly sea-level data, removal of non-tidal signals, harmonic tidal analysis, and the assessment of seasonal variability and its statistical significance. Despite limitations related to data gaps, limited temporal resolution, and the lack of complementary oceanographic data, the results reveal pronounced seasonal and interannual variability in sea level. This variability reflects the combined influence of tidal dynamics, regional ocean circulation, wind forcing, and climate-related processes. The analysis highlights the importance of continuous and homogeneous tide-gauge records along the Angolan coast for improving the detection and interpretation of sea-level variability. The findings contribute to coastal monitoring efforts and provide relevant information for coastal management, risk assessment, and the development of adaptation strategies in the context of sea-level change.

Keywords:  Sea level variability, Tide-gauge observations, Seasonal and interannual variability, Angolan coast.

How to cite: Guilherme, F., Neves, M., and Lamas, L.: Seasonal and Interannual Variability of Tide-Gauge Records along the Angolan Coast for the period 2015 – 2020 , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4120, https://doi.org/10.5194/egusphere-egu26-4120, 2026.

To understand the climate conditions in Japan before the commencement of modern official meteorological observations, it is necessary to indirectly estimate them using proxy data that serve as climate indicators.

In Japan, there is a nearly continuous annual dataset of Lake Suwa's freezing records spanning over 580 years. Furthermore, diaries from various parts of Japan contain daily weather records. By utilizing these records, daily climate data with the minimum temporal resolution can be obtained. By leveraging these proxies, it is possible to reconstruct the climate of the cold season, which has been previously less understood, across various temporal and spatial scales.

The objective of this study is to reconstruct the changes in cold-season climate in Japan over the past several hundred years with high temporal resolution.

The proxy data currently used include: lake and terrestrial sediments (Lake Suwa, approximately 1000 years), records of cherry blossom flowering and full bloom dates primarily collected in Kyoto (approximately 1000 years), tree rings (approximately 300 years), daily weather records from diaries (approximately 200 years), freezing records of Lake Suwa and Lake Jusan (approximately 580 and 150 years, respectively), early-meteorological observation data (approximately 50 years), and Japan Meteorological Agency observation data (approximately 150 years).

Firstly, the most extensive dataset, the cherry blossom flowering data, is used as a reference. Next, proxy variables are standardized after removing trends caused by human activities. Subsequently, regression analysis is performed for each period where variations either coincide or do not coincide. Furthermore, for each proxy variable, spatial correlations were calculated using 20th-century meteorological observation data to identify the regions represented by that proxy variable.

(This research was funded by JSPS Grant-in-Aid for Scientific Research (24H00118).

How to cite: Hasegawa, N., Katata, G., Hirano, J., Yonenobu, H., Yasue, K., Kumon, F., Hatano, N., Takahashi, H., Zaiki, M., and Mikami, T.: Reconstruction of Japan's Cold-Season Climate in the Past Few Hundred Years Using High-Resolution Multi-Proxy, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4208, https://doi.org/10.5194/egusphere-egu26-4208, 2026.

Tropical convectionanomaly could serve as a crucial driver of global atmospheric teleconnections and weather extremes around the world. However, quantifying the dominances of convection anomalies with regional discrepancies, relevant for the variations of global atmospheric circulations, remains challenging. By using a network analysis of observation-based rainfall and ERA5 reanalysis datasets, our study reveals that El Niño-like convection is the most primary rainfall pattern driving the global atmospheric circulation variations. High local concurrences of above-normal rainfall events over equatorial central-eastern Pacific amplify their impacts, even though the most intense rainfall anomalies are observed near the Maritime Continent. Furthermore, we find that the impacts of El Niño- like convection will be tripled by the end of this century, as projected consistently by 23 climate models. Such “rich nodes get richer” phenomenon is probably attributable to the dipolar rainfall changes over theequatorial western-central Pacific. This study highlights the dominant role of El Niño- like convection on the global climate variations, especially under the future changing climate.

How to cite: Lin, S., Cai, F., Gerten, D., Yang, S., Jiang, X., Su, Z., and Kurths, J.: Intensified dominance of El Niño-like convection relevant for global atmospheric circulation variations, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4441, https://doi.org/10.5194/egusphere-egu26-4441, 2026.

The contribution of EM4C to the TRIQUETRA project addressed a central challenge in contemporary cultural heritage protection: transforming complex scientific risk assessment knowledge into practical, operational tools that support informed decision-making by professionals and heritage authorities. From the outset, EM4C adopted an application-oriented approach, extending beyond academic research to the development of structured methodologies and digital decision-support tools aligned with real-world conservation needs.

EM4C’s involvement spanned the full project lifecycle, from methodological design and knowledge structuring (WP3) to validation and evaluation (WP6), assessment of exploitation potential and future development pathways (WP7), and contribution to reporting and documentation activities (WP1). This integrated engagement ensured continuity between research, implementation, evaluation, and long-term usability of project outcomes.

Within WP3, and particularly Task 3.6, EM4C acted as task leader for the development of tools, methods, and technologies aimed at mitigating risks to cultural heritage sites. The work recognized that heritage vulnerability results from the interaction of multiple factors, including construction materials, environmental conditions, historical interventions, patterns of use, and diverse natural hazards exacerbated by climate change. Rather than addressing these factors in isolation, EM4C developed a structured framework reflecting real-world site behavior, where risks emerge through combined and cumulative effects over time.

A major challenge identified was the extreme complexity of potential risk scenarios. Initial theoretical analysis showed that more than 1.8 million combinations could arise when accounting for all variables, rendering manual assessment impractical. EM4C addressed this through a rational reduction process, grouping construction materials into four realistic tri-material combinations commonly found in heritage sites. This filtering reduced the scenario space to 15,120 valid and prioritized cases, maintaining representativeness while ensuring usability.

These scenarios were implemented digitally through two decision-support tools: the M REPORT ENGINE for monument-scale assessments and the LS REPORT ENGINE for landscape-scale risk management. Both tools generate structured technical outputs based on user-selected parameters such as materials, hazards, and risk intensity. Crucially, the proposed conservation and protection measures are grounded in an extensive manual synthesis of scientific literature, technical guidelines, and recognized good practices, ensuring technical accuracy, consistent terminology, and non-commercial neutrality.

The developed tools were evaluated within WP6 through presentations and hands-on assessments involving conservators, engineers, and cultural heritage authorities, including representatives of the Hellenic Ministry of Culture. Feedback collected through questionnaires and qualitative observations confirmed the tools’ clarity, relevance, and capacity to support structured decision-making, while also identifying directions for future refinement.

Within WP7, EM4C assessed the exploitation potential of the model and tools, demonstrating their adaptability to diverse institutional contexts and their suitability as flexible decision-support systems. The work highlighted their potential evolution into more specialized, data-integrated applications.

Overall, EM4C’s contribution effectively bridged theory and practice, delivering scientifically robust yet operationally meaningful tools that enhance the long-term impact and applicability of the TRIQUETRA approach to cultural heritage risk management.

How to cite: Spyrakos, V.: ABSTRACT - Overall Contribution of EM4C to the TRIQUETRA Project , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5292, https://doi.org/10.5194/egusphere-egu26-5292, 2026.

The Makgadikgadi Basin (MKB), in the central Kalahari Basin of northeastern Botswana, currently consists of a wide complex playa lake system, a relic of the Paleolake Makgadikgadi. Reconstructing the Quaternary depositional, environmental, and climatic history of the lacustrine-playa system has great significance for revealing the basin's evolution. However, its sedimentary record remains largely unexplored due to methodological challenges. In this study, four sediment cores, up to 1.6 m deep, were collected along a generally E-W transect from the western to central parts of the Ntwetwe pan, MKB.  Our multi-proxy record, including sedimentology, chronology, ostracod-based biostratigraphy, and clumped (∆₄₇) isotope geochemistry from these cores, reveals three complex hydroclimatic sequences that refine the environmental and climatic evolution of the MKB for the past 29 cal ka BP. The computed Bayesian age depth model and preliminary clumped isotope analysis on ostracod valves suggest the late Pleistocene (~29-19.5 cal ka BP) hypersaline-saline phase occurred under relatively low temperature conditions (∆₄₇-T = ~18.5-21°C), aligning with global glacial cooling and supporting interpretations of severe aridity in the Kalahari during the Last Glacial Maximum. The shift to a freshwater ostracod assemblage by ~5.2 cal ka BP partly corresponds to the termination of the African Humid Period (AHP), with a mean temperature (∆₄₇-T) of ~16.8°C. However, our record reveals significant complexity during the Late Holocene. The dominance of brackish water assemblage from ~4-1.6 cal ka BP suggests a prolonged transitional phase toward aridity, consistent with the broad trend of ITCZ retreat. Most notably, the late Holocene (~1.6-1 cal ka BP) assemblage, indicating a mix of brackish and freshwater taxa alongside extreme and warmer temperatures (∆₄₇-T = ~28.5°C). This implies a period of complex hydrological variability, potentially driven by increased summer rainfall variability or episodic flood inflow. Consequently, the Late Pleistocene and Middle Holocene data align with regional patterns, while the Late Holocene sequence particularly highlights the current extreme climate in the region, suggesting ostracod growth under extreme ephemeral playa lake conditions.

How to cite: Kahsay, T., Asrat, A., Sinha, N., Marchegiano, M., and Franchi, F.: Late Pleistocene to Holocene Multi-proxy Paleoenvironmental and Paleoclimatic Reconstruction of the Makgadikgadi Basin, Central Kalahari, Botswana, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7064, https://doi.org/10.5194/egusphere-egu26-7064, 2026.

How to cite: Pan, G. and Fu, X.: Orbital eccentricity pacing of hydroclimate variability in Qinghai-Tibet Plateau during the middle–late Eocene, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7110, https://doi.org/10.5194/egusphere-egu26-7110, 2026.

The Norian-Rhaetian boundary (NRB) marks a critical interval of Late Triassic global environmental instability, ecological crisis, and climatic transition, which preceded the sustained biodiversity decline culminating in the end-Triassic mass extinction. Despite its significance, the drivers of carbon cycle and biotic disturbances across the NRB remain unresolved. In most cases, these major mass extinctions in geological history are interpreted as chain reactions triggered by volcanic activity. Interestingly, the NRB and Rhaetian intervals lack compelling evidence for synchronous, precisely dated, large-scale volcanism with demonstrable global effects. In this context, the Central Atlantic Magmatic Province (CAMP) erupted later at ca. 201 Ma, while other impact-related triggers and/or proposed large igneous provinces (LIP), such as the Angayucham LIP in Alaska (214 ± 7 Ma), remained weakly constrained in magmatic timing, magnitude, and environmental significance. In the absence of significant volcanism, the mechanisms underlying carbon cycle perturbations and ecological crises become even more enigmatic.

Here, we present a high-resolution carbonate carbon isotope (δ¹³C_carb) profile spanning the Late Triassic to Early Jurassic from South China. Through independent U-Pb dating and cyclostratigraphic analysis, a high-precision astronomical timescale was established. Carbon isotope variations are strongly controlled by orbital cycles, and the record reveals two large-magnitude negative carbon isotope excursions (CIEs) at ca. 205 Ma and 201 Ma, corresponding to the NRB and Triassic-Jurassic Boundary (TJB), respectively. Our study posits that astronomically driven climate change persistently influenced the NRB and subsequent Rhaetian intervals, triggering a series of chain reactions involving climate, vegetation, carbon burial, greenhouse gas emissions, and other factors. Ultimately, it acted as an amplifier in the NRB event, leading to carbon cycle perturbations and ecological crises during this period, thus potentially preconditioning the Earth system for the subsequent end-Triassic mass extinction. This study further highlights the significance of low-latitude coastal areas as dynamic amplifiers of carbon cycle instability and underscores the vulnerability of modern carbon reservoirs under ongoing climate change.

How to cite: Lu, T. and Fu, X.: Astronomically Driven Climate Change as an Amplifier of Carbon Cycle Instability and Ecological Crisis at the Norian-Rhaetian Boundary, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8903, https://doi.org/10.5194/egusphere-egu26-8903, 2026.

Cultural heritage is exposed to a wide range of risks arising from natural processes, extreme events and human activities, making heritage resilience a challenging and complex issue. Existing risk assessment and management approaches often lack cohesion, difficult to access or insufficiently aligned with the everyday needs of heritage managers and local communities, resulting in gaps in understanding, as well as preparedness and response capacity.

This contribution focuses on addressing these challenges by merging scientific knowledge, field-based experience, and community generated awareness through an integrated digital environment. Within the European project THETIDA, a web-based visualization and decision-support platform has been developed with the main objective of supporting a holistic understanding of cultural heritage resilience. The platform integrates hazard information, environmental monitoring data, socio-economic indicators and spatial representations within a single, accessible interface, enabling users to explore and understand how multiple risks interact and affect heritage assets and their surrounding environments.

The platform delivers three main categories of services: (i) Remote Sensing–Based Services, including inundation and flood prediction, coastal erosion monitoring, material degradation mapping, land-use change detection, and geo-hazard assessment; (ii) In-Situ Sensing Services, supporting on-site monitoring and material characterization; and (iii) a Decision Support System providing seismic hazard analysis, multi-risk assessment, and socio-economic impact evaluation. Interactive geospatial functionalities allow users to explore datasets through structured spatial representations, such as hexagonal grid systems and visualize multiple data layers simultaneously. The system operates through standard web browsers without the need for specialized GIS software, ensuring accessibility for diverse user groups, including heritage professionals, decision-makers and local communities. Multiple data formats, such as GeoJSON, TIFF, PDF, 3D models and imagery, are processed and visualized in near real time within the platform.

Τhe results demonstrate that the integration of digital tools is not only considered as a technological advancement but also as a key enabler for collaboration, participation and sustainable heritage management. Interactive and cooperative digital environments can significantly enhance the resilience of cultural heritage sites to climate and disaster-related risks, supporting informed, inclusive and actionable management strategies.

Acknowledgement:

This research has been funded by European Union’s Horizon Europe research and innovation program under THETIDA project (Grant Agreement No. 101095253) (Technologies and methods for improved resilience and sustainable preservation of underwater and coastal cultural heritage to cope with climate change, natural hazards and environmental pollution).

How to cite: Georgiou, K., Routsis, K., Michalis, P., and Amditis, A.: Digital Integration of Environmental, Socio-Economic and Hazard Data for Heritage Resilience, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9183, https://doi.org/10.5194/egusphere-egu26-9183, 2026.

The aim of this study is to explore links between large-scale atmospheric circulation and meteorological and hydrological drought events in the Danube basin.

Based on previous studies on the relationship between large-scale atmospheric circulation and the occurrence of extreme events in the Danube basin, especially in the middle and lower Danube basin, two climate indices were considered. The first index characterizes the Greenland-Balkan Oscillation (GBOI) and the second is the well-known index associated with the North Atlantic Oscillation (NAOI). For meteorological and hydrological drought, the Palmer Drought Severity Index (PDSI), with applications especially in the agricultural field, and, respectively, the Palmer Hydrological Drought Index (PHDI), especially useful for estimating drought affecting water resources on longer timescales, were taken into account.

To find the type of connection (linear/non-linear) between large-scale climate indices (GBOI, NAOI) and those at the regional scale (PDSI, PHDI), elements from information theory, such as mutual information, were applied. To get time-frequency details, bivariate and multivariate wavelet transforms were used.

The analyses were performed separately for each season. The most statistically significant results were obtained both for the link between GBOI and PDSI, in the winter season, and for that between GBOI and the two analysed Palmer indices, in the spring season.

Regarding the influence of NAOI, it is much less than that of GBOI, but it can be considered relatively significant in winter on PDSI and in spring on PHDI.

From the wavelet coherence analyses it was observed that the significant coherences between the large-scale atmospheric indices and the analysed Palmer drought indices are located in frequency bands, corresponding to ~11–year, 22-year and 33-year period bands, that can be associated with the Schwabe, Hale and even Bruckner solar activity cycles.

In exploring regional-scale droughts, for the future studies, it appeared evident the importance of taking into account of the simultaneous or delayed influence of solar activity on terrestrial climate variables.

How to cite: Mares, C., Mares, I., Dobrica, V., and Demetrescu, C.: On the  links between large-scale atmospheric circulation and extreme events in the Danube basin, identified by Palmer drought indices, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10629, https://doi.org/10.5194/egusphere-egu26-10629, 2026.

Urban areas are increasingly exposed to flood hazards due to climate change, densification and growing urbanization. Remote sensing datasets can be used to monitor the floods and warn the population. Much more, simulations of hazards, using high resolution geospatial datasets combined with meteorological data can be derived. In this case, solutions prior to the events can be implemented, so increasing the resilience of cities to natural hazards.

This study presents a high-resolution 3D urban model of Brașov, Romania, developed from an airborne photogrammetric datasets acquired in 2025, and its application in urban flood risk assessment. The 3D building models were obtained using footprints from the national topographic database and the height derived from the computed normalized Digital Surface Model (nDSM). For the flood modelling the hydrological network and land cover data were used.

To assess flood risk, time series precipitation dataset was analyzed and used in the modelling framework. The combined analysis under different scenarios, enabled the identification of flood areas and the estimation of the number of exposed buildings. The results highlight the importance of high-resolution 3D urban data for understanding flood dynamics in complex urban settings and support decision-making processes related to urban planning, risk mitigation, and climate resilience. The output also represents a starting point for a Digital Twin for Brașov.

How to cite: Pârvu, I., Cuibac, I., Pârvu, A., Pârvulescu, N., Corneanu, I., Cheval, S., Crăciunescu, V., Dumitrescu, A., Amihaesei, V., Gabrian, Ș., Dinicila, Ș., and Tudose, N.: Urban Flood Risk Assessment Using High-Resolution 3D Building Models and Multi-Temporal Meteorological Data, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11292, https://doi.org/10.5194/egusphere-egu26-11292, 2026.

Limestone cultural heritage has increasingly been threatened by the complex interplay of climatic stressors, air pollution, and biological colonization. In this STECCI study, a bio-geochemical dose-response framework was introduced to quantify and interpret the decay of stećci-medieval tombstones constructed from locally sourced limestone, across fifteen culturally significant sites in Southeastern Europe. While existing dose-response functions (DRFs) have traditionally been applied to climatic, chemical and physical weathering, biological link has often been in the Blind Spot, despite mounting evidence that lichens, mosses, and microbial taxa contribute actively to stone decay.

Two widely used DRF models Lipfert (1989) and Kucera et al. (2007) were applied to multi-decadal environmental data (1992-2023), accounting for variations in precipitation, temperature, and pollutant load (SO₂, NOₓ, PM₁₀). Bioassement surveys were conducted to record biological colonization using a modified Braun-Blanquet scale and photographic quadrat sampling. At the same toime, spatial overlays of DRF results and biological data were produced to identify zones of specific vulnerability, where climatic exposure and biodeteriogen presence were observed to overlap. As expected, the Lipfert model responded more strongly to high-precipitation karstic settings, while the Kucera model captured the cumulative effect of pollutants and humidity in urban sites. However, both models were shown to underestimate decay in areas with extensive lichen or moss coverage, highlighting the need for biotic factors to be integrated into predictive modeling. To address this, a multi-stressor approach was developed, coupling DRF-predicted surface recession with biological indicators and  introdicing b coficient within the both mathematical models, as Lithobiontic organisms, such as Lobothallia cheresina, Xanthoria elegans, and Grimmia pulvinata, were found to contribute to micro-fracturing, mineral leaching, and, most importantly, moisture retention, often acting synergistically with atmospheric deposition. Based on these insights, a STECCI Preservation Measures Assessment tool was proposed to classify heritage sites according to modeled decay, biocolonization intensity, and conservation urgency.

This integrative methodology was conducted to sharp the diagnostic capacity of DRFs and enabled the generation of science-based insights, integrating risk assessment models for heritage exposed to climatic, natural, and anthropogenic hazards. In light of projected climate shifts and persistent anthropogenic emissions, it is recommended that heritage conservation efforts adopt bio-geo diagnostics to transition from reactive toward preventive conservation strategies. The approach presented here is transferable to other limestone heritage materials and contributes to the growing discourse on climate-resilient cultural heritage preservation.

Acknowledgement: The STECCI project has received funding from the European Union’s Horizon Europe research and innovation programme under Grant Agreement No. 101094822 (STECCI), managed by the European Research Executive Agency (REA).

How to cite: Radulović, S., Anačkov, G., Radak, B., Ilić, M., Radulović, B., Novković, M., Djug, S., Smailagić Vesnić, L., Ibragić, S., and Dresković, N.: Breaking Disciplinary Boundaries: Bringing the Biological Role out of the Blind Spot in DRF-Based Assessments of Limestone Weathering under a Changing Climate, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14335, https://doi.org/10.5194/egusphere-egu26-14335, 2026.

This study develops a novel multi-criteria framework for the identification of intra-urban heat islands by integrating indicators related to three-dimensional urban morphology (e.g., height-to-width ratio and sky view factor), land cover characteristics, satellite-derived land surface temperature, and thermal comfort conditions. The proposed framework practically enables the delineation of urban areas with distinct thermal and morphological profiles, thereby providing a robust basis for targeted, site-specific intervention strategies.

Subsequently, a range of bioclimatic heat mitigation measures is assessed for selected hotspots, including nature-based solutions such as increased tree planting and green roofs, as well as the application of high-albedo (cool) materials. The effectiveness of these measures is evaluated using advanced urban climate simulation models (ENVI-met and UT&C), allowing for a comparative assessment of their performance under varying spatial configurations and microclimatic conditions.

Overall, the study provides evidence-based guidance for urban heat mitigation and supports climate-resilient urban planning in Mediterranean cities, with Athens serving as a representative case study.

How to cite: Oikonomou, M., Agathangelidis, I., and Cartalis, C.: Development of a Multi-Criteria Framework for Identifying Intra-Urban Heat Islands in Support of Urban Heat Mitigation in Athens, Greece, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14400, https://doi.org/10.5194/egusphere-egu26-14400, 2026.

The Paraná basin is composed of stratigraphic units that record distinct paleoenvironmental settings, organized into six supersequences. The Gondwana I Supersequence  (Permian) records a transgressive-regressive cycle associated with tectonic and climatic changes, in which periglacial successions (Itararé Group), coastal and marine (Guatá Group), and continental deposits (Passa Dois Group) are preserved. These sedimentary units crop out along the eastern margin of the Paraná Basin in a complex structural configuration that reveals significant tectonic displacements attributed to normal faults, resulting in the lateral juxtaposition of stratigraphically distinct units. Due to this arrangement, volcanogenic deposits play a fundamental role as stratigraphic markers, as they allow the establishment of precise geochronological correlations. This study presents geochronological data obtained from a volcanogenic deposit at Morro dos Conventos outcrop, and compares it with a compilation of the ages of volcaniclastic sediments interbedded with sedimentary deposits from the Rio Bonito Formation, aiming to constrain the evolution of depositional systems that enabled the preservation of such volcanogenic deposits within this interval. The detailed stratigraphic section of the outcrop was conducted and samples were collected for geochronological analysis. U-Pb zircon ages were determined by LA-MC-ICP-MS from a volcanogenic layer. The results reveal a unimodal zircon population with a concordia age of 286 ± 1.4 Ma (N = 9; MSWD = 1.2), allowing correlation of the deposit with the Artinskian Stage. Sedimentological and stratigraphic analysis of the section indicates a paleoenvironment of storm wave-dominated shelf, with interbedded subsystems recording high-frequency cycles associated with changes in sea level or sedimentation rates within the second-order Permian transgressive sequence. Sedimentological and geochronological data suggest that the studied succession correlates with the upper portion of the Rio Bonito Formation, in a context of progressive drowning by the Palermo Sea. At the top of the section, a progradation of subsystems is observed, characterized by the arrangement of subaerial sequences under humid backshore conditions. A similar configuration has been documented in other areas of the basin during the Cisuralian, where pelitic successions associated with coal deposits preserve centimeter-thick intercalated volcanic ash layers. The preservation of these features is attributed to paleoenvironmental conditions of subsystems developed along the margins of subaqueous bodies, dominated by low-energy settings with limited reworking, favoring the deposition of fine-grained sediments. In the studied outcrop, the preservation of the volcanogenic deposit is interpreted as a result of deposition within fine-grained sediments characterized by redoximorphic structures, indicative of fluctuating conditions between dry and wet periods typical of subaerial environments influenced by aqueous systems. A similar preservation context is observed in volcanogenic deposits recorded both in CPRM wells (Brazilian Geological Survey) and in nearby outcrops of this stratigraphic interval. The coexistence of low-energy depositional systems and episodes of high-magnitude explosive volcanism along the western margin of Gondwana enabled the preservation of ash-fall deposits in the Paraná Basin stratigraphic record, commonly associated with the Choiyoi Magmatism during the proposed Rio Bonito Formation sedimentation interval.

How to cite: Ribeiro Franqueira, A. V., Bettarel Bállico, M., Moreira Florisbal, L., Oliveira Manna, M., and Marlon dos Santos Scherer, C.: Controls on Ash-Fall Deposit Preservation in Low-Energy Depositional Systems of the Rio Bonito Formation, Paraná Basin, Brazil, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14428, https://doi.org/10.5194/egusphere-egu26-14428, 2026.

Quantitative estimates of Holocene climate conditions have been developed since the 1960s using space‑for‑time calibration approaches. Although hundreds of reconstructions are now available globally, questions persist regarding their accuracy. We evaluate quantitative reconstructions derived from three sources: site‑specific studies, regional reconstruction compilations, and regional products generated from global databases. The focus is on Holocene pollen‑based reconstructions, which remain the most widely used indicators of terrestrial paleoclimate and on North American Arctic and treeline regions. Reconstructions developed at individual sites often display substantial high‑frequency variability, including anomalous values and abrupt shifts, reflecting in part calibration-related artifacts. Regional averages (“stacks”) reduce some of this variability, yet comparisons based on different reconstruction sources reveal divergences. Conversely, studies analyzing paired cores from a single lake or from closely spaced sites frequently demonstrate strong replication and relatively low reconstruction error. Multi‑proxy analyses of Arctic cores likewise reveal both areas of agreement and persistent discrepancies. Addressing these inconsistencies remains a challenge for Holocene climate reconstruction.

How to cite: Gajewski, K.: Analysis of quantitative pollen-based reconstructions, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15510, https://doi.org/10.5194/egusphere-egu26-15510, 2026.

A realistic representation of sea surface temperature (SST) variability in climate models is essential for seasonal-to-interannual forecasting and for understanding large-scale climate oscillations. This study evaluates the impact of three initialization strategies in the NorCPM coupled climate model on the structure and temporal evolution of the leading modes of global SST variability over the 1980–2010 period. The analyzed strategies include a free-running simulation (FREE), ocean data assimilation using an ensemble Kalman filter (ODA), and atmospheric nudging of wind and temperature anomalies (NUDA_UVT). Model results are evaluated against the HadISST observational dataset.

Empirical Orthogonal Function (EOF) analysis is applied to monthly SST anomalies computed over the full globe and using all calendar months, without regional restriction or seasonal stratification. This framework enables a consistent comparison of the dominant large-scale SST variability modes across all datasets. The results indicate that ocean data assimilation (ODA) best reproduces the leading ENSO-related mode, achieving a spatial correlation of 0.98 and the lowest root mean square error in its principal component (RMSE = 1.70). For the second mode, associated with lower-frequency variability, atmospheric nudging (NUDA_UVT) shows improved spatial agreement (correlation = 0.90) compared to ODA. The free-running simulation captures the main spatial structures but displays systematically larger temporal errors.

These findings demonstrate that ocean data assimilation is the most effective strategy for representing ENSO-like variability in NorCPM, while atmospheric nudging provides added value for lower-frequency modes. As a perspective, this work will be extended to investigate the impact of initialization strategies on atmospheric fields, as well as to explore SST variability at specific seasonal and regional scales.

How to cite: Moutachaouiq, K., Bari, D., Omrani, N.-E., and Nafiri, S.: Impact of Different Initialization Strategies on the Representation of Dominant SST Variability Modes in the NorCPM Coupled Climate Model, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17596, https://doi.org/10.5194/egusphere-egu26-17596, 2026.

Cloudbursts, defined as sudden, intense rainfall episodes, are increasingly frequent extreme weather events in the Indo-Himalayan region, causing widespread devastation to human life and property; yet understanding their causal mechanisms and improving predictability remains constrained by incomplete knowledge of atmospheric and land-based precursors. Particularly, the role of soil moisture as a vital land-surface component has been underexplored in the context of cloudburst formation. This study hypothesizes that increased soil moisture from agricultural irrigation amplifies atmospheric moisture fluxes via land-atmosphere coupling and contributes to enhanced cloudburst risk. The objective here is to attribute moisture source locations, identify critical pre-event land-atmospheric indicators, and assess soil–atmosphere coupling through the analysis of IMD-specified cloudburst events from 1991 to 2020 using the Indian Land Data Assimilation System (ILDAS) dataset. We employ NOAA's Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) back-trajectory model and create Integrated Vapor Transport (IVT) maps, composited with winds, surface pressure, and sea level pressure, to trace moisture source locations. Pre-event anomaly detection and change-point analysis are performed using the Pruned Exact Linear Time (PELT) algorithm on soil moisture, precipitation, evaporation, and runoff variables across nine spatially proximate grid cells per event. Additionally, extreme percentile threshold exceedances and non-parametric persistence metrics quantify the early-warning potential. Decadal NDVI trends contextualize Land Use/Land Cover (LULC) influences. Results reveal moisture source hotspots in regions undergoing land-use transitions, with steep pressure gradients establishing strong circulation patterns that contribute moisture to multiple cloudburst events. Significant temporal anomalies occur across all four variables, with threshold exceedances and change-point detections ranging from 2 to 10 occurrences per event and anomaly persistence spanning 2 to 8 days for soil moisture. Early warning lead times of 15 to 120 days are identified for soil moisture, precipitation, evaporation, and runoff anomalies preceding the cloudburst events. These findings suggest that further quantifying the causal links among these variables can better help understand soil–atmosphere coupling and substantially improve early warning systems for detecting extreme rainfall events.

How to cite: Kaushal, A., Saharia, M., and Rajagopalan, B.: Land-Atmosphere Drivers of Cloudburst Events, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19097, https://doi.org/10.5194/egusphere-egu26-19097, 2026.

One approach to reduce the uncertaintites in the black carbon (BC) emissions estimated using the  bottom-up inventories is by integrating the atmospheric models with observational data. In this study, we estimate aerosol emissions  over Europe (15◦W–35◦E, 33–73 ◦N) by assimilating surface observations of BC from EBAS network using Local Ensemble Transform Kalman Filter (LETKF) in the LOTOS-EUROS chemical transport model. Sensitivity experiments indicate that an ensemble size of 24 and a localization distance of 300 km provide optimal performance. Furthermore, we assess the influence of CAMS BC boundary conditions on the emission estimates and find that these boundary conditions tend to overestimate BC concentrations near the domain boundaries.

Our results show that the bottom-up approach generally overestimates BC emissions across Europe. Quantitatively, the posterior emissions are found to be 21% and 30% lower than the prior emissions for the years 2011 and 2021, respectively. A reduction in both emissions and associated uncertainties is observed over central Europe, where the observations are dense. Seasonal analysis reveals that emission decreases are most pronounced over the central domain during autumn and winter. Finally, the validation of optimized BC concentrations with independent observations showed a decrease in bias and RMSE, however the correlation remains poor compared to the background concentrations.

How to cite: George, B., Thomasson, A., Roldin, P., Segers, A., and Schutgens, N.: Optimizing Aerosol Emissions over Europe using Surface Black Carbon Measurements, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19525, https://doi.org/10.5194/egusphere-egu26-19525, 2026.

Underwater cultural heritage sites are under increasing pressure from emerging environmental risks: warmer waters and changing seawater chemistry are accelerating corrosion processes in ways that remain difficult to quantify in terms of their impacts on protected archaeological metals. This work proposes an experimental approach that makes these changes measurable and comparable across sites using metallic coupons carefully selected to match materials revealed from a series of wrecks. Coupons were deployed at different depths at four different locations, and retrieved at fixed time intervals. The development of corrosion layers, concretions, and biofouling in the natural environment was investigated. Observations were integrated with results from a second experimental approach. The same set of coupons was exposed to controlled environmental conditions using a custom Micro-Environment Simulator (MES). MES was set to reproduce marine conditions at 4 bar gauge pressure (40 m depth), 20 °C water temperature, and pH 7.7, simulating ocean acidification by the end of this century according to the CMIP6 projections for the Mediterranean under the SSP5-8.5 Warming 4 °C scenario. Results have shown a significant shift in electrochemical equilibria under declining pH, significantly influencing the stability of the corrosion products, and determining a shift in the behaviour of the corrosion layers from protective barrier to pathway for continued metal loss. By linking corrosion behaviour to specific environmental settings, the approach provides indicators of when and where deterioration is likely to accelerate under future scenarios. These outputs support preventive strategies for underwater metallic heritage by identifying high-risk wreck contexts, and guiding actions before irreversible loss occurs.

Acknowledgement:

This research has been funded by European Union’s Horizon Europe research and innovation programme under THETIDA project (Grant Agreement No. 101095253) (Technologies and methods for improved resilience and sustainable preservation of underwater and coastal cultural heritage to cope with climate change, natural hazards and environmental pollution).

How to cite: Cesareo, L. P., Germinario, L., Salvemini, F., MacLeod, I. D., Marchettoni, E., Ambrosi, C., Donnarumma, L., Sorci, A., and Mazzoli, C.: How climate change rewrites metal decay: forecasting ancient shipwreck corrosion under acidified seawater, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20603, https://doi.org/10.5194/egusphere-egu26-20603, 2026.

Temperature essential for all the life form, but when the same temperature crosses its threshold limit it can become a threat for the same. In the recent decades the world has experienced a shift in temperature range both minima and maxima, as both are shifting towards the higher tail. And this is detrimental for health and air quality. Also very few studies talk about how rising temperature can impact the air quality and vice versa. Therefore, in the present work we have studied the long term heatwave pattern over Delhi, India using the ground based and satellite data observations. Delhi is known for its hot summers, landlocked geography and dense population.  During May 2022 , city experienced long heat wave event where daily maximum temperature observed higher than 40 ^OC   for consecutively  around 10 days  which not only  causes heat stress but also the pollution stress over the city as the concentration of Particulate matter (PM₁₀ and PM_2.5) observed significantly higher than the  non-heatwave period of the month. Air Quality Index (AQI) was moved from moderate to very poor during the heatwave period compared to non-heatwave where AQI showed satisfactory to poor condition. Further we observed that the Temp and Demand data increases monotonically during this period from 30 °C with demand ≈4500–5200 MW to about 38 °C with demand ≈6800–7070 MW, indicating a strong positive linear response. The regression analysis showed with 1°C increase in air temperature can increase the city demand by 97MW with r=0.61. We have further calculated night vs day slope, indicate that when night stays hot (>35°C) people might be keep cooling system running more intensely or for longer hours. And each degree increase in nighttime temperature put much larger load on demand compared to same warming during the day.

How to cite: Masiwal, R., Ganguly, D., and Kunchala, R.: Heatwave and Air pollution, a synergetic effect or not: A case Study, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21797, https://doi.org/10.5194/egusphere-egu26-21797, 2026.

The Dansgaard–Oeschger events are sudden and irregular warmings of the North Atlantic region that occurred during the last glacial period. A key characteristic of these events is a rapid shift to warmer conditions (interstadial), followed by a slower cooling toward a colder climate (stadial), resulting in a saw-tooth pattern in regional proxy temperature records. These events occurred many times during the last 100,000 years and have been hypothesized to result from various mechanisms, including millennial variability of the ocean circulation and/or nonlinear interactions between ocean circulation and other processes. Our starting point is a non-autonomous, conceptual, but process-based, model of Boers et al. [Proc. Natl. Acad. Sci. 115, E11005–E11014 (2018)] that includes a slowly varying non-autonomous forcing represented by reconstructed global mean temperatures. This model can reproduce Dansgaard–Oeschger events in terms of shape, amplitude, and frequency to a reasonable degree. However, the model of Boers et al. has instantaneous switches between different sea-ice evolution mechanisms on crossing thresholds and, therefore, cannot show early warning signals of the onset or offset of these warming events. We present a regularized version of this model by adding a fast dynamic variable so that the switching occurs smoothly and in finite time. This means the model has the potential to show early warning signals for sudden changes. However, the additional fast timescale means these early warning signals may have short time horizons. Nonetheless, we find some evidence of early warning for the transition between slow and rapid cooling for the model.

How to cite: Hobden, B., Ritchie, P., and Aswhin, P.: Regularization of a conceptual model for Dansgaard–Oeschger events, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-380, https://doi.org/10.5194/egusphere-egu26-380, 2026.

Past2Future (P2F) aims to develop, expand, and leverage the wealth of paleoclimate data to significantly improve existing Earth System Models and deepen our understanding of Earth’s climate response to various types of forcing, with a focus on abrupt climate transitions and tipping points. To achieve this, our work focuses on the compilation, integration, and re-evaluation of past sea surface temperature (SST) data with the aim of defining the states and variability of the ocean temperatures across four pivotal climate intervals: the Mid-Holocene (6.5-5.5 ka), Last Glacial Maximum (23-19 ka), Eemian (130-116 ka), and the mid-Pliocene Warm Period (3.3-3 Ma).

To date, we have identified all published global SST records spanning the Last Glacial Maximum and the Mid-Holocene, reconstructed using both geochemical techniques and faunal assemblages. For the Last Glacial Maximum, we identified 1,426 geochemical and faunal proxy records from over 1,100 cores. For the Mid-Holocene we identified 1,014 geochemical and faunal proxy records from 790 cores. Subsequently, we assessed the suitability of these records for climate model evaluation and tuning by considering: i) the robustness of each record’s age model; ii) the SST reconstruction methodology and associated uncertainties; and iii) site location and representativeness. Consequently, we have prioritised marine sediment records that feature robust age models, high-resolution SST records, low calibration uncertainties, derived from sites minimally influenced by additional climatic or environmental factors (e.g. upwelling), and, where possible, supported by alternative multi-proxy SST reconstructions. To address remaining spatial and temporal data gaps, we will generate new SST records using alkenone (U^K’₃₇) and glycerol dialkyl glycerol tetraether (TEX₈₆) proxies, generating datasets that support climate models. The resulting curated and expanded SST datasets will provide a robust benchmark for climate model evaluation and tuning, ultimately contributing to more robust and accurate simulations of past climate states and more reliable projections of future climate change.

How to cite: Cutmore, A., Sliwinska, K., and McClymont, E.: Past to Future: Defining the states and variability of the ocean, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-2331, https://doi.org/10.5194/egusphere-egu26-2331, 2026.

The risk of crossing critical thresholds in the Earth system is continuously increasing due to anthropogenic climate change, potentially leading to accelerated responses. As one of the major tipping elements, the Atlantic Meridional Overturning Circulation (AMOC) can exhibit abrupt, nonlinear shifts between distinct regimes. Evidence of such tipping behavior is found in paleo-climate records, most prominently as Dansgaard-Oeschger (DO) events. Using the Bern3D fast Earth System Model, we investigated DO events driven by AMOC variability under Marine Isotope Stage 3 (MIS3) conditions. The model exhibits unforced, self-sustained oscillations resembling DO events within a narrow parameter space defined by CO₂ concentration, wind stress forcing, and diapycnal diffusivity. We systematically explored this parameter space and its boundaries. Beyond these parameter space boundaries, the AMOC either remains in a weak regime or undergoes an abrupt transition to a stronger state. Within the parameter space, oscillations are stable, with the periodicity being strongly controlled by CO₂. The mechanism underlying DO-like oscillations is primarily oceanic and involves heat accumulation and sea ice changes in the eastern North Atlantic. Sea ice acts as an insulating barrier, allowing subsurface heat to build up until it is rapidly redistributed through the water column, melts the sea ice, is released and triggers deep convection, producing an abrupt strengthening of the AMOC. Freshwater input from sea ice melt, in turn, weakens the circulation. These results indicate that abrupt shifts in the AMOC are an inherent feature of the climate system, although the implications for the AMOC’s future evolution remain unclear due to the vastly different boundary conditions.

How to cite: de Huu, A., Pöppelmeier, F., Testorf, P., and Stocker, T.: Variability across parameter space and mechanisms of DO-like oscillations in a fast Earth System Model  , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3889, https://doi.org/10.5194/egusphere-egu26-3889, 2026.

Arctic sea ice is a critical component of the climate system, yet its long-term variability and drivers remain poorly understood due to the scarcity of direct paleoclimate records. In this study, we utilize bromine records preserved in four Greenland ice cores—NEEM, DYE-3, EGRIP, and RECAP—to reconstruct changes in Arctic sea ice cover over the past 15,000 years. Bromine in polar snow and ice is primarily derived from "bromine explosions" occurring over seasonal sea ice surfaces. These are autocatalytic photochemical reactions in which sea salt from brine and frost flowers on newly formed sea ice is activated, releasing reactive bromine into the atmosphere. Because these processes are strongly linked to the presence of first-year (seasonal) sea ice, bromine enrichment in ice cores reflects the extent and variability of seasonal sea ice cover. The combined records provide a high-resolution, multi-site perspective on sea ice variability during the late glacial–Holocene transition and throughout the Holocene. By integrating these records, we explore the spatial variability of sea ice changes and highlight the heterogeneous response of the Arctic Ocean to climatic perturbation. These results offer new insights into the mechanisms controlling past sea ice variability, providing important context for evaluating future Arctic change.

How to cite: Dey, R., Segato, D., Spolaor, A., and Kjær, H. A.: Reconstructing 15,000 years of Arctic sea ice dynamics: High-resolution bromine records from the late Glacial to the Holocene, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7352, https://doi.org/10.5194/egusphere-egu26-7352, 2026.

The Last Interglacial (LIG; 129–116 thousand years ago) experienced global mean temperatures approximately 1–2 °C above pre-industrial levels, comparable to present-day conditions and those projected for the near future. During the LIG, high and mid-latitudes were substantially warmer, Arctic sea ice was reduced, both the Greenland and Antarctic ice sheets were smaller than today, and global mean sea level was at least 5 m higher than present. Unlike modern warming, which is primarily driven by increased greenhouse gas concentrations, LIG climate anomalies were mainly caused by higher eccentricity and a precession putting NH summer closer to perihelion, with Northern Hemisphere summer insolation exceeding pre-industrial values by more than 70 W m⁻².

Many climate models struggle to reproduce the magnitude of LIG warming and the seasonally ice-free Arctic suggested by proxy evidence. Here, we present results from an abrupt-127 ka experiment following the CMIP7 Fast Track protocol, performed with ICON-XPP v1.0 (07/2024) [1], extended to allow for orbital parameter variations based on Kepler’s approximation. In this simulation, orbital parameters and greenhouse gas concentrations are set to LIG values, while all other boundary conditions (solar constant, prescribed ice sheets, prescribed vegetation, and aerosols) are kept at pre-industrial levels.

Compared to the pre-industrial control simulation, the abrupt-127 ka experiment shows top-of-atmosphere (TOA) radiation anomalies consistent with previously published LIG simulations [2] including an Arctic summer TOA increase of 50–75 W m⁻². However, for ICON-XPP v1.0 the simulated annual global mean temperature decreases by 0.3 K when only orbital parameters are changed, and by 0.47 K when LIG greenhouse gas concentrations are applied in addition. This contradicts proxy reconstructions indicating a global mean temperature increase of approximately 0.5–1.5 K during the LIG.

Exploring Arctic seasonality, we find a summer warming of 4 K in July and a winter cooling of 3 K in January, resulting in an overall Arctic cooling in the annual mean relative to pre-industrial conditions. Arctic sea ice shows little reduction in summer but increases more substantially in winter, leading to an overall annual expansion of sea ice compared to pre-industrial levels. We attribute the simulated cooling and disagreement with proxy evidence to insufficient Arctic amplification in the ICON-XPP version used, likely caused by a weak sea-ice feedback and the lack of interactive vegetation changes. We compare these results to first results obtained with the CMIP7 release of ICON-XPP (2025.10-1) and sensitivity experiments exploring the impact of prescribed vegetation changes and the inclusion of dynamic vegetation. Our findings have major implications for future simulations with ICON-XPP, as the LIG represents a climate state comparable to present-day and future warmth.

[1] Müller et al.: The ICON-based Earth System Model for Climate Predictions and Projections (ICON XPP v1.0), EGUsphere, https://doi.org/10.5194/egusphere-2025-2473, 2025.

[2] Otto-Bliesner et al.: Large-scale features of Last Interglacial climate: results from evaluating the lig127k simulations for the Coupled Model Intercomparison Project (CMIP6)–Paleoclimate Modeling Intercomparison Project (PMIP4), Clim. Past, https://doi.org/10.5194/p-17-63-2021, 2021.

How to cite: Rehfeld, K., Brugger, J., Houston, T., Racky, M., Lorenz, S., Wagner, S., and Köhler, M.: Simulations of the Last Interglacial with ICON-XPP indicate the relevance of correct sea-ice and vegetation feedback for Northern Hemisphere warming, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9552, https://doi.org/10.5194/egusphere-egu26-9552, 2026.

Accurately representing abrupt climate transitions such as Dansgaard–Oeschger (DO) events in climate models is essential for understanding past climate dynamics and improving projections of future tipping points. However, these models contain numerous uncertain parameters that are traditionally tuned manually, a process that is not only time-consuming but also subjective and limited in its ability to quantify parameter uncertainty. While systematic calibration approaches can provide rigorous parameter estimation, Bayesian inference methods such as MCMC require many sequential model evaluations, making them computationally prohibitive for complex climate models.

We present a systematic framework for climate model calibration that combines machine learning emulation with Bayesian inference to rigorously estimate model parameters and their uncertainties. Using CLIMBER-X, an Earth system model of intermediate complexity that successfully simulates DO-like oscillations in the Atlantic Meridional Overturning Circulation (AMOC) (Willeit et al., 2024), we develop a proof-of-concept for this calibration approach. We train an emulator that accurately approximates the model's AMOC response for a set of key ocean parameters, enabling efficient model evaluations.

We employ both Markov Chain Monte Carlo (MCMC) sampling and Simulation-Based Inference (SBI, Cranmer et al., 2020) techniques to estimate posterior distributions of these key model parameters. The ML emulator reduces computational cost by several orders of magnitude, making systematic parameter estimation and efficient exploration of the parameter space feasible. Since CLIMBER-X already produces realistic DO-like events, this serves as an ideal test case for validating the calibration framework. This work emphasizes the potential of ML-based emulation to accelerate systematic calibration in paleoclimate modelling.

References:

Willeit, M., Ganopolski, A., Edwards, N. R., and Rahmstorf, S.: Surface buoyancy control of millennial-scale variations in the Atlantic meridional ocean circulation, Clim. Past, 20, 2719–2739, https://doi.org/10.5194/cp-20-2719-2024, 2024.

Cranmer, K., Brehmer, J., and Louppe, G.: The frontier of simulation-based inference, Proc. Natl. Acad. Sci. USA, 117, 30055–30062, https://doi.org/10.1073/pnas.1912789117, 2020.

How to cite: Kowalczyk, K. and Boers, N.: Efficient Bayesian Calibration of Climate Models via Machine Learning Emulation: Application to Dansgaard-Oeschger Events, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12141, https://doi.org/10.5194/egusphere-egu26-12141, 2026.

Rising atmospheric carbon dioxide (CO₂) concentrations alter the vegetation composition indirectly through climate change and directly through plant physiological modifications. Both responses modulate climate through changed energy, moisture, and carbon fluxes between land and atmosphere. This makes accurate estimates of the responses important for future vegetation and climate projections. Yet, large inter-model differences regarding the magnitude of the direct response persist, leading to uncertain future projections. Here, we quantify the impact of CO₂ changes on the vegetation and climate in three Earth system models (ESMs) of varying complexity. The direct and indirect responses are separated using factorization experiments and statistical emulators. While most previous studies focus on either low or high CO₂ concentrations, we cover a large range from 150ppm to 1200ppm.

We find that plant function type (PFT) specific responses often follow a logarithmic shape except when threshold crossings create breakpoints. However, the grid box mean responses can differ from PFT-specific responses, indicating substantial modulations of the shape and amplitude by competition between PFTs. While competition amplifies the response for some variables, it dampens the response for others. For example, changes of biophysical properties like leaf area index and canopy height are amplified by competition, contributing to stronger plant-physiological impacts on some components of the terrestrial hydrological cycle than the radiative effect of rising CO₂ concentrations. The simulated long-term vegetation impacts can currently not be evaluated against present-day observations or manipulation experiments. Instead, we compare the model results with global compilations of paleobotanical data. Preliminary results for the Last Glacial Maximum indicate a model-dependent overestimation of the plant-physiological response. Future research aims at leveraging these comparisons to calibrate the modeled direct response of vegetation to CO₂, which would provide constraints for the long-term impacts of future emission scenarios on natural ecosystems.

How to cite: Weitzel, N., Valdes, P. J., Jones, C. D., and Dallmeyer, A.: A multi-model assessment of the plant-physiological response to high and low carbon dioxide concentrations, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12969, https://doi.org/10.5194/egusphere-egu26-12969, 2026.

The magnitude and spatial patterns of future changes in temperature variability remain debated. Supplementing direct observations with reconstructions of past climate has shown that CMIP-style simulations lack regional variability on decadal and longer timescales, a shortcoming that likely includes future projections. Here, we assess the range of future climate variability based on the differences between reconstructions and simulations of temperature variability during the Quaternary. The assessment uses a multi-proxy database of surface temperatures as well as long-term transient simulations of the past and possible future climates with an Earth System Model. Comparing simulations with reconstructions, we establish a relationship between warming level and local to global temperature variability for annual to millennial timescales. The identified model-reconstruction mismatch provides the basis for rescaling simulations and thus constraining future climate variability. For this, we decompose variability into its long- and short-term components. We then artificially enhance the long-term variability that is underestimated in simulations to reconstruct a possible, more realistic corresponding temperature field. Taking the uncertainty in reconstructions into account results in a wider range of possible scenarios for future climate variability given this past evidence. Our results have implications for climate indices and temperature extremes on short timescales in future scenarios, informing mitigation and adaptation efforts.

How to cite: Ziegler, E., Kapsch, M.-L., Mikolajewicz, U., and Rehfeld, K.: Constraints on future multidecadal temperature variability from climate models, reconstructions and observations of the past two million years, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13210, https://doi.org/10.5194/egusphere-egu26-13210, 2026.

The future of the Amazon basin is of great concern front projected impacts due to climate change. Current estimates of hydrological changes are said to point toward unprecedented stages in many aspects, including floods and droughts. However, hydrological monitoring is often limited in time and can mask long term natural variability, affecting conclusions regarding present and future trends. Therefore, tree-ring time series are a valuable complement to this kind of assessment and can provide insights on the long-term occurrence of small and large floods and droughts. Here, we propose a preliminary reconstruction of annual floods from 1850 to 2016 based on a tree-ring δO18 series for the Paru basin, located in the northeastern Amazon. Isotope data present strong (r between 0.6-0.9 in module) and spatially consistent correlation with annual floods and with basin-aggregated rainfall from 1980-2016 at the Paru and nearby basins. This encouraged us to explore a simple linear regression model by fitting δO18 and annual flood series. The model was fitted using simulated discharge from MGB-SA hydrological model from 1980 to 2016 (and compared with observed record), and resulted in a r² > 0.5 for more than 200 river reaches near the tree-ring data’s site. Regression models presented a success rate of >70% in classifying small flood years, while presenting >60% and >40% for regular and large floods respectively. This preliminary assessment indicates the potential of hydrological reconstruction of floods based on Paru’s δO18 data, enabling valuable insights on part of the Amazon hydrological variability since mid-19^th century. Future perspectives could include hydrological modelling based on rainfall ensembles built from this series for more detailed assessments.

How to cite: Torres Miranda, P., Cauduro Dias de Paiva, R., and Granato-Souza, D.: Potential reconstruction of 19th century flood variability in the northeastern Amazon using tree-ring δO18, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13770, https://doi.org/10.5194/egusphere-egu26-13770, 2026.

Instrumental observations only capture a short interval of the climate history of the Earth, and are insufficient to fully constrain low-frequency variability, internal dynamics, and the response of the climate system to changing background states. Paleoclimate archives, by contrast, document a wide range of past climate changes, yet translating this information into Earth System Models (ESMs) to enhance their performance and future projections remains a major challenge. Paleoclimate Data Assimilation (PDA) provides a promising pathway to bridge this gap by combining proxy records and ESMs within a physically consistent framework. Here we present a paleo reanalysis based on an adaptation of the Norwegian Climate Prediction Model (NorCPM), in which an ensemble Kalman filter is used to assimilate hundreds of annually resolved proxy records (including coral, tree-ring, and ice-core records) back to 1600 CE. Unlike many existing paleo reanalyses for past centuries, which primarily constrain the atmospheric state and only indirectly represent ocean variability, our approach explicitly accounts for ocean dynamics. By nudging three dimensional atmospheric wind fields derived from the paleo atmospheric reanalysis, we generate a dynamically consistent coupled reanalysis, by simulating the response of the ocean to large-scale wind variability, as well as their associated impacts through thermodynamical feedbacks. The climate reanalysis represents both forced and internal variability over the last four centuries and shows good agreement with independent instrumental observations. By construction, this approach yields a dynamically coherent, multivariate reconstruction that goes beyond traditional proxy reconstructions and enables direct investigation of climate dynamics and feedback. Here, we focus on methodological aspects and perspectives of PDA, highlighting how paleo reanalyses can (i) constrain modes of low-frequency variability and their stability across different climate states, and (ii) evaluate and refine the calibration of the ESMs beyond the instrumental period. Such approaches are essential for improving confidence in future climate projections, particularly with respect to long-timescale variability, feedback, and the potential for abrupt transitions in the Earth system.

How to cite: Dalaiden, Q., Counillon, F., Svendsen, L., Bethke, I., and Keenlyside, N.: Paleoclimate Data Assimilation: unlocking past climate dynamics to better constrain the future, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13882, https://doi.org/10.5194/egusphere-egu26-13882, 2026.

The Earth is experiencing an unprecedented fast climate global warming. The global mean seal level (GMSL) rise, resulting from the different SSPs scenarios leads to an increase from 0.5 to 1 m for the end of this century and about 10 m for the 23rd century.

To investigate whether this GMSL rise may act as a climate forcing by itself, an appropriate framework is the last. Indeed, the GMSL was 2 to 6m higher than today, due to a different configuration of the Greeenland and West Antarctica ice sheet. Nevertheless, most simulations of this period (127ka BP)  only account for orbital parameters and greenhouse gases (CO₂, CH₄, and N₂O) changes.

A first publication (Z. Zhang et al. Nat. Geosciences 2023), using the NORESM F1 model, demonstrated that, accounting for this GMSL rise superimposed with the insolation and greenhouse gas forcing factors helps to solve the mismatch pointed out when comparing pmip4 simulations results and SST reconstructions derived from different proxies.  Nevertheless, it was necessary to confirm this important finding with another ESM.

Therefore, we performed, using the same protocol than Zhang et al.  2023, LIG simulations With IPSL CM6 ESM. Both models were involved in PMIP4 LIG intercomparison (Otto-Bliesner  CP 2021) and show good agreement with other ESMs and reasonable agreement with data. A robust feature emerging from this intercomparison was that all models depicted an overestimation of SST for the Southern Hemisphere.

New results within IPSL CM6 demonstrate first that, indeed, GMSL is an important forcing factor for LIG.  In other words, the GMSL rise is not only a consequence of the warming but is also an important driver of this warming.  A second important result is that the spatial pattern of SST response to GMSL is model dependent.  Specifically, the AMOC decrease is enhanced in IPSL model and a difference in bathymetry of Bering strait in both models led to opposite response over Artic Ocean.

Considering the impact of GMSL for LIG depicted by both models, we also did a comparison of GMSL impact for the present interglacial period. We will also present results obtained with IPSL CM6 when using GMSL rise corresponding to 1.25 m that could be reached at the end of this century or 5 to 10m, that could be reached at the end of the 23^rd century.

How to cite: Ramstein, G., Nguyen, S., Ewart, M., Zhang, Z., and Guyonvarh, P.: Impact of Global Mean Sea level for LIG and present day climates: an intercomparison of ESMs, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17140, https://doi.org/10.5194/egusphere-egu26-17140, 2026.

Extreme weather represents one of the most significant consequences of a warming climate. Improving constraints on how such events may manifest in the future is therefore a key priority, particularly for hazards that lead to severe societal, ecological, and financial impacts, such as heatwaves, extreme rainfall, droughts and their compounding effects.

Past interglacial periods provide physically realised instances of warm-climate states that can be used to contextualise ongoing anthropogenic warming and to inform future changes. Each interglacial is characterised by a distinct combination of orbital forcing and greenhouse gas concentrations, ice-sheet configuration, and background climate. Comparing these periods allows the partial isolation of the roles played by different climate drivers and large-scale circulation patterns in shaping the frequency, intensity, variability and spatial distribution of extreme events.

Here, we compare extreme weather characteristics across four interglacial periods: the mid-Holocene (6 ka), the Last Interglacial (127 ka), Marine Isotope Stage 11 (408 ka), and Marine Isotope Stage 31 (1072 ka), alongside a pre-industrial control. The analysis is based on preliminary equilibrium time-slice simulations conducted using the HadGEM3-GC5.0 coupled climate model, which also enables an initial assessment of model performance across a range of interglacial climates. We demonstrate how distinct warm-climate conditions have affected polar and global extreme events in the past and discuss the mechanisms underpinning these changes and their relevance for future climates. 

How to cite: Neild, J., Sime, L., Zhang, X., McLaren, A., Malmierca-Vallet, I., and Diamond, R.: Extremes of the past: what interglacial periods reveal about weather of the future, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17548, https://doi.org/10.5194/egusphere-egu26-17548, 2026.

Climate models project a future weakening of the Walker circulation in the tropical Pacific in response to anthropogenic forcing, while a cooling of the eastern equatorial Pacific and a strengthening of the Walker circulation has been observed in the past decades. This discrepancy may arise from models biases in the representation of Pacific dynamics, from a transient response of the ocean atmosphere system, or from unforced decadal climate variability. Here, we propose to use paleoclimate reconstruction of ENSO variance in the mid Holocene to evaluate the skills of CMIP5 and CMIP6 models and constrain climate change projections. The 20 model ensemble shows a slight mean reduction in ENSO variance that underestimates the 50-80% reduction in reconstructions, while exhibiting a large diversity of responses ranging from a 60% decrease to a 40% increase. We show that models that best represent mid Holocene ENSO changes display a weaker modern cold tongue bias, stronger mid Holocene cooling, and a more realistic representation of the ENSO seasonality and wind response to SST. Those models also yield a stronger eastern Pacific warming and zonal gradient reduction by the end of the 21st century in global warming scenarios (SSP585 and rcp85). Although mid-Holocene climate change is driven by orbital forcing rather than GHG, the robustness of this constraint is supported by the fact that ENSO integrates large scale ocean atmosphere feedbacks, which are key to the future response of the Pacific ocean.

How to cite: David, H., Carré, M., Khodri, M., Colas, F., Vialard, J., and Braconnot, P.: Mid-Holocene ENSO constraints point to future weakening of Walker Circulation, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18085, https://doi.org/10.5194/egusphere-egu26-18085, 2026.

The recognition by 19^th Century science that glaciers not only move but were, at various times in the past, big enough to submerge continents gave rise to Ice Age Theory and revolutionised our understanding of the Earth system, demonstrating that climate can – and does – change. In the intervening years, however, glacial geology and the physical record of cryospheric growth and decay have been largely relegated to playing a supporting role to higher-resolution, better-dated palaeoclimate proxies that today dominate conversation around ice age cycles, the causes and impacts of abrupt climate change, and the nature of climate tipping points. For instance, ice cores revealed the dramatic shifts in atmospheric conditions during Dansgaard-Oeschger and Heinrich Stadial events, while marine geochemistry tells us of the ocean’s role as a dynamic CO₂ reservoir and global heat capacitor. Two key concepts to have emerged from palaeoclimatology include (1) that of North Atlantic ‘stadial’ events as periods of intense regional cooling, typically with fast onset and rapid termination, and (2) the existence of a bipolar seesaw, in which cooling (warming) in one hemisphere drives relative warming (cooling) in the other. Both are deeply rooted in modern conceptual models of abrupt climate change and incorporated in numerical model projections of future climate. Here, we draw from recent refinements in cosmogenic nuclide geochronology and a growing database of well-dated Late Pleistocene moraine records to explore how well these two seminal concepts stand up to scrutiny from a reinvigorated glacial perspective. Exploiting the sensitive yet innately straightforward relationship between melt season (summer) temperature and glacier mass balance, this emerging glacial record paints a fascinating picture of ‘stadial’ climate that contrasts with the traditional view of these severe perturbations as year-round cold anomalies driven by AMOC. We highlight strong similarities between our North Atlantic glacier records and those from other regions globally and propose that anomalous thermal seasonality in the North Atlantic is a regional by-product of overall global warming. The ramifications of this hypothesis extend far beyond the North Atlantic: the interhemispheric bi-polar seesaw hypothesis rests on the coincidence of apparent Northern Hemisphere stadial cooling and Southern Hemisphere warming. Should the pattern of globally uniform glacier (and thus climate) behaviour invoked here prove an accurate representation of atmospheric conditions during abrupt climate shifts, the physical basis for a bipolar seesaw mechanism is undermined.

How to cite: Bromley, G., Hall, B., and Putnam, A.: Exploring new glacial perspectives on tenets of abrupt climate change: North Atlantic stadials and the bi-polar seesaw, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18226, https://doi.org/10.5194/egusphere-egu26-18226, 2026.

Tipping points in the climate system mark abrupt, irreversible dynamic transitions, posing difficulties for prediction and risks for mitigation. Developing reliable early-warning indicators is therefore essential. Linear response has long been used for predicting a system's change with respect to external perturbations. The stability of the predictions depend on how far the system from tipping and, hence, it provides a candidate measure to detect tipping events. Here, we propose using the Koopman operator spectrum to quantify linear response stability through the spectral gap, which shrinks near bifurcation. We apply this approach to Veros, an ocean general circulation model, and demonstrate that spectral-gap narrowing precedes critical transitions— providing the basis for Koopman-based response prediction in complex climate models.

How to cite: Santos Gutierrez, M.: Devicing early-warning signals using linear response: A Koopman operator approach, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18521, https://doi.org/10.5194/egusphere-egu26-18521, 2026.

The Zagros mountains of Iran reside at a climatically sensitive convergence zone between three major atmospheric circulation systems: the mid-latitude Westerlies, the Indian Ocean Monsoon, and the Intertropical Convergence Zone. At a regional scale, variability in these systems has strongly shaped hydroclimate and human–environment interactions through time. More locally, as this mountain range exists adjacent to the Fertile Crescent, their dynamic interplay has implications for the very earliest of human civilizations. Thus, climate and ecological reconstructions help us to shed light on some of the most pressing archaeological questions, but they also help us to understand how humans have adapted to climatic change.

Lake Zeribar provides a well-established palaeoenvironmental archive for the central Zagros Mountains, with previous palynological analyses of lacustrine sediment cores spanning approximately the last 40,000 years BP. Building on this foundational work, we develop a quantitative climate reconstruction by integrating fossil pollen assemblages with a modern calibration framework. A regional climate space is constructed using open-access pollen data from the Eurasian Modern Pollen Database (EMPD2), while associated climate variables are derived from WorldClim. Fossil pollen assemblages from Lake Zeribar are then used to reconstruct mean annual precipitation and temperature, providing new quantitative constraints on past hydroclimate variability in this climatically sensitive region. While acknowledging the limitations inherent to classical pollen-climate transfer functions, this study represents a primary step in a larger climate modelling project (Swiss-French Sinergia MITRA Project) for the eastern Fertile Crescent region. The resulting reconstruction provides a benchmark against which more complex and mechanistic approaches will be evaluated.

How to cite: Djamali, M., Enke, S., Gandouin, E., and Guiot, J.: A continuous quantitative pollen-based climate reconstruction for Lake Zeribar, Southwest Asia since Marine Isotope Stage 3, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19104, https://doi.org/10.5194/egusphere-egu26-19104, 2026.

Obtaining a statistical description of the extreme events that occur in a system that exhibits tipping behaviour is challenging due to the strong changes that the system undergoes. In this work, we use non-stationary Generalized Extreme Value (GEV) distributions to study the statistics of the extremes while capturing their temporal variability relative to a covariate data series which can be a driver or a response to the tipping. We exemplify this methodology by employing 8000 year long CCSM4 simulations with low concentrations of atmospheric CO₂ that show spontaneous D-O oscillations. This setting allows to study the minimum annual temperatures across the globe as a function of the temporal variability of the strength of the AMOC. The parameters of the distribution convey information about how the nature of the changes observed and its spatial variability, giving an insight on how the strength of the AMOC is related with the magnitude, variability and tails of the distributions. The extrapolation capabilities of this method are discussed compared to other studies and mechanisms of AMOC collapse. 

How to cite: del Amo, I. and Ditlevsen, P.: Dansgaard-Oeschger Events as Laboratory for Extremes Variability under AMOC Collapse, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19670, https://doi.org/10.5194/egusphere-egu26-19670, 2026.

The Greenland Ice Sheet (GrIS) plays a key role in the global climate system by acting as an interglacial refrigerator closely coupled to North Atlantic ocean circulation. Global warming is presently forcing the GrIS to lose mass with at least 27 cm of sea level rise committed regardless of future climate pathways. Greenland’s total ice mass corresponds to ~7 m of sea level, and recent paleo-data indicates that at least 1.4 m of ice loss occurred during Marine Oxygen Isotope stage 11 around 420.000 years ago. Thus, it is crucial to inform Earth System Models on past GrIS dynamics, in particular when and how fast major reductions in ice volume occurred. Part of the task for P2F WP9 is to apply information from deep ocean drilling records to identify the response of the Greenland Ice Sheet to warm climate extremes. With focus on the Pleistocene “super-interglacials” and abrupt transitions, this presentation compares various proxy-data from deep drilling sites west (IODP400 Baffin Bay) and south (IODP303 Labrador Sea) of Greenland which are influenced by the warm North Atlantic surface waters. 

How to cite: Knutz, P., Monedero-Contreras, R. D., Störling, T., Perez, L. F., Sliwinska, K., Kjær, H. A., Zeppenfeld, C., Sangiorgi, F., and Nelissen, M.: Abrupt Pleistocene transitions in deep ocean drilling records west and south of Greenland, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21541, https://doi.org/10.5194/egusphere-egu26-21541, 2026.

Diversity in the responses of species to environmental variability is fundamental to building ecosystem resilience. In behavioral ecology, risk refers to variance in the outcomes of behaviors with near-term (i.e., fitness-related) consequences, and humans are especially skilled at finding innovative ways to minimize subsistence risk. Formal models for risk-sensitive decision making can reveal how particular combinations of subsistence activities minimize variance arising from climatic and environmental conditions. However, an analytical framework for assessing the extent to which the diversity of these activities promotes the capacity for human social-ecological systems (SESs) to absorb disturbances and reorganize and renew themselves is yet to be developed, and hence we are unable to reliably address resilience in ancient SESs. Here, we adapt a population dynamics model of multiple interacting and mutualistic species to simulate the impacts of external (i.e., climate) variability on equilibria. In this approach we treat three alternative, yet complementary subsistence strategies (i.e., cultivation, pastoralism, and hunting) as interacting species in a heterogeneous environment. We parameterize interaction effects between the three “species” based on payoff matrices that define the relative benefits of one strategy over another to subsistence farming economies. Different combinations of subsistence activities are expected to arise as payoff matrices are subject to variable climatic and environmental constraints on productivity. We use downscaled TRACE21K-II output variables to simulate interannual variation in returns from each strategy. The simulation produces a time series of idealized proportional contributions of each strategy to overall subsistence. We then test the model predictions against macrobotanical and faunal remains recovered from lakeside settlements (i.e., pile dwellings) in the Northern Alpine Foreland spanning to the Neolithic (6.2-4.3 kya cal. BP).

How to cite: Triozzi, N. P., Ashwin, P., Bradshaw, C., del Amo Blanco, I., Abdelkader Di Carlo, I., Hoebe, P. W., Peeters, H., Poska, A., Kolář, J., Jacomet, S., Schibler, J., and Heitz, C.: Assessing Resilience Capacities and Vulnerability in Agropastoral Societies using an adapted Lotka-Volterra modelling framework, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21842, https://doi.org/10.5194/egusphere-egu26-21842, 2026.

Terrestrial Water Storage (TWS) represents all forms of water on land, including the cryosphere (polar ice sheets and mountain glaciers), the biosphere (canopies), soil and subsurface water (groundwater), and other inland water bodies (reservoirs, rivers, lakes, and wetlands). Modelling TWS remains a challenge because of difficulties in representing the water cycle on land. Furthermore, TWS is the source of mass-driven sea level change, an increasingly important contributor to sea level variation across the globe in the 20^th and 21^st centuries, with significant implications for coastal areas.

Here, we leverage the potential of machine learning and propose a Physics-Informed Neural Networks (PINNs) framework for modeling and predicting TWS and its associated sea level impacts. Because TWS varies in space and time, we build our framework based on convLSTM, an architecture that is suitable for serially-correlated “two-dimensional images” of data. The physical constraint for our PINNs is provided by the physics of continental-ocean mass redistribution, i.e., the sea level component, as described by the gravitationally self-consistent methodology of the sea level equation. The sea level equation connects TWS and sea level change by considering the gravitational, rotational, and deformational feedbacks caused by TWS components, particularly the cryosphere.  

We train and test our PINNs based on global TWS data from 1900 up to the end of 2018 (1900-2001 for training; 2002-2018 for testing). The data have a temporal resolution of 1 year and a spatial resolution of , and were derived from an assimilation of models and satellite gravimetry observations (in the time period 2002-2018). We perform various tests and discuss the advantages and shortcomings of our PINNs framework for modeling and predicting TWS. First, we show that TWS and sea level rise can be predicted reasonably well up to 10 years ahead (relative error of less than 30% on a global scale). This might prove useful for studies of sea level rise in coastal areas. Second, we compare our predictions with those of the Ice Sheet Model Intercomparison Project (ISMIP) in CMIP6 climate models, and satellite observations of Gravity Recovery and Climate Experiment (GRACE; in the range 2015-2024). We demonstrate that our predictions are closer to GRACE observations and, therefore, more accurate (up to 40% for the lead horizon of 10 years) compared to ISMIP projections under high and low emission pathways. Finally, we discuss how the predictions could be further improved by using probabilistic deep learning approaches, particularly  so-called deep ensembles. Our results show that once trained, PINNs can provide predictions orders of magnitude faster than climate models and with better accuracy.

How to cite: Kiani Shahvandi, M., Gasparini, B., and Voigt, A.: Physics-informed neural networks predict changes in terrestrial water storage and sea level, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-594, https://doi.org/10.5194/egusphere-egu26-594, 2026.

Generating high-resolution (HR) weather and climate information at ~10 km or finer across the Himalayan regions remains a major challenge due to extremely high computational cost of forecasting models and complexity of atmospheric processes. Most operational global weather prediction systems run at low-resolution (LR) of ~25 km or coarser, that are inadequate for impact-based analyses of highly localized extreme weather events common to these regions. To bridge this gap, downscaling is essential for producing climate information at impact-relevant scales, with both statistical and dynamical approaches remaining widely used despite major shortcomings. The former is computationally efficient but often fail under future climate non-stationarity, while the latter, though physically consistent, is computationally expensive and constrained by domain-resolution trade-offs. Currently, there is no efficient data-driven approach that can produce regional-model-scale precipitation fields for the Himalayan region. This work presents WGAN, a deterministic deep neural generative adversarial network (GAN)-based emulator of the Weather Research and Forecasting (WRF) model for HR precipitation downscaling over the Himalayan region. The model is conditioned on LR meteorological variables from the European Centre for Medium-Range Weather Forecasts Re-Analysis version 5 (ERA5; 0.25°×0.25°) as input and is trained against HR precipitation from WRF (0.1°×0.1°), which uses ERA5 as boundary conditions. The architecture uses Wasserstein-1 distance (WGAN) in the generator and critic value functions with a gradient penalty for stable training. WGAN demonstrated the ability to generate fine-scale precipitation fields that closely matches WRF’s outputs, accurately capturing spatial patterns and the mean values. Incorporating terrain and an extreme aware-weighting MSE (Mean Squared Error) loss function in the model further improves precipitation magnitude representation, reduces biases, and yield ~29% reduction in RMSE in the upper decile. The model effectively captured low-frequency (large-scale) variability and better matches WRF’s power spectrum at mid-high frequency (short-scale) variability. This raises the probability of detection and lowers the false alarm rate across thresholds. With a case study, WGAN showed the ability to capture the fine-scale spatial distribution of precipitation in the mountains and foothills, at both extreme precipitation day and dry conditions, outperforming CNN-based precipitation output. These results underscore the capability of WGAN as a fast and efficient tool for precipitation downscaling for the Himalayan region, operating at only a fraction of the computational cost. The model has strong potential for operational use in early warning, risk assessment, vulnerability analysis, disaster management, and other sectors that rely on localized climate information, ultimately supporting the preparedness of communities living in and around these mountains.

How to cite: Lyngwa, R., Nikumbh, A., and Ghosh, S.: A Generative-driven Model for Precipitation Downscaling Over Himalayan Region, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-821, https://doi.org/10.5194/egusphere-egu26-821, 2026.

Exploring uncertainty and internal variability across future emission pathways remains computationally demanding with state-of-the-art Earth system models (ESMs). We present a diffusion-based machine-learning emulator trained on output from the CESM2 large ensemble dataset to reproduce absolute annual-mean temperature and year to year variability,  conditioned on anthropogenic co2 and sulfate emisisson from ssp3-7.0 scenario. The emulator employs a three-dimensional UNet architecture that learns the spatiotemporal distribution of global temperature fields in latitude–longitude–time space. Conditioning variables include cumulative CO₂ and aerosol emissions, enabling the generation of physically consistent climate responses under arbitrary emission trajectories.To enhance physical interpretability, we integrate explainable AI (XAI) methods, including gradient-based attribution and sensitivity analyses, to quantify how emission-related conditioning variables influence regional temperature responses. The emulator reduces computational cost by several orders of magnitude compared to full ESM simulations, enabling rapid scenario exploration and uncertainty assessment. This framework aims provides a scalable and interpretable pathway for fast climate response emulation

How to cite: Nordling, K.: Emulating absolute annual temperatures and variability  from the CESM2 Large Ensemble using a diffusion model, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-2500, https://doi.org/10.5194/egusphere-egu26-2500, 2026.

Earth system models (ESMs) are key tools in projecting the reponse of the Earth's climate and ecosystems to anthropogenic forcing in terms of increasing greenhouse gas concentrations and resulting temperature increases, as well as land use change. However, ESMs continue to suffer from prononced biases when compared to observations, and exhibit limited horizontal resolution due to computational constraints, mking reliable impact assessment challenging. Generative machine learning methods, such as Generative Adversarial Networks or Diffusion models, have shown great success in bias correcting and downscaling Earth system model output [1,2]. However, so far these approaches have been applied only as a postprocessing. After summarizing advances in this context, I will present recent work addressing conceptual and technical challenges in incorporating (generative) machine learning inside the architectures of process-based ESMs. These include the need for automatic differentiability of all ESM components [3], as well as physical constraints to assure that dynamics learned by machine learning components fulfills, for example, physical conservation laws [4].  

[1] P. Hess, M. Drüke, F. Strnad, S. Petri, N. Boers: Physically constrained generative adversarial networks for improving precipitation fields from Earth system models, Nature Machine Intelligence 4, 828-839 (2022)

[2] P. Hess, M. Aich, B. Pan, N. Boers: Fast, scale-Adaptive, and uncertainty-aware downscaling of Earth system model fields with generative machine learning, Nature Machine Intelligence 7, 363–373 (2025)

[3] M. Gelbrecht, A. White, S. Bathiany, N. Boers: Differentiable Programming for Earth System Modelling, Geoscientific Model Development 16, 3123–3135 (2023)

[4] A. White, N. Kilbertus, M. Gelbrecht, N. Boers: Stabilized Neural Differential Equations, NeurIPS (2023)

How to cite: Boers, N.: Machine learning for hybrid Earth system modelling, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3829, https://doi.org/10.5194/egusphere-egu26-3829, 2026.

Uncertainty in projections of future regional climate change remains large, driven by structural differences among Earth System Models and the influence of internal climate variability. Existing uncertainty-reduction approaches, including emergent constraints and Bayesian variants, primarily focus on forced climate responses derived from simple aggregate metrics, thereby requiring strong assumptions and exploiting only low-dimensional climate information. Here we propose a data-driven deep-learning framework that directly forecasts spatially and monthly resolved decadal mean climatologies of surface temperature anomalies from the 2030s to the 2090s, using only recent monthly trajectories spanning 1980-2025. The training ensemble contains 265 historical+SSP2-4.5 simulations, distributed across 40 ESMs from 25 different families (i.e., modelling centers) over which the cross validation is performed. The architecture couples pluri-annual to multi-decadal temporal convolutions with a spatial U-Net encoder-decoder and is evaluated on CMIP6 simulations using a leave-one-model-family-out cross-validation (LOMFO-CV) design to ensure generalisation across separately developed ESMs. Predictive uncertainty is quantified via LOMFO-CV errors, yielding conservative and reliable ranges that incorporate irreducible internal variability and systematic model shifts.

To further evaluate the predictive capacity beyond the CMIP6 distribution, we evaluated the network on historical+SSP2-4.5 simulations from a recent HadGEM3-GC5 model hierarchy developed within the European Eddy-Rich ESMs (EERIE) project, the European contribution to HighResMIP2 for CMIP7. In particular, the eddy-rich GC5-HH configuration explicitly simulates mesoscale ocean dynamics that are absent in CMIP6-type models, providing a rigorous test of generalisation to richer and more realistic physical representations. Despite these substantial differences, the network successfully reproduces warming trajectories and future climate patterns for all three model configurations (GC5-LL, GC5-MM, GC5-HH), with forecast errors largely contained within empirically calibrated uncertainty bounds from the LOMFO-CV, both globally and locally. These results, notably for GC5-HH and its more realistic physics, strengthens confidence in the applicability of the framework to real-world data.

When applied to observations, the extracted end-of-century global-mean surface temperature and its uncertainty range are consistent with prior estimates from Bayesian frameworks. At local scales, the network reduces uncertainty by 40% (2030s) to 30% (2090s) on average, and by up to 75% in some regions for all future decades. Importantly, these uncertainty estimates account not only for uncertainty in the forced response (as emergent constraint methods do), but also for errors associated with predicting different realisations of internal variability, providing a physically meaningful reduction of local and global climate uncertainty.

How to cite: Michel, S., Strommen, K., and Christensen, H.: Short- to long-range climate forecasts with deep learning, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4158, https://doi.org/10.5194/egusphere-egu26-4158, 2026.

Regional Climate Models (RCMs) are vital for capturing mesoscale variability, however remain too coarse for impact assessments in complex topographies like Switzerland. In this study, we bridge the "km-scale gap" by introducing a generative super resolution pipeline to downscale EURO-CORDEX ensemble to a 1 km grid over Switzerland.

We establish the added value of a deterministic residual U-Net, pixel-based as well as generative residual Latent Diffusion over operational baselines and conventional bias correction (BC) methods such as Cumulative Distribution Function - transform (CDF-t), Empirical Quantile Mapping (EQM) and dynamical Optimal Transport Correction (dOTC). Our results demonstrate that super resolved fields have superior distributional skill, better visual fidelity of fields, shows improved  trend preservation and representation of interannual variability across diverse biogeographical regions  and major population centres such as Bern, Zurich and Locarno. Further, as demonstrated by a marked reduction in bias for  20-, 50-, and 100-year return levels of multi-day precipitation totals, super resolution (SR) also complements BC for improved representation of extremes in our km-scale downscaled EUROCORDEX. Our findings establish that while BC methods remain essential for distributional fidelity, residual generative models offer a potent, actionable pathway for producing high-resolution climate information from coarse climate fields.

How to cite: Asthana, S., Koch, E., Kotlarski, S., and beucler, T.: Next-Generation Climate Projections: Insights from Blending Bias Correction with Super Resolution over Complex Terrain, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4507, https://doi.org/10.5194/egusphere-egu26-4507, 2026.

Machine learning-based weather prediction is revolutionizing weather forecasting by learning from present-day climate. However, generalization to other climates remains a major challenge. With melting sea ice, land-use change and increasing ocean temperatures, boundary conditions are changing. Therefore, generalization in time will likely only be possible if generalization in space is also given. The physics of the atmosphere is invariant in space, and as such, a model should demonstrate the same to accurately represent the real world.

Here, we present three test cases to evaluate whether machine learning-based weather and climate models generalize spatially and apply them to multiple AI weather models. The tests consist of reversing the entirety of the input data and boundary conditions in latitude (Test 1), reversing them in longitude (Test 2), as well as rotating them by 180˚ in longitude (Test 3), while keeping all aspects of the simulation physically consistent. For a deterministic model that generalizes in space, each of these test cases yields the same predictions as the baseline case, only subject to a rounding error. With these test cases, we investigate whether data-driven models hardcode representations of spatial relationships in the training data into their latent space. We show that currently, both fully data-driven and hybrid general circulation models do not pass these tests, instead performing poorly with unphysical results. This implies that they have likely not learned underlying atmospheric physics principles, but instead local spatial relationships statistically dependent on geographical location. This calls into question the ability of such models to simulate a changing regional climate. As such, we propose that machine learning-based climate models be evaluated using our spatial tests during model development to reduce overfitting on present-day regional climate.

How to cite: Höver, M., Klöwer, M., Schroeder de Witt, C., and Christensen, H. M.: Spatial Generalization Tests for Machine Learning-based Weather Models as a Requirement for Climate Predictions, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4512, https://doi.org/10.5194/egusphere-egu26-4512, 2026.

Understanding long-term trends in stratospheric species is vital for evaluating the success of the Montreal Protocol and its amendments. However, reliable trend estimation remains challenging due to the sparse spatial and temporal coverage of high-quality observations, such as those from the Atmospheric Chemistry Experiment–Fourier Transform Spectrometer (ACE-FTS).

How to cite: Dhomse, S. and Chipperfield, M.:   Machine Learning For Atmospheric Chemistry: Creating Global, Gap-Free Stratospheric Datasets for Montreal Protocol Assessments, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4514, https://doi.org/10.5194/egusphere-egu26-4514, 2026.

Machine learning models have shown great success in predicting weather up to two weeks ahead, outperforming process-based benchmarks. However, existing approaches mostly focus on the prediction task, and do not incorporate the necessary data assimilation. Moreover, these models often suffer from long-term error accumulation, limiting their applicability to seasonal predictions and climate projections. Here, we introduce Generative Assimilation and Prediction (GAP), a unified deep generative framework for assimilation and prediction of both weather and climate. By learning to quantify the probabilistic distribution of atmospheric states under observational, predictive, and external forcing constraints, GAP excels in a broad range of weather-climate related tasks, including data assimilation, seamless prediction, and climate simulation. In particular, GAP delivers probabilistic weather forecasts competitive with state-of-the-art forecasting systems, while using its own assimilated initial states from a small fraction of observations. Also, it provides seasonal predictions with skill comparable to leading operational system. Finally, GAP produces stable millennial-scale climate simulations that capture variability from daily weather fluctuations to decadal oscillations.

How to cite: Yang, S., Nai, C., Boers, N., Yuan, H., and Pan, B.: GAP: a unified deep generative framework for emulating weather and climate, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4964, https://doi.org/10.5194/egusphere-egu26-4964, 2026.

A pressing problem exacerbated by climate change is the inability to prepare for extreme climate and weather events due to the limited historical record of observed extremes. While crucial for risk assessment and informed policy-making, a better representation of the distribution of "feasible" outcomes remains largely uncertain, with  predictions ranging at variously defined confidence levels that remain sensitive to the choice of metrics and physical assumptions. This question naturally lends itself to investigating how we can engender plausible realizations of extreme events, and thereby allow for mitigation efforts, before communities are forced to confront destructive realities. We present a time-conditioned generative framework based on a computer-vision-aided diffusion model trained on 1km $\times$ 1km precipitation fields and their trajectories over time. The output of this model is n future potential realizations of possible storm events that may unfold over the San Jacinto river basin in the south coast of Texas.

Beyond unconditional sampling, we introduce control variables that make generation decision-relevant: the model is trained to be conditional on a (duration, intensity) pair, enabling users to request ensembles spanning targeted severity regimes (e.g., short–extreme vs. long–moderate) while preserving realistic spatiotemporal structure. This yields a family of distributions over storm trajectories indexed by interpretable controls, allowing systematic stress testing of infrastructure and emergency-response plans under plausible but high-impact scenarios. 

We separate the evaluation of our approach into two complementary perspectives: (i) distribution matching for in-sample generations, and (ii) physics-based alignment with storm-based properties for out-of-sample generations. Spatiotemporal structure of storms is also benchmarked against strong baselines like the analog ensemble method, quantifying the performance of our model to realistically capture intense rainfalls. To extract evolving storm geometries, we employ a kNN-based (k-nearest neighbors) computer-vision algorithm that dynamically identifies storm shapes across time steps. Due to the probabilistic nature of diffusion models, more comprehensive envelopes of the storm intensity and trajectory can be obtained for uncertainty quantification purposes. 

Finally, we introduce a metric that jointly measures physical plausibility through features like intensity–duration structure and scaling, as well as novelty relative to the raw training data. This metric works by penalizing overfitting patterns while rewarding those that respect feasible dynamics, allowing us to define a principled way to compare generative models for extremes. Therefore, we can determine not only how realistic our generated storms are, but also how much physical diversity they contribute beyond the observed data. We present an open evaluation suite for controllable storm generation, including storm-tracking, intensity–duration diagnostics, and physical-novelty scoring.

How to cite: Tsao, V., Zaniolo, M., and Veveakis, M.: Synthetic Physics-Aware Storm Generation via Diffusion Models for Risk Analysis of Catastrophic Events, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5928, https://doi.org/10.5194/egusphere-egu26-5928, 2026.

Various pollutants pose significant threats to river ecosystems. This issue is particularly critical in Taiwan, where the unique geography of short, rapid rivers makes water retention difficult, necessitating rigorous water quality monitoring. Given the complex, non-linear correlations between water quality and meteorological parameters, this study investigates the impact of different feature selection techniques and predictive models on water quality forecasting for eight rivers in Taoyuan. We utilized 14 meteorological and water quality inputs to predict six key indicators, including COD, DO, EC, NH3-N, ORP, and SS. The methodology compared four feature selection strategies—Pearson Correlation, Entropy Weight Method (EWM), Combined Weights, and Mutual Information—alongside four forecasting models: Seq2Seq LSTM, ANFIS, MLP, and Transformer.The feature selection results reveal that the Entropy Weight Method yielded the highest precision (R^2 =0.9336), surpassing the Pearson method (R^2 =0.9161). This indicates that prioritizing features based on information entropy effectively minimizes information loss during screening. Regarding predictive modeling, the Transformer model demonstrated superior stability and accuracy. While other models fluctuated, the Transformer consistently achieved the best performance with an MSE of approximately 14.86 (RMSE=3.855) and an accuracy of 82.52%, significantly outperforming the MLP and ANFIS models. This study concludes that integrating entropy-based feature selection with the Transformer model establishes a superior and highly accurate framework for water quality forecasting in Taoyuan's rivers.

How to cite: Lu, Y.-Y. and Lin, Y.-C.: Integrated Analysis of Feature Selection and Deep Learning Models for Water Quality Forecasting Based on Meteorological Parameters, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6357, https://doi.org/10.5194/egusphere-egu26-6357, 2026.

Air pollution has emerged as one of the most critical environmental health hazards globally. According to statistics from the World Health Organization (WHO) and the Global Burden of Disease Study, approximately 7 million premature deaths occur annually due to air pollution. Fine particulate matter (PM2.5), capable of penetrating deep into the lungs and entering the bloodstream, has been confirmed to be highly correlated with ischemic heart disease, stroke, chronic obstructive pulmonary disease (COPD), and lung cancer. Given its serious threat to public health, establishing high-precision PM2.5 prediction models is critical for early warning systems and health protection.

Addressing the common issue of missing values in environmental monitoring data, this study proposes a data preprocessing framework that combines Principal Component Analysis (PCA) for feature dimensionality reduction with the GRU-D model for time-series imputation. Testing confirms that this method effectively reconstructs data features without causing excessive smoothing. In terms of predictive modeling, this study incorporates East Asian-scale atmospheric pressure field data as a key environmental variable to capture the impact of large-scale weather systems on local air pollution. The performance of three advanced deep learning models—LSTM+CNN, PatchTST, and iTransformer—is evaluated and compared.

The results indicate that, when considering multivariate factors and long- and short-term dependencies, the iTransformer model demonstrates superior predictive performance with an R² of 0.91, exhibiting exceptional non-linear feature extraction capabilities. In comparison, both the LSTM+CNN and PatchTST models achieved an R² of approximately 0.86. Based on the iTransformer's advantages in handling large-scale meteorological features and high-dimensional time-series data, this study employs it as the core model to further extend PM2.5 concentration predictions across Taiwan, aiming to provide a valuable scientific reference for regional air quality management.

How to cite: Chuang, Y. and Lin, Y.-C.: Spatiotemporal Prediction of PM2.5 in Taiwan Using iTransformer and Large-Scale Atmospheric Pressure Features, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6363, https://doi.org/10.5194/egusphere-egu26-6363, 2026.

In recent years, climate change has led to a clear increase in both the frequency and intensity of extreme weather events. Taiwan lies along major typhoon tracks in the western North Pacific, where typhoons represent one of the most significant natural hazards. The strong winds and heavy rainfall associated with typhoons frequently cause flooding, agricultural losses, and damage to critical infrastructure. In practice, however, the severity of typhoon-related disasters does not always correspond to traditional typhoon intensity classifications based primarily on central pressure and wind speed, indicating that wind-based classifications alone may not adequately represent actual disaster impacts.

This study utilizes hourly meteorological station observations to investigate the wind and rainfall characteristics of historical typhoon events in Taiwan. Multiple machine learning and regression models are applied, together with residual analysis, to quantify typhoon characteristics and construct a Typhoon Type Index (TTI). Based on the relative behavior of wind and rainfall during individual events, different typhoon types are further examined to identify their occurrence patterns and characteristic differences across historical cases.

The results indicate that the TTI derived from machine learning–based classification models can effectively improve upon previous TTI formulations established using regression models alone. Moreover, typhoons with different wind–rainfall characteristics are associated with distinct patterns of disaster impacts, and in some cases, rainfall intensity better reflects disaster severity than wind speed. By offering an alternative perspective to conventional intensity-based classifications, this study contributes to improved typhoon disaster risk assessment and provides useful insights for future disaster mitigation and preparedness strategies.

How to cite: Huang, S.-H. and Lin, Y.-C.: Improving the Typhoon Type Index by Integrating Strong Wind and Heavy Rainfall Using Machine Learning, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6364, https://doi.org/10.5194/egusphere-egu26-6364, 2026.

We introduce Field-Space Attention, a novel, scalable, interpretable, and flexible attention module designed for Earth system machine learning models. The key concept involves computing attention directly within physical space on the HEALPix sphere. This approach ensures that all intermediate states remain as globally defined geophysical fields rather than as abstract latent tokens. This field-centric design maintains the physical meaning of internal representations, renders layer-wise updates interpretable, and offers a simple interface for integrating scientific constraints and prior knowledge throughout the network (see Figure). Field-Space Attention is based on a fixed, non-learned, multiscale, spherical decomposition. It learns structure-preserving deformations that coherently couple information across coarse and fine scales. This enables global context without sacrificing local detail.

We demonstrate the module's effectiveness in representative Earth system learning experiments on spherical grids. We focus on global near-surface temperature super-resolution on a HEALPix grid using ERA5 reanalysis data and benchmark it against widely used Vision Transformer and U-Net–style baselines. Our Field-Space Transformer model trains more stably, converge faster, achieve strong accuracy with substantially fewer parameters, and yield physically interpretable intermediate fields.

By keeping computation in field space and explicitly separating scales, Field-Space Attention is particularly well-suited for high-resolution Earth system modeling. It supports scale-aware inductive biases, principled cross-scale consistency, and the efficient coupling of large-scale dynamics with fine-scale variability. These properties position Field-Space Attention as a compact building block for next-generation, high-resolution Earth system prediction and generative modeling. This includes downscaling, spatiotemporal forecasting, infilling, and data assimilation under stronger physical constraints.

How to cite: Witte, M., Meuer, J., Plésiat, É., and Kadow, C.: Field-Space Attention for Structure-Preserving Earth System Transformers, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7473, https://doi.org/10.5194/egusphere-egu26-7473, 2026.

The impact of droughts on vegetation is commonly assessed through correlational analysis of satellite-derived variables, such as NDVI, precipitation anomalies, soil moisture, and more (Hao & Singh 2015, Park et al. 2016, Joiner et al. 2018). However, these correlation-based approaches cannot disentangle the true causal drivers from their confounded associations (Zhang et al. 2022). This limits our ability to understand and attribute the scale of vegetation stress to specific drought mechanisms (e.g. soil moisture deficits versus irrigation resilience), and our ability to design effective interventions that address the primary drivers.

As such, we propose a novel causal inference framework to estimate the impact of drought on vegetation health using satellite time-series data, and demonstrate its application to the Iberian Peninsula. We firstly define a graphical causal model based on established eco-hydrological pathways, and then integrate multi-sensor remote sensing data (MODIS NDVI, SPEI, etc.) and climate reanalysis (ERA5). By extending traditional causal inference methods for georeferenced time-series raster data and controlling for well-established confounding variables (temperature, solar radiation, precipitation, soil moisture, land cover, and irrigation), we isolate the effect of drought severity on vegetation. We also implement novel visualisation methods to display these causal influence estimates.

While causal inference allows us to move beyond correlation and understand the impact on vegetation from each of these key variables, counterfactual intervention is also essential to understand how varying conditions would otherwise change the outcome (Schölkopf et al. 2021), i.e. the severity of the drought impact. Therefore, by leveraging these interventions, our results go from descriptive analytics to actionable insights on drought severity under the changing climate. This enables more effective drought impact assessment for scientists, policymakers, and industry experts.

References:
1. Hao, Z. and Singh, V.P., 2015. Drought characterization from a multivariate perspective: A review. Journal of Hydrology, 527, pp.668-678.
2. Park, S., Im, J., Jang, E. and Rhee, J., 2016. Drought assessment and monitoring through blending of multi-sensor indices using machine learning approaches for different climate regions. Agricultural and forest meteorology, 216, pp.157-169.
3. Joiner, J., Yoshida, Y., Anderson, M., Holmes, T., Hain, C., Reichle, R., Koster, R., Middleton, E. and Zeng, F.W., 2018. Global relationships among traditional reflectance vegetation indices (NDVI and NDII), evapotranspiration (ET), and soil moisture variability on weekly timescales. Remote Sensing of Environment, 219, pp.339-352.
4. Zhang, X., Hao, Z., Singh, V.P., Zhang, Y., Feng, S., Xu, Y. and Hao, F., 2022. Drought propagation under global warming: Characteristics, approaches, processes, and controlling factors. Science of the Total Environment, 838, p.156021.
5. Schölkopf, B., Locatello, F., Bauer, S., Ke, N.R., Kalchbrenner, N., Goyal, A. and Bengio, Y., 2021. Toward causal representation learning. Proceedings of the IEEE, 109(5), pp.612-634.

How to cite: Jancauskas, V., Garske, S., and Espinoza Molina, D.: A Causal Inference Framework for Analysing Drought Drivers, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7552, https://doi.org/10.5194/egusphere-egu26-7552, 2026.

Earth system model (ESM) intercomparison is essential for assessing model performance and identifying future challenges in climate modeling. The Taylor diagram [1] is one of the most widely used tools for this purpose, as it provides an intuitive summary of standard evaluation metrics — such as correlation, root-mean-square error, and standard deviation — by comparing multiple simulated datasets against a reference, typically observations or a ground truth, within a single plot.

However, in several relevant applications, including the development of new ESM parameterizations, the comparison of conceptual models, or the evaluation of simulated statistical distributions, classic linear correlation and RMSE metrics may be insufficient. Here, we propose a set of extensions to the Taylor diagram based on a generalization of cross-covariance using kernels, allowing both nonlinear relationships and distributional aspects of similarity to be taken into account. Nonlinear similarity is characterized through a kernel-space analogue of rotational alignment, while distributional similarity can be quantified using metrics such as maximum mean discrepancy, as originally introduced in [2], as well as alternative kernel-based measures. Using controlled synthetic experiments, we show that the proposed kernel Taylor diagrams can resolve differences in model skill that remain indistinguishable under the classical Taylor diagram. These results indicate that the kernel-based extensions provide complementary diagnostic information to standard metrics and can support more informative Earth system model evaluation and development.

[1] Taylor, K. E. (2001), Summarizing multiple aspects of model performance in a single diagram, J. Geophys. Res., 106(D7), 7183–7192, doi:10.1029/2000JD900719.

[2] Wickstrøm, K., Johnson, J. E., Løkse, S., Camps-Valls, G., Mikalsen, K. Ø., Kampffmeyer, M., & Jenssen, R. (2022). The Kernelized Taylor Diagram. doi:10.48550/arXiv.2205.08864

How to cite: Gavrilov, A., Mankovich, N., Link, M., Huang, F., and Camps-Valls, G.: Kernel Taylor Diagram for Earth System Model Evaluation, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7678, https://doi.org/10.5194/egusphere-egu26-7678, 2026.

Understanding the driving forces behind mesoscale cloud organization is fundamental to reducing uncertainties in cloud climate feedbacks. Traditional climate models cannot explicitly resolve mesoscale cloud structures due to their limited resolution, leading to large uncertainties in cloud climate feedback estimates. Storm-resolving models that simulate the atmosphere at kilometre resolution have the potential to reduce these uncertainties. Yet, these models are still biased in their organizational structure when compared to satellite observations. Approaches constraining cloud feedbacks directly from the satellite records are promising but often rely on manually chosen cloud controlling factors (CCFs) that do not necessarily capture all the information necessary to explain mesoscale organizational structures and generally only utilise linear models to predict cloud radiative properties from CCFs.

We present CloudDiff, a probabilistic machine learning model that generates mesoscale cloud structures at kilometre resolution conditioned on environmental conditions in the atmosphere, namely the temperature and humidity profiles as well as vertical and horizontal winds. The model is trained on MODIS Level 1 satellite data and environmental conditions from ECMWF ERA5 reanalysis data. CloudDiff is able to reconstruct realistic MODIS observations from matching ERA5 environmental conditions and achieves a lower reconstruction error compared to generating MODIS observations solely from pre-defined CCFs. In CloudDiff’s generation stage, the environmental conditions are compressed into a latent representation using an attention mechanism. This latent representation can be interpreted as a set of CCFs that have been learned purely from data. We’ll discuss the properties of the learned CCFs including how they relate to existing CCFs, their geographical distribution, and their predictive power of the radiative properties of cloud fields.

How to cite: Reichelt, T. and Stier, P.: CloudDiff: A Conditional Diffusion Model to Generate Mesoscale Cloud Structures, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7971, https://doi.org/10.5194/egusphere-egu26-7971, 2026.

How to cite: Dewey, M., Wilcox, L., Samset, B., and Ekman, A.: Deep-AeroGP: deep kernel learning for projecting the regional climate response to anthropogenic aerosol emission changes , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8307, https://doi.org/10.5194/egusphere-egu26-8307, 2026.

Accurate initiation of deep convection remains a persistent challenge in weather and climate models. Most general circulation models (GCMs) operate at coarse resolution and therefore cannot explicitly resolve convective events; instead, they rely on convective parameterizations in which triggering is diagnosed from environmental thresholds, commonly based on convective available potential energy (CAPE). Convection-permitting models (CPMs) alleviate some of these structural limitations by resolving grid-scale convective spectrum while leaving behind sub-grid scale events. On the other hand, machine learning (ML)-based convection trigger functions have emerged, but still with uncertainty, whose causes are rarely examined. Here, we diagnose the atmospheric states associated with “blind spots” in ML predictors of deep convection initiation, leveraging the Department of Energy Atmospheric Radiation Measurement constrained variational analysis (VARANAL) product and the CPM-based CONUS404 hydroclimate dataset over the Southern Great Plains (SGP). We train a conventional artificial neural network (ANN) and a controlled abstention network (CAN), evaluate their skill in identifying deep convection, and use CAN to quantitatively isolate low-confidence samples while understanding the associated physical conditions in which the models are least reliable. ANN and CAN show comparable baseline performance, and for both models, skill increases when low-confidence samples are excluded, indicating that abstention identifies systematically difficult conditions rather than random noise. Across both VARANAL and CONUS404 datasets, low-confidence samples preferentially occur under weak-to-moderately negative mid-level vertical velocity (−10 to −5 hPa hr⁻¹) and dynamic generation rate of CAPE (dCAPE; 0–200 J kg⁻¹ hr⁻¹). Additionally, these cases are dominated by short, convective episodes that persist for only a few hours, dominantly occurring during the afternoon. These abstention samples also exhibit locally forced, non-equilibrium environments characterized by larger convective adjustment time (τ), consistent with reduced predictability relative to regimes controlled by broader synoptic forcing with smaller τ. Collectively, our results quantitatively identify the regimes and associated physical mechanisms in which ML-based convection predictors are least robust, providing actionable guidance for operational forecasters to treat predictions with greater caution when these low-confidence conditions are present.

How to cite: Bhattarai, A. and Zheng, Y.: Uncertainty-Aware Machine Learning for Deep Convection Initiation: Insights from ARM Observations and Kilometer-Scale Hydroclimate Reanalysis, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8710, https://doi.org/10.5194/egusphere-egu26-8710, 2026.

The emergence of global km-scale climate models challenges traditional model evaluation approaches, which typically rely on long climatological averages. The substantial computational costs and enormous data volumes associated with km-scale simulations often constrain simulation length, limiting the availability of long-term averages. As a result, conventional analysis methods become less practical and less informative when assessing short, high-frequency model output that is potentially dominated by internal variability. At the same time, recent advances in machine learning (ML), particularly in deep neural networks, offer new and innovative ways to efficiently extract information from large climate datasets. Building on this progress, we present an ML-based framework for evaluating climate models on a regional scale over short periods, focusing on daily near-surface air temperature fields over Europe.

We train a convolutional neural network (CNN) to distinguish spatial temperature fields from a large set of climate models. We employ 28 regional simulations from EURO-CORDEX and two global km-scale models from nextGEMS and Destination Earth. Beyond the classification based on climate model simulations, the pre-trained CNN is applied to observation-based test datasets. This setup allows us to build towards an evaluation metric, as the model, the observation-based datasets are more frequently assigned to, might be considered most similar to observed climate. Despite the regional focus of EURO-CORDEX, observation-based samples are most frequently classified as the global km-scale model IFS-FESOM. This suggests that this global km-scale model may capture regional temperature patterns more accurately than regional climate model simulations. Although our results are consistent with traditional metrics in identifying IFS-FESOM as the best-performing model, they also indicate that CNN-based evaluation provides additional information about the similarity between models and observations. 

To better understand which spatial features influence the CNN’s classification for observation-based samples, we apply explainable artificial intelligence (XAI) methods, specifically layerwise relevance propagation (LRP), to the classification outcomes. The resulting relevance patterns indicate that static features such as orography and coastlines, as well as relevance hotspots potentially linked to regions of dynamic variability, play a dominant role in the classification. This highlights that the CNN is sensitive to physically meaningful structures that define model-specific spatial fingerprints.

Using our ML-based framework, we show that a CNN can robustly distinguish between climate models on regional and short time scales as well as identify the model closest to observations. More broadly, we demonstrate that ML, combined with XAI, offers a scalable and physically interpretable approach for evaluating high-resolution climate models, thereby complementing established evaluation frameworks.

How to cite: Meindl, M., Kornblueh, M., Brunner, L., and Voigt, A.: Using Explainable AI to uncover physically meaningful features in km-scale climate models on a regional scale, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9427, https://doi.org/10.5194/egusphere-egu26-9427, 2026.

Following recent advances of foundation models in natural language processing and computer vision, there is growing interest in leveraging geospatial foundation models (GFMs) for Earth system monitoring and climate-relevant applications. In particular, GFMs promise to support large-scale observation of climate-driven extreme events such as wildfires, floods and landslides. However, despite strong benchmark results, recent studies indicate that GFMs for land-cover modelling and hazard mapping models can behave unreliably under real-world conditions. Pretraining datasets often underrepresent rare or extreme environmental regimes, leading to degraded model performance precisely in situations where robust predictions are most critical for climate risk assessment and disaster response. Furthermore, GFMs are often surpassed by simple supervised baselines, highlighting the need for systematic reliability analysis, including out-of-distribution (OOD) detection and uncertainty quantification.

We present SHRUG-FM (systematic handling of real-world uncertainty in geospatial foundation models), a reliability-aware prediction framework that integrates three complementary signals: (1) OOD detection in the input space, (2) OOD detection in the embedding space and (3) task-specific predictive uncertainty obtained from decoder ensembles. We evaluate SHRUG-FM on climate-relevant extreme-event applications, including burn-scar, flood and landslide segmentation. Our results show that elevated OOD scores consistently co-locate with degraded model performance, while uncertainty-based indicators successfully capture many low-confidence and erroneous predictions. By linking these reliability signals to hydro-environmental descriptors from HydroATLAS, we further demonstrate that model failures cluster in distinct geographic and hydroclimatic regimes, revealing interpretable gaps in the pretraining distribution and guiding future dataset design.

SHRUG-FM delivers practical, operationally relevant diagnostics for Earth system monitoring and prediction. It enables selective prediction, rejection strategies, and reliability-aware quality control. These capabilities are essential for integrating GFMs into real-world workflows for climate impact assessment, hazard monitoring and early warning systems. Future work will extend the framework to additional foundation models and climate-driven hazards.

How to cite: Cohrs, K.-H., Gonzalez-Calabuig, M., Nedungadi, V., Osika, Z., Cartuyvels, R., Knoblauch, S., Massant, J., Nath, S., Ebel, P., and Sitokonstantinou, V.: SHRUG-FM: Reliability-Aware Foundation Models for Earth Observation, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9940, https://doi.org/10.5194/egusphere-egu26-9940, 2026.

Statistical causality methods are becoming increasingly widespread in climate teleconnection analysis, but they typically require a prior reduction of high-dimensional, multivariate climate fields. Most common aggregation techniques, such as spatial averaging or Principal Component Analysis (PCA) (largely known as Empirical Orthogonal Functions, EOF, in the climate community) [1], are not designed to preserve causal structure and can mask spatially complex or low-variance causal signals.

We introduce Granger PCA [2], a novel dimensionality reduction method that explicitly extracts components that are influenced by a causal driver. Instead of maximizing variance, Granger PCA identifies spatial patterns whose associated time series are maximally Granger caused by an external variable, such as a large-scale climate mode. This is achieved by optimizing spatial weights to maximize the Granger causality F-statistic and yields a low-dimensional representation that captures the Granger causal information present in the field.

The method is particularly effective in cases where causal effects are spatially heterogeneous, have low variance, or are hidden by strong local autocorrelation. In such cases, variance-based methods can fail even when robust causal influence exists.

We apply Granger PCA to several teleconnection problems, including the influence of the North Atlantic Oscillation on precipitation and the impact of ENSO on vegetation variability. Granger PCA recovers physically interpretable patterns that are not captured by PCA or correlation-based approaches.

In summary, Granger PCA provides a simple and interpretable framework for causally oriented dimensionality reduction and offers a new tool for teleconnection analysis in climate science.

References

• [1] A. Hannachi, I. T. Jolliffe, D. B. Stephenson et al., “Empirical orthogonal functions and related techniques in atmospheric science: A review,” International Journal of Climatology, vol. 27, no. 9, pp. 1119–1152, 2007.
• [2] G. Varando, M.-Á. Fernández-Torres, J. Muñoz-Marí, and G. Camps-Valls, “Learning causal representations with Granger PCA,” in UAI 2022 Workshop on Causal Representation Learning, 2022.

How to cite: Durand, H., Varando, G., and Camps-Valls, G.: Granger PCA: Extracting Granger-causal patterns in climate fields, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10055, https://doi.org/10.5194/egusphere-egu26-10055, 2026.

How to cite: Bouabid, S., Womack, C., Flierl, G., Selin, N., Ferrari, R., Souza, A., Giani, P., and Lutjens, B.: Advances in generative climate emulation to support impact-assessment, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10825, https://doi.org/10.5194/egusphere-egu26-10825, 2026.

Africa and South America together account for more than 70 % of the global burned area representing nearly 65 % of global fire-related carbon emissions (van der Werf et al., 2017). Beyond carbon release, wildfires emit large amounts of dust and aerosols that influence regional climate through radiative processes. More generally, wildfires strongly modify land surface properties, including vegetation composition, soil carbon stocks, or surface albedo, with far-reaching consequences for regional carbon, water, and energy cycles.

In the ISBA land surface model (Delire et al., 2020), burned area is currently parameterized using grid-cell surface characteristics, a fire-resistance coefficient, soil moisture, and available biomass. While computationally efficient, this simplified formulation may contribute to persistent regional biases in simulated fire activity. To overcome these limitations, we develop a data-driven fire modeling framework based on two artificial neural network architectures: one addressing a regression task and the other a classification task. The models use meteorological conditions, vegetation states, and anthropogenic factors to estimate the daily burned area fraction.

The proposed framework reproduces the spatiotemporal variability of burned areas with some fidelity. It is specially the case in important areas such as Africa, South America, and Australia. These results highlight the potential of deep learning approaches to enhance wildfire representation and prediction in Earth system models. That would be the very future of our research project.

How to cite: Rougier, H., Decharme, B., and Mallet, M.: Modeling of burned areas on a global scale using statistical learning methods, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11185, https://doi.org/10.5194/egusphere-egu26-11185, 2026.

Application of deep learning has proved useful in many scientific domains and has also gained increased interest as a tool for weather and climate modeling in recent years. Deep Learning weather models have already demonstrated competitive prediction performance to state-of-the-art methods while hybrid models and emulators have shown some promise for climate simulation. However, the realism of simulated climate variability, and climate modes of pure deep learning models trained only on observational or reanalysis data, has not received as much attention.
As one example of these models, we investigate DLESyM, an autoregressive deep learning model based on the U-Net architecture and originally trained on ERA5 reanalysis data from 1981 to 2017 (REF1). Unlike many weather-generating deep learning models, DLESyM does not draw on sea-surface temperatures as boundary conditions, but learns to generate ocean surface patterns. Its applications could, therefore, extend to free-running simulations. The original authors showed its ability to generate stable climate simulations for time-spans up to three millenia, with the absence of spurious drifts and unphysical smoothing in the annual cycle. Here we test how realistic the simulated climate variability of DLESyM is, focusing on interannual to centennial spatio-temporal modes of internal climate variability. We seek to identify whether it is able to generalize to the underlying physical processes of the climate system, or if it is only capable of reproducing spatio-temporal statistical patterns of its training data. We compare the unforced variability of the deep learning model to that in equilibrium simulations out of General Circulation Models out of the Coupled Model Intercomparison Project phase 6 (CMIP6 GCMs), and palaeoclimate reconstructions (REF2). We focus on regional and global power spectra of surface temperatures, and gradients between land and ocean, tropics and extratropics, as well as the high latitudes. To assess the model’s ability to generalize outside the distribution of the training data we perform simulations from varying initial conditions, and comparing them with the output of CMIP6 GCMs. Based on this we discuss potentials and limitations of such a purely data-driven model for climate simulations and future climate risk assessment, where characteristics beyond mean state and slow changes become relevant.

REF1 Cresswell-Clay, N., Liu, B., Durran, D. R., Liu, Z., Espinosa, Z. I., Moreno, R. A., & Karlbauer, M. (2025). A deep learning Earth system model for efficient simulation of the observed climate. AGU Advances, 6, e2025AV001706. https://doi.org/10.1029/2025AV001706

REF2 Laepple, T., Ziegler, E., Weitzel, N. et al. (2023) Regional but not global temperature variability underestimated by climate models at supradecadal timescales. Nat. Geosci., 16, 958–966. https://doi.org/10.1038/s41561-023-01299-9

How to cite: Jansen, H., Racky, M., and Rehfeld, K.: Testing the realism of interannual to centennial climate variability in a generative coupled atmosphere-ocean deep learning model, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11509, https://doi.org/10.5194/egusphere-egu26-11509, 2026.

Numerical Earth system model (ESM) simulations require bias correction and downscaling to assess regional climate impacts due to their coarse resolution (50-100km) and systematic errors. Recent generative machine learning-based downscaling methods show promise in capturing small-scale spatial patterns, as well as multivariate and temporal dependencies [1,2,3]. However, making these approaches efficient and scalable to high resolutions globally remains challenging.

Here, we present a generative machine learning method for multivariate and temporally consistent downscaling of global climate fields at daily and 0.25° spatial resolution.  An autoregressive consistency model [4] is trained using Patch Diffusion [5] as an efficient probabilistic emulator of the ERA5 reanalysis and applied to downscale 8 key climate impact variables, including precipitation, temperature, wind speed, and radiation.
We downscale five 100-year simulations per ESM, including pre-industrial control,  historical, and 2K warming scenarios with and without tipping of the Atlantic meridional overturning circulation and the Amazon rainforest, from three CMIP6-class ESMs (MPI-ESM1-2-HR, HadGEM3-GC31-MM, and CESM1-CAM5).

The approach accurately reproduces small-scale variability and extremes, outperforms statistical baselines, substantially reduces biases, and preserves the large-scale response of the tipping dynamics in the ESMs.

   
[1] Mardani, M., Brenowitz, N., Cohen, Y., Pathak, J., Chen, C. Y., Liu, C. C., ... & Pritchard, M., Residual corrective diffusion modeling for km-scale atmospheric downscaling, Communications Earth & Environment, 6(1), 124, 2025. 
[2] Schmidt, J., Schmidt, L., Strnad, F. M., Ludwig, N., & Hennig, P., A generative framework for probabilistic, spatiotemporally coherent downscaling of climate simulation. npj Climate and Atmospheric Science, 8(1), 270, 2025.
[3] Hess, P., Aich, M., Pan, B., & Boers, N.,  Fast, scale-adaptive and uncertainty-aware downscaling of Earth system model fields with generative machine learning, Nature Machine Intelligence, 1-11, 2025.
[4] Wang, Z., Jiang, Y., Zheng, H., Wang, P., He, P., Wang, Z., ... & Zhou, M., Patch diffusion: Faster and more data-efficient training of diffusion models, Advances in neural information processing systems, 36, 72137-72154, 2023. 
[5] Song, Y., & Dhariwal, P., Improved techniques for training consistency models, In The Twelfth International Conference on Learning Representations, 2024.

How to cite: Hess, P., Bathiany, S., and Boers, N.: Generative Machine Learning for Dynamically Consistent Multivariate Downscaling of Tipping Point Simulations from Global Earth System Models, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11688, https://doi.org/10.5194/egusphere-egu26-11688, 2026.

Air pollution, particularly fine particulate matter (PM2.5), poses a significant risk to public health, necessitating accurate high-resolution monitoring. While global Chemical Transport Models (CTMs) like the Copernicus Atmosphere Monitoring Service (CAMS) provide continuous worldwide coverage, their coarse spatial resolution (~40 km) limits their utility for assessing local exposure relative to regional models (~10 km) that are restricted to specific domains, such as Europe. To bridge this gap, we present a novel deep learning approach for global downscaling of pollutant concentrations based on the Adaptive Fourier Neural Operator (AFNO), benchmarking its performance against a standard U-Net baseline.

We adapted the Modulated AFNO architecture for spatial super-resolution, using low-resolution CAMS Global PM2.5 and dynamic meteorological fields (wind, temperature, dew point, boundary layer height). A key innovation is integrating these inputs with high-resolution static data: orography and population density. We demonstrate that directly inputting static features into the network backbone outperforms separate spatial conditioning, effectively leveraging the Fast Fourier Transform to capture long-range dependencies while respecting local physical constraints.

The model was developed using daily forecasts from 2020 to mid-2025. Training used a sequential split into 2021–2024, preserving 2020 (COVID-19 anomalies) and 2025 as a held-out test set. The model effectively reconstructed fine-scale details and corrected global model biases. Verification against European Environment Agency observations (2020) confirmed performance comparable to high-resolution CAMS Europe regional forecasts. Crucially, the AFNO model consistently outperformed the U-Net baseline and traditional linear interpolation in spatial correlations and error rates. Finally, transferability tests in North America (AirNow data) confirmed the model generalizes effectively to unseen regions, maintaining lower errors than both the original global forecast and the baseline.

How to cite: Monsalvez-Pozo, K., Granell-Haro, F., Martinez-Roig, M., Galván Fraile, V., Plaza-Martín, N. P., Paul Ramacher, M. O., Bieser, J., Flemming, J., Razinger, M., Harder, P., Azorin-Molina, C., and Camps-Valls, G.: AFNO-based downscaling of global air pollution fields, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12008, https://doi.org/10.5194/egusphere-egu26-12008, 2026.

Providing high-resolution climate information by downscaling future climate projections from the Coupled Model Intercomparison Project (CMIP6) remains a central challenge for the regional climate modeling community. CMIP6 includes a wide range of global climate model (GCM) simulations across multiple Shared Socioeconomic Pathways (SSPs), resulting in substantial computational demand for dynamical downscaling if each member is to be fully regionalized. To address this challenge, we propose a computationally efficient statistical downscaling framework based on a U-Net architecture trained over Europe. The model learns high-resolution spatial mappings directly from reanalysis data, offering a low-cost complement to regional climate models (RCMs) for large-ensemble downscaling.

We demonstrate that the climate downscaling U-Net achieves performance comparable to the HCLIM RCM when applied to unbiased EC-Earth3-Veg simulations for both the historical period and the low-emission SSP1-2.6 scenario up to 2100. The model captures spatial temperature patterns, seasonal variability, and the amplitude of warming remarkably well in these cases, providing confidence in its ability to translate GCM-scale information into higher regional climate scales.

When the U-Net is trained exclusively on reanalysis data, its extrapolation behavior under stronger forcing scenarios becomes an important aspect to evaluate. In the high-emission SSP3-7.0 scenario, after the regional climate warms by approximately +2.0 °C beyond the conditions represented in the training data, typically during 2060-2080, the model begins to diverge modestly from the warming magnitude simulated by both the driving GCM and the HCLIM downscaling. This divergence is most pronounced during summer months, while winter temperature trends remain in close agreement. These deviations are not presented as shortcomings of the method, but rather as a clear illustration of the limits of extrapolation when statistical models are trained solely on historical climate states. Highlighting these limits is essential for understanding the robustness of statistical downscaling within and beyond the training domain, particularly for applications involving strong climate-change signals.

Finally, we investigate how the model’s capabilities evolve when future regional climate information is included in the training set. Incorporating a subset of future data markedly improves the extrapolation performance, enabling the U-Net to recover long-term warming trends and seasonal patterns consistent with HCLIM even under strong forcing. This demonstrates that the U-Net architecture can effectively learn and generalize high-resolution climate transformations when provided with an extended training domain. Overall, our findings underscore the potential of deep-learning-based downscaling for scalable, ensemble-wide applications while also clarifying the conditions under which historical-only statistical training remains reliable.

How to cite: Ivanov, M., Fuentes Franco, R., and Koenigk, T.: Can ML-based statistical downscaling models reliably extrapolate into the future?, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12050, https://doi.org/10.5194/egusphere-egu26-12050, 2026.

How to cite: Rodriguez Pinilla, M., Benitez Benavides, M., Exarchou, E., Margalef, T., and Panadero, J.: Deep Learning Downscaling of Precipitation and Temperature Climate Data for Future Wildfire Risk Assessment , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12404, https://doi.org/10.5194/egusphere-egu26-12404, 2026.

Climate models typically operate at coarse spatial resolution (~100 km) due to computational constraints, yet many climate-change impact assessments require fine-scale information (<10 km). In this study, we systematically benchmark three state-of-the-art machine-learning approaches for statistical downscaling, using the storm-resolving ICON NextGEMS dataset as reference. All methods take coarse-resolution climate fields as input and generate physically plausible high-resolution predictions. We compare: (1) UNet, a deterministic encoder–decoder architecture; (2) CorrDiff, which augments the UNet backbone with a diffusion model to produce probabilistic ensembles; and (3) CorrDiff++, which replaces diffusion with flow-matching to improve sampling efficiency. We perform 10× downscaling (0.56° to 0.056°) over central Europe for six surface variables, including temperature, wind, and precipitation. The models are evaluated along multiple dimensions: deterministic accuracy (bias, correlation), probabilistic skill (ensemble reliability and sharpness), and physical realism (energy spectra, temporal coherence, representation of extremes). Our results highlight fundamental trade-offs between computational cost, physical consistency, and uncertainty quantification. These insights provide guidance on when the additional complexity of generative models is justified for climate science applications.

How to cite: Debeire, K., Eyring, V., and Thuerey, N.: Benchmarking Deterministic and Generative Machine Learning Models for Statistical Climate Downscaling over Europe, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12407, https://doi.org/10.5194/egusphere-egu26-12407, 2026.

The timing of the first ice-free Arctic summer is a key indicator of climate change, yet projections remain highly uncertain due to inter-model spread, internal variability, and systematic model biases. We develop a prototype framework that combines machine-learning-based methods with causal diagnostics to assess how different bias-correction and emulation approaches influence projections of the first year of ice-free Arctic conditions. Linear scaling is used as a statistical baseline to provide a transparent reference for evaluating more complex machine-learning-based approaches.

Building on recent analyses of the drivers of summer Arctic sea-ice extent at the interannual time scale, we analyse CMIP6 multi-model large ensembles to quantify relationships between September Arctic sea-ice extent and its dominant drivers, including preceding winter sea-ice volume, Arctic near-surface air temperature, and ocean heat transport. Machine-learning-based regression and emulation models are applied to refine model output, while causal diagnostics based on information flow are used to evaluate the physical consistency of inferred driver–response relationships.

We focus on two CMIP6 large ensembles with contrasting historical Arctic temperature biases over 1980–2014. Ensemble uncertainty is explored by partitioning ensemble members into bias-based subsets to assess the sensitivity of projected ice-free timing and inferred driver relationships. Results show that linear scaling shifts projected timing without altering causal structure, whereas machine-learning-based methods can modify ice-free year distributions and induce state-dependent changes in inferred causal relationships. These findings highlight the value of causal diagnostics for interpreting machine-learning-based climate projections and underscore the need for physically interpretable frameworks when applying data-driven methods to critical Arctic climate transitions.

How to cite: Tian, T., Richards, B., and Docquier, D.: Toward more reliable projections of an ice-free Arctic: Integrating machine learning and causal diagnostics in CMIP6 ensembles, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12525, https://doi.org/10.5194/egusphere-egu26-12525, 2026.

Probabilistic risk assessment requires large ensembles of high-resolution climate scenarios, yet generating such data is often computationally intractable. This study introduces a scalable generative framework designed to overcome the scarcity of high-fidelity climate data. We introduce the Field-Space Autoencoder, a geometric compression model that preserves the causal structure of atmospheric fields without forcing them onto regular lat-lon grids. Unlike standard deep learning approaches fixed to a single resolution, our method utilizes a multi-scale decomposition that stores a resolution-invariant latent representation. This flexibility unlocks a novel hybrid training strategy for generative diffusion: we combine the statistical robustness of multi-century, low-resolution simulations with the structural precision of limited high-resolution datasets. The resulting Compressed Field Diffusion model is capable of synthesizing atmospheric states that inherit the internal variability of the large ensemble and the spectral sharpness of the high-res ground truth. By bridging these data sources, we present a pathway to democratizing access to exascale-quality climate data through efficient, physically consistent emulation.

How to cite: Meuer, J., Witte, M., Plésiat, É., and Kadow, C.: Generative Emulation on the Sphere: Bridging the Resolution Gap with Field-Space Diffusion, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12592, https://doi.org/10.5194/egusphere-egu26-12592, 2026.

Cloud feedback is one of the key sources of uncertainty in the sensitivity of climate projections to anthropogenic forcing in Earth system models (ESMs). Improving its representation remains challenging because clouds sit at the intersection of radiation, dynamics, and microphysics, and small errors in any of these can strongly affect climate sensitivity.. Consequently, analysing and understanding errors in simulated cloud feedback, evaluated against observations, is essential for advancing cloud parameterizations in ESMs.

In this work, we explore methodological frameworks for evaluating cloud feedback in climate models that move beyond simple model–observation comparisons toward physically interpretable insights into model properties and dynamics. We propose two advances: (1) improved cloud regime identification by extending standard k-means clustering to Wasserstein k-means, and (2) the use of explainable machine-learning methods to evaluate the extent the ESMs capture the realistic sensitivity between the cloud radiative anomalies and key cloud-controlling factors. We demonstrate these approaches by evaluating different versions of the HadGEM model in AMIP experiments against observations, illustrating their potential to support more physically grounded diagnosis of cloud-feedback behaviour in climate models.

How to cite: Mankovich, N., Gavrilov, A., Huang, F., Camps-Valls, G., Lan, F., and Bodas-Salcedo, A.: Explainable Cloud Feedback Evaluation in Earth System Models, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13326, https://doi.org/10.5194/egusphere-egu26-13326, 2026.

Reliable causal inference is a central challenge in Earth and climate sciences: observational records are limited, interventions are rare or impossible, and process representations in models rely on parametrizations that can introduce strong asymmetries between variables and the causal mechanisms [1,2]. Leveraging these asymmetries, rather than treating them as nuisances, can offer a principled route to causal discovery that is directly aligned with scientific modeling practice [2].

We address bivariate causal discovery from the standpoint of equation discovery using the Bayesian Machine Scientist (BMS) framework [3]. Our key contribution is to formalize the theoretical link between Symbolic Regression (SR) and Algorithmic Information Theory (AIT) via the Minimum Description Length (MDL) principle: the more plausible causal direction is the one that admits a shorter joint description in terms of a mechanism plus independent inputs [4]. Building on this connection, we characterize the mathematical properties of the resulting causal criterion, including identifiability and asymptotic consistency, and we analyze the role of core assumptions—most notably the Principle of Independent Causal Mechanisms (ICM)—in the context of geophysical data and climate-model parametrizations [5].

We demonstrate the approach on simulated benchmarks and on real Earth-system examples covering both i.i.d. settings and time-series climate data. The results illustrate when and why asymmetric parametrizations help disambiguate causal direction, and they provide a practical pathway to turn discovered governing equations into testable causal hypotheses for Earth and climate science.

References

[1] Jonas Peters, Dominik Janzing, and Bernhard Schölkopf. Elements of causal inference: foundations and learning algorithms. The MIT Press, 2017.

[2] Gustau Camps-Valls, Andreas Gerhardus, Urmi Ninad, Gherardo Varando, Georg Martius, Emili Balaguer-Ballester, Ricardo Vinuesa, Emiliano Diaz, Laure Zanna, and Jakob Runge. Discovering causal relations and equations from data. Physics Reports, 1044:1–68, 2023.

[3] Roger Guimera, Ignasi Reichardt, Antoni Aguilar-Mogas, Francesco A. Massucci, Manuel Miranda, Jordi Pallares y Marta Sales-Pardo. A Bayesian machine scientist to aid in the solution of challenging scientific problems. Science Advances, 6(5):eaav6971, 2020.

[4] Dominik Janzing, Joris Mooij, Kun Zhang, Jan Lemeire, Jakob Zscheischler, Povilas Daniūsis, Bastian Steudel und Bernhard Schölkopf. Information-geometric approach to inferring causal directions. Artificial Intelligence, 182:1–31, 2012.

[5] Sascha Xu, Sarah Mameche, and Jilles Vreeken. Information-theoretic causal discovery in topological order. In The 28th International Conference on Artificial Intelligence and Statistics, 2025.

How to cite: Camps-Valls, G., Guimerà, R., Varando, G., Diaz, E., Cohrs, K.-H., and Sales-Pardo, M.: Causal discovery from equation discovery, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13632, https://doi.org/10.5194/egusphere-egu26-13632, 2026.

Drought is a slowly developing natural hazard that can affect all climatic zones and is commonly defined as a temporary but significant decrease in water availability. In Europe alone, drought impacts over the last decades have generated very large economic losses, and recent summer events have been exceptional in a long-term historical perspective. Despite extensive research on drought monitoring and management, accurately characterizing how drought drivers evolve into impacts is still a key unresolved challenge, especially when impacts result from the cumulative and interacting effects of multiple hydroclimatic anomalies rather than a single precursor.

In this work, we introduce a machine learning procedure named DRIER (Drought Detection via Regression-based Interpretable Extraction and Causal Relationships) to develop interpretable, impact-based drought indices. Unlike traditional indices that primarily look at meteorological anomalies (e.g., precipitation deficits), DRIER is designed to capture the compound nature of drought impacts, such as prolonged dry periods occurring alongside high temperatures and reduced snowpack. DRIER is a fully data-driven and automated framework that integrates: (i) non-linear feature aggregation for dimensionality reduction to preserve an interpretable representation of candidate hydro-meteorological predictors, while reducing their dimension; (ii) conditional mutual information-based feature selection to identify the most informative drought drivers; (iii) multi-task linear regression to upscale learning across multiple sub-regions, leveraging shared drought processes while preserving local heterogeneity; (iv) causal validation using the Transfer Entropy Feature Selection algorithm to confirm that the relationships identified between hydroclimatic variables and drought impacts are not merely correlative but grounded in robust causal mechanisms.

We demonstrate DRIER in the Po River Basin (Italy) by considering 10 sub-basins and using vegetation stress quantified through the Vegetation Health Index (VHI) as an impact proxy. The application shows that DRIER can capture spatially heterogeneous drought–impact relationships across sub-regions while benefiting from multi-task learning to share information where responses are correlated. Importantly, because the framework is interpretable end-to-end, the resulting impact-based index is not a black-box score: each step produces transparent, auditable outputs that identify the key hydroclimatic drivers, how they are aggregated into the index, and how they contribute (in sign and magnitude) to vegetation stress. The integrated causal discovery component further strengthens confidence in real-world use by privileging predictors consistent with robust physical mechanisms, reducing the influence of spurious correlations and supporting transferability across space and time.

How to cite: Bonetti, P., Giuliani, M., Bucci, T., Cardigliano, V., Metelli, A. M., Restelli, M., and Castelletti, A.: Impact-based drought detection via Interpretable Machine Learning and Causal Discovery, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13676, https://doi.org/10.5194/egusphere-egu26-13676, 2026.

Rapid climate scenario exploration remains constrained by a fundamental tension: General Circulation Models and Earth System Models provide comprehensive representations of atmosphere-ocean-carbon interactions but impose computational demands prohibitive for iterative policy evaluation, while Energy Balance Models offer tractability at significant cost to predictive fidelity. Conventional machine learning approaches, though computationally efficient, exhibit excessive data dependence and lack the mechanistic transparency essential for regulatory compliance and evidence-based climate policy. This methodological gap motivates our development of a scientific machine learning framework that augments coupled climate-carbon dynamics through Universal Differential Equations (UDEs), achieving simultaneous forecasting accuracy and interpretability for rapid scenario assessment.

We formulate a three-state coupled dynamical system governing surface temperature anomaly, deep ocean temperature anomaly, and atmospheric CO₂ concentration, incorporating radiative forcing, ocean-atmosphere heat exchange, and temperature-dependent carbon uptake feedback mechanisms. Our investigation proceeds through systematic experimental evaluation. First, we assess Neural Ordinary Differential Equations (Neural ODEs) as black-box dynamical system learners across three random initializations under 1% observational noise. Neural ODEs exhibit substantial forecasting errors—12.45% for surface temperature, 64.08% for ocean temperature, and 5.17% for CO₂ concentration at t=50 years—with progressive error amplification throughout the forecast horizon, demonstrating fundamental limitations in capturing climate dynamics without physical constraints.

Subsequently, we construct a UDE architecture that preserves known energy balance and carbon cycle physics while replacing the temperature-dependent carbon uptake term (βTC) with a neural network component. This hybrid formulation achieves forecasting errors below 0.2% across all climate variables for three distinct initializations, representing order-of-magnitude improvement over Neural ODEs while requiring 57.5% fewer training iterations. Comprehensive robustness analysis across six noise levels (1–25%) demonstrates exceptional stability, with percentage errors remaining below 0.74% up to 20% observational noise, degrading catastrophically only at the 25% threshold.

To ensure mechanistic transparency—critical for climate policy applications—we employ Sparse Identification of Nonlinear Dynamics (SINDy) for symbolic regression on learned neural network outputs. SINDy successfully recovers the correct functional form β·T·C across all noise regimes up to 20%, achieving 100% functional form recovery rate with average relative error of 25.22% at 1% noise. Performance metrics degrade systematically with increasing noise: R² decreases from 0.9985 (1% noise) to 0.7812 (20% noise), with complete interpretability breakdown at 25% noise (R²=0.4028). This characterizes operational bounds for symbolic recovery under realistic measurement uncertainty.

Comparative benchmarking against statistical baselines—Vector Autoregression (VAR) and Autoregressive Integrated Moving Average (ARIMA)—confirms UDE superiority in data-scarce regimes with known physical constraints. While VAR and ARIMA exhibit computational parsimony (21 and 10 parameters respectively versus 8,577 for UDE), they incur prediction errors exceeding 19% for temperature variables, rendering them unsuitable for high-fidelity forecasting. The UDE framework uniquely achieves the accuracy-efficiency-interpretability tradeoff essential for climate scenario exploration, enabling policymakers to evaluate interventions through mechanistically transparent simulations satisfying quantitative risk assessment requirements.

Our results establish that physics-informed machine learning enables accurate climate trajectory prediction while symbolic regression maintains interpretability, yielding a computationally efficient framework for rapid exploration of emission scenarios, carbon taxation policies, and adaptation strategies with explicit uncertainty quantification.

How to cite: Pani, T., Dinesh Joshi, P., Abhijit Dandekar, R., Dandekar, R., and Panat, S.: CLIMASIM — Climate Simulation with Scientific Machine Learning, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13744, https://doi.org/10.5194/egusphere-egu26-13744, 2026.

As climate change amplifies precipitation extremes and their societal and economic impacts, downscaling precipitation provides valuable local-scale information for risk assessment and adaptation planning.
However, deep-learning based statistical downscaling methods typically rely on high-resolution training data (e.g., radar observations), which are scarce and unevenly distributed globally, making geographic generalization a central challenge. Prior work shows large performance drops of deep-learning based downscaling models under geographic distribution shifts -- effects that remain even when considerably increasing the training data [1].
We view the geographic distribution shift as a form of subpopulation shift, where training and target samples are drawn from the same set of geographic domains but differ in their sampling frequencies. Consequently, the shift is driven primarily by changes in the prevalence of climatic regimes, rather than by changes in the conditional relationship between predictors and targets.
To improve robustness under cross-region transfer, we inject additional geographic context through Earth embeddings from geospatial foundation models (e.g., SatCLIP [2]). Potential strategies for integrating these embeddings into diffusion-based downscaling models include attention-based conditioning, feature modulation, and auxiliary conditioning networks.

[1] Harder, P., Schmidt, L., Pelletier, F., Ludwig, N., Chantry, M., Lessig, C., Hernandez-Garcia, A. and
Rolnick, D. [2025], ‘Rainshift: A benchmark for precipitation downscaling across geographies’, arXiv
preprint arXiv:2507.04930 .

[2] Klemmer, K., Rolf, E., Robinson, C., Mackey, L. and Rußwurm, M. [2025], Satclip: Global, general-
purpose location embeddings with satellite imagery, in ‘Proceedings of the AAAI Conference on Artificial
Intelligence’, Vol. 39, pp. 4347–4355.

How to cite: Schmidt, L., Lemaire, P.-L., Ludwig, N., Hernandez-Garcia, A., and Rolnick, D.: Leveraging Earth Embeddings for Generalizable Precipitation Downscaling Across Geographies, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14198, https://doi.org/10.5194/egusphere-egu26-14198, 2026.

Many climate variables are naturally defined on the sphere and exhibit strong anisotropy and directionality (e.g., fronts, jets, boundary currents). Yet most deep-learning forecasting models still rely on planar projections and Euclidean convolutions, which introduce geometric distortions and artificial discontinuities. Graph-based spherical models alleviate some of these issues, but typically remain isotropic and do not explicitly represent local orientation, a key ingredient to model directional transport-like patterns.

Here we introduce and evaluate a gauge-equivariant spherical U-Net implemented directly on the HEALPix grid, designed to encode local orientation consistently across the sphere. Our approach leverages gauge-equivariant convolutions that transform predictably under changes of local reference frame, allowing the network to learn directional filters while preserving spherical geometry. This provides a principled alternative to both planar U-Nets (with longitude-periodic padding) and graph U-Nets, and addresses a core limitation of most spherical models: the lack of explicit orientation handling. This work benchmarks this model against two strong baselines: a planar U-Net with longitude-periodic padding and a spherical graph U-Net defined on the same HEALPix discretization.

We apply this architecture to multi-horizon forecasting of global sea-surface temperature (SST) anomalies at NSIDE=32, using a controlled experimental design with matched training protocols and comparable parameter budgets, with emphasis on low-capacity regimes relevant to data-limited climate settings (≈30–40 years of monthly observations). We report quantitative metrics across horizons and analyze qualitative error modes, showing how gauge-equivariant spherical convolutions mitigate projection artefacts while enabling orientation-aware feature extraction on the sphere. Our results highlight when and why encoding orientation through gauge equivariance provides added value beyond “spherical-but-isotropic” baselines, and offer practical guidance for deploying spherical equivariant models in climate forecasting pipelines.

How to cite: Delouis, J.-M., Odaka, T., and Tétaud, S.: Gauge-Equivariant Spherical U-Nets on HEALPix for Global SST Forecasting: Encoding local orientation on the sphere, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14352, https://doi.org/10.5194/egusphere-egu26-14352, 2026.

Tropical cyclones (TCs) are among the most destructive natural hazards worldwide. While several decades of satellite and reanalysis products now provide relatively large observational datasets of TCs, these datasets remain small by modern deep-learning standards and, crucially, are extremely imbalanced and do not sufficiently cover the tails of the distribution, with Category 5 cyclones being several orders of magnitude rarer than tropical storms. This severe data scarcity and imbalance poses fundamental limitations for supervised learning approaches to tasks such as intensity estimation, rapid intensification forecasting, or impact modeling, where performance on extremes is often the primary objective.

In this context, generative artificial intelligence offers a promising alternative. Diffusion models, in particular, have recently demonstrated state-of-the-art performance in modeling complex, high-dimensional data distributions. By learning the full probability distribution of TC-related fields rather than a single conditional mapping, diffusion models have the potential to generate physically plausible samples across the entire intensity spectrum, including rare but high-impact extremes. However, most existing applications of diffusion models—both within and outside the geosciences—are evaluated using perceptual or distributional metrics originally developed for natural images, such as visual inspection or feature-space distances. These metrics are poorly aligned with the physical constraints and scientific objectives that govern atmospheric phenomena, and may obscure important deficiencies in dynamical or thermodynamical realism.

Here, we present a diffusion-based generative framework for tropical cyclone spatial fields and propose a comprehensive evaluation strategy grounded in physically meaningful diagnostics. Rather than relying on perception-oriented scores, we assess generated samples using a suite of metrics designed to capture key aspects of TC structure and behavior, including radial symmetry, intensity–structure relationships, spatial gradients, and consistency with known climatological distributions across intensity classes. This allows us to directly interrogate whether the model reproduces physically coherent storm morphologies, particularly in the poorly sampled tails of the distribution. Beyond evaluation, we also explore multiple strategies for embedding physical realism directly into the model design. Together, these results highlight both the opportunities and the limitations of diffusion models as scientific tools for tropical cyclone research, and provide a framework for using generative AI not merely as a data-augmentation device, but as a principled instrument for studying rare and extreme atmospheric phenomena.

How to cite: Ascenso, G., Scoccimarro, E., and Castelletti, A.: Assessing physical realism in diffusion models for tropical cyclones, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15130, https://doi.org/10.5194/egusphere-egu26-15130, 2026.

Accurate forecasting of precipitation remains a central challenge in climate science, primarily due to the strong temporal and spatial variability of rainfall, a difficulty that is further intensified by the ongoing impacts of climate change. Recent developments in machine learning have facilitated the design of more accurate and robust predictive frameworks. In this context, the present study implements and evaluates three deep learning architectures; Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Extended Long Short-Term Memory (xLSTM); to forecast monthly precipitation at 27 meteorological stations distributed across Morocco, for lead times ranging from 1 to 4 months. The models are trained using a heterogeneous set of large-scale climatic predictors, including sea surface temperature (SST) over the Atlantic Ocean and the Mediterranean Sea, the East Atlantic pattern (EA), the Madden–Julian Oscillation (MJO), the El Niño–Southern Oscillation (ENSO), the Mediterranean Oscillation (MO), the North Atlantic Oscillation (NAO), and the Western Mediterranean Oscillation (WeMO). To identify the most influential predictors at each station, a principal component analysis (PCA)-based feature selection procedure is implemented. The results indicate that precipitation variability across the study area is predominantly controlled by the MO, NAO, and WeMO indices. Probabilistic forecasts are then generated using Monte Carlo dropout, enabling the networks to approximate Bayesian inference and thereby quantify predictive uncertainty and associated confidence intervals. Relative to conventional LSTM and GRU configurations, the xLSTM architecture exhibits superior predictive performance across all stations and lead times, with notably reduced uncertainty, particularly in the representation of extreme precipitation events. Overall, the models demonstrate robust skill in northern Morocco, with coefficients of determination (R²) ranging from 0.82 to 0.96 for a 1‑month lead time. However, predictive skill degrades toward the southern region, characterized by arid to semi-arid climatic conditions, where R² values decrease to 0.36–0.86. These results indicate that xLSTM effectively captures long-range temporal dependencies and low-frequency, high-intensity rainfall events, thereby representing a promising framework for improving probabilistic monthly precipitation forecasts in climatically heterogeneous regions such as Morocco.

How to cite: Zellou, B., Agdoud, F., and Ouatiki, H.: Probabilistic Monthly Precipitation Forecasting over Morocco Using xLSTM and Large-Scale Climate Predictors, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15554, https://doi.org/10.5194/egusphere-egu26-15554, 2026.

Machine learning–based weather prediction models have recently surpassed traditional numerical weather prediction systems on many skill metrics at regional and global scales, yet there is limited progress towards models operating on hectometric-scale resolutions. This setting is challenging both due to the cost of generating high-quality training data and the complex dynamics of important small-scale processes.

We introduce a graph neural network with Large-Eddy Simulation (LES) capabilities, to operate at hectometer horizontal resolution and sub-hourly time steps. Using 42 days of high-resolution realistic model output for the trade-wind regime over the western Atlantic, we train and evaluate the network on its ability to reproduce key mesoscale processes, with particular emphasis on cold-pool dynamics and convective triggering.

Cold pools are a crucial driver of low-level thermodynamic variability and cloudiness, and thus provide a stringent physical consistency test for models targeting hectometer scales, as they require accurate coupling between the cloud layer and the surface. Through a targeted ablation study, we quantify the relative importance of different input variables for reproducing surface temperature perturbations associated with cold pools, offering guidance for future parameterization and data selection strategies.

Finally, we show that the model can deterministically predict the evolution of cold pools over multiple successive generations, indicating that graph-based LES emulators can robustly capture the nonlinear feedbacks governing mesoscale organization in shallow convective regimes.

How to cite: Schulz, H., Oskarsson, J., and Denby, L.: ML-LES: Modeling cold-pool dynamics with graph-based neural network at hecto-meter grid-spacings, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16552, https://doi.org/10.5194/egusphere-egu26-16552, 2026.

The majority of supervised machine learning pipelines, particularly in the popular domains of natural language processing and computer vision, rely on manually annotated data. In geoscience applications, however, reference data are not necessarily derived from human annotation but could come as the output of explicit physical models or algorithms. These algorithms typically rely on simplifying hypotheses about the underlying physical processes and may be computationally expensive or applicable only to a limited subset of observations. In such circumstances, machine learning can be used to emulate explicit algorithms, with the objective of reproducing their outputs while potentially exploiting wider information pathways present in the data.

Beyond computational considerations, this hypothesis-light, data-driven framework allows for counterfactual testing by selectively removing input information and evaluating the model’s ability to recover similar predictions. For instance, in computer vision, color information can be removed by averaging RGB channels, while semantic or contextual information can be limited by progressively reducing the input patch size or by exploiting the inductive biases of different neural network architectures. In this way, one can identify additional cues in the input data linked to the physical property of interest, but also assess whether the model reproduces biases inherent to the reference algorithm. 

We explore this approach for high-resolution cloud-top height (CTH) estimation within the Clouds Decoded project, which uses Sentinel-2 (S2) multispectral observations (originally intended for land monitoring) to retrieve cloud properties. CTH can be estimated from S2 imagery using a stereo-based method that leverages the instrument’s geometry and inter-band delays. While effective, this approach is computationally demanding and relies on assumptions that restrict its applicability across diverse cloud scenes. We assess whether a neural network can learn to approximate this stereo-based CTH retrieval and analyse which textural, spectral, high-level semantic, or even geolocation-related cues the model might use to infer cloud height.

How to cite: Borne--Pons, P., Francis, A., Czerkawski, M., Campbell, J., and Bertozzi, B.: Machine learning emulation of stereo-based cloud-top height retrieval from Sentinel-2 , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18374, https://doi.org/10.5194/egusphere-egu26-18374, 2026.

Machine learning based dynamic numerical climate multi-model ensemble
weighting for high impact weather affecting the energy infrastructure
The infrastructure for renewable energy production and electrical grid itself is affected
by weather and climate conditions and vulnerable to high impact weather and
cascading events. A reliable representation of the meteorological conditions leading to
such events including their uncertainty is therefore needed for both weather and climate
time scales. Individual numerical weather and climate models exhibit systematic
strengths and weaknesses across scales, and geographic regions, despite the
differences in model physics and parametrizations. One way to tackle this and avoid
running single-model ensemble climate simulations are multi-model or poor-man
ensembles consisting of a set of different climate or weather models combined.
Ensemble mixing offers a way to mitigate these weaknesses while providing uncertainty
quantification. Simply ensemble averaging can dilute forecast and climate signals and
penalize outliers and rare extremes. Different approaches have been proposed to tackle
this problem by assigning non-uniform weights to individual model fields and
parameters, however, these methods often rely on domain knowledge such as model
dependencies [1,2].

Here, we propose a machine learning-based multi-model ensemble model-mixing
framework that is domain-agnostic and assigns spatially and temporally dynamic
weights, in addition to an error metric. The domain of interest is the Alps, which exhibit
challenging terrain and localized extreme events, e.g. precipitation extremes that are
difficult to capture in conventional climate models. The CERRA reanalysis data at ~5.5
km resolution serves as the target grid. We build a multi-model ensemble by combining
dynamically downscaled simulations of 2 m air temperature, precipitation, and wind
speed from COSMO-CLM (6 km) and WRF (10 km). Each regional model is driven by two
CMIP6 global climate models (MPI-ESM and EC-Earth) under two scenarios (SSP1-2.6
and SSP5-8.5), with an additional historical period used for training. Static information
such as orography and seasonal dependencies are considered. We evaluate the
ensemble’s performance on selected extreme events (e.g., heavy precipitation,
windstorms, heatwaves) that can (and did) harm energy infrastructure, such as the
European derecho 2022.

[1] Christensen, Jh, E Kjellström, F Giorgi, G Lenderink, and M Rummukainen. 2010.
‘Weight Assignment in Regional Climate Models’. Climate Research 44 (2–3): 179–94.
https://doi.org/10.3354/cr00916.

[2] Merrifield, Anna Louise, Lukas Brunner, Ruth Lorenz, Iselin Medhaug, and Reto
Knutti. 2020. ‘An Investigation of Weighting Schemes Suitable for Incorporating Large
Ensembles into Multi-Model Ensembles’. Earth System Dynamics 11 (3): 807–34.
https://doi.org/10.5194/esd-11-807-2020.

How to cite: Thiele, P., Baier, K., Hasel, K., Schellander-Gorgas, T., Lehner, S., Spiekermann, R., Lampert, J., Lexner, A., and Schicker, I.: Machine learning based dynamic numerical climate multi-model ensemble weighting for high impact weather affecting the energy infrastructure, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18613, https://doi.org/10.5194/egusphere-egu26-18613, 2026.

 Ocean physical and biogeochemical extremes, such as marine heatwaves (MHWs), deoxygenation, and acidification events have significant impacts on the marine environment, ecosystems, and economic livelihoods. In recent decades, the frequency, intensity and spatial extent of these extremes have been amplified (Capotondi et al., 2024; Shu et al., 2025; Gruber et al., 2021). Hence, a deeper understanding of the processes and precursors leading to extreme events remains crucial for improving and forecasting risk assessment.

In this study, we apply interpretable machine learning approaches to investigate which oceanic and atmospheric variables, as well as their lag effects, are most relevant for the extreme events in the North Atlantic, a relevant region for their occurrence in recent decades (England et al., 2025). Our framework combines high-resolution ocean model simulations with explainable artificial intelligence (XAI) techniques (He et al., 2024, Camps-Valls, 2025), allowing us to examine where, when, and which model variables are more important when identifying extreme events.

Rather than focusing on predictive skill, the emphasis of this study lies on identifying the underlying physics of precursor patterns leading to ocean extremes across different spatial and temporal scales. By integrating XAI into the analysis, this approach provides a more transparent and interpretable perspective on the decision-making processes of machine learning models, offering insights into the key variables and structures associated with the occurrence of ocean extremes. The outcomes of this study improve the interpretable assessment of potential precursors of MHWs, ocean deoxygenation and acidification extremes.

Camps-Valls, G., Fernández-Torres, M. Á., Cohrs, K. H., et al. (2025). Artificial intelligence for modeling and understanding extreme weather and climate events. Nature Communications, 16, 1919. https://doi.org/10.1038/s41467-025-56573-8

Capotondi, A., Rodrigues, R. R., Sen Gupta, A., et al. (2024). A global overview of marine heatwaves in a changing climate. Communications Earth & Environment, 5, 701. https://doi.org/10.1038/s43247-024-01806-9

England, M. H., Li, Z., Huguenin, M. F., et al. (2025). Drivers of the extreme North Atlantic marine heatwave during 2023. Nature, 642, 636–643. https://doi.org/10.1038/s41586-025-08903-5

Gruber, N., Boyd, P. W., Frölicher, T. L., et al. (2021). Biogeochemical extremes and compound events in the ocean. Nature, 600, 395–407. https://doi.org/10.1038/s41586-021-03981-7

He, Q., Zhu, Z., Zhao, D., Song, W., & Huang, D. (2024). An interpretable deep learning approach for detecting marine heatwave patterns. Applied Sciences, 14(2), 601. https://doi.org/10.3390/app14020601

Shu, R., Wu, H., Gao, Y., et al. (2025). Advanced forecasts of global extreme marine heatwaves through a physics-guided data-driven approach. Environmental Research Letters, 20(4). https://doi.org/10.1088/1748-9326/adbddd

How to cite: Radin, C., Mathis, M., Li, H., and Ilyina, T.: Explainable AI for Identifying Precursors of Extreme Oceanic Events in the North Atlantic, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19617, https://doi.org/10.5194/egusphere-egu26-19617, 2026.

High-resolution gridded climate datasets are essential for Earth system modelling and impact assessments, yet generating them from sparse, irregularly distributed station networks remains a significant challenge, particularly in regions with complex topography. This study evaluates the Spatial Multi-Attention Conditional Neural Process (SMACNP), a probabilistic deep learning framework, for the daily spatial interpolation of air temperature and precipitation, marking the first application of its localized encoder variant to the challenge of gridding climate data from a sparse station network. We investigate two distinct encoder configurations—Global and Localized—to determine the optimal structural prior for capturing spatial dependencies in data-scarce regimes. The models were developed and evaluated using data from a sparse network of meteorological stations in Romania from 2020 to 2023. To ensure applicability for long-term historical reconstruction, the input features were restricted to static topographic predictors derived from a Digital Elevation Model (DEM). Performance was benchmarked against Regression Kriging (RK), a standard geostatistical baseline that incorporates these same topographic covariates. Results demonstrate that the SMACNP architectures substantially outperform the RK baseline for both variables.

The SMACNP (Localized) configuration, which utilizes an attention mechanism, emerged as the most robust model, achieving the lowest Mean Absolute Error (MAE) and the highest correlation across the majority of seasons. The performance gains were particularly pronounced for precipitation, where the deep learning models effectively captured fine-scale spatial heterogeneity and non-linearities that traditional methods tended to over-smooth. These findings indicate that localized neural process-based models offer a powerful, scalable, and physically plausible alternative to geostatistical methods for generating high-quality gridded climate datasets in complex, data-sparse environments.

This research was supported by the project “Cross-sectoral Framework for Socio-Economic Resilience to Climate Change and Extreme Events in Europe (CROSSEU)” funded by the European Union Horizon Europe Programme, under Grant agreement n° 101081377.

How to cite: Dumitrescu, A.: A deep learning framework for gridding daily climate variables from a sparse station network, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19697, https://doi.org/10.5194/egusphere-egu26-19697, 2026.

How to cite: Miranda, V., Castro, M., Paixão, J., Girão, I., Marques, B., Magnus Koktvedgaard Zeitzen, R., Cunha, R., Fonteneles, C., Pereira, É., Khudynian, M., Thejll, P., Jomo Danielsen Sørup, H., Paletta, Q., and Oliveira, A. P.: CLIM4cities - from Citizen Science, Machine Learning and Earth Observation towards Urban Climate Services , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19718, https://doi.org/10.5194/egusphere-egu26-19718, 2026.

Global Climate Models (GCMs) provide critical insights into future climate variability, yet their coarse spatial resolution limits their utility for regional and local-scale impact assessments. AI-driven downscaling techniques have emerged in the last few years as a cost-effective and viable alternative to traditional methods to enhance the spatial resolution of climate projections. Nevertheless, establishing their reliability in unseen climate states remains a priority. This study applies and evaluates a deep generative Latent Diffusion Model, leveraging a residual approach (LDM_res, Tomasi et al., 2025) to downscale GCM outputs (~1°) to high-resolution (~4 km) 6-hourly precipitation and 2-m minimum and maximum temperature fields.

The LDM is developed as an emulator of the COSMO-CLM dynamical model, trained on VHR-REA_IT data (Raffa et al., 2021 - a dynamical downscaling of ERA5). By using aggregated ERA5 data as low-resolution predictors (along with high-resolution static data), the LDM_res model is required to learn to mimic the computationally expensive physics of dynamical downscaling. The model, trained over the past 40 years, is subsequently applied to generate high-resolution climate projections based on the input from four selected CMIP6 GCMs across four different emission scenarios. This modeling chain establishes a hybrid ML-Physics-based system to provide impact assessors with cost-effective, high-resolution climate information.

A central challenge addressed in this work is the evaluation of the model's out-of-distribution generalization—specifically its ability to perform in unseen future climate states and under predictor configurations characteristic of CMIP6 projections. We evaluate the emulator's reliability by comparing its outputs against VHR-PRO_IT, a "twin" dataset of VHR-REA_IT produced using COSMO_CLM to dynamically downscale projections (Raffa et al., 2023), providing a rigorous test of the ML system’s reliability in out-of-domain scenarios.

Furthermore, we compare the LDM_res against traditional statistical (e.g., quantile mapping) and dynamical approaches. Comparative results over the Italian peninsula indicate that while the LDM preserves large-scale seasonal signals from CMIP6 models, it significantly enhances spatial realism and local variability in topographically complex areas. Unlike purely statistical methods, the hybrid ML approach demonstrates superior ability to represent fine-scale heterogeneity in mountainous and coastal regions while maintaining consistency with the original signal.

How to cite: Tomasi, E., Franch, G., Tomezzoli, G., Calmanti, S., and Cristoforetti, M.: Deep learning for high-resolution climate projections: a Latent Diffusion Model emulating dynamical downscaling over Italy, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19912, https://doi.org/10.5194/egusphere-egu26-19912, 2026.

The accurate assessment of extreme flood events and their associated losses requires massive sample sizes (e.g., 50,000+ years of weather data) for statistical robustness and a comprehensive coverage of event characteristics. Generating such a large dataset using dynamical Earth System Models would be extremely computationally intensive, so instead, we propose a lightweight and computationally efficient climate emulator built upon a video diffusion architecture. 

The model is trained to reproduce the statistical properties and physical dynamics of the Community Earth System Model version 2 (CESM2) over Europe. It operates autoregressively to generate synthetic, multivariate, daily atmospheric data (including temperature, specific humidity, wind vectors, and surface pressure) at ~100 km resolution. The model utilizes a U-Net architecture that is conditioned on previous time-steps to produce and evolve weather patterns with spatial and temporal consistency. To enhance the stability of long-term generation and improve the faithful reproduction of extremes, we employ a seasonality-aware standardization scheme, training the model to learn the dynamics in anomaly space rather than physical space.

We demonstrate that this approach successfully reproduces the complex spatiotemporal dependencies within CESM2, captures atmospheric dynamics, including the frequency and persistence of dominant circulation types, and can maintain stability over multi-decadal generation windows. Furthermore, the output of this emulator can be fed into existing downscaling models to produce higher resolution multivariate meteorological data fields to drive downstream impact models. We validate this full modeling chain by demonstrating that the resulting hydrological statistics exhibit physical characteristics consistent with the CESM2-driven benchmark.

This computationally efficient generative model offers a pathway to generating thousands of years of physically consistent flood events. 

How to cite: Marshall, A., Lucas, C., Addor, N., Lord, N., Moraga, J. S., Hoch, J., and Wing, O.: Fast emulation of climate models for precipitation and flood impact modelling using autoregressive video diffusion, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19946, https://doi.org/10.5194/egusphere-egu26-19946, 2026.

Remote sensing datasets for land cover classification are mostly distributed in UTM projection which introduce significant geometric distortions—particularly at high latitudes—and fail to respect the spherical geometry of Earth. These distortions propagate into deep learning models trained on such data, leading to latitude-dependent biases, edge artifacts in tile-based processing, and poor generalization across geographic boundaries. While convolutional neural networks (CNNs) have achieved state-of-the-art performance on benchmark datasets like BigEarthNet, they operate on Euclidean grids and cannot naturally handle the structure of a sphere.

Here we introduce a comprehensive pipeline for transforming the BigEarthNet dataset—comprising 549,488 multispectral image patches from its original UTM projection into the HEALPix (Hierarchical Equal Area isoLatitude Pixelization) representation. HEALPix, originally developed for cosmic microwave background analysis, offers equal-area partitioning of the sphere, ensuring uniform statistical treatment of pixels regardless of latitude, and provides a natural hierarchical structure for multi-resolution analysis.

We implement and evaluate spherical CNNs architectures designed for data on spherical manifolds—against traditional planar CNN baselines (Unet/Resnet50) trained on the HEALPix-transformed data, benchmarking classification performance for multi-label land cover prediction using the 19-class BigEarthNet nomenclature with metrics suited to imbalanced settings (F1-macro/micro, precision, recall, average precision).

This work represents the first large-scale application of HEALPix projection to Remote Sensing classification and validates the effectiveness of spherical deep learning for real-world remote sensing beyond traditional climate science domains. Our experimental design employs matched training protocols and comparable model capacities, demonstrating that spherical representations eliminate projection-induced artifacts, enable seamless cross-boundary analysis, and provide rotation equivariance that reduces the need for extensive spatial data augmentation—key advantages for global-scale Earth observation applications.

How to cite: Tétaud, S. and Delouis, J. M.: BigEarthNet-HEALPix: Spherical CNNs for Land Cover Classificatiom, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20756, https://doi.org/10.5194/egusphere-egu26-20756, 2026.

Surface methane emissions can be estimated from atmospheric observations using inverse modelling systems, which often rely on Lagrangian Particle Dispersion Models (LPDMs) to simulate how the gas is transported through the atmosphere using meteorological fields. However, LPDM-based techniques struggle to scale to the size of modern satellite datasets, as one LPDM run is needed for each observation, taking on the order of 10 CPU-minutes to complete. Previously, we introduced the Machine Learning model GATES (Graph-Neural-Network Atmospheric Transport Emulation System), which can replicate LPDM outputs 1000x faster than the physics-based model, and demonstrated its application to infer emissions over South America. Training GATES over other world regions and comparing cross-regional performance shows that the learnt transport is domain-specific, consistent with the strong heterogeneity in wind patterns and topography across continents. In this presentation, we discuss transfer learning techniques and characterisation of regional differences in wind patterns, topography, data availability and the shape and magnitude of LPDM outputs, to increase transfer learning performance. This work builds capabilities towards efficient, global methane emissions emulation. 

How to cite: Clark, J., Fillola, E., Keshtmand, N., Santos-Rodriguez, R., and Rigby, M.: From regional to global emulation: characterising regional differences to increase transfer learning performance  , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20966, https://doi.org/10.5194/egusphere-egu26-20966, 2026.

Deep convection exhibits substantial variability even under fixed large-scale forcing, challenging deterministic descriptions of convective organization. Using idealized radiative–convective equilibrium simulations with imposed low-level shear, we quantify this intrinsic variability through a reduced-order stochastic framework. Convective transport is characterized by isentropic mass flux and embedded in a low-dimensional latent space using a variational autoencoder. The temporal evolution of convection in this space is modeled as a Markov chain, yielding a data-driven representation of convective states and their transition probabilities.

This framework demonstrates that internal feedbacks alone generate a broad ensemble of admissible convective trajectories within a single environment, which we interpret as the system’s intrinsic stochasticity. The leading latent dimensions correspond to the convective life cycle and degree of organization, while state transitions identify the constrained pathways through which organized convection emerges and evolves. Comparison of individual storm trajectories in latent space exposes systematic differences in dynamical behavior that are difficult to diagnose in physical space. However, departures from strictly Markovian behavior indicate that the instantaneous state representation does not fully capture slow memory effects associated with convective organization, which likely condition transition probabilities.

These results show that organized convection is best understood as one realization drawn from a constrained distribution of possible trajectories and establish a general machine-learning-enabled framework for quantifying variability and limits of predictability in multiscale atmospheric systems.

How to cite: Abramian, S., Olivier, P., and Pierre, G.: How Organized Convection Evolves in Latent Space, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21859, https://doi.org/10.5194/egusphere-egu26-21859, 2026.

Sub-grid parameterisations in climate models are traditionally static and tuned offline, limiting adaptability to evolving states. This work introduces FedRAIN-Lite, a federated reinforcement learning (FedRL) framework that mirrors the spatial decomposition used in general circulation models (GCMs) by assigning agents to latitude bands, enabling local parameter learning with periodic global aggregation. Using a hierarchy of simplified energy-balance climate models, from a single-agent baseline (ebm-v1) to multi-agent ensemble (ebm-v2) and GCM-like (ebm-v3) setups, we benchmark three RL algorithms under different FedRL configurations. Results show that Deep Deterministic Policy Gradient (DDPG) consistently outperforms both static and single-agent baselines, with faster convergence and lower area-weighted RMSE in tropical and mid-latitude zones across both ebm-v2 and ebm-v3 setups. DDPG's ability to transfer across hyperparameters and low computational cost make it well-suited for geographically adaptive parameter learning. This capability offers a scalable pathway towards high-complexity GCMs and provides a prototype for physically aligned, online-learning climate models that can evolve with a changing climate. Code accessible at https://github.com/p3jitnath/climate-rl-fedRL.

How to cite: Nath, P., Schemm, S., Moss, H., Haynes, P., Shuckburgh, E., and Webb, M.: FedRAIN-Lite: Federated Reinforcement Algorithms for Improving Idealised Numerical Weather and Climate Models, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-351, https://doi.org/10.5194/egusphere-egu26-351, 2026.

This study presents a data-driven parameterization of deep convection, implemented and tested within the global climate model ARP-GEM at 50 km resolution. Initially, a 'naive' neural network was used to replace ARP-GEM's traditional physical parameterization. A 30-year simulation with this data-driven approach revealed significant biases, particularly in the representation of high clouds.
To adress these biases, we developed a two-fold neural network architecture: one component responsible for detecting the triggering of the convection and another responsible for computing convective tendency terms. This refined parameterization substantially improved performance compared to the initial version. Furthermore, the enhanced parameterization was evaluated under warmer climate conditions, demonstrating online stability and consistent overall fidelity.

How to cite: Balogh, B., Germain, H., Geoffroy, O., and Saint-Martin, D.: Online test of a data-driven parameterization of deep-convection: evaluation in present and future climate, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1118, https://doi.org/10.5194/egusphere-egu26-1118, 2026.

Earth System Models of Intermediate Complexity (EMICs) are essential tools for investigating climate dynamics on millennial to orbital time scales, which are computationally prohibitive for high-resolution CMIP-class models. The computational efficiency of EMICs is primarily achieved by reduced spatial resolution of the atmosphere and ocean components. However, EMICs often couple ice-sheet and terrestrial vegetation components, which require much higher spatial resolution. The coupling of these components therefore remains a major challenge and often results in inadequate climatic forcing for these sub-modules, particularly regarding precipitation patterns. Generative machine learning, specifically diffusion models and their variants, has emerged as a powerful technique to bridge this resolution gap. Here, we present the integration of a consistency model-based approach to facilitate efficient, online downscaling of temperature and precipitation within the Bern3D EMIC with negligible computational overhead.

To achieve this, the consistency model was trained on monthly ERA5 ensemble output to learn the mapping from the coarse Bern3D grid to high-resolution fields.  This approach successfully reconstructs high-resolution spatial variability while maintaining inference speeds compatible with long model integration times, effectively avoiding additional runtime costs. This framework therefore allows for the representation of small-scale heterogeneity in surface boundary conditions which is critical for realistic ice sheet and vegetation dynamics. 

Ultimately, this approach opens new avenues to investigate complex climate-ice-vegetation feedback on orbital time scales, such as during the Last Glacial Cycle or the Mid-Pleistocene Transition.

How to cite: Wirths, C., Hofmann Elizondo, U., Hess, P., and Pöppelmeier, F.: Towards Generative Machine Learning-based Downscaling for Atmosphere-Surface Coupling in the Bern3D EMIC, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-2982, https://doi.org/10.5194/egusphere-egu26-2982, 2026.

In weather and climate models, momentum, heat, humidity and tracer fluxes between the Earth’s surface and atmosphere strongly depend on surface roughness. The roughness length depends on space and time-dependent surface properties over ocean, sea-ice and land. For example, surface winds impact wave height over sea-ice free oceans; vegetation and orography determine roughness length over land, where its effect on near-surface turbulence strongly impacts the surface fluxes. Here, we present a set of machine learning models trained on reanalysis data to predict surface roughness over both land and ocean grid cells in SpeedyWeather, a Julia-based climate model. More accurately representing the surface roughness has been shown to significantly improve model bias against observations over a range of variables such as surface air temperatures and near-surface wind speed. We explore the downstream impacts of using this parameterisation in the climate model, and test the generalisability of an offline-learned surface roughness scheme in future climates with reduced sea ice and land-use change. Spatial generalisation is achieved through surface roughness being a function of local variables only. We discuss efficient inference on CPU and GPU for every grid cell on each integration time-step. So-called model distillation via symbolic regression minimises the trade-off between speed versus accuracy, enabling another route to rapid inference on a grid-cell basis. Further, we investigate online learning through differentiable physics parameterisations to calibrate the learned parameterisation to surface variables from ERA5 reanalysis. We generally propose machine-learned schemes of individual climate processes towards interpretable, data-driven climate modelling. 

How to cite: Munday, G., Klöwer, M., Mansfield, L., and Gelbrecht, M.: A learned surface roughness scheme for climate prediction, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4851, https://doi.org/10.5194/egusphere-egu26-4851, 2026.

Hybrid Earth system models (ESMs) that combine physical laws with machine learning (ML) demonstrate great potential to reduce uncertainties in climate projections, particularly for subgrid processes like clouds. However, widespread adoption faces critical challenges: deep learning "black boxes" often lack interpretability and physical consistency, and coupling them with standard ESMs remains difficult due to stability issues and the need for complex re-calibration. Here, a two-step method is presented to improve a climate model with data-driven parameterizations. First, we incorporate a physically consistent cloud cover parameterization, derived from storm-resolving simulations via symbolic regression, into the ICON atmospheric climate model. We refer to this hybrid configuration, which retains the interpretability and efficiency of the traditional model, as ICON-A-MLe. Second, we address the coupling and tuning bottleneck by introducing an automated, gradient-free calibration procedure based on the Nelder-Mead algorithm. This method efficiently calibrates ICON-A-MLe without requiring differentiable physical components, making it easily extendable to other ESMs. Our results show that the tuned ICON-A-MLe substantially reduces long-standing biases. Specifically, it reduces cloud cover errors over the Southern Ocean by 75% and in subtropical stratocumulus regions by 44%. These improvements also lead to a better top-of-atmosphere radiative budget. Crucially, the model demonstrates strong generalization capabilities: it remains robust and physically consistent under significantly warmer climate scenarios. These results demonstrate that interpretable machine-learned parameterizations, paired with practical tuning, can efficiently and transparently strengthen ESM fidelity.

How to cite: Grundner, A., Beucler, T., Savre, J., Lauer, A., Schlund, M., and Eyring, V.: Reduced cloud cover errors in a hybrid AI-climate model through equation discovery and automatic tuning, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5525, https://doi.org/10.5194/egusphere-egu26-5525, 2026.

Accurately simulating the terrestrial carbon cycle remains a major challenge in climate science, due in part to uncertainties in how slow-varying land-surface boundaries and fast-varying biophysical states are represented and coupled in Earth-system models.  We introduce a unified data-driven framework designed to generate high-resolution (1 km) historical reconstructions and future projections of Land Use (LU), Land Cover (LC), and Leaf Area Index (LAI) for real-time coupling with digital twin platforms, such as those deployed in the Destination Earth framework.

Moving beyond sequential downscaling, this framework treats the generation of boundary conditions as a cohesive multi-task learning problem. We benchmark two distinct modeling strategies: (1) Architectures trained from scratch, where we compare the performance of convolutional baselines (U-Net) against attention-based Vision Transformers (ViT) in capturing spatial heterogeneity; and (2) Foundation Model (FM) Adaptation, where we leverage state-of-the-art Earth FMs (such as  TerraMind and Prithvi) as backbones. Within this second strategy, we evaluate the trade-offs between full fine-tuning, parameter-efficient techniques using adapters, and models trained from scratch.

By integrating static geophysical features with high-frequency climate reanalysis (ERA5) and atmospheric CO2​ concentrations, the framework ensures that vegetation dynamics remain phenologically consistent with environmental forcing. We assess these approaches based on their computational efficiency, generalization across sparse data regimes, and physical consistency between categorical (LU/LC) and continuous (LAI) variables. The final output is a suite of open-source interoperable emulators designed to act as dynamic, on-demand boundary condition generators. 

How to cite: Duarte, A., Mozaffari, A., Castaño, M., Materia, S., and Castrillo Melguizo, M.: A Unified Data-Driven Framework for High-Resolution Land Surface Boundary Conditions, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9688, https://doi.org/10.5194/egusphere-egu26-9688, 2026.

Global land surface and hydrological models are crucial components of Earth System Models (ESMs). In addition to providing realistic boundary conditions for the atmosphere and ocean components, they also play a key role in understanding Earth’s changing energy imbalance and the response of the terrestrial carbon and water cycles to anthropogenic climate change. The land surface components of most ESMs typically rely on reduced-complexity parameterizations of land processes in order to efficiently resolve the transient coupling of the land surface to the atmosphere at global scales. The complexity of such models is therefore limited by the coarse spatial resolution of the atmosphere and thus they are not easily constrained by in situ and remote sensing observations of land surface parameters. As a result, offline downscaled and bias-corrected climate models and reanalysis products are often used as forcings when calibrating land surface and hydrological models at local and regional scales. We argue that this lack of online coupling in the downscaling step is one of many factors contributing to persistent biases in modern ESMs. As such, there is a need for a new generation of land models which can support more flexible coupling with the atmosphere as well as the incorporation of data-driven components. Here we present Terrarium.jl, a Julia-based land modeling framework for GPU-accelerated and automatically differentiable simulations of soil, snow, and vegetation dynamics, along with their corresponding land-atmosphere exchange fluxes. We demonstrate the value of GPU acceleration and differentiability through a series of performance benchmarks and sensitivity analyses. We further present our initial experiments in achieving stable coupling to a reduced-complexity atmosphere model, SpeedyWeather.jl, as well as a proof-of-concept for online downscaling from the scale of an intermediate-complexity ESM (~5°) to that of ERA5 (~0.25°). We discuss the main challenges encountered thus far and outline a roadmap for future development.

How to cite: Groenke, B., Badri, M., Lin, Y., Gelbrecht, M., and Boers, N.: Terrarium.jl: A framework for fully differentiable and GPU-accelerated land modeling to enable online downscaling in coarse-scale ESMs, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11235, https://doi.org/10.5194/egusphere-egu26-11235, 2026.

Subgrid-scale (SGS) parameterizations remain a leading source of uncertainty in weather and climate models, where they represent the effects of unresolved processes occurring at scales smaller than the model’s grid resolution on the resolved fields. Similar closure problems arise in computational fluid dynamics (CFD), where turbulence models are needed to represent the impact of unresolved scales on the resolved flow. Despite recent progress toward differentiable, hybrid climate models enabled by automatic differentiation, most operational Earth system models (ESMs) remain effectively non-differentiable, limiting systematic online calibration and training.

While data-driven closures trained offline can perform well a priori, their performance often deteriorates a posteriori, once coupled to the solver because the coupled setting introduces feedbacks that are absent during offline training. In this study, we treat a controlled CFD turbulence setting as a benchmark for climate-relevant SGS learning and compare two classes of online training strategies for data-driven closures in non-differentiable models: (i) gradient-free ensemble Kalman inversion (EKI), leveraging the robustness and parallelism of ensemble-based inverse methods, and (ii) gradient-based optimization enabled by a learned differentiable emulator. For the emulator, we train a fast neural ODE surrogate of the forward model dynamics that preserves its structure and is differentiable by construction, enabling gradient-based training without modifying the original solver. We then evaluate both approaches using metrics such as accuracy, computational cost, and scalability.

How to cite: Badri, M., Groenke, B., Gelbrecht, M., and Boers, N.: Comparison of Online Training Methods for Data-Driven Subgrid-Scale Parameterizations in Non-Differentiable Models, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11475, https://doi.org/10.5194/egusphere-egu26-11475, 2026.

The Pelagic Interaction Scheme for Carbon and Ecosystem Studies (PISCES) is a marine biogeochemical model that is used in several IPCC-Class Earth System models. PISCES simulates the distribution of nutrients (four macronutrients and one micronutrient) that regulate the growth of two phytoplankton classes (nanophytoplankton and diatoms). It also simulates the ocean carbon cycle with a complete representation of the marine carbonate systems. PISCES includes 24 state variables, and increases the runtime of NEMO, the physical ocean model with which it is coupled, by a factor of 3.4, indicating a high computational cost.

PISCES-AI has been developed as a U-Net based machine learning PISCES emulator, which takes a small number of input variables (TOS, ZOS, SOS, PAR, atmospheric CO₂), and predicts two output variables: surface chlorophyll and the difference in partial pressure of CO₂ between the atmosphere and the ocean. These are the only outputs which have a direct influence on climate simulations by Earth system models. Previous work has shown the predictive power of PISCES-AI across multiple timescales, and in an out-of-domain setting.

In this work, we couple the AI emulator of PISCES to NEMO, using Eophis and Morays for Python-Fortran interaction. We evaluate its performance, as well as its computational efficiency, to give a holistic picture of the challenges and opportunites for AI emulation of ocean biogeochemistry. With a particular interest in the computational speed, we find that inference for a single time-step to be around 10ms, with a much larger preliminary bottleneck due to CPU-GPU transfer (200ms per timestep). Even with this bottleneck, with our implementation we obtain a speed-up of factor 3 compared to PISCES, and we explore ways in which the data transfer bottleneck could be reduced.

How to cite: Gow-Smith, E. and Séférian, R.: Coupling of NEMO to a neural network emulator of PISCES, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12174, https://doi.org/10.5194/egusphere-egu26-12174, 2026.

Reliable high resolution precipitation fields are essential for hydrology flood risk management agriculture and climate impact assessment yet remain difficult to reconstruct from sparse and irregular rain gauge networks. Reanalysis products such as ERA5 provide physically consistent estimates but are constrained by coarse effective resolution temporal smoothing and weak local observational constraints. By formulating interpolation of spatiotemporal precipitation fields as a probabilistic context to target regression problem using neural process (NP) models, this study assesses whether NP-based approaches can outperform reanalysis and classical interpolation for local to regional rainfall reconstruction. Using high quality UK rain gauge observations combined with gridded auxiliary variables from ERA5 we implement convolutional NPs within the DeepSensor framework and compare them with a transformer based NP variant.

Models are jointly conditioned on dense meteorological fields and sparse precipitation observations and output full predictive distributions using a Bernoulli–Gamma likelihood to capture intermittency and extremes. Training is performed using random sensor masking to enforce location agnostic learning and enable zero shot prediction at unseen coordinates. Performance is evaluated against ERA5 and Kriging using identical data splits with emphasis on interpolation accuracy as well as calibration robustness to sensor sparsity. Generalisation is further assessed through few shot and zero shot transfer across regions with contrasting regimes including England, Scotland and selected GHCN domains in the US.

Using NPs, this work aims to recover sharper spatial structure with improved uncertainty calibration and higher frequency precipitation estimates relative to ERA5 under sparse observation scenarios and also evaluates their potential as robust uncertainty aware additions to physics-based models for high resolution environmental monitoring.

How to cite: Pazola, A., Ladopoulou, D., Morris-Wiltshire, C., Nath, P., and Coca-Castro, A.: Learning Spatiotemporal Precipitation Fields with Probabilistic Neural Processes, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12294, https://doi.org/10.5194/egusphere-egu26-12294, 2026.

The integration of machine learning parameterisations within climate models is paving the way for the next generation of Earth System models. Machine learning parameterisations are being developed to represent ocean and atmosphere processes such as turbulence, vertical mixing, and cloud and precipitation. These parameterisations typically require large volumes of high-resolution data for their training. This training data is often derived from the same numerical model that the parameterisation is intended for. This has the advantage that the machine learning model is only exposed to one set of numerical discretisation schemes.

Recently, global km-scale models have been introduced that simulate climate processes at remarkable detail. Explicitly resolving mesoscale and sub-mesoscale eddies and filaments enables these models to capture heat, carbon, and salt fluxes without the need for parameterisations. Global km-scale models are therefore promising training data sets for machine learning parameterisations.

In this work we intend to examine two global km-scale models that could be employed for oceanic turbulence parameterisations: NEMO ORCA36 and ICON-O. The ORCA36 model uses the tripolar grid and ICON-O uses an icosahedral grid. The question is, can either model be used to inform new ML parameterisations that can be employed in any numerical model? Therefore, a key assessment of these models will be done by exploring and contrasting their energetics, as well as the heat, salt, and carbon transports. This work will take the first step towards model-independent machine learning parameterisation development, while facilitating further cross modelling centre collaboration.

How to cite: Wilder, T. and Li, H.: Towards model-independent machine learning parameterisations of meso-scale eddies, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12770, https://doi.org/10.5194/egusphere-egu26-12770, 2026.

Representing and quantifying uncertainty in physical parameterisations is a central challenge in weather and climate modelling, and approaches are often developed separately for different timescales. Here, we consider the separation of uncertainty by source using machine learning frameworks for subgrid-scale parameterisations. In this context, aleatoric uncertainty arises from internal variability in the training data, and epistemic uncertainty, arises from poorly constrained parameters during training. Using the Lorenz 1996 system as a testbed for simplified chaotic dynamics, we deal with uncertainties through a unified framework using Bayesian Neural Networks, to explore how the different sources of uncertainty evolve over different prediction timescales.

How to cite: Mansfield, L. and Christensen, H.: Separating Epistemic and Aleatoric Uncertainties in Weather and Climate Models, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13183, https://doi.org/10.5194/egusphere-egu26-13183, 2026.

In an Atmospheric General Circulation Model (AGCM), the representation of subgrid-scale physical phenomena, also referred to as physical parameterizations, requires computational time which constrains model numerical efficiency. However, the development of emulators based on Machine Learning offers a promising alternative to traditional approaches.

We have developed offline emulators of the atmospheric component named ICOLMDZ (for DYNAMICO and LMDZ) of the IPSL climate model, in an idealized aquaplanet configuration, with the aim of emulating all the parameterizations, i.e. the LMDZ atmospheric physics component. While the results are quite promising, some fundamental questions are raised, particularly in terms of the generalization of the emulation process to meteorological conditions not seen by the emulator. This step is important for adopting the emulator as a substitute for traditional parameterizations.

This question of generalization, which relates to the ability of emulators to infer and adapt to new system states, has been studied in experiments linked to climate change. Indeed, we first investigated the performance of our emulators, trained on an aquaplanet configuration, in extrapolating the emulation process to new aquaplanets where boundary conditions are modified in order to simulate climates that are warmer and colder than the climate on which emulators are trained. The results reveal the potential of our aquaplanet emulators to reproduce the physical parameterizations of new climates. However, we also showed the limitations of these aquaplanet emulators since they encountered difficulties to generalize on a realistic configuration, i.e. when continents, topography and sea ice area are included.

This study encourages the coupling of emulators with the dynamic parts called DYNAMICO in order to better assess the relevance of the learning process, while analyzing the stability of the simulations obtained.

How to cite: Crossouard, S., Kageyama, M., Vrac, M., Dubos, T., Thao, S., and Meurdesoif, Y.: Generalization ability of emulators in reproducing the physical parameterizations of the IPSL model, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17216, https://doi.org/10.5194/egusphere-egu26-17216, 2026.

Atmospheric ozone is a crucial absorber of solar radiation and an important greenhouse gas. However, most climate models participating in the Coupled Model Intercomparison Project (CMIP) still lack an interactive representation of ozone due to the high computational costs of atmospheric chemistry schemes. Here, we introduce a machine learning parameterization (mloz) to interactively model daily ozone variability and trends across the troposphere and stratosphere in common CMIP simulations, including pre-industrial, abrupt-4xCO₂(Ma et al. 2025), historical and future Shared Socioeconomic Pathway (SSP) scenarios simulations. We demonstrate its high fidelity on decadal timescales and its flexible use online across two different climate models -- the UK Earth System Model (UKESM) and the German ICOsahedral Nonhydrostatic (ICON) model. With meteorological variables and forcing data as inputs, mloz produces stable ozone predictions around 31 times faster than the chemistry scheme in UKESM, contributing less than 4% of the respective total climate model runtimes. In particular, we also demonstrate its transferability to different climate models without chemistry schemes by transferring the parameterization from UKESM to ICON in standard climate sensitivity simulations. This highlights mloz’s potential for widespread adoption in CMIP-level climate models that lack interactive chemistry for future climate change assessments, where ozone trends and variability will significantly modulate atmospheric feedback processes.

Reference:
Ma Y, Abraham N L, Versick S, et al. mloz: A Highly Efficient Machine Learning-Based Ozone Parameterization for Climate Sensitivity Simulations[J]. arXiv preprint arXiv:2509.20422, 2025.

How to cite: Ma, Y., Abraham, N. L., Versick, S., Ruhnke, R., Schneidereit, A., Niemeier, U., Back, F., Braesicke, P., and Nowack, P.: mloz: A Highly Efficient Machine Learning-Based Ozone Parameterization for CMIP Simulations, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17576, https://doi.org/10.5194/egusphere-egu26-17576, 2026.

Are AI-powered climate models intrinsically more efficient than traditional climate models?

While progress is still needed before they become operational, hybrid AI-physics climate models and AI emulators of climate models have the potential to sharply reduce inference cost relative to traditional CPU-based models, allowing larger ensembles to explore different scenarios and sharpen uncertainty estimation. Yet this apparent efficiency becomes less obvious when the comparison includes GPU-ported dynamical climate models, and when efficiency is assessed against the effective complexity of the simulated climate system.

As a first step, recognizing that a perfect apple-to-apple comparison is rarely possible from reported configurations, we synthesize reported performance for leading AI climate model emulators (e.g., ACE2, CAMulator), hybrid AI-physics models (e.g., CliMA, NeuralGCM), and GPU-accelerated traditional models (e.g., SCREAM, ICON). We examine two complementary scaling views. The first compares throughput (simulated years per day) per accelerator (GPUs or TPUs) and per prognostic variable, as a function of horizontal grid spacing. The second compares the same normalized throughput against an effective complexity proxy, defined as the number of vertical levels divided by the product of the time step and the squared horizontal grid spacing, to account for the simulated vertical structure and, importantly, time-step constraints imposed by numerical stability.

We find that AI-powered models can show favorable apparent scaling with horizontal resolution in raw throughput, but that the advantage becomes modest once effective complexity is accounted for: at comparable complexity, AI climate models do not appear intrinsically more efficient than GPU-ported dynamical models. Hybrid approaches occupy a distinct middle ground: their acceleration and added value come primarily from learned parameterizations that improve the representation of unresolved processes while the overall model retains a physically-based dynamical core, including explicit conservation laws. AI climate model emulators, by contrast, offer their clearest computational advantage through task-targeted prediction, where a limited set of climate-relevant variables can be directly simulated on the grid of interest. This avoids integrating the full high-frequency, multivariate state at the short time step traditionally required for numerical stability, which is especially advantageous when emulating a fine-resolution reference model with a coarser emulator. Diverse downscaling or targeted post-processing strategies can further substitute for explicit fine-scale resolution when observations are available, enabling inexpensive local or hazard-specific risk assessment at decadal to multi-decadal time horizons.

How to cite: Beucler, T., Neelin, D., Su, H., Bretherton, C., Chapman, W., Christopoulos, C., Grover, A., Lopez-Gomez, I., Schneider, T., Subel, A., Watt-Meyer, O., and Zanna, L.: Reassessing the Scaling of AI-Powered Climate Models Against Dynamical Counterparts, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18440, https://doi.org/10.5194/egusphere-egu26-18440, 2026.

Persistent systematic errors in Earth system models (ESMs) arise from difficulties in representing the full diversity of subgrid, multiscale atmospheric convection and turbulence. Machine learning (ML) parameterizations trained on short high-resolution simulations show strong potential to reduce these errors. However, stable long-term atmospheric simulations with hybrid (physics + ML) ESMs remain difficult, as neural networks (NNs) trained offline often destabilize online runs. Training convection parameterizations directly on coarse-grained data is challenging, notably because scales cannot be cleanly separated. This issue is mitigated using data from superparameterized simulations, which provide clearer scale separation. Yet, transferring a parameterization from one ESM to another remains difficult due to distribution shifts that induce large inference errors. Here, we present a proof-of-concept where a ClimSim-trained, physics-informed NN convection parameterization is successfully transferred to ICON-A. The scheme is (a) trained on adjusted ClimSim data with subtracted radiative tendencies, and (b) integrated into ICON-A. The NN parameterization predicts its own error, enabling mixing with a conventional convection scheme when confidence is low, thus making the hybrid AI-physics model tunable with respect to observations and reanalysis through mixing parameters. This improves process understanding by constraining convective tendencies across column water vapor, lower-tropospheric stability, and geographical conditions, yielding interpretable regime behavior. In AMIP-style setups, several hybrid configurations outperform the default convection scheme (e.g., improved precipitation statistics). With additive input noise during training, both hybrid and pure-ML schemes lead to stable simulations and remain physically consistent for at least 20 years, demonstrating inter-ESM transferability and advancing long-term integrability.

How to cite: Heuer, H., Beucler, T., Schwabe, M., Savre, J., Schlund, M., and Eyring, V.: Beyond the Training Data: Confidence-Guided Mixing of Parameterizations in a Hybrid AI-Climate Model, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18986, https://doi.org/10.5194/egusphere-egu26-18986, 2026.

Ensuring that deep learning models generalize across distinct regimes remains a fundamental challenge in Earth system modeling. Due to the inherent violation of the independent and identically distributed (i.i.d.) assumption, models optimized for local conditions rarely exhibit robust performance on unseen domains. While Unsupervised Domain Adaptation (UDA) is a well-established technique for mitigating such distribution shifts in computer vision, its application to Earth system modeling remains underexplored. In this study we investigate the efficacy of UDA for the super-resolution of atmospheric fields, utilizing kilometer-scale COSMO simulations [1] and the RainShift benchmark dataset [2] to quantify model robustness across different regions. We apply residual learning to jointly super-resolve precipitation and surface pressure, incorporating static predictors such as topography. To quantify transferability, we propose a systematic framework that trains on source domains and evaluates on unseen target domains, treating spatial transfer as a proxy for model robustness under distribution shifts. We introduce a consistency metric to evaluate model adaptation by comparing mean performance on seen versus unseen domains. We assess a hierarchy of adaptation methods, ranging from simple regularization to physics-informed approaches. These include domain-specific regularization and distribution alignment methods, domain adversarial training, and geometry-robust training via group-equivariant convolutions. Preliminary results on the COSMO simulations demonstrate that even elementary adaptation strategies, such as dropout and data augmentation, improve cross-domain consistency. This work establishes a controlled setup for benchmarking generalization, suggesting that UDA offers a viable pathway to bridge the gap between locally trained models and global applicability.

[1]: Cui, R., Thurnherr, I., Velasquez, P., Brennan, K. P., Leclair, M., Mazzoleni, A., et al. (2025). A European hail and lightning climatology from an 11-year kilometer-scale regional climate simulation. Journal of Geophysical Research: Atmospheres, 130, e2024JD042828. https://doi.org/10.1029/2024JD042828

[2]: Paula Harder et al. RainShift: A Benchmark for Precipitation Downscaling Across Geographies. 2025. arXiv: 2507.04930 [cs.CV]. url: https://arxiv.org/abs/2507.04930.

How to cite: Quarenghi, F. and Beucler, T.: Transferring knowledge across regions: unsupervised domain adaptation for km-scale super-resolution, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19396, https://doi.org/10.5194/egusphere-egu26-19396, 2026.

Data driven parameterisations for sub-grid processes unlocks the ability to surpass the current computational constraints of Earth system models. However, machine learning (ML) can be brittle. State-of-the-art ML approaches can reliably perform on in-distribution data, exceeding human ability across a diverse range of tasks. Yet, when faced with shifts in data distribution, performance degrades. In climate modelling, when the task is predicting the state of a non-stationary system, this is evidently a substantial issue. We illustrate this with the ClimSim dataset, forming spatio-temporal groups and quantitatively show how even small shifts in distribution affect performance.

Next, we use the theory of compositional generalisation to build models which are less susceptible to these shifts in distribution. Compositional generalisation is the formation of novel combinations of observed elementary components. That is, the ability to decompose data into building blocks that are reused across both the in- and shifted-domains, such that a model can capture a domain shifted state through a set of in-domain, learnt abstractions. Inspired by these concepts we propose various architectural and regularisation changes to standard ML parameterisations to improve generalisation. Preliminary results in sub-grid process emulators suggest new insights into if and how CG can reduce model sensitivity to domain shifts.

How to cite: Stanley-Clamp, B., Posner, I., and Christensen, H.: Beyond In-Distribution Skill: Towards Robust ML Parameterisations for Non-Stationary Climate Systems, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20353, https://doi.org/10.5194/egusphere-egu26-20353, 2026.

Under greenhouse gas forcing, the global climate exhibits a long-term warming trend superimposed with quasi-periodic multidecadal oscillations (~60–70 years) closely linked to the Atlantic Meridional Overturning Circulation (AMOC). As a pivotal component of global ocean circulation, the AMOC regulates the distribution of oceanic heat and freshwater, exerting profound influence on global climate variability. Conventional views posit a positive correlation between AMOC strength and global mean surface temperature (GMST) on multidecadal timescale. However, our analysis reveals a significant phase shift of approximately 45°–90° between AMOC and GMST on multidecadal timescale under anthropogenic warming. This shift arises as enhanced vertical ocean heat transport within the subpolar North Atlantic’s mid-depth layers modulates the surface energy budget balance under increasing radiative forcing, thereby disrupting the equilibrium between horizontal meridional heat transport and surface net heat flux. External radiative forcing perturbs internal climate variability, driving a substantial reduction in mean-state density in the subpolar North Atlantic’s mid-depth ocean. Crucially, the intensified vertical heat transport associated with AMOC strengthening emerges as the key mechanism facilitating heat sequestration into the ocean interior.

How to cite: Bi, H., Chen, X., Li, X., and Tung, K.-K.: Phase Shift of AMOC and Multidecadal Global Mean Surface Temperature Under Anthropogenic Forcing, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-603, https://doi.org/10.5194/egusphere-egu26-603, 2026.

Atmospheric heat and carbon uptake and storage by the ocean are controlled by seawater stratification, which is also linked to Atlantic Meridional Overturning Circulation (AMOC) through ocean heat distribution that can modify density stratification. The effects of a potential weakening of the AMOC on ocean stratification, and therefore on heat uptake and storage, remain an open question. To gain insight into these dynamics, we used marine sedimentary archives of the last deglaciation (last 27000 years) to reconstruct temperatures at intermediate (GeoB9512-5, 793 m water depth) and deep (GeoB9508-5) water masses of the eastern Atlantic off the coast of Senegal (northwestern Africa). During this time, marked changes in AMOC strength took place: the Last Glacial Maximum (LGM, 23,000 – 19,000 years ago), a time of shallower meridional overturning; and the Heinrich Stadial 1 (HS1, 18,200–14,900 years ago) and Younger Dryas (YD, 12,800–11,700 years ago), when AMOC was weaker than today. Our benthic foraminifera-based Mg/Ca (seawater temperature) and δ18O (ocean density) show that a persistently shallow and strong (LGM) or weak (HS1 and YD) meridional overturning led to a mid-depth warming at the same time deep-ocean heat uptake was paused, leading to a strong density stratification in the Atlantic. These results are compatible with previous temperature reconstruction across the tropical and north Atlantic, and also show that with a Holocene AMOC strengthening, mid-depth cooling and resumed deep-ocean heat uptake resulted in a weaker stratification. Our findings show that the AMOC state sets the depth of heat storage and that the depth of the upper AMOC cell is tightly related to deep ocean stratification.

How to cite: Barragán Montilla, S., Mulitza, S., Johnstone, H. J. H., and Pälike, H.: Deglacial ocean density de-stratification with a weaker Atlantic Meridional Overturning Circulation, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1299, https://doi.org/10.5194/egusphere-egu26-1299, 2026.

During the Ice Ages, abrupt climate changes co-occurred with switches in Atlantic Meridional Overturning Circulation (AMOC) strength. The thermal bipolar seesaw has served as a seminal conceptual framework to explain the global extent of these events, calling on interhemispheric redistribution of heat to explain the observed north-south temperature pattern. Here we summarize an emerging alternative framework centered instead on the global ocean heat content (OHC) and planetary energy budget, which we illustrate using simulations of spontaneous abrupt climate change in three climate models. In all models, the AMOC strength sets the OHC trend via the rate of North Atlantic heat loss, coupled to the top-of-the-atmosphere energy budget through radiative feedbacks. Antarctic and Greenland temperatures, as recorded in ice cores, are shown to reflect OHC and the rate of North-Atlantic heat loss, respectively. Under intermediate glacial climate states, global ocean heat uptake cannot reach steady-state with the bimodal rate of North Atlantic heat loss causing instability. Our synthesis suggests that the AMOC serves as a heat valve that alters planetary temperature by changing the radiative balance. This implies amplified planetary heat uptake in response to projected future AMOC weakening.

How to cite: Buizert, C., Abe-Ouchi, A., Vettoretti, G., Zhang, X., Kuniyoshi, Y., Shackleton, S., Rasmussen, S., Pedro, J., Galbraith, E., and Stocker, T.: The ocean heat valve: AMOC and planetary energy budget during abrupt glacial climate change, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1465, https://doi.org/10.5194/egusphere-egu26-1465, 2026.

The Atlantic Meridional Overturning Circulation (AMOC) is a key regulator of global climate and has been a subject of major scientific interest. Observational studies have raised concerns about its ongoing weakening and potential collapse this century. While climate models generally show an overall cooling over Europe as a result of this weakening, confirmation based on observations is lacking due to difficulties in assessing causality in data. Here, we overcome this problem by constructing causality maps and tracking AMOC’s impact over Europe in observations. First, the causal link between AMOC and its SST fingerprint is established. Then, decomposing the SST fingerprint of AMOC into a decreasing centennial trend and a multidecadal oscillation (AMO), we find the trend impacts only winter and AMO only summer. In winter, the weakening warms north-central Europe and increases northern precipitation, with no overall cooling being observed nor expected. In summer, AMO induces multidecadal oscillations in temperature and precipitation. These quantitative results can be an observational benchmark for future model simulations, inform policy making, and national security.

How to cite: Nichita, D., Dima, M., Vaideanu, P., and Ionita, M.: A weakened AMOC warms winters and drives summer multidecadal variability over Europe, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1849, https://doi.org/10.5194/egusphere-egu26-1849, 2026.

The Atlantic Meridional Overturning Circulation (AMOC) is a key component of the climate system with far-reaching effects on global climate. Here, we investigate the influence of ocean basins beyond the Atlantic on both AMOC dynamics and surface climate variability, using simulations with the coupled climate model MPI-ESM-LR. We apply an AMOC upwelling pathways framework to quantify the influence of the Indo-Pacific and Southern Ocean on AMOC strength over the 58-year time period 1958-2014 in three model setups: a historical simulation, an atmosphere-only assimilation, and a coupled atmosphere-ocean assimilation. Through regression analysis, we reveal the relationship between the AMOC upwelling pathways in the different ocean basins and sea-surface temperature (SST). Preliminary results show distinct SST patterns on a global scale for each setup, suggesting teleconnections between the AMOC and its upwelling components, and global surface climate dynamics. By comparing the different model setups, we assess the impact of the assimilation of observational data on the representation of the AMOC, the SST and their relationship, and improve our understanding of the role of the AMOC as part of the global climate system.

How to cite: Bühl, T., Brune, S., and Baehr, J.: An analysis of the imprint of the global ocean circulation on AMOC dynamics and surface climate during the time period 1958-2014, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3811, https://doi.org/10.5194/egusphere-egu26-3811, 2026.

For climate predictions on decadal to multi-decadal time scales, the ocean circulation has been found to carry a substantial portion of the memory from initialisations. In this study, we analyse the global ocean overturning circulation, in particular the Atlantic meridional overturning circulation (AMOC), in climate simulations with the global coupled model MPI-ESM for the time period 1960-2100. We compare an ensemble of multi-decadal predictions, initialised from a coupled assimilation simulation, and an ensemble of uninitialised simulations, both with CMIP6 historical and SSP2-45 external forcing. We find three distinct time scales for the evolution of the AMOC strength at 26N after the initialisation time. On a time scale up to 5 years after initialisation, the AMOC reacts to the initialised state with a rapid under- or overshooting when compared to uninitialised simulations, depending on the initialisation time. On a time scale of 30 to 140 years after initialisation, the AMOC by and large maintains this bias between initialised predictions and uninitialised simulations. We also find these distinct time scales in the characteristics of the AMOC cells, in both the overturning and re-circulation cells. In addition, we show that the AMOC evolution is related to the global ocean circulation. Specifically, we find a strong connection of the AMOC cell with the global Southern Ocean circulation, and we also find that multi-decadal AMOC trends are being partly compensated by changes in the strength of the Indo-Pacific meridional overturning. Our results show that the ocean circulation, in particular the AMOC, may carry the information about initialisation over multi-decadal time scales, up to 140 years. While this does not necessarily imply good prediction skill on the multi-decadal time scale, it adds another dimension on how we asses the uncertainty of climate projections until 2100.

How to cite: Brune, S., Hansen, J., Bühl, T., Uddin, M. B., Düsterhus, A., and Baehr, J.: The Atlantic meridional overturning circulation in multi-decadal end of century climate predictions, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5291, https://doi.org/10.5194/egusphere-egu26-5291, 2026.

The Atlantic Meridional Overturning Circulation (AMOC) is weakening in response to global warming, while Nordic Seas Overturning Circulation (NOC) is projected to increase. So far, no causal link has been proposed between these two opposing trends. Here we propose that a density reduction in the subpolar North Atlantic will weaken the AMOC by reducing the density difference with lighter waters further south, while at the same time strengthening the NOC by increasing the density difference with the heavier waters further north. Using high resolution climate model data and a box model, we find that in response to combined global warming and freshwater input the NOC initially increases moderately as the AMOC weakens, while a tipping point may be reached later if deep convection in the Nordic Seas shuts down and the NOC collapses together with the AMOC.

How to cite: Roewer, S., Fiedler, L., Årthun, M., Huiskamp, W., and Rahmstorf, S.: Nordic Overturning Increases as AMOC Weakens in Response to Global Warming, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5641, https://doi.org/10.5194/egusphere-egu26-5641, 2026.

The Atlantic Meridional Overturning Circulation (AMOC) has shown signs of decline over the last two decades. Climate models project that a continued slowdown of the AMOC will increase precipitation over parts of northern Europe, particularly in the Irish-British Isles¹, with potential impacts on agriculture and related systems. However, the ability of climate models to predict when such changes might occur remains limited, calling for the use of paleoclimate archives. Here, we present a stalagmite‑based paleoclimate record from the west coast of Ireland spanning 11.1–7.7 ka (b2k). Combined Sr/Ca and stable isotope data indicate a sudden increase in precipitation at ~8.6 ka, coincident with the collapse of the Hudson Bay Ice Saddle (HBIS)² and a reduction in eastern North Atlantic bottom and surface currents^3,4. We interpret this hydroclimatic shift as a response to the slowdown of the AMOC caused by the HBIS freshwater discharge, indicating a minimum time lag (of decadal scale) between ocean circulation disruption and atmospheric response. Due to enhanced thermal and pressure gradients over the North Atlantic, a weakened AMOC can favour positive North Atlantic Oscillation (NAO+) conditions, which typically bring wetter and stormier weather over northern Europe. We therefore associate the ~8.6 ka precipitation increase with the development of NAO+ conditions in the region, which aligns with existing work⁵. In addition, our record evidences sustained precipitation throughout the '8.2 ka' cooling anomaly, suggesting that, regardless of temperature direction, heightened precipitation is a persistent consequence of AMOC reduction in northwest Europe.

1: Jackson et al. (2015) Climate Dynamics, 45. 2: Lochte et al. (2019) Nature Communications, 10, 586. 3: Ellison et al. (2006) Science, 312. 4: Thornalley et al. (2009) Nature, 457. 5: Smith et al. (2016) Scientific Reports, 6, 24745.

How to cite: Ansberque, C., Schenk, F., Mark, C., Hällberg, P., Kylander, M., and McDermott, F.: Rapid climatic response to the Hudson Bay Ice Saddle collapse (~8.6 ka) recorded in Ireland, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7116, https://doi.org/10.5194/egusphere-egu26-7116, 2026.

This study performs uncertainty quantification on the regional mean surface temperature response to changes in the Atlantic Meridional Overturning Circulation (AMOC) and allows the investigation of novel AMOC scenarios. ESMs/GCMs primarily show gradual AMOC slowdown in the 21^st and early 22^nd century while other approaches suggest that a “tipping point” may be present which could lead to faster decline in the AMOC during this period. This study aims to estimate the impacts of a rapid decline or other AMOC scenarios and the range of possible outcomes which can be inferred from the current ensemble of climate models and approaches.  Changes in temperature and AMOC will be analysed under a range of forcing scenarios including CMIP6 SSP scenarios for global warming, freshwater hosing scenarios from NAHosMIP, and ClimTip runs showing a combination of global warming and freshwater hosing. The relationships between AMOC change, global mean surface temperature and regional mean surface temperature are described, as well as our uncertainty in these values based on the model ensembles.  These relationships are used to generate annual mean regional/ national temperature trajectories under a range of potential AMOC scenarios, with uncertainty ranges given for each scenario and location. These methods can be extended to both seasonal temperature and annual precipitation, and the data produced is highly consequential for economic impact assessments and adaptation planning.

How to cite: Rosser, J. and Stainforth, D.: Uncertainty Quantification of the regional temperature consequences of a large AMOC decrease and use in AMOC scenario exploration, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8954, https://doi.org/10.5194/egusphere-egu26-8954, 2026.

How to cite: van Westen, R., Börner, R., and Dijkstra, H.: Radiative Forcing Path Dependent Temperature Thresholds for AMOC Tipping, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9875, https://doi.org/10.5194/egusphere-egu26-9875, 2026.

We identify a millennial-scale oscillatory eigenmode of the Atlantic Meridional Overturning Circulation (AMOC) in a conceptual two-hemisphere box model. To isolate the governing mechanism, we examine two idealized cases that represent situations where AMOC variability arises exclusively from the North Atlantic Deep Water (NADW) cell or from the Antarctic Bottom Water (AABW) cell. 

In the NADW-influenced case, the AMOC anomaly is parameterized as positively related to the north-south salinity difference. Linear analysis shows that the oscillation period increases as the mean AMOC strength decreases. Thus, a weaker mean AMOC produces slower oscillations, and the dominant time scale can shift from multicentennial to millennial. For example, when the mean AMOC strength is near 10 Sv, the model yields a dominant millennial-scale oscillation. 

In the AABW-influenced case, the AMOC anomaly arises from AABW-related processes and exhibits a negative linear dependence on the north-south salinity difference. The resulting millennial oscillation is driven by upward transport from the deep to the upper South Atlantic, a process that responds sensitively to local surface freshwater fluxes. 

Taken together, these results highlight internal ocean dynamics that can generate millennial-scale AMOC variability through two distinct pathways, associated with northern and southern overturning processes, respectively. Finally, we discuss the implications of these findings for interpreting observed millennial-scale climate variability during the last glacial period and the Holocene. 

How to cite: Zhou, X. and Yang, H.: Millennial-Scale Oscillation of the AMOC in a Two-hemisphere Box Model, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10253, https://doi.org/10.5194/egusphere-egu26-10253, 2026.

The Atlantic Meridional Overturning Circulation (AMOC) plays a central role in regulating Earth's climate, and is widely considered to be a vulnerable tipping element of the climate system. The Bering Strait Throughflow (BST) can play a key role in the AMOC's stability. Through this narrow passage relatively fresh Antarctic Intermediate Water from the Pacific basin enters the Arctic Ocean and eventually ends up in the deep-water formation zones in the North Atlantic. Moreover, an open Strait enhances the freshwater exchange between the Arctic and North Atlantic. All in all, the Throughflow's net effect is a freshening of the North Atlantic, and hence a weakening of the AMOC. Recent research has indicated that the AMOC is weakening and may reach its tipping point before the end of this century. Since the Bering Strait has limited width and is relatively shallow (approximately 80 km across and on average 50 m deep) constructing a barrier is technically feasible. In this work we show that such a barrier can prevent an AMOC collapse in three levels of the model hierarchy. Firstly, a conceptual model of the World Ocean is extended to include the BST and Arctic amplification, showing that for a low freshwater forcing in the North Atlantic a closure of the Strait prevents an AMOC tipping under climate forcing. Moreover, the conceptual framework allows us to test the sensitivity of the results with respect to BST parametrization and rate of forcing. Next, the conceptual results are reproduced in an Earth system Model of Intermediate Complexity (EMIC). Here we have investigated the AMOC's safe carbon budget for either an open or closed Strait for various freshwater hosing strengths. This reveals an increased carbon budget under a closure given -again- a sufficiently low strength of North Atlantic hosing. Lastly, the closure's effectiveness is tested in a CMIP5 model, namely CESM1. Here an AMOC collapse occurs under RCP8.5 forcing for both a low and high freshwater hosing. In the former the AMOC strength matches observations, while in the latter the overturning-induced freshwater transport through the Atlantic's southern boundary is realistic. In both scenarios a closure of the Bering Strait prevents an AMOC collapse on the condition that this closure occurs sufficiently early. In the strong hosing scenario a closure has to occur at least as early as 2050, while in the low hosing case a closure as late as 2080 is still sufficient. Hence, we have shown throughout the model hierarchy that a closure of the Bering Strait can prevent a collapse of the AMOC, and that it is a potential climate intervention strategy should emissions mitigation fail.

How to cite: Soons, J., van Westen, R., and Dijkstra, H. A.: A Constructed Closure of the Bering Strait to prevent an AMOC tipping, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10492, https://doi.org/10.5194/egusphere-egu26-10492, 2026.

Paleoclimate evidences and coupled model studies suggested that the Atlantic Meridional Overturning Circulation(AMOC) has significant multicentennial variability. In this study, we use simplified two-dimensional and three-dimensional ocean model to extend previous theoretical and coupled model studies on the multicentennial oscillation(MCO) of AMOC, providing clearer physical insights and bridging the gap between idealized conceptual model and high-complexity numerical models. Our results demonstrate that stochastic salinity forcing effectively excites AMOC MCO, with the oscillation primarily driven by the tropical-subpolar advection feedback. Sensitivity experiments show that the period of the AMOC MCO is largely determined by the strength and vertical structure of the climatological AMOC: a stronger AMOC leads to a shorter oscillation period, while a deeper AMOC maximum results in a longer period. Under weak AMOC conditions, the oscillation timescale can extend to millennial scales. We also explore the role of wind-driven circulation and find that, although it has little influence on the MCO period, it slightly modifies the amplitude of variability by suppressing low-frequency components and enhancing high-frequency fluctuations. These simplified ocean model enables a systematic exploration of key physical mechanisms underlying AMOC MCO, offering valuable insights into long-term climate variability.

How to cite: Wang, S., Yang, H., and Zhou, X.: Investigating the multicentennial oscillation of the AMOC using simplified ocean model, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10948, https://doi.org/10.5194/egusphere-egu26-10948, 2026.

The Atlantic Meridional Overturning Circulation (AMOC) regulates large-scale heat and freshwater transport, and strongly influencing global climate patterns. Beyond its role in shaping mean climate conditions, the AMOC background state also modulates climate variability. The AMOC is a tipping element of the climate system and a collapse of the AMOC alters atmospheric circulation patterns such as the Hadley circulation, polar jet stream, and tropical trade winds, with consequences that extend far beyond the Atlantic basin. These changes affect atmospheric and oceanic variability, thereby reshaping global teleconnection patterns. Using the results of a full hysteresis simulation of the AMOC in the CMIP5 version of the Community Earth System Model (CESM), we study the importance of the present-day AMOC mean state in shaping the large-scale atmospheric circulation, the global oceanic circulation, and internal climate variability. By comparing equilibrium climate states under AMOC on- and off conditions, we investigate the role of the AMOC in climate variability phenomena, such as the El Niño-Southern Oscillation and the midlatitude patterns of sea surface temperature variability. Our results highlight the AMOC as a critical regulator of global climate variability, emphasising the importance of understanding its stability in a warming climate.

How to cite: Smolders, E., van Westen, R., and Dijkstra, H.: The Role of the AMOC in Shaping Internal Climate Variability, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11388, https://doi.org/10.5194/egusphere-egu26-11388, 2026.

Cooling across Europe is the most widely mentioned impact of a weakened AMOC. However, we find that the end-of-century net temperature change over Europe, including both the AMOC-induced cooling and global warming, remains surprisingly undetermined in the existing literature. In our study, using both new Earth system model simulations and existing multi-model evidence, we show that certain parts of Europe could cool below preindustrial temperatures in scenarios with both a substantial AMOC weakening and low emissions. Under continued emissions, however, most regions would either not face the risk of net cooling or only at very high amounts of AMOC weakening. Simulations under combined scenarios of AMOC weakening and global warming reveal that the effect of a given amount of AMOC weakening on European temperatures is remarkably linear and independent of the underlying emissions scenario. This relationship circumvents the large uncertainties around the AMOC’s future evolution by instead inferring the amount of AMOC weakening that would cool a specific European region or country for any global warming scenario.

How to cite: Alastrué de Asenjo, E. and Schaumann, F.: Could Europe actually cool if the AMOC weakens in a warming climate?, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11518, https://doi.org/10.5194/egusphere-egu26-11518, 2026.

The projected weakening of the Atlantic Meridional Overturning Circulation (AMOC) poses substantial risks for global and regional climate stability. While the large-scale cooling associated with a weakened AMOC is well-documented, how weather and climate extremes respond to such changes remains little examined. Here, we investigate how recent European summer and winter temperature extremes (2018-2022) would change under different weakened AMOC states using the Alfred Wegener Institute Climate Model (AWI-CM3). We generate three sets of five-member ensemble simulations, each representing a different AMOC state: a factual (present-day AMOC) state and two counterfactual states with a weakened and a shut-down AMOC. All simulations are spectrally nudged to the large-scale winds observed during 2017-2022. We thus focus primarily on the thermodynamic impacts induced by AMOC weakening within the same realization of atmospheric variability. Our research indicates that a weakened AMOC generally reduces the occurrence of summer hot days, though this response is spatially heterogeneous, implying a flow-dependence of the AMOC-related impact. For instance, Eastern Europe remains comparatively less affected even when AMOC strength is reduced by 60% relative to the present day conditions. In contrast, winter cold extremes are substantially intensified. We observe a drastic increase in cold days, with daily minimum temperatures during these events decreasing by more than 6 °C in several northwestern European capital cities. These findings highlight the nonlinear and seasonally asymmetric responses of European temperature extremes to AMOC weakening and provide important insights for regional climate risk assessment and adaptation strategies.

How to cite: Ma, Q., Athanase, M., Sanchez-Benitez, A., Streffing, J., Goessling, H., Jung, T., Lohmann, G., and Ionita, M.: Response of European Temperature Extremes to a Weakened AMOC, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12946, https://doi.org/10.5194/egusphere-egu26-12946, 2026.

The Atlantic meridional overturning circulation (AMOC) is a vitally important component of the global ocean circulation because of its impact on the environment, weather, and ecosystems. The South Atlantic is a key gateway for water mass exchanges between the Atlantic and other basins as southward overturning freshwater transport at 34.5°S increases the likelihood of an AMOC collapse in the future. In two-thirds of state-of-the-art coupled climate models, the overturning freshwater transport at 34.5°S is northward and AMOC is monostable, whereas most observations find that freshwater transport is southward suggesting AMOC is bistable. The upper limb of the AMOC and Deep Western Boundary Current (DWBC), a major element of AMOC’s lower limb, control freshwater transport at 34.5°S. It is therefore crucial to observe the daily strength of both of these circulation systems and use these observations to validate numerical models.

We examine AMOC and DWBC variability from over fourteen years of South Atlantic MOC Basin-wide Array (SAMBA) measurements between South America and South Africa along 34.5°S . These observational records enable concurrent examination of the temporal variations of the upper and lower limbs of AMOC. During 2009-2022, the AMOC volume transport weakened by -0.6 Sv/yr, but this trend is obscured by significant higher frequency variability (± 10 Sv standard deviation with respect to the 18.6 Sv long-term mean) and a 3-year data gap on the eastern boundary during 2010-2013. The inclusion of more years of data has shifted the AMOC seasonal cycle from semi-annual to quasi-annual, and has improved agreement with Argo-altimetry based estimates on seasonal timescales. SAMBA transports are more energetic than Argo-altimetry on intraseasonal and interannual time scales, with the largest differences occurring when SAMBA density-driven variations are strong. The SAMBA DWBC has a mean southward transport of -17 Sv and a standard deviation of 22 Sv, with a significant negative trend of -0.3 Sv/year (DWBC increasing in strength). AMOC and DWBC variations are modestly correlated along 34.5°S on monthly and longer timescales, such that a weaker AMOC corresponds to stronger DWBC anomalies. This covariability will be explored further to better establish the connectivity between AMOC and the DWBC in the South Atlantic.

How to cite: Perez, R. C., Dong, S., Ansorge, I., Campos, E., Chidichimo, M. P., Garcia, R., Lamont, T., Louw, G., Le Henaff, M., Piola, A., Sato, O., Speich, S., Tuchen, F. P., van den Berg, M., and Volkov, D.: Observed Variability of the Atlantic Meridional Overturning Circulation and the Deep Western Boundary Current along 34.5°S, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13099, https://doi.org/10.5194/egusphere-egu26-13099, 2026.

Statistical methods generally predict a possible tipping of the Atlantic Meridional Overturning Circulation (AMOC) in the near future, suggesting that the climate models overestimate the stability of AMOC. Conversely, observations show a stable AMOC during the past decades, suggesting otherwise. Based on the MITgcm-ECCO2, here we show that the biases in the simulated Arctic sea ice, freshwater content, and the water transport across various straits/passages around the Arctic play a key role in the future stability of AMOC in the climate models. Specifically, most climate models project an increased freshwater export from the Arctic across the Fram Strait in the future. In contrast, our model, with minimal bias for the present day, simulates a decrease in freshwater export across the Fram Strait but an increase across the Lancaster Strait. This shift of location increases AMOC stability as the freshwater coming out of Fram Strait has a direct impact on the surface density over the North Atlantic deepwater formation region.

How to cite: Gavilan Pascual-Ahuir, E. and Liu, Y.: A Resilient Atlantic Meridional Overturning Circulation in the Near Future, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14045, https://doi.org/10.5194/egusphere-egu26-14045, 2026.

Previous studies indicate that cooling in the Subpolar North Atlantic (SPNA), known as the “Cold Blob,” may influence European summer heat extremes. However, how internally generated ocean–atmosphere variability and anthropogenic forcing jointly shape this relationship remains poorly understood. Here, we use the 50-member Max Planck Institute Grand Ensemble (MPI-GE) under the SSP2–4.5 scenario to assess how SPNA sea surface temperature (SST) anomalies affect the likelihood of exceptionally persistent European heatwaves.

We analyze ensemble-member differences in Atlantic Meridional Overturning Circulation (AMOC)–driven heat transport, SPNA SST evolution, and associated atmospheric circulation over Europe. We find that declining AMOC heat transport enhances ocean heat divergence in the subpolar gyre, promoting SPNA surface cooling and the emergence of the Cold Blob, although the magnitude and persistence of this cooling vary strongly across ensemble members. Persistent European heatwaves are favored primarily when subpolar cooling coexists with subtropical warming, strengthening the inter-gyre SST gradient and promoting stationary large-scale pressure systems over Europe. In mid-century projections, the relationship between cold SST anomalies and heatwaves is highly sensitive to the evolving oceanic background state.

Overall, our results demonstrate that internal coupled ocean–atmosphere variability strongly modulates near-term European summer heatwave risk under climate change and identify SPNA SSTs as a promising source of seasonal-to-multiyear predictability.

How to cite: Cerato, G., Lohmann, K., von Hardenberg, J., Bellomo, K., and Matei, D.: The Subpolar Gyre as Ocean–Atmosphere Bridge Between AMOC Variability and European Summer Temperature Extremes, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14331, https://doi.org/10.5194/egusphere-egu26-14331, 2026.

Although the Atlantic Meridional Overturning Circulation (AMOC) is considered a critical climate tipping element, its impacts on the terrestrial carbon cycle in Earth system models remain uncertain. Using the Earth system model, CLIMBER-X, we investigate the response of vegetation carbon to idealized AMOC collapse under pre-industrial conditions. We assess the role of carbon-climate feedback by comparing simulations incorporating interactive carbon cycles with experimental results set at atmospheric CO₂ concentrations.

The results indicate that AMOC collapse leads to a large-scale change of vegetation carbon, with significant differences in responses between the Northern and Southern Hemispheres. The simulated global vegetation carbon response depends on whether the carbon-climate interaction is considered in the model, highlighting the importance of interactive carbon cycle processes. Our findings indicate the sensitivity of terrestrial vegetation carbon to AMOC changes and suggest that it is important to account for ocean-terrestrial-carbon coupling in Earth system models when assessing potential AMOC tipping events.

How to cite: Nian, D., Willeit, M., and Rockström, J.: Terrestrial Vegetation Carbon Responses to an AMOC Collapse in an Earth System Model, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14940, https://doi.org/10.5194/egusphere-egu26-14940, 2026.

Shallow subsurface Atlantic nutrient and carbonate chemistry respond to many oceanic processes, including deep ocean circulation changes. Changes in the Atlantic Meridional Overturning Circulation (AMOC) since the Last Glacial Maximum (LGM) affected the shallow Atlantic nutrients and carbonate chemistry, but high-resolution records from the shallow tropical Atlantic are from only a few sites and not replicated, and hence the timing and significance of millennial changes are poorly constrained. After reevaluating the optimal benthic foraminifera species for seawater nutrients and carbonate ion concentration ([CO₃^2-]) reconstructions with more core-top data and down-core inter-species comparisons, we present eight nutrient and five [CO₃^2-] reconstructions from the upper to intermediate-depth western Atlantic Ocean (~500 – 2000 m) to document changes in the shallow tropical-subtropical North Atlantic and trace changes in their southern- and northern-sourced regions.

The Demerara Rise and Florida Margin are among the first replicate nutrients and [CO₃^2-] records for the deglaciation, and are remarkably similar, confirming that these millennial changes represent regional signals of shallow tropical-subtropical North Atlantic. These high-resolution nutrients and [CO₃^2-] records provide new evidence for a weakened AMOC during Allerød and the Younger Dryas (YD), a diminished nutrient stream in upper North Atlantic during YD, and a millennial event at ~ 9 ka. We confirm that AMOC changes from ~18 to 8 ka are likely the main cause of nutrient and [CO₃^2-] changes in the shallow tropical North Atlantic. The long-term changes in [CO₃^2-] were additionally affected by rising atmospheric CO₂ since the LGM. Our results support the notion that changes in nutrients and carbonate chemistry can be affected by multiple factors, but a better understanding of their driving mechanisms and a combination of reconstructions may provide a more complete picture of AMOC changes since the LGM.

How to cite: Lu, W., Oppo, D., Lynch-Stieglitz, J., and Hess, A.: Evolution of shallow subsurface Atlantic nutrient and carbonate saturation state since the Last Glacial Maximum, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15210, https://doi.org/10.5194/egusphere-egu26-15210, 2026.

This project aimed to investigate how southern Australia was impacted by the AMOC driven millennial scale climate events of the Last Glacial Period.

Paleoclimate studies have demonstrated that abrupt millennial-scale climate events during the Last Glacial Period coincided with variations in the strength of the Atlantic Meridional Overturning Circulation (AMOC). These include Dansgaard-Oeschger events, which coincide with periods of AMOC strengthening, and Heinrich events, which coincide with AMOC weakening or collapse.  

Whilst numerous paleoclimatic studies have examined the global climatic and environmental consequences of these events, relatively few of these studies are based in the southern hemisphere, even fewer in Australia, with southern Australia largely overlooked. This is a problem, as there is currently very little understanding of how the southern Australian hydroclimate, fire regimes and vegetation was impacted by AMOC slowdown and/or shutdown in the past. Moreover, the scarcity of high resolution, temporally extensive paleoclimatic records in southern Australia constrains our capacity to understand interhemispheric leads & lags as well as the local response to rapid climate events.

To address these knowledge gaps, this project produced three new southern hemisphere mid-latitude paleoclimatic datasets and improved the age-constraints and proxy resolution on one existing published paleoclimatic dataset.

Three speleothems were analysed for this project- from Mammoth Cave (Southwest Western Australia), Kubla Khan Cave (Tasmania, Australia) and Hollywood Cave (South Island, New Zealand). We investigated the paleohydrology of these sites using stable isotope analysis (δ¹⁸O and δ¹³C), trace element analysis and geochronology (U-Th dating). The datasets from Mammoth Cave (38-14ka) and Kubla Khan (75-23 ka) have demonstrated hydroclimate excursions associated with millennial climate events, likely due to the meridional displacement of the South Westerly Winds. Extensive U/Th dating of the Hollywood Cave speleothem (73-11ka) has altered the pre-existing, published age model, with implications for the current interpretation of millennial climate event timing in the southern mid-latitudes.

A lake sediment sequence was also analysed as part of this project, to determine the vegetation, fire regime and hydroclimate impacts of AMOC driven millennial climate events. Lake Bullen Merri (western Victoria) was cored in early 2025, yielding 15m of lake sediment and ~36,000 years of climate history. Thirty-one 14C dates have been returned, providing a robust age-depth model. A suite of analyses have been applied to this sediment core; X-RF, magnetic susceptibility, loss on ignition, palynology, macroscopic and microscopic charcoal counting, biomarkers (n-alkanes, sterols, PAH’s) and leaf wax H-Isotope analysis. These results show significant hydroclimate & fire activity excursions throughout the past ~36,000 years, with higher resolution proxy analysis underway to highlight millennial/centennial scale excursions.

These results provide one of the first insights into the way southern Australia is impacted by millennial scale climate events, offering a valuable regional insight, as well as a point of comparison for interhemispheric studies.

How to cite: Sheridan, L., Fletcher, M.-S., Drysdale, R., and Korasidis, V.: Investigating the impact of millennial scale climate events on southern Australia during the Last Glacial Period, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17273, https://doi.org/10.5194/egusphere-egu26-17273, 2026.

A potential Atlantic Meridional Overturning Circulation (AMOC) slowdown, possibly caused by external forcings, is widely debated, and its historical drivers and future evolution remain uncertain. Here we disentangle the effects of greenhouse gases and anthropogenic aerosols on the AMOC and on other relevant processes in the high-latitude North Atlantic (NA) over 1850–2014. We analyze a multi-model ensemble of experiments from the Large Ensemble Single Forcing Model Intercomparison Project, specifically: hist-GHG (varying concentrations of greenhouse gases, other forcings constant) and hist-aer (same as hist-GHG, but for anthropogenic aerosols), and we compare these to the respective CMIP6 historical simulations (all forcings varying) and observational datasets.

Robust AMOC weakening under hist-GHG and strengthening under hist-aer is found across the respective multi-model ensembles with various accompanying changes, exhibiting a high degree of spatial antisymmetry. In both sets of experiments, the same causal pathway (yet with opposite sign) occurs. We describe the key role of subpolar upper-ocean salinity and connect its variations to changes in sea ice and air–sea heat fluxes. Our results indicate that the primitive radiative forcing directly impacts sea-ice mass, and thereby drives upper-ocean salinity variations, while accompanying changes in surface freshwater fluxes further modulate salinity. The resulting variations in salinity induce changes in upper-ocean density and stratification in the subpolar NA that, in turn, determine the simulated AMOC trends. We further discuss key mechanisms in play, including the positive AMOC–salinity and AMOC–evaporation feedbacks, describing the dominant processes of the causal pathway.

By offering insights onto the respective roles of external forcings in the context of climate change and by advancing our understanding of key NA ocean–atmosphere interactions, our results also highlight models limitations in the representation of coupled processes that are critical for reliable projections.

How to cite: Giaquinto, D., Nicolì, D., Smith, D. M., Iovino, D., Frierson, D., and Athanasiadis, P. J.: Understanding AMOC changes resulting from varying historical radiative forcings, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17467, https://doi.org/10.5194/egusphere-egu26-17467, 2026.

A weakening of the Atlantic Meridional Overturning Circulation (AMOC) is often portrayed as economically beneficial - leading to a reduction in the social cost of carbon. The reduced social cost of carbon is attributed to the reduction of temperatures in large parts of the globe. However, the existing literature relies on integrated assessment models (IAMs) without an explicit representation of AMOC strength, and is therefore unable to consider the implicit AMOC weakening that is already included in projected temperature patterns. This study accounts for the amount of AMOC weakening that is implicit in pattern scaling procedures within the IAM when considering the effects of AMOC weakening. The implicit AMOC weakening is teased out from the pattern scaling as a function of global mean temperature change across CMIP6 models. Additionally, we recalibrate the temperature response to AMOC weakening at the country level by analysing simulations from the North Atlantic Hosing Model Intercomparison Project (NAHosMIP). The new temperature response, as well as four already implemented responses, are considered using the META IAM. We then analyse the change in social cost of carbon caused by AMOC weakening along seven different AMOC projections, taking into account the AMOC response implicit in pattern scaling. Overall, we find that AMOC weakening-induced temperature changes lower the social cost of carbon. Contrary to previous assumptions, this reduction in the social cost of carbon is driven only by global mean cooling, whereas the pattern of the temperature responses increases the social cost of carbon.

How to cite: Hansen, J., Alastrué de Asenjo, E., Schaumann, F., and Baehr, J.: Assessing the effect of AMOC-induced temperature patterns on the global social cost of carbon, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18114, https://doi.org/10.5194/egusphere-egu26-18114, 2026.

The Atlantic Meridional Overturning Circulation (AMOC) plays an important role in regulating the global climate. The AMOC change in response to global warming has important environmental and, potentially, societal impacts but remains an issue with large uncertainty. Here we use a series of coupled climate model experiments to reveal the overlooked role of Atlantic subtropical salinification, a robust consequence of an intensified hydrological cycle, in inhibiting AMOC weakening under global warming. Without subtropical salinification, the AMOC weakening more than doubles in response to a doubling of CO₂, primarily driven by a reduced zonal salinity gradient that weakens the geostrophic component of AMOC through the thermal wind relation. This larger AMOC weakening reduces surface warming in the Northern Hemisphere by as much as 1–3 K at northern high latitudes when subtropical salinification is inhibited.

How to cite: Liu, M., Soden, B., and Vecchi, G.: Greenhouse gas-induced Atlantic subtropical salinification partly offsets a large decline in the AMOC, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18907, https://doi.org/10.5194/egusphere-egu26-18907, 2026.

The collapse of the Atlantic Meridional Overturning Circulation (AMOC) has long been classified as a low-probability, high-impact event. However, recent evidence suggests the probability of such a collapse may be significantly higher than previously estimated. From a disaster and risk management perspective, this shift calls for a re-evaluation of preparedness strategies and a deeper inquiry into how a drastic weakening or a complete shutdown would reshape the global risk landscape.

Central to these concerns is the role of AMOC in modulating Northern Hemisphere precipitation. An anthropogenic weakening could significantly alter future drought dynamics, further complicating the management of drought risk, a hazard already characterised by extensive socio-economic impacts.

To address these changing dynamics, we examine four sets of paired climate model simulations, each comparing a weakened AMOC state with a control run featuring a stable, stronger AMOC. Three of these experiment pairs employ the EC-EARTH3.3 model, where freshwater perturbations in the North Atlantic induce an artificial AMOC slowdown under fixed pre-industrial, present-day (2025), and future (2050, SSP5-8.5) forcing. The fourth pair employs the NASA GISS ModelE, simulating a spontaneous AMOC collapse under an extended SSP2-4.5 scenario without external freshwater forcing. Using an advanced Meteorological Drought Tracking approach based on the Standardized Precipitation Index (SPI) we quantify shifts in drought duration, severity, and spatial coherence, highlighting where significant changes would be expected.

How to cite: Volpi, D., Acosta Navarro, J. C., Bellucci, A., Caporaso, L., Corti, S., Fioravanti, G., Hrast Essenfelder, A., Meccia, V. L., Romanou, A., Toreti, A., and Zampieri, M.: Understanding Drought Risk in the Northern Hemisphere under AMOC weakening, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20190, https://doi.org/10.5194/egusphere-egu26-20190, 2026.

The strength and variability of the Atlantic Meridional Overturning Circulation (AMOC) are closely linked to deep water formation (DWF) in three key regions: the Labrador Sea, the Irminger Sea, and the Greenland Sea. However, quantifying the relative contributions of these regions to the AMOC in climate models and how these contributions evolve under future climate scenarios remains challenging. While CMIP6 models consistently project a weakening of the AMOC, they show wide inter-model spread in the rate of decline. This highlights the need for robust metrics that enable more informative intercomparison. The commonly used mixed layer depth metric captures some aspects of convection, but does not directly quantify DWF. Here, we introduce a volume-conservation-based diagnostic that serves as an index for quantifying DWF, enabling robust comparison across models with differing resolution and complexity. It further quantifies the regional contributions of the Labrador, Irminger and Greenland Seas to the AMOC.  

When applied to EC-Earth3 at standard and high resolutions, the diagnostic suggests that DWF in the Labrador Sea is the main cause of the projected weakening of the AMOC. Meanwhile, the Irminger Sea emerges as the AMOC's largest overall contributor, experiencing only a modest decline and remaining essential for sustaining the circulation. At the same time, the contribution from the Arctic increases. We assess inter-model differences in DWF magnitude and examine their relationship to AMOC changes by extending the analysis to a suite of CMIP6 models. This allows us to evaluate the robustness of these processes across models. Overall, our results provide new insight into the factors underlying differences in AMOC projections among models and into the mechanisms that may influence the risk of an AMOC slowdown or tipping point.

How to cite: Karami, M. P., Navarro-Labastida, R., Koenigk, T., and Chafik, L.: A diagnostic framework for deep water formation and AMOC variability in selected CMIP6 models, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20496, https://doi.org/10.5194/egusphere-egu26-20496, 2026.

It has been shown that predictability of the North Atlantic Oscillation (NAO) in seasonal forecasts is better than models suggest, a consequence of the signal-to-noise paradox, whereby individual ensemble members contain a smaller proportion of the predictable variance than seen in observations. We intend to use two seasonal forecast models, GloSea6 and CESM-SMYLE, to study whether ‘NAO-matching’, where we select only the ensemble members that most closely resemble the ensemble mean NAO, can produce more accurate seasonal forecasts of the Atlantic Meridional Overturning Circulation (AMOC) than the full seasonal forecast ensemble. This method has been shown to improve predictability of other aspects of the North Atlantic climate, such as the Atlantic Multidecadal Variability pattern and Northern European Precipitation. The skill of AMOC predictability in seasonal hindcasts will be assessed against the RAPID array observations as well as historical reconstructions of the overturning circulation to determine whether it too is subject to signal-to-noise errors, and consequently if ‘AMOC-matching’ is a potentially useful calibration tool for improving predictability of its related climate impacts.

How to cite: Hay, S., Walsh, A., Screen, J., Scaife, A., and Robson, J.: Leveraging the signal-to-noise paradox to improve seasonal forecasts of the AMOC and its impacts, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21430, https://doi.org/10.5194/egusphere-egu26-21430, 2026.

Abrupt changes in the Atlantic meridional overturning circulation (AMOC) can cause dramatic global climate changes, but their impacts on tropical hydroclimate remain uncertain. Heinrich Stadial 1 (HS1, ~18 to 14.5 thousand years ago) involves the largest AMOC reduction in recent geological time, providing a unique opportunity to investigate the influence of AMOC reduction on tropical hydroclimate. Our proxy data-model simulation synthesis reveals a zonal hydroclimate mode characterized by widespread drought in tropical East Africa and generally wet but spatially heterogeneous conditions in the Maritime Continent, analogous to the modern negative phase of the Indian Ocean Dipole. We propose that North Atlantic cooling associated with a weakened AMOC drives millennial-scale tropical Indian Ocean hydroclimate variations by affecting both the latitudinal position of the ITCZ, and the strength of the Indian Ocean Walker circulation.

In addition, we conducted new sensitivity experiments using iCESM1.3 that show glacial boundary conditions, especially changes in sea level and the exposure of Sunda and Sahul shelves, strongly modulate the tropical Indian Ocean response to North Atlantic cooling by altering the background state and interannual SST variability in the tropical Indian Ocean. Furthermore, we explored the response of the tropical Indian Ocean to a potential AMOC weakening under a global warming scenario and its relationship to interannual variability over the Indian Ocean to assess the implications for future climate change and extreme events in this region.

How to cite: Du, X., Russell, J., Liu, Z., Otto-Bliesner, B., Zhu, J., Zhu, F., and Zhu, C.: Response of the tropical Indian Ocean to past AMOC weakening and implications for the future, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-22186, https://doi.org/10.5194/egusphere-egu26-22186, 2026.

The Atlantic Meridional Overturning Circulation (AMOC) is a key component of the climate system, yet the consequences of a pronounced weakening under emission pathways consistent with the Paris Agreement remain poorly understood. Using the comprehensive GFDL ESM2M Earth System Model with the Adaptive Emissions Reduction Approach, we impose a freshwater-induced strong AMOC weakening to 20% of its preindustrial strength starting in year 2026. These simulations otherwise follow a pathway in which global warming stabilizes at 2°C and the AMOC weakens only modestly and partially recovers. Relative to the modest-weakening scenario, a strong AMOC weakening cools global mean surface air temperature by −0.8°C (5-member ensemble range: −0.7 to −0.9) by 2171-2200, with pronounced regional cooling in the North Atlantic, reaching up to −6.8 °C (−4.1 to −9.7) in winter over Iceland. The ocean stores an additional 385 ZJ (331–428) of heat, primarily south of 20°N, associated with reduced northward heat transport and enhanced heat uptake in the North Atlantic. The additional heat increases global thermosteric sea level rise by 10% (8–12). Atmospheric CO2 declines by 13 ppm due to anomalous land carbon uptake of 44 GtC (33–53), dominated by enhanced carbon storage in the Amazon under cooler and wetter conditions. In contrast, global ocean carbon storage decreases by 14 GtC, mainly north of 20°N, although carbon uptake increases in the northern North Atlantic. The AMOC-induced cooling breaks the near-linear relationship between cumulative CO2 emissions and warming, increasing the remaining carbon budget for limiting warming to 2°C by 63% (54–72). Compared to identical freshwater forcing under preindustrial conditions, the surface temperature, ocean heat content, and sea-level responses are substantially damped, indicating reduced climate sensitivity to AMOC collapse in a warmer world. These results demonstrate that a strong AMOC weakening would profoundly alter future climate–carbon cycle interactions and underscore the importance of explicitly accounting for AMOC risks in long-term climate assessments.

How to cite: Frölicher, T. L., Maier, P., Burger, F. A., Silvy, Y., Swingedouw, D., and Elizondo, U. H.: Climate and Carbon Cycle Responses to a 21st century AMOC collapse under a 2°C stabilization pathway, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-22272, https://doi.org/10.5194/egusphere-egu26-22272, 2026.

In the summer of 2024, Nam Co, one of the oldest lakes on the Tibetan Plateau, was the focus of the ICDP NamCore scientific drilling campaign aimed at reconstructing the Quaternary climate history of the region. Within this framework, the SNSF-funded DIGESTED project investigates biosphere-geosphere interactions across the entire lake system, encompassing water column conditions and deep sedimentary records. By integrating sedimentology, lake physics, biogeochemistry, and microbiology, the project seeks to assess the extent to which biological processes influence the sedimentary archive used to reconstruct paleoclimates and understand the ecological trajectory of the lake over the past million years.

We present the biogeochemical results from modern waters, recent and ancient sediments from the drill site. The water column is fully oxidized, with oxic conditions extending  8 cm below the sediment-water interface. Below this zone, microbially produced methane (supported by C and H isotopic ratios) shows a successive increase (0 to 7.8 mmol/L) to a depth of ~80 mblf. Methane is abundant in measurable quantities down to a depth of ~250 mblf, which marks a change in lithology from sand to non-calcareous mud. Biomarker ratios associated with methane cycling indicate a pronounced shift in microbial activity at this depth. Both the GDGT-0/Crenarchaeol ratio and the methane index (Zhang et al., 2011) increase sharply at and below 200 m, consistent with limited methanogenesis and methanotrophy above this boundary and substantially more active microbial processes below, despite the absence of detectable methane. This transition also coincides with changes in the composition of preserved and extractable subsurface microbial DNA. Our 16S rRNA gene sequence analyses reveal communities associated with fermentation and C1-based metabolisms below 200 m, whereas sediments above this depth are dominated by archived or transported taxa that are rarely active in such anoxic sedimentary environments.

With this study, we begin to piece together how microbial processes and their suppression, fluid migration, and paleoenvironmental conditions collectively shape the integrity of this climatic archive. A pronounced lithological and biogeochemical boundary at ~200 m separates a likely once-active methane cycling system from an overlying, energy-limited deep biosphere that permits methane accumulation and slow diffusive transport toward geological boundaries. Our ultimate goal is to disentangle the paleoenvironmental conditions leading to such strong shifts by coupling an age model with sedimentological, chemo-physical, and biological characterization of those archives.

Zhang, Y. G., Zhang, C. L., Liu, X.-L., Li, L., Hinrichs, K.-U., & Noakes, J. E. (2011). Methane Index: A tetraether archaeal lipid biomarker indicator for detecting the instability of marine gas hydrates. Earth and Planetary Science Letters, 307(3), 525–534. https://doi.org/10.1016/j.epsl.2011.05.031

How to cite: Thomas, C., Ceriotti, G., Raemy, E., Kou, Q., Bauersachs, T., Laakkonen, A., Shore, M., Adolph, M.-L., Moser-Roeggla, P., Picard, M., Schubert, C. J., Kipfer, R., Berg, J., Henderson, A. C. G., Clarke, L., Zhu, L., Wang, J., Ju, J., Haberzettl, T., and Vogel, H.: Deep subsurface shifts in microbial processes in Nam Co (Tibet) revealed by multidisciplinary investigations of an ICDP drill core, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3499, https://doi.org/10.5194/egusphere-egu26-3499, 2026.

Hadal ocean trenches are among the least explored environments on Earth, yet they host the largest and most hazardous earthquakes. Formed at subduction zones where megathrust earthquakes and tsunamis originate, hadal trenches act as terminal sinks for sediment and carbon. Because instrumental and historical records are too short to capture the full range and recurrence of giant (Mw ≥9) earthquakes, hadal trench basins provide unique geological archives to reconstruct long-term earthquake behavior, including rare slip-to-the-trench events that generate large tsunamis. These basins also host extreme subseafloor ecosystems and play an unresolved role in Earth’s carbon cycle, making them key targets for integrated scientific ocean drilling and Earth system research.

IODP³ Expedition 503 (November–December 2025) drilled a trench-fill basin in the central Japan Trench at Site C0028 using the D/V Chikyu. Coring in five holes at water depths of up to 7,608.5 m reached a maximum depth of 178 m below seafloor (mbsf), recovering a complete trench-fill succession and providing the first continuous full record from the depositional center of a hadal trench basin. Initial results demonstrate that drilling successfully penetrated the full trench-fill sequence and its base. Lithostratigraphy documents a systematic transition from basal volcaniclastic-rich deposits to mixed detrital sediments and overlying biosiliceous oozes, reflecting basin initiation, growth, and progressive migration toward the trench axis. Structural data show increasing bedding dips and a normal-fault regime in the lowermost section, consistent with horst-and-graben formation related to bend faulting of the incoming Pacific Plate. An angular unconformity at depth, together with paleomagnetic observations and initial stratigraphic correlations to IODP and DSDP sites sampling the sedimentary cover of the Pacific oceanic crust, confirms recovery below the trench-fill base.

Event stratigraphy is exceptionally well preserved. Numerous thick turbidites, replicated between holes and tied to seismic reflectors, form a robust framework for paleoseismic interpretation. Distinct variability patterns in radiolarian fossil taxa abundances, together with frequent tephra layers, provide strong potential for high-resolution chronological control. Paleomagnetic data indicate a polarity reversal in the deepest cores, tentatively correlated with the Matuyama–Brunhes boundary (~773 ka), implying that the recovered sequence spans several hundred thousand years.

Geochemical analyses largely confirm previous results from giant piston coring during IODP Expedition 386 in 2021 down to a depth of approximately 40 mbsf. A decrease in alkalinity, previously hypothesized from shallow subsurface records, is confirmed, with significant changes in pore-water profiles observed below ~80 mbsf down to the base of the trench-fill sequence. Integrated sedimentological, mineralogical, physical property, headspace gas, and pore-water data document depth-dependent reaction zones, compaction trends, and early diagenesis linked to dynamic element cycling in the hadal subseafloor. Importantly, Expedition 503 successfully recovered high-quality core material suitable for microbiological investigations, enabling assessment of subseafloor microbial activity and its coupling to geochemical processes.

Together, these initial results demonstrate that hadal trench basins preserve long, continuous archives of tsunamigenic megathrust behavior and associated biogeochemical processes, opening new perspectives on earthquake recurrence, geohazards, and carbon cycling along subduction zone systems.

How to cite: Strasser, M., Ikehara, K., and Maeda, L. and the IODP3 Expedition 503 Scientists: Recovering the Long-Term Record of Subduction-Zone Tsunamigenic Slip and Element Cycling in a Hadal Trench Basin at the Japan Trench: Initial Results of IODP³ Expedition 503, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4209, https://doi.org/10.5194/egusphere-egu26-4209, 2026.

A central objective of the NamCore ICDP project is to understand Quaternary biotic dynamics—specifically species diversity, distribution, and evolution—in relation to Asian monsoon variability and orbitally driven climate change. Lacustrine ostracodes are therefore ideal indicators to assess (1) whether Nam Co served as a glacial refugium for cold-adapted species during glacial periods, (2) how biota responded to glacial–interglacial environmental transitions, and (3) whether the lake exhibits a high ecological resilience to environmental change.

To address these objectives, a multi-scale analytical approach was applied. Ostracode valve analyses were conducted on 43 core catcher samples spanning depths from 8 m to 470 m b.l.f., corresponding to a stratigraphic resolution represented by intervals of 3–35 m, to provide an overview of broad-scale changes in ostracode distribution and abundance. To obtain higher-resolution data on species distribution and morphological variability, additional samples from core sections within the upper 33 m b.l.f. were analyzed at 16 cm intervals. Morphometric analyses of valve outline shape and size are intended to identify either gradual or abrupt changes in morphological variability. Environmentally driven morphological responses are expected to manifest as gradual shifts in size and/or shape, whereas re-colonization from other lakes may produce distinct morphological signatures, resulting in discontinuous variation in size or shape.

Preliminary results indicate that ostracode abundance and species composition are highly variable, with ostracodes absent below 470 m b.l.f. In total, ten species were identified, with a maximum of five species per sample. Generally, samples from the uppermost 30 m contain four species that are absent in the lower sections of the record. Although Leucocytherella sinensis and ?Leucocythere dorsotuberosa represent the most abundant taxa, no species occurs continuously throughout the sedimentary record.

Detailed analyses of species composition, combined with morphometric investigations, are expected to elucidate whether the discontinuous ostracode distribution pattern reflects repeated lake colonization events associated with, e.g. glacial–interglacial cycles. Such findings would have significant implications for understanding the role of the Tibetan Plateau as a biodiversity refugium during Quaternary climate oscillations and for reconstructing paleoenvironmental conditions from ostracode assemblages in high-altitude lake systems.

How to cite: Wrozyna, C., Hoehle, M., Adolph, M.-L., Frenzel, P., Schmitz, O., Clarke, L., Henderson, A. G., Vogel, H., Wang, J., Zhu, L., and Haberzettl, T.: Lacustrine ostracodes from Nam Co (Tibetan Plateau) indicate biotic responses to Quaternary climate change (NamCore ICDP project), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5230, https://doi.org/10.5194/egusphere-egu26-5230, 2026.

How can rocks obtained by scientific drilling increase our understanding of deformation, stress, and strength in one of Earth’s classic collisional orogenic belts? By integrating scientific drilling data with high-resolution laboratory measurements, this study presents a combined structural, mineralogical and geomechanical characterization of the Scandinavian Caledonides, based on data from the COSC-2 borehole acquired during the ICDP logging campaign in 2022 . From the surface to approximately 1200 m depth, the borehole intersected an extensive Early Phanerozoic sedimentary succession, dominated primarily by wacke, shale and siltstone. Beneath this succession, extending to 2276 m, lies a crystalline basement comprising a volcanic sequence, including porphyry, gabbro and gabbroid rocks, intruded by dolerite dykes. The contact between the sedimentary succession and the crystalline basement is relatively undisturbed, with a thin regolith covering the altered top of the porphyry.

A key objective of our study is to investigate the physical properties and in-situ stress state of the COSC-2 rocks using laboratory tests on selected core samples. Specifically, we examine how stress magnitudes vary with depth, which stress regime dominates the area, how rock stiffness varies with lithology, mineralogy, and depth, and whether laboratory-derived elastic properties are consistent with downhole sonic log measurements (Vp and Vs). To address these questions, a suite of laboratory measurements was conducted on 19 core samples, including Brazilian tensile strength (BTS), uniaxial compressive strength (UCS), P- and S-wave velocities, Poisson’s ratio, Young’s, bulk, and shear moduli, grain and bulk density, and quantitative mineralogical analyses using X-ray diffraction (XRD) and electron microprobe analysis (EPMA). Our findings show that crystalline rocks exhibit in general a higher stiffness and compactness, reflected in elevated wave velocities and elastic moduli, combined with greater densities and lower porosity resulting in greater mechanical strength, both in compression and tension loading. This behaviour is reflected in specific samples, which record some of the highest BTS and UCS values. In contrast, three samples in doleritic and gabbroic rocks display unexpectedly low BTS values (19–20 MPa) and UCS values (180–211 MPa) compared to the other crystalline basement samples. Analysing the mineralogical composition, we found the presence of primary and secondary phyllosilicates in these rocks, which likely weaken the rock fabric and can be responsible for the reduced strength. In contrast, the overlying sedimentary rocks exhibit lower stiffness and strength but greater variability largely controlled by porosity and internal heterogeneity.

Of course, such geomechanical properties are also controlled by the presence of microcracks, open and cemented veins, mineral alignment and the precipitation of secondary minerals reflecting enhanced fluid flow and fluid-rock interaction processes. Especially the occurrence of secondary mineral phases identified through XRD and EMPA further reveal a complex tectono-metamorphic history. Together, these findings provide a solid framework for geomechanical modelling and advance our understanding of the evolution of collisional orogens.

How to cite: Pierdominici, S., Lopera Restrepo, A. P., Kottkamp, W., Schleicher, A. M., Wilke, F. D. H., and Schmitt, D. R.: The influence of geomechanical properties on rock strength in the ICDP COSC-2 borehole, at Are, Sweden, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5610, https://doi.org/10.5194/egusphere-egu26-5610, 2026.

SPARC (Scientific Projects using Ocean Drilling Archives) is an IODP³ programme to utilize legacy cores by large-scale research groups. Three projects (Exp. 504S, Exp. 505S, Exp. 506S) were launched in the first year, with a start date in summer or fall and will last for three years.

The North Atlantic plays a crucial role in regulating global climate due to its proximity to major ice sheets and sensitivity to changes in the Atlantic Meridional Overturning Circulation (AMOC). Over millennial and orbital timescales, the region has experienced abrupt climate shifts with significant global implications. The Exp. 506S SIGNALS (Stratigraphic InteGration of North Atlantic Legacy Sites) project aims to synthesize and integrate legacy records into a coherent, four-dimensional stratigraphic framework to provide a regional reconstruction of past climate variability on millennial to orbital timescales since the late Miocene.

SIGNALS will enhance stratigraphic correlation, refine age models, and synchronize proxy datasets for multiple legacy sites across the North Atlantic spanning a wide range of climatic and bathymetric gradients. The project will capitalize on advanced methods, including machine learning and signal correlation algorithms, to rapidly produce high-resolution data by automated processing of core images, point counting, and precise stratigraphic correlation.

SIGNALS will address methodological issues associated with estimating uncertainty in stratigraphic correlations and the limits of temporal resolution at each site given varying sedimentation rates, bioturbation, and sampling frequency. Furthermore, we will develop process models to understand how orbitally-driven climatic changes are expressed as cycles in the stratigraphic record of each site. By analyzing high-resolution geochemical and sedimentological proxies in a robust stratigraphic framework, the project seeks to reconstruct climate evolution and ocean circulation changes across the North Atlantic since the late Miocene. The project will focus on major climatic transitions and provide robust regional paleoclimate data for numerical modeling and assimilation studies. Beyond research advancements, SIGNALS will also foster collaboration by developing user-friendly computational tools, training early-career researchers, and making data publicly accessible through open repositories.

Although the exact implementation plan will not be decided until the science team has been selected, we will present objectives and general plans of Exp. 506S SIGNALS as one of the first SPARC projects.

How to cite: Seki, A., Hodell, D., Herbert, T., Obrochta, S., and Voelker, A.: Perspectives of IODP3 Expedition 506S SIGNALS - Stratigraphic InteGration of North Atlantic Legacy Sites, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6279, https://doi.org/10.5194/egusphere-egu26-6279, 2026.

The Tibetan Plateau (TP), often referred to as the “Third Pole” and “Asian Water Tower”, serves about two billion people downstream with its water resources; thus, investigations of past climate changes on the TP have significant socio-economic implications for both the scientific community and governmental concerns. Numerous lakes on the plateau provide valuable archives to carry out paleoenvironmental change studies on different time scales by drilling sediment cores. With the support of the ICDP (NamCore project, Expedition 5073) and other funding, we accomplished a drilling campaign in a high-altitude, deep lake (Nam Co, 4718 m) on the Tibetan Plateau in the summer of 2024. In total, ~950 m of cores was recovered from seven holes at one site, with a deepest drilling depth of 510 m b.l.f., making NamCore a great success among ICDP lake drilling projects in the past several decades with respect to its altitude and maximum penetration depth. These achievements enable us to study past climate changes in this area potentially back to ~1 Ma and their linkages with other regions globally. Three core opening and sampling parties of the NamCore project have been organized in Beijing, where the cores were stored, to complete the splitting of all cores. Core descriptions, magnetic susceptibility scanning of the entire sequence combined with other analyses (e.g., grain size, organic/inorganic carbon content, biomarkers and pollen, etc.) on core catcher samples have revealed sediment variations, which can distinctly show the fluctuations between glacial and interglacial cycles, although the chronology using various approaches is still challenging. The results show four major lithologies throughout the drilled cores including calcareous mud, non-calcareous mud, sand and calcareous mud with ferric staining. Calcareous mud dominates the upper ~120 m and ferric-stained mud mainly appear in the sections deeper than ~320 m. Many sand layers with different thickness occur in the entire sequence but mostly in the middle part. Nothing has been retrieved in a section greater than 30 m in thickness in the lower part, which probably indicates a remarkable change in the sedimentary environment associated with a glacial period. Time series analysis on the magnetic susceptibility data shows two prominent cycles at 10.1 m and 21.4 m, which potentially correspond to the orbital precession and obliquity forcing of 21 ka and 41 ka, respectively. This cyclostratigraphic approach will be helpful to constrain the chronology and, by comparison with stalagmites in monsoonal areas and ice cores in polar regions, plays an important role in discovering the different drivers of climate change from low and high latitudes. However, more efforts are still needed to obtain absolute ages to establish a precise timeframe for these cores.

How to cite: Wang, J., Adolph, M.-L., Cidan, Z., Zhu, L., Haberzettl, T., Vogel, H., Clarke, L., Henderson, A., Spiess, V., Ju, J., Ma, Q., Kou, Q., and Daut, G.: Scientific Drilling on the Third Pole: achievements of the highest ICDP lake drilling project on the Tibetan Plateau (Nam Co, 4718 m.a.s.l), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7514, https://doi.org/10.5194/egusphere-egu26-7514, 2026.

  During the last 5 Ma (Pliocene-Holocene) the Earth’s climate system has undergone a series of marked changes, including; (i) the shift from the warm state Pliocene to the cold state Pleistocene, (ii) the evolution in frequency, magnitude and shape of glacial-interglacial cycles at the Early Middle Pleistocene Transition (~1.25-0.65 Ma), and (iii) the appearance of millennial-scale climate variability. While much of this paleoclimatic narrative has been reconstructed from marine proxy records, relatively little is known about the expressions of these major changes in continental areas and their impact on terrestrial environments and biodiversity, thus resulting in a significant knowledge gap surrounding a fundamental component of the Earth’s climate system. In the framework of the Mediterranean area, a region that is sensitive to changes in temperature and hydrological cycle, the Fucino Basin in Central Italy stands out as one of the few sites that meets the necessary requirements to fill this gap. The geophysical evidence and the stratigraphical, geochronological and multi-proxy data for multiple sediment cores acquired in recent years, indicate that the Fucino lacustrine succession (i) spans continuously for at least 4.6 Ma, (ii) is highly sensitive to climate change, and (iii) contains an outstanding number of volcanic ash layers, which facilitate an independent, high-resolution time-scale. With respect to the half-graben, wedge-shape geometry of the basin, three drilling targets were identified: MEME-1, located in the middle of the basin, would intersect the whole Quaternary infill and the upper part of the Pliocene continental sequence at ~400-500 m depth; MEME-2, which is located ca. 1.8 km west of MEME-1, where the sedimentation rate is lower, and is ~400-500 m deep, allows recovery of the entire Pliocene-Quaternary infill reaching the Messinian substratum; MEME-3 (~250-300 m depth), located for tectonics objectives on the footwall of the basin master fault and covering, though discontinuously, the lake history back to ~4.6 Ma. Through a multi-method dating approach, and a multi-proxy analysis of sedimentary physical and biogeochemical properties, the MEME project will provide a detailed record of changes in the Earth climate system and the environmental-ecological response, independent of any a priori assumptions on response times to climate forcing and feedback mechanisms. Furthermore, the Fucino sedimentary succession has enormous potential to reconstruct a uniquely comprehensive long-term, high-temporal resolution record of peri-Tyrrhenian explosive volcanism and of the post-orogenic extensional tectonics in this area of the Apennine chain.  

How to cite: Giacco, B. and the MEME Team: ICDP Fucino paleolake project: the longest continuous terrestrial archive in the MEditerranean recording the last five Million years of Earth system history (MEME), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8013, https://doi.org/10.5194/egusphere-egu26-8013, 2026.

The Middle Miocene represents one of the warmest intervals in Earth’s recent geological history. Understanding the climate dynamics of this period can provide valuable insight into how the climate system may respond to future anthropogenic forcing. The study of tropical regions during the Miocene is particularly important, because these environments are underrepresented in climate records. Since the Miocene, Australia’s climate has undergone substantial changes driven by the northward drift of the continent, its collision with Southeast Asia, and the associated reorganisation of oceanic circulation around the Maritime Continent. Northern tropical Australia is presently influenced by a monsoonal system that forms part of the broader Asian Monsoon; however, the Australian Monsoon remains poorly understood, particularly with respect to its onset and variability. Investigating monsoon dynamics across different climatic states in this region may therefore improve our understanding of how large-scale circulation patterns could evolve under anthropogenically driven climate change. As a result of persistent arid conditions and lack of tectonic subsidence, evidence of fluvial and lacustrine activity has been destroyed, meaning terrestrial records of palaeoclimatic change in Australia are sparse. This study focuses on a marine core: Ocean Drilling Program (ODP) Hole 1195B on the Marion Plateau. Hole 1195B preserves an erosional and oceanographic record extending back to ~21 Ma and provides an opportunity to examine links between climate variability and continental weathering since the Middle Miocene.

This study employs XRF core scanning, GDGT biomarker analysis, and elemental analysis using ICP-OES. The results indicate that the highest delivery of clastic material to the Marion Plateau occurred during the Miocene Climatic Optimum (~17 Ma), coinciding with peak sea surface temperatures. The most pronounced cooling is observed between 11 and 9 Ma and was accompanied by significant changes in sediment input to the site. These changes were likely associated with shifts in Coral Sea circulation, potentially reflecting a strengthening of the East Australian Current. Notably, this regional response occurs ~2 Myr after the global cooling event observed elsewhere at ~13 Ma, suggesting that tropical climate systems may respond independently, or with some delay, in comparison to global climate perturbations. This highlights the importance of understanding climate dynamics in the tropics when considering potential future responses to anthropogenic climate change.

How to cite: McGanity-Smith, B., Clift, P. D., and Petrick, B.: Sediment cycling on the Marion Plateau since the Miocene, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10351, https://doi.org/10.5194/egusphere-egu26-10351, 2026.

How to cite: Rulff, P., Abels, H., Fulton, P., and Bretaudeau, F. and the extended SEE-MORE team: Towards SEE-MORE - A multi-use borehole for optimisation of Subsurface Energy Exploration and MOnitoring of low-enthalpy geothermal REsources , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10916, https://doi.org/10.5194/egusphere-egu26-10916, 2026.

The mineral abundance, their properties and geometrical arrangement on small spatial scales directly affect the physical characteristics of the continental crust at large scales. Consequently, the mineral assemblages determine to a large extent how geophysical methods respond to these rocks. Determining the mineral volume fractions is an essential first step for modelling and interpreting geophysical data, constraining crustal structure, and understanding the evolution of the Earth’s lithosphere. In this study, we develop a Bayesian inversion framework that integrates petrophysical information from downhole well logs and multi-sensor core logging data with X-ray fluorescence (XRF) data to estimate continuous mineral fraction profiles along two ICDP-DIVE boreholes (Greenwood et al. 2026) drilled through the exhumed lower continental crust of the Ivrea–Verbano Zone (IVZ) with almost 100% core recovery. The framework involves two schemes: (1) an overdetermined inversion of relative sparse XRF oxide weight fraction data from powdered rock samples combined with core density logs, and (2) a severely underdetermined inversion of potassium, magnetic susceptibility, and core density logs, conducted by groups derived from a cluster analysis of these logs. The latter scheme is constrained by the first scheme, which allows to retrieve a continuous mineral fraction estimates along both boreholes from the limited number of 3 petrophysical logs. An ensemble Markov Chain Monte Carlo algorithm (Cheng et al. 2022) is adapted to recover the posterior mineral fraction distributions while quantifying uncertainties. An essential input is the prior knowledge of the minerals present and their chemical formula, which may require supplementary measurements, especially for minerals such as amphibole, whose chemical formula is difficult to determine. The results show that the XRF Oxide–density inversion approach provides robust mineralogical estimates that are consistent with independently obtained modal estimates from section observations. The constrained inversion of the petrophysical logging data successfully captures mineral fractions across most lithologies despite the underdetermined nature of the problem. The study demonstrates that combining XRF-derived oxide fractions with continuous downhole and core logging data within a Bayesian framework provides a powerful approach for obtaining quantitative, mineral fractions in a range of lower crustal lithologies.

Cheng, L., Jin, G., Michelena, R., & Tura, A. (2022). Practical Bayesian Inversions for Rock Composition and Petrophysical Endpoints in Multimineral Analysis. SPE Reservoir Evaluation & Engineering, 25(04), 849–865. https://doi.org/10.2118/210576-PA

Greenwood, A., Venier, M., Hetényi, G., Ziberna, L., Heeschen, K., Pacchiega, L., Lemke, K., Dutoit, H., Bonazzi, M., Degen, S., Li, J., Secrétan, A., Trabi, B., Tholen, S., Lefeuvre, N., Auclair, S., Mariani, D., Del Rio, M., Černok, A., Bhattacharyya, A., Narduzzi, F., Mansouri, H., Urueña, C., Beltrame, M., Hawemann, F., Velicogna, M., Toy, V., Dominique, J., Longo, A., Tonietti, L., Barosa, B., Brusca, J., Nappi, N., Gallo, G., Esposito, M., Diana, S. C., Bastianoni, A., Eckert, E. M., Confal, J. M., Pondrelli, S., Piana Agostinetti, N., Tertyshnikov, K., Caspari, E., Truche, L., Wiersberg, T., Baron, L., Giovannelli, D., Pistone, M., Zanetti, A., Müntener, O. (2025): Drilling the Ivrea-Verbano zonE: DIVE 1 – ICDP Operational Report, Potsdam: GFZ Data Services, 109 p. doi:10.48440/ICDP.5071.001

How to cite: Li, J., Secrétan, A., Degen, S., Caspari, E., Greenwood, A., Venier, M., Lemke, K., Ziberna, L., Hetényi, G., and Müntener, O.: Integrating Petrophysical Logging and XRF Data for Mineral Fraction Estimation of Lower Crustal Rocks from the ICDP-DIVE Project using a Bayesian Inversion Framework, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11216, https://doi.org/10.5194/egusphere-egu26-11216, 2026.

The Trans-Amazon Drilling Project (TADP) aims to reconstruct the Amazonian physical landscape, climate, and rivers during the Cenozoic, in parallel with the evolutionary history of the tropical forests. Scientific drilling was carried out in the western Acre Basin and the eastern Marajó Basin to recover Cenozoic sediments up to 2000 m and 1280 m depth, respectively. Multiple episodes of drill-string imprisonment hindered the achievement of target depths, none-the-less the TADP recovered an 870-m drill core (TADP-1A) in the Acre Basin (923 m depth) and a 735-m drill core (TADP-2A) in the Marajó Basin (924 depth) between June 2023 and September 2024. Each core comprises a sequence of poorly consolidated sandstones, siltstones, and mudstones representing Amazonian fluvial sedimentation during the Late Cenozoic. Sandstones and mudstones, respectively, of the Acre Basin are distinctive in their immature feldspathic composition and intense paleopedogenesis in comparison with analogous facies of the Marajó Basin. The TADP-1A core was described and sub-sampled for laboratory analyses, whereas the detailed description and sub-sampling of the TADP-2A core is scheduled for July 2026. This presentation will describe drilling operational issues, outreach activities, and initial results from ongoing geochronologic, geochemical, mineralogical, geophysical, and biotic analyses of the TADP-1A core. 

How to cite: Sawakuchi, A., Fritz, S., Baker, P., Silva, C., Noren, A., Jaramillo, C., Bezerra, I., Martinez, A., and Garcia, M. D. G.: Drilling operations and initial results of the Trans-Amazon Drilling Project (TADP), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11871, https://doi.org/10.5194/egusphere-egu26-11871, 2026.

The C-Drill Core Repository, currently under development at the CNR - Territorial Research Area of Rome 1, is designed to address this gap by establishing a national reference facility for the conservation, management and scientific reuse of continental drilling cores. The repository will accommodate approximately 50–70 km of cores stored in controlled environments (+4 °C, room temperature, −20/−80 °C), supported by high-density mobile racking, automated climate control and continuous environmental monitoring. The facility will also host dedicated laboratories for core splitting and handling, high-resolution and multispectral imaging, XRF scanning, physical property logging (MSCL), wet and dry sample preparation, and microscopy. These capabilities will enable non-destructive characterisation and advanced analytical workflows directly linked to the archived materials.

A fully integrated digital workflow will be implemented, including systematic core digitisation, LIMS-based traceability, a standardised sample request system and interoperability with international data repositories in compliance with FAIR principles (e.g. PANGAEA, EarthChem, mDIS).

Conceived as a modular and scalable infrastructure, C-Drill will ensure high standards of climatic stability, data integrity, safety and user accessibility. The repository will provide services to national research institutes, universities and public agencies; support training activities for early-career scientists and technical staff; and generate interoperable datasets aligned with ICDP/IODP³, EPOS and other international frameworks.

By overcoming the current fragmentation of continental core archives in Italy, C-Drill will harmonise procedures for core acquisition, documentation and access with the best practices of leading European and international repositories. At the same time, it will enhance research efficiency, foster multidisciplinary collaboration and strengthen Italy’s capacity to participate in major international scientific drilling initiatives.

This contribution presents the design rationale, functional requirements, technological solutions and planned user services of the C-Drill Core Repository. The EGU platform will be used to engage potential partners, gather community feedback and refine the development roadmap towards full integration into the international scientific drilling network.

How to cite: Iadanza, A., Tentori, D., Mazzini, I., and Giaccio, B.: A national hub for continental scientific drilling: the C-DRILL core repository at CNR (Italy), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12539, https://doi.org/10.5194/egusphere-egu26-12539, 2026.

Ostracods from sediment cores of the ICDP NamCore drilling project were analysed to document their distribution, abundance, and preservation, and to explore their potential for reconstructing past lacustrine conditions on the central Tibetan Plateau. Selected core catcher samples covering sediment depths from ~8 to 470 m were investigated for their microfossil content. Sediment samples (10–15 g each) were wet-sieved at 63 µm and 200 µm, and the >200 µm fraction was examined under a stereomicroscope. Ostracods were assessed semi-quantitatively and assigned to five abundance categories (absent, 1-10, >10, >100, >1000 valves per sample). Preservation states were evaluated qualitatively, and taxonomic identifications were based on established regional faunal keys.

Ostracods represent the only fossils observed in the sand-sized fraction of the analysed samples. Their abundance varies strongly, ranging from complete absence to more than 1000 valves per sample, with approximately half of the samples containing more than 100 valves. Preservation is generally good, although weakly etched, fragmented, compacted, or deformed valves occur, particularly below ~180 m core depth. Assemblages are of low diversity, with a maximum of five species per sample. At least six ostracod taxa were identified, including Leucocytherella sinensis, ?Leucocythere dorsotuberosa, ?Leucocythere postilirata, Ilyocypris ?bradyi, a smooth Ilyocypris species of uncertain taxonomic status, and juvenile Candona spp. The taxonomic assignment of ?Leucocythere dorsotuberosa and the smooth Ilyocypris species is the subject of ongoing investigations.

Variations in ostracod abundance, species level assemblage composition, and variable preservation suggest changes in depositional and post-depositional conditions through the core. While the presence of ostracods throughout most sections is consistent with predominantly lacustrine settings, intervals with low abundances or poor preservation may reflect a range of factors, including lake-level changes, sedimentation dynamics, or taphonomic overprinting. Further quantitative analyses, improved taxonomic resolution, and integration with independent proxies are required to refine palaeoenvironmental interpretations.

How to cite: Schmitz, O., Frenzel, P., Pint, A., Adolph, M.-L., Clarke, L., Henderson, A., Vogel, H., Wang, J., Zhu, L., Höhle, M., Wrozyna, C., and Haberzettl, T.: Ostracods from sediment cores of the ICDP NamCore drilling project provide insights into long-term lacustrine evolution on the Tibetan Plateau, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13464, https://doi.org/10.5194/egusphere-egu26-13464, 2026.

The Trans-Amazon Drilling Project (TADP) provides a unique opportunity to investigate subsurface light-hydrocarbon dynamics in Amazonian sedimentary basins through the integration of continuous real-time gas monitoring during drilling, discrete gas sampling from sediment cores, and laboratory incubation experiments. This study combines gas geochemical data from Cenozoic sediments of the Acre Basin in western Amazonia and the Marajó Basin in eastern Amazonia, to characterize the occurrence, composition, origin, migration of light gaseous hydrocarbons, and potential microbial production of methane (CH₄), within continental siliciclastic successions, contributing to a refined understanding of subsurface carbon cycling.

In the Acre Basin, drilling penetrated a 923-m-thick sedimentary sequence dominated by interbedded claystones, siltstones, and sandstones. Continuous online gas analysis (OLGA) revealed CH₄ as the dominant hydrocarbon throughout the drilled profile, accompanied by recurrent detections of ethane (C₂H₆), propane (C₃H₈), iso-butane (i-C₄H₁₀), and n-butane (n-C₄H₁₀). Elevated concentrations of CH₄, C₂H₆, and C₃H₈ were preferentially associated with sandstone and siltstone layers sealed by claystones, indicating stratigraphic trapping of migrated gas. Bernard parameter values (CH₄/(C₂H₆+ C₃H₈)) range from 2 to 1904, reflecting strong compositional variability and mixing between gas sources. Carbon isotopic signatures of CH4 (δ¹³C- CH4 between −35‰ and −25‰ VPDB) indicate a dominant thermogenic contribution.

In the Marajó Basin, continuous gas monitoring during drilling to 924.3 m depth revealed higher? CH4 concentrations than in the Acre Basin with a general increase toward greater depths. Heavier hydrocarbons (C2–C4) show co-occurring concentration maxima indicating stratigraphically discrete gas migration and accumulation. CH₄ carbon isotopic compositions document a clear vertical transition in gas origin, from microbial hydrogenotrophic methanogenesis in the upper 250 m (δ¹³C- CH4 between −80‰ and −60‰), to mixed microbial–thermogenic gas between 250 and 300 m depth, and dominantly thermogenic gas below 300 m depth (δ¹³C- CH4 approaching −35‰), coinciding with increased C2–C4 concentrations.

Laboratory incubation experiments conducted on core sediment samples from both basins under anoxic conditions reveal a progressive increase in CH₄ concentrations over time, indicating active microbial methanogenesis. Incubation results show higher CH₄ yields in deeper samples, suggesting that, despite the strong influence of migrated thermogenic gas at depth, in situ microbial CH₄ production also contributes to the subsurface methane pool and is modulated by depth, substrate availability, and redox conditions.

Overall, the integrated results demonstrate that light hydrocarbon distributions in both basins are governed by the interaction between upward migration of thermogenic gas from deeper sources, stratigraphic trapping in permeable units sealed by fine-grained sediments, and active microbial processes identified through incubation experiments. The combined use of real-time gas monitoring, isotopic analyses, and incubation experiments provides a robust framework for disentangling gas origin and transformation processes, offering new insights into subsurface carbon cycling in Amazonian sedimentary basins.

How to cite: Martinez, A., Oliveira Sawakuchi, A., Oliveira Sawakuchi, H., Bertassoli Junior, D. J., Wiersberg, T., Tsai, S. M., Azevedo Bezerra, I. S., Noren, A., Guizan Silva, C., Fritz, S., and Baker, P. A.: Characterization of light hydrocarbons in subsurface Cenozoic sediments of western and eastern Amazonia drilled by the Trans-Amazon Drilling Project (TADP), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13841, https://doi.org/10.5194/egusphere-egu26-13841, 2026.

The sea floor drill rig MARUM-MeBo is a robotic drill rig that is deployed on the sea floor in order to collect cores from sediments and hard rocks. It can be deployed from multipurpose research vessels. The first generation MeBo70 was designed to drill down to 70 m below sea floor (mbsf) and is operated for about 20 years since 2005. The second generation MeBo200 had its first deployments in 2014. So far, a maximum drilling depth of 146 mbsf was reached. In summary, we have conducted 30 research expeditions and drilled 7465 m. Core recovery rates were in average about 67 % and strongly depend on the drilled lithology. A better control on flush water circulation by using a mud mixing system will be needed to improve the drilling results especially in sandy deposits and crystalline rocks. Knowledge on the expected geology combined with a hydroacoustic survey including high resolution bathymetry in rough terrain, high resolution seismics and sub-bottom profiling are needed for safe operations and optimizing the drilling strategy. A variety of research targets were addressed during the drilling campaigns with MeBo including paleoenvironmental research, gas hydrates and associated processes like authigenic carbonate and pockmark formation, slope stability, geothermal gradient and fluid circulation as well as mafic and ultra mafic rock alteration. Next to core drilling, the sea floor drill rigs are used for bore hole logging and the installation of instrumented borehole observatories.

How to cite: Freudenthal, T.: 20 years of operational experiences with the MARUM-MeBo sea floor drill rigs: scientific applications and lessons learned, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14426, https://doi.org/10.5194/egusphere-egu26-14426, 2026.

How to cite: Horgan, H., Patterson, M., van de Flierdt, T., Levy, R., Dunbar, G., Kulhanek, D., Gasson, E., Grant, G., Marschalek, J., Power, P., Tetard, M., Ulfers, A., Vadman, K., Venturelli, R., Coenen, J., Heins, M., Harwood, D., and Leventer, A. and the SWAIS2C Science Team: SWAIS2C – The Sensitivity of the West Antarctic Ice Sheet to 2 degrees of Warming - Results from Crary Ice Rise.  , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15363, https://doi.org/10.5194/egusphere-egu26-15363, 2026.

Understanding the chemical and physical processes governing the formation and evolution of the Earth’s continental crust is fundamental for the Earth system and other planets. The upper crust is accessible to direct geological observation and sampling, but deeper portions, especially the lower crust and the crust–mantle transition zone (“Moho”) are usually beyond reach. The lower continental crust (LCC) is one of the most important, but also most enigmatic regions of Earth’s lithosphere and its composition and physical properties are strongly debated. Here we report some of the initial results from the first phase of the ICDP-funded “Drilling the Ivrea-Verbano zonE” (DIVE) project (site 5071_1) in Val d’Ossola (northern Italy). From October 2022 to April 2024 two boreholes of respectively 578.5 (Ornavasso, 5071_1_B) and 909.5 m (Megolo, 5071_1_A) depth were drilled using continuous diamond double tube wireline coring. During and after drilling, geophysical logs were acquired, providing natural and spectral gamma ray, magnetic susceptibility, electrical resistivity (SPR and DLL), spontaneous potential, sonic, acoustic and optic televiewer data, complemented by multi-sensor core logging data (with focus on density) acquired in the core repository of the BGR in Berlin-Spandau (Germany). In addition, continuous real-time mud gas logging provides evidence of varying gas mixtures including He, H₂, CH₄, and CO₂, indicating diverse fluid sources and possible microbial activities in the deep crust.

The two drillholes sampled two fundamentally different compositions of the lower continental crust: the first hole (5071_1_B) drilled the upper part of the lower continental crust and mostly consists of metasedimentary rocks and a few amphibolites. The second hole (5071_1_A) drilled the lowermost continental crust and mostly captured a variety of garnet and/or orthopyroxene bearing gabbroic rocks with intercalations of garnet granulite facies metasediments, pyroxenite, and intrusive gabbronorite including frequent pseudotachylites. Combining multi-sensor core-logging data with petrophysical information and whole rock geochemical data provides mineral modes of the drilled cores, which can be used to calculate densities and seismic velocities. These calculations together with direct observations of drilled rock types indicate that the lowermost part of the Ivrea Verbano Zone continental crust is enriched in garnet.

Bulk compositions of the two different drillholes of the lower crust show fundamental differences. 5071_1_B is felsic, similar to global upper crust, while 5071_1_A is dominantly mafic, and similar to the more depleted estimates of global lower continental crust. There is about a 10-fold difference in radiogenic heat producing and volatile elements between the two drillholes, and highly variable thermal properties. Extrapolating the observed datasets beyond the scale of the drillholes suggests both intrinsic and structural variability caused anisotropy of the continental lower crust.

How to cite: Müntener, O., Hetényi, G., Andrew, G., Ziberna, L., Zanetti, A., Pistone, M., Giovanelli, D., and Venier, M. and the The DIVE Drilling Project Science Team: The variability of lower continental crust: Initial and advanced results from the ICDP DIVE project, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16927, https://doi.org/10.5194/egusphere-egu26-16927, 2026.

The NEPTUNE (Noto Peninsula Earthquake Drilling Project for Understanding Fluid Triggered Slip Events) initiative aims to elucidate the mechanisms underlying the 2024 Noto Peninsula Earthquake (January 1st, 2024: Mw 7.6), a major seismic event characterized by a complex rupture sequence across multiple fault segments. This earthquake began with a slow initial rupture that evolved into a dynamic rupture extending over 150 km, highlighting the critical need to understand the interactions between fault behavior and pre-seismic crustal processes.

A central focus of the project is the influence of elevated pore fluid pressure, which promotes fault slip by lowering effective normal stress. The migration and accumulation of fluids—likely derived from the mantle—have been identified as key factors that triggered the preceding earthquake swarms. Geochemical signatures, including high ³He/⁴He ratios, support this interpretation. The event further demonstrated that rupture propagation was facilitated by both segmented fault structures and fluid-induced weakening.

The project plans to drill from the coastal region of the Noto Peninsula, targeting the fault plane of the 2024 earthquake. Core objectives include retrieving fluid, gas, and rock samples to investigate fluid sources, chemical interactions, and fault zone microstructures. Long-term monitoring of fluid and gas behavior near the fault zone is also planned to track post-seismic evolution and enhance preparedness for future seismic events.

Three primary research areas are emphasized:

• Observing fluid migration and pressure fluctuations through direct sampling, numerical simulations, and seismic analysis.
• Characterizing the origins of fault zone rocks and fluids to evaluate their role in earthquake generation.
• Assessing mineralogical and geochemical transformations within the fault zone to understand their impact on fault strength and slip behavior.

The outcomes of NEPTUNE are expected to deepen our understanding of earthquake nucleation, particularly the transition from swarm activity to rapid fault rupture. Aligned with the geohazard priorities of the ICDP Science Plan 2020–2030, the project aims to improve forecasting capabilities for intraplate seismic hazards. In addition, the project includes a complementary proposal for Land-to-Sea (L2S) drilling, aiming to access and study the tsunami-generating fault system from an onshore platform, to be submitted to IODP.

How to cite: Otsubo, M. and the NEPTUNE Proponents: NEPTUNE Project: Exploring Fluid-triggered Slip Mechanisms through Scientific Drilling in the Noto Peninsula, Japan, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20515, https://doi.org/10.5194/egusphere-egu26-20515, 2026.

As part of a multidisciplinary effort to characterize the deep continental crust, two scientific boreholes were drilled in the Ivrea Verbano Zone (IVZ, Western Alps, Italy), one of the few near-complete continental crustal sections exposed on Earth's surface (Pistone et al. 2020). The boreholes were drilled within the Drilling the Ivrea Verbano ZonE (DIVE) project, supported by the International Continental Scientific Drilling Program (ICDP-5071; Li et al. 2024; https://gfzpublic.gfz.de/pubman/item/item_5037328) . Among various well log measurements, time-domain induced polarization (TDIP) logs with two electrode spacings (16″ and 64″) were collected in both wells, from which chargeability data is inferred. The boreholes intersect a wide range of lithologies hosting sulfides and oxides, either disseminated or concentrated along veins and fractures, which represent potential sources of chargeability. A set of eleven samples from these boreholes were analyzed using both scanning electron microscopy (SEM) and micro-computed tomography (μCT). The following mineralogical and microstructural characteristics have been evaluated so far: the type and abundance of metallic minerals (expressed as volume and area fractions); the perimeter-to-area and surface-to-volume ratios and the preferred orientation of these conductive phases. These parameters were compared to the TDIP response signal at the corresponding depths of the borehole, resulting in the following findings:

i) Borehole chargeability is not necessarily proportional to the abundance of metallic minerals;

ii) The total surface area (which is high for fine grain sizes) plays a dominant role over the total volume fraction of metallic minerals;

iii) The shape preferred orientation of conductive phases appears to be a key factor influencing the measured chargeability;

iv) The presence of other mineral phases, such as graphite, may mask or amplify the response of metallic minerals depending on their structural relationship.

While no deterministic relationship has been identified at this stage, this work outlines a potential path to improve the interpretation of TDIP data in mineralized systems and to define complementary yet efficient tools for assessing the economic potential of mineral deposits.

References

Li, J., E. Caspari, A. Greenwood, et al. 2024. “Integrated Rock Mass Characterization of the Lower Continental Crust Along the ICDP‐DIVE 5071_1_B Borehole in the Ivrea‐Verbano Zone.” Geochemistry, Geophysics, Geosystems 25 (12): e2024GC011707. https://doi.org/10.1029/2024GC011707.

Pistone, Mattia, Luca Ziberna, György Hetényi, Matteo Scarponi, Alberto Zanetti, and Othmar Müntener. 2020. “Joint Geophysical‐Petrological Modeling on the Ivrea Geophysical Body Beneath Valsesia, Italy: Constraints on the Continental Lower Crust.” Geochemistry, Geophysics, Geosystems 21 (12): e2020GC009397. https://doi.org/10.1029/2020GC009397.

How to cite: Ventola, I., Caspari, E., Greenwood, A., Hawemann, F., Venier, M., and Virginia, T.: Investigating Borehole TDIP Response in the Ivrea-Verbano Zone (ICDP-DIVE project):Linking Chargeability to Mineral Distribution from SEM and MicroCT Data, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20662, https://doi.org/10.5194/egusphere-egu26-20662, 2026.

Half-precession cycles (HPs), first identified in the 1980s, have been increasingly recognized as an important driver of tropical hydroclimate variability during the Quaternary. However, continuous long-term proxy records capturing their imprint on monsoon systems remain scarce. Here, we investigate the presence and evolution of HPs signals in a high-resolution inorganic geochemical record from ODP Site 663 (Eastern Equatorial Atlantic). This record provides a continuous perspective on West African Monsoon (WAM) variability over the last 1.2 Myr, spanning the Mid-Pleistocene Transition (MPT). Our results indicate that WAM variability during the MPT does not exhibit clear glacial–interglacial pacing. While this contrasts contemporaneous records from the Mediterranean and North Africa, the influence of the WAM becomes more pronounced after ~600 kyr, with intensified interglacial conditions and weaker glacial phases. In contrast, HPs show a stronger imprint on monsoon variability after the MPT, particularly during interglacial intervals. These findings are consistent with runoff records from the tropical American ICDP sites, suggesting a coherent low-latitude hydroclimate response. We propose that modulation of the Atlantic Meridional Overturning Circulation may provide a mechanistic link between HPs forcing, West African Monsoon variability, and tropical American precipitation.

How to cite: Martinez-Abarca, R., Ulfers, A., Zeeden, C., Westerhold, T., Vinnepand, M., De Vleeschouwer, D., Röhl, U., and Kaboth-Bahr, S.: Half-precession modulation of the West African Monsoon and possible links to continental hydroclimate records since the Mid-Pleistocene Transition., EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21573, https://doi.org/10.5194/egusphere-egu26-21573, 2026.

In the past four decades, high latitude regions have been warming 2 to 4 times more rapidly than the global average, and their ocean and cryosphere are rapidly changing. Arctic sea-ice loss, complex Antarctic sea-ice variability, instability of the Greenland and Antarctica continental icesheets, and melting have consequence for the entire planet including sea-level rise, changing ocean currents, and impacts on polar species, ecosystems and biodiversity. The International Ocean Discovery Program (IODP) completed 8 expeditions in high latitude locations during the past ~10 years. One aim that these expeditions share is to study ocean and cryosphere responses to warm climates in the geological past to get insights into cryosphere instability thresholds in a (future) warm climate scenario. Obtaining continuous high latitude records is challenging, but even snapshot views of the past offer important insights into the interaction among climate, ocean and cryosphere (in)stability, and ecosystem responses.

On behalf of numerous collaborators, I will present an overview of what we have learned so far about ocean and cryosphere variability during warm periods of the Neogene (Miocene Climatic Optimum and Pliocene) and the Quaternary. I will discuss (preliminary) results, mostly centered on palynology, obtained from Expeditions 374 (Ross Sea) and 400 (NW Greenland) in the context of additional sedimentological and geochemical data, and link them to results from previous (I)ODP expeditions and on-going projects. 

How to cite: Sangiorgi, F., Hou, S., Koene, B., Nelissen, M., Sriwichai, M., Steinsland, K., Bijl, P., Peterse, F., Kulhanek, D., McKay, R., de Santis, L., Knutz, P., Jennings, A., Hillenbrand, C.-D., and Larter, R. and the IODP Exp 374 & Exp 400 scientists: High-latitude ocean and cryosphere during warmer than present climates of the Neogene and Quaternary: a view from Antarctic and NW Greenland (I)ODP expeditions, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-22430, https://doi.org/10.5194/egusphere-egu26-22430, 2026.

The mobility of sulphur and chalcophile metals through the lithosphere remains poorly constrained, yet it is likely to have a significant impact on metal budgets and the localisation of ore systems that underpin the supply strategic commodities for the energy transition.

Increasing evidence suggests that sulphur and chalcophile metals can be redistributed via multiple, potentially overlapping processes, including sulphide melt migration, fluid-mediated transport, partial melting, and deformation-assisted remobilisation. These mechanisms operate across a wide range of pressure-temperature conditions and may decouple metal transport from large-volume magmatic fluxes, producing complex metal redistribution patterns within the lithosphere.

In this framework, deep mafic-ultramafic cumulates in lower crustal zones act as major reservoirs and transfer hubs where sulphide melts can sequester a large fraction of the metal budget, while being episodically mobilised within melt-bearing cumulate frameworks, enabling upward transfer to upper‑crustal levels (i.e Holwell et al. 2022). A complementary mechanism for enriching and moving sulphur and copper is provided by devolatilization and wall-rock assimilation. Fluids can already effectively mobilise sulphur and copper under subsolidus conditions, enhancing mobilisation that may have already accompanied partial melting and produce Cu-rich sulphide droplets that can attach to fluid bubbles (i.e Blanks et al. 2020) or carbonate melt droplets (i.e Cherdantseva et al. 2024) which can be transported buoyantly within silicate melt (i.e Virtanen et al. 2021). Deformation potentially introduces an additional mechanism of redistribution, which is highly relevant for interpreting sulphide signatures in deep crustal rocks, and metamorphism commonly overprints ore systems, creating favourable conditions for further mobilization of critical metals (i.e Cugerone and Cenki 2025).

Newly acquired continuous drill core from the Ivrea-Verbano Zone provides an exceptional opportunity to investigate these processes in a well-constrained lower-crustal setting. The core samples mafic and ultramafic lithologies across documented igneous, metamorphic, and structural domains, allowing sulphide occurrence, texture, and chemistry to be examined in their primary context, while also distinguishing deep-crustal magmatic processes from later deformation- or fluid-assisted remobilisation.

Borehole 5071_1_A is dominated by gabbroic lithologies, with intercalations of granulite-facies metasediments and pyroxenites, as well as intrusive gabbronorites. We combine (i) XRF elemental mapping on the flat split core surfaces to track core-scale variations and identify sulphide-rich intervals and their structural/lithological controls, with (ii) SEM-EDS analyses to characterise sulphide mineralogy and (iii) LA-ICP-MS trace-element analyses of the major sulphide phases to constrain phase-dependent trace-element budgets and variations among various host lithologies as well as different textures. The sulphide assemblage is dominated by Fe-Ni-Cu sulphides (pyrrhotite-pentlandite-chalcopyrite), occurring both as disseminated interstitial grains and as foliation- and fracture-related networks associated with localised deformation and late-stage fluid pathways. Across these textural populations, trace-element systematics display different variations consistent with sulphide-silicate equilibration as well as later stage remobilisation.

By linking centimetre-scale elemental maps from continuous core to micro-analytical sulphide fingerprints, the ICDP-DIVE record allows us to advance the exploration and resource assessment for critical raw materials in complex crustal systems.

How to cite: Venier, M., Caruso, S., Fiorentini, M., Müntener, O., Ziberna, L., and Toy, V. and the DIVE Science Team: Lower continental crust sulphides from the Ivrea-Verbano Zone (ICDP-DIVE project 5071): textures, trace-element chemistry and mobility, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-22443, https://doi.org/10.5194/egusphere-egu26-22443, 2026.

The West Antarctic Ice Sheet (WAIS) is currently experiencing accelerated mass loss and contains enough ice to raise global sea levels by up to five meters if it were to melt completely. The objective of the international and interdisciplinary SWAIS2C project (Sensivity of the West Antarctic Ice Sheet to 2 Degrees Celsius of Warming) is to understand past and present factors influencing WAIS dynamics and to reconstruct WAIS response to warmer temperatures, including those exceeding the +2°C target outlined in the Paris Climate Agreement.

In its third season, the SWAIS2C project targeted the second drilling location Crary Ice Rise Site 1 (CIR), a grounded ice sheet upstream the Ross Ice Shelf. After hot water drilling through ~516 m of ice, rotary coring retrieved a sediment succession of 228 m length comprising different lithological units.

The LIAG Institute for Applied Geophysics and the German Helmholtz Centre for Geosciences (GFZ) are in charge of geophysical downhole logging operations and retrieved the first such dataset below grounded ice. The spectrum gamma radiation (SGR) tool records the natural radiation and its components – the K-, Th-, and U-concentration – of the surrounding sediments. The data indicate distinct boundaries between the main lithological units, but minor variations and ratios of the measured elements indicate smaller differences within the units. Particularly in transitions between major units, patterns in the data may reflect changing paleo-environmental conditions. This data set will be valuable for the ongoing project as it is an in-situ, continuous record of drilled sediment succession with high accuracy depth measurements.

We give a brief overview of the SWAIS2C project, focus on the downhole logging data measured as part of the project and relate the results to other data sets from below/around the Ross Ice Shelf.

How to cite: Ulfers, A., Horgan, H., Patterson, M., Dunbar, G., Kulhanek, D., Levy, R., van de Flierdt, T., Pierdominici, S., and Zeeden, C. and the SWAIS2C Science Team: Characteristics of spectrum gamma radiation (SGR) data from geophysical downhole logging in the SWAIS2C project – West Antarctica , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-23159, https://doi.org/10.5194/egusphere-egu26-23159, 2026.

Urban populations are frequently exposed to complex mixtures of air pollutants, a critical public health challenge as compound exposures often produce nonlinear, synergistic health impacts greater than the sum of individual risks. This study presents a high-resolution, satellite-based assessment of population exposure to concurrent exceedances of multiple air pollutants in Gujarat’s major metropolitan areas—Ahmedabad, Surat, Vadodara, and Rajkot—from 2019 to 2024.

We leverage advanced remote sensing data from the TROPOspheric Monitoring Instrument (TROPOMI) on Sentinel-5P and the Moderate Resolution Imaging Spectroradiometer (MODIS) on Aqua to monitor key pollutants, including Aerosol Optical Depth (AOD), Nitrogen Oxides (NOx), Sulfur Dioxide (SO2) and Methane (CH4) as a proxy for particulate matter. By analyzing spatiotemporal patterns, we identify and characterize episodic events where multiple pollutants simultaneously exceed baseline thresholds, creating potential ‘pollution cocktails’.

These multi-pollutant exceedance events are then integrated with high-resolution gridded population data to quantify the number and demographic distribution of residents exposed to compounded air quality risks. The methodology enables a shift from single-pollutant monitoring to a holistic exposure assessment framework.

Preliminary findings reveal significant temporal and spatial heterogeneity in compound exposure events, strongly influenced by urban form, industrial activity, and meteorological conditions. The analysis identifies recurring pollution hotspots and temporal patterns (e.g., seasonal, episodic) where populations face elevated health risks from concurrent pollutants. The results underscore that mitigation strategies focused on single pollutants may underestimate population health risks in these urban centers.

This study provides a critical evidence base for designing targeted, health-centric air quality management policies. By mapping compound exposure risks, it empowers urban planners and public health officials in Gujarat to prioritize interventions, optimize monitoring networks, and develop early warning systems that address the real-world, multi-pollutant environments experienced by urban populations, thereby strengthening resilience and advancing sustainable urban development goals.

Key Words: Compound Risk, Air Pollution, Satellite Data, Population Exposure

How to cite: Gadekar, K. and Kandya, A.: Beyond Single Pollutants: Quantifying Urban Population Exposure to Concurrent Air Pollution Hazards in big cities of Gujarat, India , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1252, https://doi.org/10.5194/egusphere-egu26-1252, 2026.

The concurrent occurrence of temperature and precipitation extremes, known as compound temperature-precipitation extreme events (CTPEEs), leads to more pronounced consequences for human society and ecosystems than when these extremes occur separately. However, such compound extremes have not been sufficiently studied, especially during boreal spring. Spring is an important transition season, during which the CTPEEs plays a pivotal role in plant growth and revival of terrestrial ecosystems. This study investigates the spatio-temporal variation characteristics of spring CTPEEs in China, including warm-dry, warm-wet, cold-dry and cold-wet combinations. The compound cold-wet extreme events occur most frequently, followed by warm-dry, warm-wet and cold-dry events. The frequency of CTPEEs associated with warm (cold) extremes shows a marked interdecadal increase (decrease) since the mid-to-late 1990s. It is found that the interdecadal change in CTPEEs is primarily determined by the variation in temperature extremes. This interdecadal shift coincides with the phase transitions of the Atlantic Multidecadal Oscillation (AMO) and the Interdecadal Pacific Oscillation (IPO). After the mid-to-late 1990s, the configuration of a positive AMO and a negative IPO excited atmospheric wave trains over mid-high latitudes, causing high-pressure and anticyclonic anomalies over East Asia. This leads to less cloudiness, allowing an increase in downward solar radiation, which enhances surface warming and contributes to an increase (decrease) in warm-dry and warm-wet extremes. The above observations are confirmed by the Pacemaker experiments. The results of this study highlight a significant contribution of internal climate variability to interdecadal changes in CTPEEs at the regional scale.

How to cite: Wang, L., Chen, S., Chen, W., Wu, R., and Wang, J.: Interdecadal Variation of Springtime Compound Temperature-Precipitation Extreme Events in China and its Association with Atlantic Multidecadal Oscillation and Interdecadal Pacific Oscillation, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1776, https://doi.org/10.5194/egusphere-egu26-1776, 2026.

Access to water is crucial for all aspects of life. Anthropogenic global warming is projected to disrupt the hydrological cycle, leading to water scarcity. However, the timing and hotspot regions of unprecedented water scarcity are unknown. Here, we estimate the Time of First Emergence (ToFE) of droughtdriven water scarcity events, referred to as “Day Zero Drought” (DZD), which arises from hydrological compound extremes, including prolonged rainfall deficits, reduced river flow, and increasing water consumption. Using a probabilistic framework and a large ensemble of climate simulations, we attribute the timing and likelihood of DZD events to human influence. Many regions, including major reservoirs, may face high risk of DZD by the 2020s and 2030s. Despite model and scenario uncertainties, consistent DZD hotspots emerge across the Mediterranean, southern Africa, and parts of North America. Urban populations are particularly vulnerable at the 1.5 °C warming level. The length of time between successive DZD events is shorter than the duration of DZD, limiting recovery periods and exacerbating water scarcity risks. Therefore, more proactive water strategies are urgently needed to avoid severe societal impacts of DZD.

How to cite: Franzke, C. and Ravinandrasana, V.: The first emergence of unprecedented global water scarcity compound extremes in the Anthropocene, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-2293, https://doi.org/10.5194/egusphere-egu26-2293, 2026.

Compound events often exert more severe and widespread consequences than isolated extremes, and this effect is especially pronounced for those occurring at a regional scale. Here, we define and apply the concept of regional compound drought and heatwave events (RCDHEs) on daily scale to investigate spatiotemporally contiguous compound drought and heatwave events (CDHEs) across China during 1961–2022. We use homogenized observational data adapted from the China Meteorological Administration (CMA), and assess the performance of the state-of-the-art ERA5-Land reanalysis for its potential in supporting future studies. Rather than identifying events within fixed regions, we extract RCDHEs considering spatial and temporal coherence and characterize their dominant patterns through cluster analysis. The results reveal a marked increase in RCDHEs severity and duration across China over the recent decades. Over the mainland China RCDHEs can be grouped into eleven patterns. Most of these RCDHE patterns exhibit larger spatial extent, longer durations, and greater intensities during the recent decades. ERA5-Land effectively reproduces the various spatiotemporal features of these events; however, it consistently overestimates the frequency of RCDHEs across all region types since the late 1990s, limiting its reliability for assessing long-term trends. These findings enhance understanding of regional compound extremes in China and inform the appropriate application of ERA5-Land in future investigations of compound drought and heatwave events.

How to cite: Deng, J., Duan, Y., Ma, Z., Li, Z., Li, M., Wei, W., and Yang, Q.: Regional compound drought and heatwave events over China and evaluation of ERA5-Land dataset based on classification approach, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3230, https://doi.org/10.5194/egusphere-egu26-3230, 2026.

European countries share essential firefighting equipment and personnel to manage wildfires. Nevertheless, climate change is increasing the likelihood of synchronous fire danger—periods when wildfire conducive weather conditions occur simultaneously across multiple European regions. Synchronous fire weather conditions could strain existing resource sharing plans and overwhelm firefighting capacities. To ensure an effective wildfire response in a warming climate, it is essential to understand how climate change influences the occurrence and spatial scale of synchronous fire danger.

Here, we analyze future fire weather synchronicity across ten European regions using the Canadian fire weather index (FWI) with three complementary perspectives—regional, inter-regional, and continental. Our analysis is based on a regional Single Model Initial condition Large Ensemble (SMILE), i.e. the CRCM5-LE, spanning the period from 1990 to 2099. This enables a robust quantification of both internal variability and forced response, as well as a sufficient sampling of extreme events. To identify impact-relevant fire weather conditions, we derive regional thresholds of FWI anomaly according to its cumulative distribution function (CDF) on burned area during 2001-2020, using CERRA reanalysis data and FireCCI burned area observations. Thereby, we focus on two levels of fire danger: moderate (FWI anomaly corresponding to 50% of burned area) and extreme (FWI anomaly corresponding to 90% of burned area). We then apply these regional thresholds to the FWI anomaly from the CRCM5-LE ensemble for each grid cell within the respective region. To quantify inter-regional and continental synchronicity, we compute weekly block maxima of the regional land area exceeding these thresholds as a proxy for regional fire danger.

From a regional perspective, we find that the magnitude (i.e., spatial extent) of fire danger increases with increasing global warming level (GWL) in all ten European regions. The increase is larger for extreme fire weather conditions than for moderate fire weather conditions. We find the strongest increases (fivefold) in the magnitude of extreme conditions in France and the Alps under 4 °C GWL. From an inter-regional perspective, we find an increasing pair-wise dependence of fire danger between regions under climate change, for both the moderate and extreme conditions. France, the Alps and Central Europe will become strongly dependent on each other in their weekly fire danger under 4 °C GWL. All region pairs show an emergence of synchronous fire danger between 2 and 3 °C GWL compared to 1990-2019, with southern regions emerging earlier (or at lower GWLs) than northern regions. From a continental perspective, we find that increasing GWLs also increases the odds of more than five European regions co-experiencing fire danger in one week, with an even stronger increase for extreme than moderate conditions.

Our results point toward increasing fire weather synchronicity in Europe under climate change and underscore the urgency to adapt current fire management strategies and collaboration in a warming climate. This is especially relevant for France, the Alps and Central Europe, that have historically low wildfire activity but will undergo a strong increase in fire danger under climate change.

How to cite: Li, X., Wood, R., Miller, J., and Brunner, M.: European fire weather synchronicity under climate change: three perspectives from regional to continental scales, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-3975, https://doi.org/10.5194/egusphere-egu26-3975, 2026.

Extreme humid heatwaves (HS) have emerged as one of the most threatening compound disasters under climate change, posing severe risks to human health and socio-economic security. Yet their dynamic evolution and mechanism, especially the relation with antecedent precipitation, remain insufficiently understood. Based on global reanalysis data from 1985 to 2024, this study presents a systematical assessment for the spatiotemporal evolution of humid heatwaves, their thermodynamic drivers, and the modulation effects of preceding precipitation. Results reveal a significant intensification trend in HS frequency, duration, and intensity, which can be statistically significantly attributed to anthropogenic forcing. The occurrences of extreme humid heatwaves are mainly driven by humidity anomaly in 94.45% of global land areas, while the influences of temperature and humidity changes on HS trends exhibit larger spatial heterogeneity. The relations between antecedent precipitation and subsequent HS are strengthening, evidenced by their increasing synchrony and high HS triggering probability. Moreover, HS exhibit distinct patterns based on preceding precipitation: HS following light precipitation are most frequent, while long-duration or heavy precipitation are likely to trigger most intense HS. Notably, HS tends to occur more rapidly after the cessation of long-duration heavy rainfall, demonstrating a differentiated regulatory mechanism from land-atmospheric coupling and necessitating context-specific adaptation strategies tailored to these divergent precipitation-HS relationships.

How to cite: Fang, J. and Tu, Y.: Increasing risk of global compound humid heatwaves and the impacts of antecedent precipitation, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4325, https://doi.org/10.5194/egusphere-egu26-4325, 2026.

Global warming has intensified the frequency of synchronous extreme climate events, posing severe threats to the water-food-energy-ecosystem nexus and challenging regional sustainability. Current studies overlook the inverse symbiotic relationship of the droughts and wet events and the complex, nonlinear spatiotemporal correlations underlying transregional extreme climate events. Here, using complex network, we systematically identify the synchronous structure of drought, pluvial, and drought–pluvial dipole (DPD) events within the Western Route of the South-to-North Water Diversion Project in China. Our analysis reveals a distinct wet-dry co-variability between the Yangtze and Yellow River basins. From the perspectives of atmospheric circulation and local weather systems, we elucidate the physical coupling between extreme hydroclimatic events and circulation anomalies as well as moisture transport pathways. We identify remote coupling zones of DPD events and highlight a pronounced spatial asymmetry in cross-basin hydroclimatic behavior. Drought and pluvial synchronicity is predominantly characterized by short-to-medium spatial scales, compared to DPD events exhibiting robust cross-basin teleconnections. Notably, the signal sources for these extremes are anchored in the southwestern portion of the study area. We show that positive geopotential height anomalies, airflow subsidence, and monsoon disruption drive drought conditions, whereas the transport of warm, moist air generates pluvial events -together forming a “drought-pluvial seesaw” at the climatic scale. This study provides critical scientific foundation for cross-basin water resource management and offer vital insights for developing climate-resilient infrastructure and optimizing adaptive spatial planning under a changing climate.

How to cite: Li, W. and Xie, J.: Climate Network-Based Synchronized Structures Identification of Extreme Droughts and Pluvials in Cross-basin Regions, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4417, https://doi.org/10.5194/egusphere-egu26-4417, 2026.

Spatially synchronized drought or flood events, that is the co-occurrence of drought/flood in multiple locations, can have severe impacts that challenge water and emergency management because they require resources in multiple places at once. Climate change can affect the frequency of such compound events because of its influence on drought and flood generation processes. While the impacts of climate change on local hydrological extreme events are well studied, its impact on event synchronicity remains uncertain. Here, we investigate how hydrological drought and flood synchronicities have changed in Europe during the period 1981-2020 using observations from 4299 streamflow stations. Our results show that drought synchronicity has grown significantly, most strongly in Central Europe, and that years with spatially extensive drought tend to follow one another. In contrast, flood synchronicity has remained relatively stable. Regionally, regions of growing drought synchronicity show decreasing flood synchronicity, and vice versa. Synchronicity trends are mostly in line with those of local frequencies suggesting that trends in synchronicity are mainly driven by overall frequencies, rather than by the spatial distribution of events. The observed growth in drought synchronicity highlights the need to develop adaptation measures to more frequent large-scale droughts.

How to cite: Brunner, M. I., Berghuijs, W. R., Janzing, J., and Bruno, G.: Hydrological drought but not flood synchronicity increases over Europe, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4770, https://doi.org/10.5194/egusphere-egu26-4770, 2026.

Climate change is intensifying the frequency, severity, and interactions of extreme heat, drought, wildfires, and air pollution, increasing risks to both ecosystems and human populations worldwide. These risks can emerge from the accumulation of multiple hazards over time (multi-hazard risk), but also from their simultaneous co-occurrence (compound events). Here, we present a global, spatially explicit assessment of human risk from wildfires and air quality associated with hot and dry extremes, explicitly integrating multi-hazard and compound risk representations, following a hazard, exposure, and vulnerability perspective.

Using global datasets at 0.75° spatial resolution for the period 2003–2022, hazards are quantified based on the number of hot days per month (derived from exceedances of daily maximum temperature above the 90th percentile of a 1991–2020 climatology), drought occurrence (as depicted by the 6-month Standardized Precipitation–Evapotranspiration Index, SPEI), wildfire activity (characterized using MODIS Fire Radiative Power, FRP), and the number of days with PM_2.5concentrations exceeding World Health Organization air quality thresholds. Human exposure is represented exclusively by gridded population density, while vulnerability is characterized using indicators capturing human sensitivity and adaptive capacity, e.g., the Human Development Index (HDI) and Water Stress Index (WSI).

Human risk is quantified by combining hazard intensity, population exposure, and vulnerability, following both a multi-hazard and a compound formulation. In a multi-hazard formulation, hazards are aggregated without requiring temporal co-occurrence, capturing the cumulative burden of climate extremes. In parallel, compound risk is assessed by explicitly accounting for the co-occurrence of hazards within the same temporal windows, enabling a direct comparison between cumulative and compound representations of risk. In addition, we quantify the global population affected by different risk classes. Our estimates indicate that approximately half of the world’s population is currently exposed to high to very high risk, while a substantially smaller fraction resides in low or extremely low risk conditions. High and very-high risk classes together account for several billion people, underscoring the widespread nature of climate-related human risk. When aggregated at the country level, risk levels exhibit a clear socioeconomic gradient, with higher average risk values concentrated in lower-income countries, low life expectancy at birth, and high infant mortality rate.

The results illustrate how a compound events perspective can alter the spatial distribution and relative intensity of human risk compared to a multi-hazard one, highlighting regions where hazard interactions may further amplify societal impacts. This work provides a generalized framework for global human risk assessment, offering new insights into how different representations of climate extremes shape risk patterns and supporting the development of more effective adaptation and risk reduction strategies.

This work is supported by FCT, I.P./MCTES through national funds (PIDDAC): LA/P/0068/2020 - https://doi.org/10.54499/LA/P/0068/2020, UID/50019/2025, https://doi.org /10.54499/UID/PRR/50019/2025, UID/PRR2/50019/2025. This work was performed under the scope of project https://doi.org/10.54499/2022.09185.PTDC (DHEFEUS) and the Horizon Europe research and innovation programmes under grant agreement number 101081661 (WorldTrans).

How to cite: Bento, V. A., Köberle, A. C., Trigo, R. M., Lima, D. C. A., and Russo, A.: Global patterns of human risk from hot and dry climate extremes, wildfires and poor air quality: insights from multi-hazard and compound analyses, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6731, https://doi.org/10.5194/egusphere-egu26-6731, 2026.

The sequence of compounding winter storms of 2013/14 in the United Kingdom (UK) caused a range of significant impacts across the country, totalling an economic cost of approximately £1.3 billion (Environment Agency, 2016). Heavy rain, totalling 545mm over the season, caused widespread flooding, and coastal impacts were exacerbated by high spring tides and strong winds. The Somerset Levels particularly felt the impact of the flooding, accounting for 30% of the total UK area of flooded agricultural land. This event is a case study for the COMPASS project, where the main goal is to produce a flexible and harmonised methodological framework for such compound extremes with a focus on impact attribution.

Using the flood model SFINCS (Super-Fast Inundation of Coasts), developed by Deltares, we model the total flood extent for a small region of the Somerset levels for the season. We drive this model with both factual and counterfactual (“natural”, with anthropogenic warming removed) simulations of the winter precipitation, using the HadGEM3-A large-ensemble attribution runs (Ciavarella et al., 2018). We find the likelihood of the magnitude of the observed flood extent to be 1.21 times more likely due to climate change, based on return periods. We also find that a flood event under “natural” forcing but with the same return period as the factual event would be slightly less severe in its extent, 113.40km² compared to 114.02km².

Although these results are not statistically significant, this agrees with generally inconclusive results from other studies on the 2013/14 UK winter storms, such as that of Schaller et al. (2016) who found a small, non-significant increase due to climate change in the number of properties impacted by flooding in the Thames river catchment. Potential modelling improvements to refine results for Somerset include increasing resolution and adding flood defences to better represent the coastal inundation. Investigation of attribution of the flood extent to post-industrial sea level rise also opens another avenue for exploring the compound nature of the event.

The Horizon-Europe COMPASS project (Compound events attribution to climate change: towards an operational service) is exploring climate and impact attribution of different complex extreme events, and scoping an operational attribution service. It aims to develop transferable attribution methods for operational attribution of compound extremes to support climate change evidence and policy.

How to cite: Matthews, E., Munday, G., Perks, R., Cotterill, D., Bernie, D., Couasnon, A., and Vertegaal, D.: Climate change attribution of the compound 2013/14 winter storms flooding in Somerset, UK, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7906, https://doi.org/10.5194/egusphere-egu26-7906, 2026.

The Carpathian Basin is identified as one of the climate change hotspots in Europe. According to the latest data from the Copernicus Climate Change Service (C3S), the European continent – including Hungary – has already warmed by approximately 2.4 °C compared to the pre-industrial period (1850-1900), accompanied by more frequent extreme weather events. This substantial warming justifies the aim to focus on the detailed analysis of summer heat waves and droughts, especially their simultaneous occurrence. As demonstrated by the exceptionally hot and dry summer of 2022 in Hungary, the cumulative impact of these events poses severe consequences for agriculture, water management, and public health.

The main goal of our research is to explore the relationship between hot and dry periods in Hungary using homogenised, gridded daily maximum temperature and precipitation data from the HuClim database (0.1° spatial resolution) for the period 1971-2025. To investigate the spatial behaviour of the dependence strength between the monthly extremes of the base variables, a detailed cross-correlation analysis was completed. First, we analysed the spatial structure of monthly extreme temperature and precipitation fields separately using cross-correlation matrices based on different percentile values often used as extreme thresholds (i.e. the 75th, 90th, 95th, and 99th percentiles). In addition, we used anomaly maps to identify regions where extreme heat occurs with precipitation deficit at the same time. To investigate the duration of dry periods, we selected the Consecutive Dry Days (CDD) index calculated from daily precipitation data.

Our preliminary results indicate substantial differences in the spatial structure of the monthly variables. The analysis of the cross-correlation matrices demonstrates that while temperature fields follow a quite uniform, homogeneous pattern even in extremes, precipitation fields show a more heterogeneous structure. The joint evaluation of spatial anomalies (calculated as the difference between grid-point values and the regional mean) revealed substantial spatial heterogeneity. While mountainous regions show lower values due to orographic effects, the Great Hungarian Plain emerges as the most vulnerable 'hotspot' regarding the combined impact of heat waves and droughts, where the most pronounced positive temperature anomalies coincide with the greatest precipitation deficits. This is especially important due to the dominance of agriculture in the region, and suggests a clear necessity of adaptation strategies depending on further future climatic changes.

Acknowledgements. This work has been implemented by the National Multidisciplinary Laboratory for Climate Change (RRF-2.3.1-21-2022-00014) project within the framework of Hungary's National Recovery and Resilience Plan supported by the Recovery and Resilience Facility of the European Union. 

How to cite: Fritz, P., Kis, A., and Pongrácz, R.: Analysis of hot and dry compound events in Hungary in 1971-2025, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8320, https://doi.org/10.5194/egusphere-egu26-8320, 2026.

Compound hazards represent a major challenge in hydrogeological risk analysis, as the co-occurrence of extreme conditions can generate complex and non-intuitive impacts, sometimes exceeding those produced by isolated extreme events. This preliminary study investigates the statistical relationship between droughts and landslides in Italy, with the aim to quantify the marginal and conditional probabilities associated with their co-occurrence and, thus, assess the relevance of drought–landslide compound events. The proposed analysis uses the historical series of Standardized Precipitation Evapotranspiration Index (SPEI), provided by the European Drought Observatory at multiple temporal scales, and the historical landslide occurrences provided by the ITALICA national catalogue. Specifically, for the analyzed period 1996-2021, a landslide–SPEI database is constructed by associating each grid cell in which at least one landslide occurred in the corresponding SPEI time series. As expected, a decrease in landslide frequency is observed during drought conditions. However such a frequency remains non-negligible, highlighting the need for multiple risk management strategies.

How to cite: Zofei, R. D., Palazzolo, N., Cancelliere, A., and Peres, D. J.: A statistical analysis of compound drought-landslide events in Italy, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9599, https://doi.org/10.5194/egusphere-egu26-9599, 2026.

Compound dry and hot events (CDHEs), which arise from the co‑occurrence of heatwaves and droughts, represent one of the most critical and rapidly intensifying climate‑related hazards worldwide, particularly in climate change hotspots like the Mediterranean Europe. The consequences of these CDHEs often exceed those associated with isolated occurrences.

Despite growing recognition of their importance, CDHEs remain challenging to characterize due to their multivariate structure, requiring methodological approaches that differ from those typically employed in univariate analyses. As a result, advancing the study of CDHEs is essential, especially given expectations of their increasing frequency and severity under continued warming.

In this study, we employ a compound severity index based on the product of marginal probabilities of individually standardized hot and dry indicators, providing a meaningful measure of compound hot–dry severity across Europe. These indicators rely on well‑established metrics for defining heatwaves and drought conditions, including commonly used heatwave indices and the SPI, both derived from ERA5 data. The severity index is used to evaluate the spatial and temporal patterns of CDHEs for the period 1979–2025, with particular emphasis on distinct severity classes and the percentage of area affected by events.

The results show distinct spatial and temporal variations in CDHE severity and in the extent of the areas impacted. This perspective on joint magnitude and spatial extent allows for a consistent comparison of events, helping to identify those that were both exceptionally strong and unusually widespread across the domain, uncovering information that would be missed by analyses limited to event frequency.

Overall, this investigation advances the emerging field of compound‑event research by providing a detailed climatological assessment of heatwaves, droughts, and their co‑occurrence in a region already experiencing substantial climate pressures. The proposed framework offers a robust way to improve the representation of multivariate hazard characteristics and is expected to offer useful insights for climate‑impact assessment and risk management under continued warming. It further provides a solid starting point for expanding the analysis to include additional variables and processes linked to compound events, supporting more comprehensive evaluations of climate‑related risks.

Acknowledgements: This work is supported by FCT, I.P./MCTES through national funds (PIDDAC): LA/P/0068/2020 - https://doi.org/10.54499/LA/P/0068/2020, UID/50019/2025,  https://doi.org /10.54499/UID/PRR/50019/2025 ,UID/PRR2/50019/2025, and DHEFEUS (https://doi.org/10.54499/2022.09185.PTDC).

How to cite: Santos, R., M. Gouveia, C., Bento, V., and Russo, A.: Spatiotemporal Patterns of Compound Heatwave–Drought Severity Across Europe, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10559, https://doi.org/10.5194/egusphere-egu26-10559, 2026.

Compound drought-heatwave events (CDHEs), defined by the co-occurrence of soil moisture droughts and heatwaves, are among the most damaging climate extremes due to their impacts on ecosystems, agriculture, and humans. Previous studies have reported increasing CDHE occurrence in many regions. However, the extent to which CDHE trends are driven by long-term changes in the soil moisture–temperature (SM–T) feedback remains unclear, compared to their roles in single heat or drought events alone. In particular, traditional correlation-based approaches to quantify SM–T feedback are limited in their ability to resolve its causal roles.

We investigate how land–atmosphere feedback drives CDHEs using the normalized non-stationary Liang–Kleeman information flow. This framework allows us to quantify the strength of the coupling in both directions of the soil moisture–temperature feedback while considering common confounders to assess how these couplings have evolved over recent decades at the global scale. Using ERA5 and ERA5-Land, we find that the widespread increases in CDHE frequency cannot be fully explained by changes in heatwave or drought frequency alone. We identify significant trends in the coupling strength for directions of the SM-T feedback, with generally stronger trends in regions with higher water availability.

We further combine this causal analysis with anthropogenic attribution to disentangle the respective roles of anthropogenic forcing and natural climate variability, using an ensemble of CMIP6 models under historical and natural-only forcings. We find significant effects of anthropogenic emissions on CDHE frequency across most land areas. On the contrary, we find heterogeneous spatial patterns of the anthropogenic impact on the frequency of univariate extremes, emphasizing the need to investigate anthropogenic impacts on the dependence between climate variables. We therefore investigate and detect anthropogenic influences on the trends of the SM-T feedback across most land areas for both coupling directions, highlighting the role of evolving land–atmosphere feedbacks in shaping compound event likelihood. Our study provides a physically interpretable and reanalysis-based pathway toward improved understanding of compound extremes under ongoing climate change using causal, multivariate methods for identifying the compounding physical drivers of high-impact climate events.

How to cite: Therville, T., Hagan, D., and Tabari, H.: Drivers of compound drought-heatwave events: assessments of univariate extremes and causal soil moisture-temperature feedback, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13090, https://doi.org/10.5194/egusphere-egu26-13090, 2026.

Weather extremes are becoming more frequent and intense across Europe, as climate change transforms once-rare events into more common and severe occurrences with major consequences for society, calling for revised adaptation and mitigation strategies. When extremes such as terrestrial heatwaves and droughts occur in combination (CDHWs), their compound effects may lead to amplified impacts, creating complex, multiscale challenges. At the same time, marine heatwaves (MHWs) are rising in intensity, duration, and frequency, profoundly affecting marine ecosystems and showing potential relationship to terrestrial extreme weather. In Europe, both oceanic and land-based heat extremes display parallel warming trends, underscoring the connectivity of Earth’s subsystems, yet the regional teleconnections that drive this connectivity remain insufficiently explored. Understanding the relationship among heatwaves, droughts, and MHWs requires robust detection and characterisation of compound events, drawing on statistical, empirical, high-dimensional, and network analysis methods. Within the ESA XHEAT project, we are leveraging Earth Observation data to expose common spatiotemporal signatures of these extremes and to test the hypothesis that North Atlantic MHWs modulate the persistence and intensity of terrestrial heatwaves and droughts, focusing on the Iberian Peninsula, the Mediterranean basin, and Scandinavia. Early results reveal coherent patterns that suggest strong linkages between oceanic heat extremes and concurrent atmospheric extremes, supporting improved probabilistic seasonal forecasting. In addition, we are integrating machine learning techniques into traditional MHWs detection workflows to develop a mechanistic, spatiotemporal approach that captures the connectivity of these anomalies. Our work aims to enhance the understanding of how ocean–atmosphere interactions contribute to interconnected risks, enabling better prediction of such events, anticipating their impacts and promoting timely response measures to mitigate them, and thus aiming to support improved preparedness and resilience in Europe.

How to cite: Silva, F., Oliveira, A., Lopes, B., Paixão, J., Baeta, R., Barros, L., Girão, I., Cunha, R., Garcia, T., Lourenço, A., Kristiansen, J., Yang, C., Bartucca, C., Martins de Araujo, J., and Riaz, A.: Connecting Marine and Terrestrial Extremes: Oceanic Drivers of Temperature and Precipitation in Europe, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13419, https://doi.org/10.5194/egusphere-egu26-13419, 2026.

Human-induced climate change is leading to an increase in the intensity and frequency of some severe extreme weather and climate events, including compound extreme events (Seneviratne et al. 2021; Seneviratne et al., in preparation). This presentation will provide an overview of recent literature on this topic in the context of climate projections, as well as in relation to climate adaptation and mitigation. A recent study assessing projected changes in concurrent extreme events at country level at different levels of global warming reveals an increasing probability of near-permanent extreme conditions in most countries of the world with increasing global warming (Batibeniz et al. 2023). An analysis of changes in spatially compounding hot, wet and dry events with increasing greenhouse gas forcing additionally reveals that the spatial extent of top-producing agricultural regions threatened by climate extremes will increase drastically if mean global warming shifts from +1.5 ^∘C to +2.0 ^∘C, and possibly higher levels of global warming (Biess et al. 2024). Some compounding changes also come from the clear increase in the number of extreme events affected by human-induced climate change, as recently shown for heatwaves on global scale (Quilcaille et al. 2025). Further new results highlight how changes in climate extremes and compound events constrain potential future options for climate mitigation and adaptation, and why they need to be integrated in the development of plausible emissions and adaptation scenarios for the coming decades.

References:

Batibeniz, F., M. Hauser, and S.I. Seneviratne, 2023: Countries most exposed to individual and concurrent extremes and near-permanent extreme conditions at different global warming levels. Earth Syst. Dynam., 14, 485–505, 2023 https://doi.org/10.5194/esd-14-485-2023

Biess, B., L. Gudmundsson, and S.I. Seneviratne, 2024: Future changes in spatially compounding hot, wet or dry events and their implications for the world’s breadbasket regions. Environ. Res. Lett. 19, 064011, https://doi.org/10.1088/1748-9326/ad4619.

Quilcaille, Y., L. Gudmundsson, D.L. Schumacher, T. Gasser, R. Heede, C. Heri, Q. Lejeune, S. Nath, P. Naveau, W. Thiery, C.-F. Schleussner, and S.I. Seneviratne, 2025: Systematic attribution of heatwaves to the emissions of carbon majors. Nature, https://doi.org/10.1038/s41586-025-09450-9.

Seneviratne, S.I., X. Zhang, M. Adnan, W. Badi, C. Dereczynski, A. Di Luca, S. Ghosh, I. Iskandar, J. Kossin, S. Lewis, F. Otto, I. Pinto, M. Satoh, S.M. Vicente-Serrano, M. Wehner, and B. Zhou, 2021: Weather and Climate Extreme Events in a Changing Climate. In Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change [Masson-Delmotte, V., P. Zhai, A. Pirani, S.L. Connors, C. Péan, S. Berger, N. Caud, Y. Chen, L. Goldfarb, M.I. Gomis, M. Huang, K. Leitzell, E. Lonnoy, J.B.R. Matthews, T.K. Maycock, T. Waterfield, O. Yelekçi, R. Yu, and B. Zhou (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, pp. 1513–1766, doi:10.1017/9781009157896.013.

Seneviratne, S.I. et al., in preparation: Extreme climate events from past to future: A 5-year update since the IPCC AR6 report. Manuscript in preparation.

How to cite: Seneviratne, S. I., Batibeniz, F., Biess, B., Schöngart, S., Schumacher, D., Bauer, V., Gudmundsson, L., Hauser, M., Hirschi, M., Quilcaille, Y., Seeber, S., and Windisch, M.: Compound extreme events in a warming climate: Implications for climate change adaptation and mitigation, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14314, https://doi.org/10.5194/egusphere-egu26-14314, 2026.

In early 2016, Kenya experienced a whiplash between two opposing extreme events: extreme heat in March followed by heavy rainfall in April. In particular, the North-East region (Mandera, Wajir, Isiolo, Marsabit, and Samburu counties) endured an ‘ultra extreme’ 20-day heat event, defined using the Heat-Wave Magnitude Index daily (HWMId), followed closely by a 4-day heavy rainfall period. This type of compound event, which involves a succession of individual events, is termed a ‘temporally-compounding event’ and can be particularly devastating as the initial event ‘preconditions’ the human and physical environment, thereby exacerbating the impacts of the second event.

There is a dearth of literature on compound events in East Africa, despite their increasingly common nature. Here we present an attribution methodology to disentangle the mechanisms driving temporally-compounding events to fill this gap.  While attribution studies are still predominantly performed on individual extreme events, those which do consider compound events tend to focus on co-occurring multivariate events. The attribution of temporally-compounding events is, however, still in its infancy.

There are an additional range of factors to consider when attributing the drivers of a succession of hazards when compared to an individual extreme event. We build upon existing proposed methodologies to navigate these complicating factors, such as deciding between univariate or multivariate thresholds for event definitions, and deciding the ‘reasonable’ time interval between the cessation of the first event and the instigation of the second.

This research aims to contribute to the shared understanding of the interactions between the mechanisms driving compound events, specifically temporally-compounding events, within an East African context. This improved understanding can be used to inform locally-specific compound event definitions which can ultimately inform effective early-warning systems. By determining the relative contributions of anthropogenic climate change and natural variability on the 2016 Kenyan event, this research also hopes to lay the foundation for future attribution studies on compound events in the region.

How to cite: Eyles, H., Otto, F., Kimutai, J., Barnes, C., and Keeping, T.: The Attribution of Temporally Compounding Events: A Study on North-East Kenya, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14496, https://doi.org/10.5194/egusphere-egu26-14496, 2026.

Compound Heatwaves (CoHots), characterized by persistent day-night combined high temperatures, have intensified in recent decades, posing growing threats to human health, productive activities, and socioeconomic systems. Although much research has focused on the evolution of CoHots, high-resolution mapping of their changes in large metropolitan areas remains limited by sparse observational networks and coarse-resolution reanalysis data. Additionally, the influence of urbanization on the onset timing of CoHots has received little attention.

This study compares the start dates of CoHots across more than 700 urban–rural station pairs worldwide, revealing a significantly earlier onset in urban areas. Using machine learning and SHAP interpretability analysis, we demonstrate that this effect is primarily driven by urban building volume and height, rather than by the fraction of impervious surfaces. The influence is further amplified in climates with warm nights and strong daytime solar radiation.

To quantify urbanization's impact at a spatially and temporally continuous scale, we developed the Urban-informed Heatwave Ensemble AI Downscaling (U-HEAD) framework. This model integrates dynamic urbanization factors through an ensemble machine learning approach to downscale 0.25° ERA5 reanalysis data to 1 km resolution. Compared to the original product, U-HEAD substantially improves the simulation of spatiotemporal patterns and long-term trends of compound heat events. The framework can also be integrated with statistical downscaling methods to generate future high-resolution projections of CoHot evolution under combined climate change and urbanization scenarios. This research provides a robust, high-resolution modeling tool to quantify urbanization’s role in shaping compound heat extremes. Future work will focus on applying U-HEAD to project CoHot risks under various climate and urban development pathways, and to inform climate-resilient urban planning and heat adaptation strategies.

How to cite: Ji, P.: Harnessing machine learning for quantifying and attributing compound heatwave changes in metropolis, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16419, https://doi.org/10.5194/egusphere-egu26-16419, 2026.

In a rapidly changing climate, various types of compound climate events (CCEs) have been widely analysed at both global and regional scales recently. Yet in the Baltic States, they have scarcely been studied. In this research three different CCEs were analysed: compound drought and heatwave events (CDHE), late spring frost events (FS events), and compound precipitation amount and wind speed extremes (CPWE). The aim of this study was to examine the recurrence, intensity, and spatial distribution of these CCEs from 1950 to 2022, and to assess projected changes in their characteristics by the end of the 21^st century in the eastern part of the Baltic Sea region.

ERA5 reanalysis data were used to identify CDHEs, FS events, and CPWEs during the 1950–2022 period. Future projections were derived from five CMIP6 climate models using the NASA Earth Exchange Global Daily Downscaled Projections (NEX–GDDP–CMIP6) dataset under the SSP2–4.5 and SSP5–8.5 scenarios. Changes were assessed by comparing the period from 2081 to 2100 with a baseline period from 1995 to 2014. CDHEs were identified by calculating daily Standardised Precipitation Index (SPI) values to distinguish droughts and by defining heatwaves using the 90^th percentile of daily maximum air temperature. CDHEs occurred when drought and heatwave conditions coincided. FS events were detected when the last spring frost occurred after the start of the growing season. Finally, CPWEs were defined as days when both precipitation and maximum wind speed exceeded their respective 98^th percentiles at the same grid cell.

During 1950–2022, the number of CDHE days increased in over 75% of grid cells, mainly driven by a widespread rise in heatwave days (> 99% of grid cells). Also, FS events increased across more than 80% of the study area, while CPWEs became more frequent in 70.2% of grid cells. However, in most cases, the observed changes were small. They were statistically significant (p < 0.05) in less than 10% of the study area. Depending on the model and scenario, future projections indicate an increase in the number of days with CDHEs by the end of the century, with an average rise of 0.8–18.3 days/year. These events are also projected to become longer and more intense. CPWEs are expected to increase by 0.7–4.5 events/decade. Only the projections for FS events are uncertain, with different models indicating either increases or decreases in both frequency and intensity.

Distance from the Baltic Sea was found to have a strong influence on the spatial distribution of CCEs, with the highest number of CPWEs occurring in the western part of the study area. On the contrary, FS events and CDHEs occurred more frequently farther from the Baltic Sea coast. The results of this study suggest a potential increase in risks associated with CCEs in the Baltic States, underscoring the need for evaluations of climate adaptation strategies.

How to cite: Klimavičius, L. and Rimkus, E.: Spatiotemporal variability and future projections of compound climate events in the eastern part of the Baltic Sea region, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17338, https://doi.org/10.5194/egusphere-egu26-17338, 2026.

Recent studies have revealed strong trends in humid heat, including the nearing of human physiological limits in some regions. Understanding of past extremes and their meaningfulness for contextualizing future possibilities, especially in the near-term, is limited by the absence of a global analysis focused on the most extreme humid-heat-anomaly events. Here we identify record-setting humid-heat days for 216 global regions and assess the likelihood of these records being broken even under present-day climate forcing. We use several reanalyses as a historical catalogue, and large climate-model ensembles to represent other statistically plausible events. Unlike the spatial pattern of large temperature anomalies, we find that humid-heat anomalies are most intense, and most seasonally and interannually concentrated, in the deep tropics and arid subtropics. Many top events have attracted little if any prior attention. The eastern United States is especially susceptible to record-breaking humid heat due to modest current records (>1% inferred annual exceedance probability) contrasting with numerous simulated large-anomaly days. Australia and eastern China are also prone to locally exceptional episodes, with >40% of ensemble members simulating events exceeding the ERA5-based distribution maximum. Model biases for key characteristics, together with the observed record-setting day affecting its estimated return period by >2.5x in half of regions, underline several valuable aspects of a joint observation/model perspective on humid heat. This approach aids in evaluating the plausibility of as-yet-unseen extremes; identifying regions of concern that might otherwise be overlooked and underprepared; and gauging regionally specific correlations between event magnitudes and societal impacts.

How to cite: Thompson, V., Raymond, C., Suarez Gutierrez, L., and van der Wiel, K.: Distinct Favoured Regions for Historical Record-Setting and Future Record-Breaking Humid Heat, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17662, https://doi.org/10.5194/egusphere-egu26-17662, 2026.

In May 2023, two severe hydrometeorological events affected Emilia-Romagna (Italy), on May 2–3 and May 16–17, respectively. These events led to widespread, concurrent flooding across multiple river basins, triggered by levee overtopping and embankment failures, and impacted 37 municipalities throughout the region. This study presents a compound-event analysis of the two events, employing a dual methodological framework. First, a bottom-up, impact-based assessment was conducted, starting from documented damages and tracing back to the underlying meteorological drivers. Subsequently, a top-down, model-based analysis was performed to investigate the dynamics of these events and quantify the relative contributions of hydrometeorological and geomorphological factors to the flood events. In addition, a sensitivity analysis was conducted to assess the influence of the considered forcings and parameter choices on the robustness of the results. This integrated framework provides new insights into the dynamics and drivers of compounding flood hazards.

How to cite: De Michele, C., Banfi, F., and Russomando, M. P.: The May 2023 Flood Events in Emilia-Romagna, Italy: A Compound-Event Perspective, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17670, https://doi.org/10.5194/egusphere-egu26-17670, 2026.

Water supply systems face their greatest challenge when stream flows decline while demand surges during extreme heat. When heatwaves coincide with low stream flows, low flows diminish thermal capacity and wetted surface area, amplifying thermal sensitivity to atmospheric forcing. The resulting impacts cascade across multiple sectors: elevated river water temperatures stress aquatic ecosystems beyond critical thresholds, thermal power plants struggle to access adequate cooling water during peak energy demand, concentrated pollutants in warm, stagnant water degrade drinking water quality, and irrigation withdrawals intensify competition among agricultural, industrial, and municipal users.

Low flow data traditionally inform how much water stakeholders can safely withdraw for domestic use, industry, agriculture, and energy production while preserving river ecosystems. However, examining low flows in isolation fails to capture how concurrent extreme heat amplifies these stresses and triggers cascading failures across interconnected water-dependent systems.

Here we quantify the spatial and temporal evolution of compound low-flow—heatwave events across European rivers from 1960–2020 using observed air temperature and high-resolution pan-European hydrological reanalysis data. We identify regional hotspots characterized by the most frequent, longest, and most intense compound events and assess changes in event frequency, duration, and intensity between historical (1961–1990) and recent (1991–2020) climatic periods. We further analyze the dominant drivers of these changes across different European regions, including increases in the number and duration of low-flow events and the frequency of heatwaves occurring during low-flow periods.

Our analysis reveals that Central and Eastern Europe exhibit the most pronounced increases in compound event frequency, duration and severity, potentially experiencing the largest impacts from these events. We find a marked escalation in compound event severity, with heatwaves increasingly coinciding with low-flow conditions in recent decades. Critically, the longest-duration compound events—which pose greatest risk to aquatic ecosystems and water-dependent economic activities—have become significantly more frequent in recent decades. These results reveal expanding spatial coverage of simultaneous low flow and heatwave hazards, with implications for water resource management under continued warming.

How to cite: Dengri, A. and Greve, P.: More frequent and intense compound low-flow and heatwave events in European rivers since 1960, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17822, https://doi.org/10.5194/egusphere-egu26-17822, 2026.

Understanding the spatial and temporal evolution of high-impact compound droughts, arising from interacting water-supply and atmospheric-demand drivers, is crucial for assessing ecosystem risks under climate change. Here, we analyze global compound droughts from 1983 to 2021 by integrating the Standardized Precipitation Evapotranspiration Index (SPEI) and vapor pressure deficit (VPD), representing water-supply and atmospheric-demand dimensions, respectively. A compound drought event is defined when SPEI < –1.1 and VPD ≥ the 90th percentile within the same grid cell and time period.

Results reveal a significant intensification and expansion of compound droughts over the past four decades. The probability multiplication factor (PMF) between SPEI and VPD exceeds 1 across most subtropical and continental interior regions, indicating strong co-occurrence of soil and atmospheric dryness. Subtropical high-pressure zones (15°–40° N/S)—including southwestern North America, the Mediterranean Basin, southern Africa, and southern Australia—emerged as global hotspots, with 2020 marking the historical peak in affected area.

Distinct land-cover contrasts were also observed. Forests and grasslands experienced the highest exposure frequencies, whereas tundra and cropland were less affected. After 2005, compound drought areas in forests and grasslands expanded markedly, consistent with the global rise in atmospheric aridity. These findings underscore the growing dominance of compound droughts as a global climate hazard and highlight the importance of jointly considering multiple interacting drivers in ecosystem risk assessments and future adaptation strategies.

How to cite: Yang, J., Zhang, R., Schaub, M., and Hagedorn, F.: Emerging Hotspots and Land-Cover Contrasts of Global Compound Droughts (1983–2021), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18035, https://doi.org/10.5194/egusphere-egu26-18035, 2026.

The 1997 New Year's Flood was the most costly flood in California history, a compound extreme event driven by a category 5 atmospheric river carrying extreme precipitation, and amplified by snowmelt and elevated antecedent soil moisture. Previous work has successfully recreated the event using regionally-refined model meshes, identifying the major drivers of the flood, demonstrating the importance of model horizontal resolution in representing runoff totals, and demonstrating the warming sensitivity of these flood drivers. However, analyses have stopped short of estimating the likelihood of a comparable flood occurring under future climate conditions. Estimating the likelihood of single variable, much less compound extremes, under future climate is challenging, due to the multiplicity of uncertainty sources, the challenges in navigating deep uncertainties, and the limitations of comparable large ensembles and simulation capabilities between climate models. Building on recent work proposing a new conceptual methodology for "probabilistic storylines", the work presented here addresses this gap by estimating conditional probabilities for the 1997 flood storyline using the Energy Exascale Earth System Model (E3SM). Univariate and multivariate thresholds are calculated using the E3SM reanalysis, and return likelihoods and risk ratios under 1.5°C, 2°C, and 3°C global warming levels are estimated using the E3SM historical and SSP370 large ensemble. Multivariate joint probabilities of compound flood drivers are evaluated using copulas. Results quantify how the likelihood of a 1997 flood-like compound extreme evolves under different warming levels, conditional on the global climate sensitivity and regional structural uncertainty of E3SM. 

How to cite: Longmate, J. and Rhoades, A.: Probabilistic Storylines: Conditional Likelihoods of the 1997 California New Year’s Flood Under Global Warming Levels, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18635, https://doi.org/10.5194/egusphere-egu26-18635, 2026.

Romania is a country situated in Southeast Europe, and due to its geographical position, it is exposed to different climatic hazards (heatwaves, floods, droughts). In the context of climate change, extreme weather phenomena are becoming increasingly frequent and intense, including in Romania. Compound events (CEs) are a combination of different hazards/climate drivers that can pose a significant risk to society and the environment. One type of CEs is multivariate events, when multiple hazards/climate drivers are co-occurring in the same geographical region. In this study, we analyzed multivariate events in Romania from 1940 to 2024 using CETD (Compound Events Toolbox and Dataset). This tool can generate the duration, frequency, and severity of compound events. Daily maximum temperature (tasmax), daily minimum temperature (tasmin), total precipitation (pr), mean surface wind speed (sfcWind), and mean wind speed at 500 hPa (preWind500) were extracted from ERA5 reanalysis dataset obtained from the Copernicus Climate Change Service (C3S) Climate Data Store (CDS), with a spatial resolution of 0.25°x0.25°. We selected three hazard pairs related to extreme hot temperature: hot-dry, hot-stagnation, and hotday-hot night. Each hazard was defined as follows: hot (tasmax ≥95th percentile), dry (pr<5th percentile), windy (sfcWind ≥ 95th percentile), and stagnation (sfcWind<3.2 m/s and preWind500 < 13 m/s). The results indicate that specific areas in Romania are vulnerable to these three compound events, with a significant trend over recent decades, pointing to the need for effective risk mitigation implementation.

This study was carried out within Nucleu Program, contract number 24N/03.01.2023 (SOL4RISC), project no. PN23360202 and Catalina – Roxana Bratu’s PhD project at the Faculty of Physics, University of Bucharest.  Contact: Drd. Catalina-Roxana BRATU, catalina.bratu@infp.ro. This work was supported by a grant of the Ministry of Education and Research, CCCDI - UEFISCDI, project number PN-IV-P6-6.1-CoEx-2024-0042, within PNCDI IV.

How to cite: Bratu, C.-R., Antonescu, B.-A., and Ene, D.: Temporal evolution of extreme weather in Romania (1940-2024), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-19037, https://doi.org/10.5194/egusphere-egu26-19037, 2026.

The anthropogenic shifts in the climate have triggered an unprecedented rise in climate extremes which have impacted millions of lives and caused trillions of dollars’ worth of damage. The climate drivers that cause these high impact events are usually spatially or temporally compounded. These compound climate extremes are extreme events that occur simultaneously, in close succession or due to drivers that are not implicitly extreme but become extreme when combined. The impact depends on the vulnerability and exposure of the stakeholders that define the risk. Compound climate extremes are exemplified through hot-dry such as compounding heatwaves and drought or hot-wet extremes such as compounding heatwaves and extreme precipitation. Pakistan is not a new to the occurrence of heatwaves and extreme precipitation however, the compounding of these extremes is a relatively novel field of study.

Pakistan’s climate adaptation and disaster management strategies predominantly focus on these extremes in isolation while compound climate extremes have been overlooked. To assess the scientific gap, we quantified the historical occurrences and intensities of these compounding extremes, we assessed sequential heatwaves and extreme precipitation in Pakistan from 1980 to 2024 over 47 meteorological stations across the country by employing daily observational datasets for daily maximum temperatures (˚C) and precipitation (mm).

The analysis recorded a total of 599 events for the study period. A rise in the frequency of these compounding extremes was recorded since 1980 for extreme precipitation following heatwaves events within 7 days, 5 days, 3 days and 1 day. These events are consolidated in the North and Northeastern stations of Pakistan. The highest duration for these events is recorded for 7 day interval event.

These compounding extremes are especially high risk as compared to isolated extreme events because of the smaller gap between their occurrences which leaves little to no time to respond. Moreover, these events not only cause heatwave associated morbidities and losses and damages but also lead to pluvial and flash flooding. The occurrence of these events especially in southern provinces, as depicted by the study, highlights the potentially high risk of impact as a consequence of the large population, underdevelopment, pervasive poverty, social inequalities, and crippling infrastructure which increases the exposure and vulnerability of the people to such events.

How to cite: Ijaz, S., Ullah, A., Ahmad, R., Khan, M. S., and Saeed, F.: Historical Evidence of Compound Heatwave and Extreme Precipitation in Pakistan, 1980-2024, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21490, https://doi.org/10.5194/egusphere-egu26-21490, 2026.

Although European agriculture faces escalating threats from climate extremes, current risk assessments focus on single hazards, overlooking compounding effects that drive devastating agricultural losses. This study presents a novel framework for assessing compound agroclimatic extremes in potato production across Europe, utilizing crop-specific physiological thresholds rather than generic meteorological definitions. We employed copula methods for extreme precipitation-temperature events and vine copula approaches for dry-cold-heat and wet-cold-heat compound events to characterize 32 compound extreme combinations across duration, intensity, magnitude, and frequency dimensions using ERA5 reanalysis data (1990-2024). Our analysis reveals striking spatial heterogeneity with pronounced north-south gradients. Mediterranean regions experience persistent hot-dry events lasting 3-4 days on average, but Northern Europe faces brief but intense cold-dry and hot-dry extremes. Key findings reveal nonlinear risk amplification under triple compound events, which exhibit intensity values 4-13 times higher than double events and magnitude anomalies 2-4 times greater. This amplification stems from synergistic interactions among temperature, precipitation, and land-atmosphere processes that generate cascading feedback exceeding individual hazard impacts. Cold-dry extremes emerge as the dominant threat, occurring 5-10 times more frequently than cold-wet combinations and at least 10 times more frequently than hot-wet extremes across central and northern Europe. Joint return period analysis reveals that severe hot-dry events occur every 1-3 years in Mediterranean hotspots, while moderate cold-dry events occur every 1-5 years across most of Europe. These results fundamentally challenge single-hazard agricultural risk frameworks and underscore the urgent need for adaptation strategies accounting for compound events. Our methodology, integrating crop-specific thresholds with comprehensive temporal characterization, provides a scalable and transferable approach for assessing agricultural climate risk across diverse cropping systems. The findings offer actionable insights for European potato cultivation planning and climate adaptation policies, highlighting critical hotspot regions where compound extremes pose the greatest threat to agricultural productivity and food security.

How to cite: Gohari, A., Saboori, M., Ghadimi, S., and Torabi Haghighi, A.: A mutivariate Framework for Assessing Compound Agroclimatic Extremes, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-21668, https://doi.org/10.5194/egusphere-egu26-21668, 2026.

Learning Compound Climate Extremes: Generative AI for Hot–Dry Event Risk

In recent years, there has been a growing interest in the applications of machine learning methods to multi-hazard events, mainly due to their ability to ingest large amounts of data and capturing the relationships between variables. Compound weather and climate events (CEs) are of significant societal importance, as they present greater risks, and better understanding their response to climate change is crucial. This response has mostly been explored through dynamical climate model ensemble methods. However, accurately estimating the uncertainty of climate scenarios often requires very large ensemble simulations to be conducted, which can be computationally costly.

Generative deep learning methods offer a cost-effective alternative by enabling the generation of large sets of synthetic events which follow the joint distribution of high-dimensional data.

This work builds on the HazGAN framework, an ML framework which generates synthetic event sets for risk analysis of specific CEs in defined regions, capturing the dependence structure among variables. We utilise a HadGEM3 Large Ensemble to further condition the model on the large-scale climate background state, allowing the characteristics of the synthetic events to vary with different climate regimes. This approach aims to account for non-stationarity in compound event behaviour and to provide a physically consistent framework for exploring changes in hot–dry extremes under a changing climate. We also address uncertainties associated with using machine learning for extrapolation by rigorously testing out-of-distribution predictions. This work enhances the understanding of compound events, their risks, and future impacts under climate change scenarios

How to cite: Breitburd, T. and Colfescu, I.: Learning Compound Climate Extremes: Generative AI for Hot–Dry Event Risk, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-22138, https://doi.org/10.5194/egusphere-egu26-22138, 2026.

Compound climate and weather extremes are increasingly recognized as key drivers of high-impact events, yet existing frameworks to assess their risk are often sector-specific and thus not broadly applicable. In this study, we develop an impact-based framework for characterizing and quantifying compound events and apply it to two case studies in the Netherlands. Our approach links multivariate meteorological conditions (the physical drivers) to sector-specific vulnerabilities, employs cut-offs based on stakeholder expertise, and draws on publicly available datasets (such as ERA5) to evaluate compound event risk. We apply the framework to two case studies: first, in the renewable-energy sector, we assess the occurrence of wind and solar daily "droughts" that lead to critical shortfalls in renewable power generation; second, in the agricultural sector, we analyze compound temperature–moisture constraints on crop primary productivity, quantifying the probability of extreme conditions detrimental to plant growth. Across both sectors, we employ empirical statistical methods to evaluate the frequency and co-occurrence of critical impacts, producing spatial estimates of return intervals for critical events, and assessing the tail-dependence of critical variables. We produce spatial representations of risk for both cases, which allow for the minimization of compound hazard potential in future planning where specific renewable energy mixes or crop types are assessed. For the energy sector, we identify the critical energy shortfalls as 2-, 3-, 4-, and 5-consecutive-day resource scarcity extremes and find their overall risk in terms of average return interval to be of 0.3-, 1-, 4-, and 5-year events based on 40 years of observations. The framework's reliance on identifying an impact of interest and characterizing key variables that control its associated risks enables its application across different sectors and regions, ideally supporting stakeholder engagement and decision-making.

How to cite: Fernandez Jimenez, A.: Impact-based analysis of compound hydroclimatic events in The Netherlands: case studies in energy and agriculture sectors, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-22701, https://doi.org/10.5194/egusphere-egu26-22701, 2026.

Compound hydroclimatic extremes, particularly dry-to-wet transitions, represent a growing climate risk due to their cascading impacts on flooding, agriculture, and water resources. Under climate change, shifts in the frequency and severity of such compound events are expected, yet large uncertainties remain in their detection, characterization, and future evolution. This study presents a probabilistic, ensemble-based assessment of compound dry-to-wet events across Pakistan, with explicit attention to event detection, severity classification, and uncertainty in climate projections. Compound dry-to-wet events are detected using the Standardized Precipitation Evapotranspiration Index (SPEI), capturing transitions from sustained dry conditions to subsequent wet extremes. To systematically characterize event severity, we develop a compound magnitude index that integrates the severity and duration of both the dry and wet phases of each event. This index enables the classification of compound dry-to-wet events into mild, moderate, severe, and extreme categories, facilitating robust comparisons across regions, models, and emission scenarios. The analysis is based on historical and future simulations from CMIP6 global climate models and CORDEX dynamically downscaled regional climate models under multiple Shared Socioeconomic Pathways (SSPs). Changes in the frequency, duration, intensity, and severity distribution of compound dry-to-wet events are evaluated relative to a historical reference period. Probabilistic metrics are used to quantify ensemble agreement and spread, while uncertainty is decomposed into contributions from model structure, scenario choice, and internal climate variability. Differences between CMIP6 and CORDEX ensembles are further examined to assess the role of regional downscaling in representing compound event characteristics. Results indicate an increased likelihood and severity of compound dry-to-wet events under higher-emission scenarios, with pronounced spatial heterogeneity across Pakistan. In particular, severe and extreme events show more robust increases than mild and moderate events. Model uncertainty dominates projections of compound event magnitude, while scenario uncertainty becomes increasingly important toward the late 21st century. Regional climate models enhance the representation of localized extremes but exhibit larger inter-model variability. This study advances compound event research by introducing a SPEI-based compound magnitude framework and a comprehensive uncertainty assessment, providing valuable insights for climate risk assessment and adaptation planning in climate-vulnerable regions.

How to cite: Imran, A.: Probabilistic Changes of Compound Dry-to-Wet Events: Detection and Uncertainty from Climate Model Ensembles, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-23179, https://doi.org/10.5194/egusphere-egu26-23179, 2026.

Exposure to fine particulate matter (PM_2.5) has been linked with adverse birth outcomes in Nigeria. Emerging evidence suggests that high temperatures may also be associated with these outcomes. However, this association, as well as whether temperature modifies the effects of PM_2.5 on these outcomes, has not been explored in Nigeria.
Using data from the 2018 Nigerian Demographic and Health Survey, we examined the association between maternal exposure to maximum temperature (Tmax) during pregnancy and adverse birth outcomes, including Low Birth Weight (LBW), Preterm Births (PTB), and Stillbirths (SB). A daily maximum near-surface air temperature gridded dataset (2012-2018) at 1-km² resolution was obtained from Zhang et al. (2022) and linked to birth clusters based on geographic coordinates. Temperature metrics (hot days and heatwave events) were derived from the 90th percentile threshold of the daily Tmax values, based on all pregnancy periods. Logistic regression analysis was used to estimate the association between these metrics and birth outcomes. The intensity, frequency, and duration of these temperature metrics in relation to the birth outcomes were also evaluated. We then estimated the Relative Excess Risk due to Interaction (RERI) using interaction terms for each temperature metric during the corresponding PM_2.5 exposure period.
We observed a strong correlation (r=0.93) between the model temperature data and observational data (2012-2018). An increasing positive association was observed between the duration of hot days and PTB, while an increase in heatwave events was positively associated with LBW. Intensity in hot days was positively associated (1.59; 95% CI: 1.28-1.96) with LBW. At the same time, frequency in hot days showed no significant relationship with any of the birth outcomes. Positive additive interaction between high temperature and PM_2.5 was observed across exposure categories for LBW and SB. The magnitude of interaction was greater at moderate PM_2.5 levels (Q2) for LBW, while the highest levels (Q3) had a greater effect for SB. As global temperatures rise, these findings provide evidence that maximum temperature can intensify the health burden of ambient PM_2.5 during pregnancy, underscoring the need for climate-adaptive maternal health interventions.

How to cite: Seyinde, D. and Dey, S.: Adverse Birth Outcomes Attributable to High Heat in Nigeria , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-694, https://doi.org/10.5194/egusphere-egu26-694, 2026.

Compound heatwave–high ozone pollution events (CHOPs) represent an emerging climate–health challenge, yet their drivers and long-term population impacts remain insufficiently quantified. Using 2000–2022 high-resolution climate and environmental datasets, together with updated epidemiological evidence for compound heat–ozone risks and machine-learning diagnostics, we show that CHOP occurrences in Eastern–Northern China (ENC) have risen by nearly 3.7‐fold since 2013—far exceeding the increases in isolated heatwaves (1.85-fold) and ozone events (2.66-fold). We identify Western Pacific Warm Pool (WPWP) warming as a dominant climatic precursor that strengthens tropical–midlatitude ocean–atmosphere coupling and reinforces a persistent barotropic high-pressure ridge over ENC. This circulation pattern produces simultaneous heat accumulation, stagnant ventilation, and enhanced photochemical ozone formation, thereby amplifying compound extremes beyond the sum of their individual components. The intensified CHOPs have markedly elevated health burdens. Among older adults, CHOP-related mortality risks have nearly quadrupled, while the associated economic losses now exceed 14.3 billion CNY annually—an increase of more than threefold compared to the early 2000s. These disproportionate impacts highlight the vulnerability of aging populations to compounding climate and air-quality stressors. By revealing the teleconnection pathways that modulate CHOP variability and quantifying their escalating human and economic costs, this study provides a scientific foundation for climate-informed seasonal forecasts, targeted early-warning systems, and equitable adaptation strategies. Our findings underscore the necessity of integrating large-scale climate precursors into compound-risk assessments to safeguard public health under a warming climate.

How to cite: Zhu, S.: Western Pacific Ocean Warming Intensifies Heat–Ozone Compound Extremes and Population Health Risks in China, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-705, https://doi.org/10.5194/egusphere-egu26-705, 2026.

Dengue continues to expand across Brazil under increasingly variable climatic conditions, and anticipating where infections may spread is essential for effective public health preparedness. However, most existing early warning systems focus on local case trajectories alone and overlook the spatial redistribution of infection risk driven by human mobility. This gap leaves planners without the ability to foresee where cases are likely to be imported before local transmission accelerates.

In this study, we develop a generalizable forecasting framework that couples climate-informed dengue incidence predictions with a multimodal mobility network covering all 5,570 Brazilian municipalities. Weekly dengue cases are forecasted using a long short-term memory (LSTM) model that incorporates temperature and humidity dynamics. These forecasts are combined with a composite mobility matrix spanning road, river, and air flows, allowing us to estimate the expected volume of imported infections between cities for every epidemiological week of 2024.

The resulting importation-risk surfaces reveal well-defined corridors of movement-mediated dengue spread, including strong directional asymmetries between major source regions (e.g., large urban hubs with intense outbound flows) and peripheral sink municipalities that depend heavily on external seeding. We find that high importation risk often precedes subsequent local increases in incidence, highlighting the added value of capturing human mobility in early warning systems.

This framework advances dengue surveillance by integrating climate variability, human mobility, and short-term predictive modeling into a unified pipeline. Beyond dengue and Brazil, the approach is modular and transferable to other climate-sensitive infectious diseases and mobility-rich settings. By quantifying how infections may spread through movement pathways before they emerge locally, this work provides a scalable tool for proactive, spatially targeted public health response in an era of intensifying climate-health risks.

How to cite: Chen, X. and Moraga, P.: Forecasting Dengue Importation Risk in Brazil Using Deep Learning and Multimodal Mobility Networks , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-843, https://doi.org/10.5194/egusphere-egu26-843, 2026.

Black Carbon Exposure as a Risk Factor for Child Health  in India

Rajesh Bag^1,2, Debajit Sarkar², Ram Pravesh Kumar¹, Sagnik Dey^2,3

¹School of Environmental Sciences, Jawaharlal Nehru University, New Delhi, India.

²Centre for Atmospheric Sciences, Indian Institute of Technology (IIT) Delhi, New Delhi, India.

³Adjunct faculty, Korea University, Seoul, South Korea.

Email: rajeshgeovu@gmail.com

Keywords: Black carbon; stunting; wasting; low birth weight.

Introduction

Black Carbon (BC), a short-lived climate pollutant and a major light-absorbing component of fine particulate matter (PM_2.5) plays a dual role in driving climate change and adversely impacting human health. In India, persistently high levels of ambient PM_2.5 are compounded by household air pollution from biomass combustion, resulting in chronic BC exposure across large sections of the population. Child undernutrition manifested as stunting, wasting, and Low Birth Weight (LBW) continues to be a critical public health challenge in India, contributing to elevated child morbidity, mortality, and long-term developmental deficits. Despite the biological plausibility linking BC exposure to quantifying associated health effects in the Indian context is limited. Addressing this gap, the present study investigates the association between chronic BC exposure and three key indicators of child undernutrition, thereby providing novel insights into the intersection of air pollution and child health.

Methodology

We utilized nationally representative data from the National Family Health Surveys (NFHS-4: 2015-16 and NFHS-5: 2019-21), comprising 437,908 children under five years of age. Among them 10,362 observations had missing mean BC exposure and 35,386 had missing information on fuel type, wealth index, mother Body Mass Index (BMI), mother age, mother education, residence, child sex and mother smoking status. These records were excluded from the analysis. After removing all missing values, the final analytic sample included 402,508 children. Monthly mean BC exposure (2010-2021) at 1 km × 1 km resolution was merged with geocoded DHS cluster coordinates (Dey et al., 2020). For stunting and wasting exposure was averaged from child birth to the month of interview. For in-utero exposure related to LBW, we averaged BC concentrations from 9^th months prior to birth through the month of birth. Generalized Linear Model (GLM) and Generalized Linear Mixed Models (GLMM) were used to estimate associations between long-term BC exposure and odds of stunting, wasting, and LBW, adjusting for household fuel type, mother education, mother wealth index, residence, mother age, mother BMI, child gender and mother smoking status. We estimated the exposure response relationship using a Generalized Additive Model (GAM) incorporating a cubic spline for BC. Effect modification by all covariates was evaluated using multiplicative interaction terms. Stratified ORs with 95% uncertainty intervals were reported only for significant interactions. All models were adjusted for the same covariates.

Results & discussions

 After adjusting for confounders, the odds of stunting and wasting increased to 1.03 (95% UI 1.026-1.032) and 1.04 (95% UI 1.026-1.032) respectively for each 1 μg/m³ increase in long-term ambient BC exposure . Under the GAM framework the exposure response curves for stunting and wasting showed a monotonic increase with rising BC levels.

How to cite: Bag, R.: Black Carbon Exposure as a Risk Factor for Child Health in India, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1311, https://doi.org/10.5194/egusphere-egu26-1311, 2026.

Healthcare systems across Europe face growing challenges from climate-related hazards such as heatwaves, extreme precipitation, poor air quality, allergen exposure, and vector-borne diseases. To support outpatient medical practices in adapting to these risks, the AdaptNet project developed the AdaptNet Climate-Health Toolbox, a comprehensive, practice-oriented suite of tools designed to build climate resilience within primary and specialist care. Developed jointly with ambulatory physicians, the toolbox integrates scientific evidence with pragmatic operational guidance and is freely accessible online (https://www.gesundheitsnetznuernberg.de/adaptnet-klima-toolbox/).

The toolbox consists of several complementary modules. An interactive nationwide risk map enables users to assess present and future climate-related health risks for any German region, covering hazards such as heat, floods, air pollution, allergens, wildfires, and vectors. Downloadable checklists provide actionable recommendations for extreme weather events, power outages, and heat preparedness, supporting structured team-based adaptation planning. A basic online training introduces essential climate-health knowledge, while advanced training modules deepen practical implementation through case-based learning and support for quality circles and workshops.

To enhance clinical management, the toolbox includes a heat-focused medication review tool, helping practitioners identify and adjust risk-relevant drugs during heat periods. For patient communication, customizable “info-prescriptions” on heat and pollen, posters, flyers, and waiting room videos convey clear behavioural guidance and increase awareness during high-risk periods. All components are designed for simple integration into routine workflows and can be adapted to local needs. Collectively, the toolbox provides a structured pathway for practices, from risk assessment to team coordination, patient counselling, and medical decision support, to strengthen resilience to climate change impacts.

AdaptNet is funded by the G-BA Innovation Fund (01VSF22044).

How to cite: Kaspar-Ott, I., Álvarez, F., Köhn, P., Gäbel, P., Herrmann, A., Hueber, S., Klanke, M., Lindenthal, J., Nieder, J., Shimada, D., Stark, S., Quitmann, C., Wambach, V., and Hertig, E.: The AdaptNet Climate-Health Toolbox: A Comprehensive Multi-Component Framework to Strengthen Climate Resilience in Outpatient Healthcare in Germany, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1560, https://doi.org/10.5194/egusphere-egu26-1560, 2026.

How to cite: Raffetti, E., Martellini O Nocentini, M., and Wybrant, M.: Hydroclimatic Extremes and Climate Variability as Drivers of Malaria Risk in Sub-Saharan Africa, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1635, https://doi.org/10.5194/egusphere-egu26-1635, 2026.

The effect of environmental parameters on glycaemic trends undoubtedly has clinical relevance, which needs to be managed appropriately by understanding the responses of patients treated with different therapeutic approaches. In general, it is possible to assess how glycaemic trends in diabetic patients respond to external temperatures, humidity and Humidex. This study presents the preliminary results obtained as part of the project "Innovation Ecosystem: innovation, digitalisation and sustainability for the widespread economy in central Italy (VITALITY)", funded by NextGenerationEU, on patients with type 2 diabetes and the findings of analyses carried out on children and young adults (type 1 diabetes) followed at the UOSD Regional Paediatric Diabetes Service Hospital, ‘SS. Annunziata’ Hospital. The study was performed to assess the effect of climate change on diabetic patients;  to this end, a correlation analysis between atmospheric temperature, humidity and Humidex trends with respect to blood glucose patterns was carried out both on the entire sample of patients followed at the Lanciano-Vasto-Chieti (Abruzzo, Italy) Local Health Authority (approximately 200,000 subjects) over 5 years (2019-2023), and on precision basis, following a subset of approximately 50 patients with type 2 diabetes, intensively for one week during this year (2025). A parallel analysis was conducted over a period of one year (Autumn 2022 - Summer 2023) on 219 patients with type 1 diabetes, evaluating their glycaemic trends in relation to outdoor temperatures [1,2]. Results showed a close correlation between atmospheric conditions and blood glucose levels at every stage of analysis, highlighting the importance of considering environmental parameters, such as outdoor temperatures and humidity in the study of chronic diseases like diabetes.

[1] Mascitelli, Alessandra, Stefano Tumini, Piero Chiacchiaretta, Eleonora Aruffo, Lorenza Sacrini, Maria Alessandra Saltarelli, and Piero Di Carlo. 2025. "Effect of Atmospheric Temperature Variations on Glycemic Patterns of Patients with Type 1 Diabetes: Analysis as a Function of Different Therapeutic Treatments" International Journal of Environmental Research and Public Health 22, no. 12: 1850. https://doi.org/10.3390/ijerph22121850

[2] Chiacchiaretta, Piero, Stefano Tumini, Alessandra Mascitelli, Lorenza Sacrini, Maria Alessandra Saltarelli, Maura Carabotta, Jacopo Osmelli, Piero Di Carlo, and Eleonora Aruffo. 2024. "The Impact of Atmospheric Temperature Variations on Glycaemic Patterns in Children and Young Adults with Type 1 Diabetes" Climate 12, no. 8: 121. https://doi.org/10.3390/cli12080121

How to cite: Mascitelli, A., Chiacchiaretta, P., Staropoli, M. C., Aruffo, E., Tumini, S., Ferretti, A., Franciotti, R., La Fratta, I., Lucarelli, F., and Di Carlo, P.: Climate change and diabetes: preliminary results on patients with type 1 and type 2 diabetes in the Abruzzo Region (Italy), EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-2147, https://doi.org/10.5194/egusphere-egu26-2147, 2026.

Air pollution has been investigated as a potential risk factor for breast cancer [1]; however, its quantitative impact on malignancy risk stratification remains uncertain, particularly when integrated with radiological features. In this study, we investigate whether long-term exposure to air pollution — a climate-sensitive environmental stressor — derived from Copernicus Atmosphere Monitoring Service (CAMS) reanalysis data provides complementary information for predicting breast lesion malignancy in a screened population.

We analysed mammographic and clinical data from 906 women undergoing breast cancer screening, classified as benign (BI-RADS B2) or malignant (BI-RADS B5). Individual exposure to NO₂, PM₂.₅, PM₁₀ and O₃ was estimated by linking the zip code of residence to CAMS gridded concentrations, computing both annual mean levels and cumulative exposure over the three years preceding diagnosis. Environmental exposure metrics were integrated with radiological descriptors, including lesion morphology, margins and breast density patterns, together with demographic information.

To reduce model complexity and limit overfitting, univariate feature selection was applied using an ANOVA F-test (p < 0.05) prior to training a feed-forward neural network. Model performance was assessed using independent validation data and compared with models excluding environmental exposure variables.

The integrated model achieved a ROC-AUC of 0.78, with balanced accuracy and a weighted F1-score of 0.73. Radiological features such as spiculated margins and irregular lesion shape remained the strongest predictors of malignancy; however, cumulative NO₂ and PM₂.₅ exposure metrics retained independent statistical significance and contributed to model performance. Limiting partially redundant air-quality metrics decreased apparent predictive power but improved model stability and interpretability, highlighting the potential impact of spatial and exposure-related confounding in observational datasets.

These findings suggest that long-term air-pollution exposure, as quantified using Copernicus atmospheric reanalysis products, provides a modest but consistent contribution to breast lesion malignancy risk stratification when combined with mammographic features [2]. This study demonstrates the feasibility of integrating atmospheric reanalysis data with clinical imaging information for exploratory environmental health applications, while underscoring the need for geographically robust validation and cautious interpretation of causality.

[1] White AJ, Bradshaw PT, Hamra GB. Air pollution and Breast Cancer: A Review. Curr Epidemiol Rep. 2018 Jun;5(2):92-100. doi: 10.1007/s40471-018-0143-2. Epub 2018 Mar 27.  

[2] Fiore, M.; Palella, M.; Ferroni, E.; Miligi, L.; Portaluri, M.; Marchese, C.A.; Mensi, C.; Civitelli, S.; Tanturri, G.; Mangia, C. Air Pollution and Breast Cancer Risk: An Umbrella Review. Environments 2025, 12, 289. https://doi.org/10.3390/environments12050153

How to cite: Chiacchiaretta, P., Dotta, F., Staropoli, M. C., Aruffo, E., Mascitelli, A., Sallese, I., Delli Pizzi, A., and Di Carlo, P.: Spatiotemporal Machine Learning Integration of Atmospheric Reanalysis and Mammographic Data for Breast Lesion Malignancy Prediction, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-2149, https://doi.org/10.5194/egusphere-egu26-2149, 2026.

Mosquito-transmitted diseases, particularly dengue, chikungunya, and Zika pose an increasing public health challenge in Southeast Asia where climate change and rapid land-use change are altering transmission dynamics and associated risks. As primary vectors of these diseases, Aedes aegypti predominates in densely urbanized and peri‑urban environments, exploiting artificial containers for oviposition, while Ae. albopictus, historically found in rural and suburban areas, tends to expand its ecological range worldwide and occupies a broader range of landscapes, including forested and peri‑urban areas. Temperature, rainfall, and humidity influence their survival and reproduction, shaping where each species can thrive under different climatic conditions. These contrasting preferences reflect specific climatic tolerances and landscape associations observed along gradients throughout the region.

Climatic factors such as temperature, precipitation, and relative humidity are identified as determinants for distribution and abundance of Ae. aegypti and Ae. albopictus, although their effects vary seasonally and geographically. Remote sensing and GIS-based studies have further highlighted the role of vegetation indices and urban land cover in shaping vector suitability. Geographic gaps exist in Southeast Asia, where most species distribution modeling studies are limited to local or national scales. This is largely due to the lack of standardized and comprehensive mosquito occurrence data, the concentration of studies in more easily accessible areas, and the challenges of harmonizing data across countries. As a result, regional models remain limited, impeding comprehensive assessment of the environmental drivers of Aedes at broader scales. Many existing models lack justification for variable selection and rarely address multicollinearity among predictors, limiting interpretability and robustness. To fill these gaps, standardized methodologies must be put in place that rigorously test correlations between environmental determinants in order to improve the predictive capacity of distribution models relevant to public health planning and vector control.

Here, we develop a species distribution modeling (SDM) framework that combines statistical and machine learning approaches to quantify the environmental drivers of Ae. aegypti and Ae. albopictus across mainland Southeast Asia. By combining entomological data from Global Biodiversity Information Facility (GBIF) and local datasets provided by project partners, and integrating satellite-derived land-cover classifications, landscape metrics, and high-resolution bioclimatic variables, we evaluate the importance of climatic and landscape predictors while considering collinearity and scale effects. Model performance is evaluated using spatial cross-validation to ensure transferability across countries. Our results provide spatially explicit maps of Aedes mosquitoes habitat suitability and identify key environmental determinants driving current distributions across Southeast Asian countries. We discuss how these determinants may evolve under ongoing and future climate change and on the potential consequences for Aedes suitability patterns and implication for climate-sensitive disease risk. This perspective highlights the relevance of our findings for surveillance prioritization, targeted vector control strategies, and the development of data-driven early warning systems supporting climate-resilient public health planning in Southeast Asia.

How to cite: Teillet, C., Hoeun, S., Huynh, T. T. T., Boyer, S., and Herbreteau, V.: Climatic and landscape drivers of Aedes aegypti and Aedes albopictus mosquito distributions in mainland Southeast Asia, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-4826, https://doi.org/10.5194/egusphere-egu26-4826, 2026.

Heat conditions pose a substantial threat to population health, with certain groups being particularly vulnerable. People with mental health disorders may be especially at risk due to structural and social stressors (e.g., living environment, limited access to cooling), physiological factors (e.g., medication effects on thermoregulation), and psychological factors (e.g., reduced self-care). This study is part of the Austrian climate–health project Parahsohl, in which we develop a data-driven workflow to assess health-relevant heat indicators, focusing on individuals with mental health disorders in the federal state of Tyrol. The objective is to develop regression models linking hospitalizations to weather conditions while accounting for relevant confounders, enabling interpretation in an impact-based weather-warning context and providing a basis for subsequent climate risk assessments.

The analytical workflow comprises three stages: (i) data preparation, (ii) model development, and (iii) model evaluation. Daily hospital admissions (n = 83,673) between May and September from 2007–2023 were used as the response variable for the nine administrative districts of Tyrol, focusing on mental and behavioural disorders (ICD-10 diagnosis codes: F00–F99). Weather predictors were derived from high-resolution (1×1 km) gridded observation data (SPARTACUS) for the same time period and aggregated using a population-weighted approach. Thus, we could account for differences in exposure between densely and sparsely populated areas. Lag variables over multiple temporal windows were generated for key meteorological metrics to capture delayed health effects. Hospitalization counts were modelled using generalized additive mixed models (GAMMs) with a negative binomial distribution. Weather variables were included as fixed effects, while day-of-week, year, and district were treated as random effects. Population offsets allowed incidence-based interpretation. Model performance was evaluated using standard statistical criteria for fit and predictive accuracy, and predictive skill was further assessed through temporal cross-validation across years and months. Initial results indicate that higher daily mean temperatures are significantly associated with increased hospitalization counts, with lagged temperature effects further enhancing model performance. Partial-effect plots and relative risk estimates provide interpretable quantitative measures of heat-related impacts on mental health outcomes. For instance, the hottest day in the study period was associated with an estimated increase in hospitalization risk exceeding 10% compared with average summer conditions.

As a next step, the analysis will be extended to a more detailed examination of diagnostic subgroups to better identify particularly vulnerable populations. These results will be presented at EGU2026. This study provides a quantitative assessment of heat-related impacts on mental health hospitalizations and contributes to the development of evidence-based indicators applicable for short-term applications (e.g., user specific impact-based weather warnings) as well as long-term climate risk assessments.

How to cite: Steger, S., Wittholm, J., Baier, K., Schneider, M., Bügelmayer-Blaschek, M., Hofer, L., Kienberger, S., and Brugger, K.: Linking Heat Conditions to Mental Health Hospitalizations: A Data-Driven Analysis for Tyrol, Austria, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5303, https://doi.org/10.5194/egusphere-egu26-5303, 2026.

Extreme temperatures are the leading cause of climate-related mortality worldwide. To inform mitigation and adaptation strategies, it is crucial to have accurate feels-like temperature measures that quantify thermal stress on human physiology. The Universal Thermal Climate Index (UTCI) is among the most widely used feels-like temperature metrics in climate-health research, applicable in a large range of weather conditions. UTCI is also used by several national and international weather forecasting services to predict thermal stress and issue warnings. However, because of the high complexity of the full UTCI model and associated computational cost, it is operationally approximated by a high-order polynomial to increase computational efficiency.

Here, we demonstrate that a carefully trained and robustly tested neural network model calculates UTCI with significantly greater accuracy compared to the polynomial approximation used in the literature. The neural network model substantially outperforms the polynomial model with a similar computational cost, reducing the approximation error by 86%— from 2.78°C to 0.38°C— and thermal stress misclassification by 76%. We eliminate the need to exclude wind speeds above 17m/s from UTCI calculation, which currently limits the global application of the polynomial approximation. When applied to ERA5 reanalysis data, our model reveals a 25% operational difference in daily heat stress categorization between the two methods in Rome, Italy during the 2003 European heatwave. We provide our UTCI model as openly accessible software, as a more accurate way to calculate UTCI in operational procedures. The neural network UTCI model has the potential to enhance climate-health risk research and improve the accuracy of public weather warning systems. 

How to cite: Pastine, B., Klöwer, M., Tang, T., Wilson Kemsley, S., and Slater, L.: An upgraded neural network-based operational procedure for the Universal Thermal Climate Index (UTCI) , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5838, https://doi.org/10.5194/egusphere-egu26-5838, 2026.

Introduction
Climate change is increasingly associated with extreme weather events that disrupt healthcare delivery, yet the system-wide health consequences of these disruptions remain poorly quantified. While damage to health facilities following extreme events is well documented, far less is known about how climate-related disruptions to service accessibility propagate through health systems and affect population health. In Malawi, for example, Cyclone Freddy in 2023 led to the closure of at least 79 healthcare facilities, in some cases for several months, substantially reducing access to care in already resource-constrained settings.

Methods
We use Malawi, one of the world’s most climate-vulnerable countries, as a case study to investigate the interactions between extreme precipitation, health system functioning, and population health. We integrate empirically-estimated damage functions for the impact of precipitation on healthcare service delivery into Thanzi La Onse, an all-disease, health-system model calibrated to Malawi. Using two climate and socioeconomic futures (SSP2-4.5 and SSP5-8.5), we project the impacts of climate-related disruptions on healthcare access between 2025 and 2040, accounting for heterogeneous healthcare-seeking behaviour and changes in service accessibility.

Results and Discussion
We estimated, for the first time, the population health impact of precipitation-mediated disruptions to healthcare services. We estimate that up to 4% of healthcare appointments may be disrupted by precipitation-related events over the study period. Disruptions disproportionately affect conditions requiring continuous or long-term engagement with care, such as chronic pain, mental health conditions, and contraceptive services (Figure), where interruptions increase the likelihood of individuals falling out of care entirely. Additionally, acute care such ante- and postnatal care were disrupted. Despite these effects, we project only modest changes in aggregate DALYs, reflecting both pre-existing barriers to healthcare access and conservative assumptions regarding the scope of service disruption. Notably, our analysis does not yet capture complete precipitation-driven changes in disease prevalence, suggesting that our estimates likely represent a lower bound of true impacts. Nonetheless, the projected scale of disruption highlights a substantial and growing strain on healthcare systems under climate change, particularly in rural and infrastructure-poor areas. Future work will extend this framework to explicitly model facility closures, transport disruptions, and climate-sensitive diseases, providing a more comprehensive assessment of health system vulnerability and resilience.

Figure: Disruption to appointments due to precipitation-mediated disruptions under the SSP2.45 scenario (compared with the "no dispruption" Baseline) between 2025 and 2040. Those services that either required a long-term engagement with care, or were acute, were most affetced. 

How to cite: Murray-Watson, R. and the TLO Modelling Team: Healthcare Disruptions and Health System Resilience under Climate Change in Malawi, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6736, https://doi.org/10.5194/egusphere-egu26-6736, 2026.

Early warning systems (EWS) for environmental hazards are increasingly implemented across regions and play an important role in protecting population health. However, most existing systems remain predominantly hazard-based rather than health impact-based, relying on simple threshold exceedances and colour-coded alerts. While such warnings provide a useful first indication of elevated risk, they often lack direct relevance for health-protective decision-making, action and behaviour by, e.g., local authorities, health services, care facilities, employers, and the public. This limitation is particularly important for heat and air pollution, which are typically addressed through separate warning systems despite strong epidemiological evidence that their health effects interact. Numerous studies show synergistic effects on death and disease from concurrent exposure to high temperatures and air pollution (PM₂.₅ and ozone), especially from cardiovascular and respiratory causes, implying that health risks during compound events exceed the sum of individual hazards. Failing to consider these interactions may therefore result in underestimation of risk during such events. Moreover, most available evidence on joint heat and air pollution health risks comes from temperate, high-income settings, while many of the world’s hottest regions are those also experiencing the highest air pollution levels.

Beyond improving risk detection, joint consideration of heat and air pollution offers a major opportunity for health co-benefits. Analyses from the Horizon 2020 RIA project EXHAUSTION show that accelerated air-pollution reduction can function as a powerful adaptation strategy to extreme heat. Integrating heat–air pollution interaction effects into regional mortality and welfare-cost projections revealed that achieving WHO’s annual PM₂.₅ guideline level could reduce heat-related cardiopulmonary mortality by nearly 40% in Europe over the coming decades. Particularly large benefits were found for Balkan and Mediterranean regions where high heat exposure and air pollution coincide and the annual welfare economic costs from heat-related mortality reach billions of Euros.

In the ongoing COPE project in India – a country experiencing increasingly frequent and intense heatwaves and home to many of the world’s most polluted cities – we directly respond to the evidence on joint heat and air pollution effects. Working with local partners, we aim to develop an Early Warning and Decision-support System for heat and air pollution in Delhi and Kolkata. We investigate whether alert thresholds should be dynamically adjusted when heat and air pollution co-occur, and whether vulnerability varies by season, warranting differential season-specific alert thresholds. We draw upon insights from the Horizon CSA project ENBEL, which highlight key technical (data and modelling constraints), structural (institutional capacity and funding), and societal (risk communication and equity) barriers to effective heat-health warning systems, as these lessons are directly applicable to the development of integrated heat–air pollution warnings. In COPE, the research is co-designed and conducted in collaboration with user groups from vulnerable populations and stakeholder partners from the health and governmental sector to ensure that alerts reach all relevant users and include meaningful and tailored actionable guidance for different users. We suggest that integrated, impact-based and action-oriented early warning systems are essential for effective, equitable climate-health adaptation in a warming and, in many places, increasingly polluted, world.

How to cite: Aunan, K. and Chowdhury, S.: Beyond single-hazard alerts: Rethinking heat and air pollution Early Warnings Systems in high-exposure settings, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6928, https://doi.org/10.5194/egusphere-egu26-6928, 2026.

How to cite: Urban, A., Quispe-Haro, C., and Mühlhaus, R.: Effects of Temperature and Air Pollution on Urgent Hospital Admissions in the Czech Republic, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7085, https://doi.org/10.5194/egusphere-egu26-7085, 2026.

The impacts of weather, weather changes, climate and climate change do not only affect the physical but also the mental health of humans. It ranges from (post traumatic) stress disorders, depression and anxiety to cognitive and behavioural maladaptation and disorders. Form and characteristics of the impact depend on personal and social factors. Personal predispositions like psychological disorders, gender, age and genetics can influence psychical resilience against environmental impacts.
In PsyCourse x Weather we conduct a cross-sectional study. We compare the impact of weather and weather changes on the quality of life (QOL) of people with affective and psychotic disorders, like schizophrenia and bipolar disorder, with the QOL of a healthy control group. The objective is to find out whether there are differences in the impact of weather and climate on the QOL between patients and a control group and if gender and genetic factors influence the impacts. Health data was gathered from the 17 locations in Germany and Austria of the PsyCourse study (PsyCourse 2015), like the WHOQOL, age, gender and the polygenic risk score. As predictors we use meteorological and air hygienic reanalysis data from ERA5 and CAMS. We include parameters like precipitation, air pressure, ozone, particulate matter, wet bulb globe temperature (WBGT), heat wave and cold stress wave indices, summarised to periods of 14 days and 28 days to reflect the time span of the WHO quality of life questionnaire (WHOQOL) and to include longer-term weather conditions. As a result of regression analysis using generalized additive models, we find that meteorological and air hygienic variables have a rather marginal impact. The fact of having a psychiatric disease has in general a strong influence on QOL compared to weather. The mean QOL (scale ranging from 4 to 20, the higher the number the higher the QOL) of the groups is  17 for the control group and 13.1 for the patients. Nevertheless, we find connections between atmospheric changes and the QOL. For our control group we identify heatwave and WBGT as relevant parameters. While for the patients we find ozone, precipitation and particulate matter as influencing factors. During the 14-day periods there are two significant parameters for the control group with reduced influence in the 28-day periods. In contrast, patients are impacted by more parameters with increasing impacts from the 14-day to the 28-day periods. We also identify differences between male and female. In the control group, heatwaves have negative impacts on the group, while males are more affected compared to females. Males in the patient group are also negatively impacted by heatwaves, yet not significantly, females however have increased QOL during heatwaves.

PsyCourse (2015): Home. Available online at http://www.psycourse.de/, updated on 1/13/2015, checked on 8/25/2025.

How to cite: Rückle, K., Greiner, S.-K., Senner, F., and Hertig, E.: PsyCourse x Weather: the impact of weather changes on mental health, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7274, https://doi.org/10.5194/egusphere-egu26-7274, 2026.

Environmental variables such as air pollution, noise, floods, green space and conflict , shape human health and disease. An important concept is the human exposome, the totality of an individual’s exposure to the environment over their lifetime. The exposome can explain a large proportion of our health, yet quantification across the world population remains surprisingly limited. As we are facing global climate and population change, it becomes increasingly important to understand and quantify the exposome. However, existing studies are often small-scale, do not integrate human mobility affecting personal exposure or focus on narrow sets of variables, thereby failing to capture the full range of socioeconomic and physical variables and values. Harmonized global scale quantitative assessments of the entire exposomes of the world population remain limited.  We address this by collecting data sets on environmental variables for a wide range of geo-domains at high resolution (<= 1-10 km²) with global coverage. Seven geo-domains characterizing the external exposome were defined: meteorological and hydrological, biological, geological, air, soil, technological (built environment), and societal.  Human mobility in tandem with differing exposure pathways (e.g. passive, through inhalation vs. active, by selecting food stores) are represented globally by aggregating exposures within the spatial context of individuals. The relevant spatial context is an area surrounding the residential locations. Exposure values are first calculated in distance rings centred at the residential location and then aggregated using distance dependent weights. The function determining the weight and maximum distance depends on the exposure pathway and, following this, the relevant human mobility characterization for exposure. To this harmonized dataset, population density is attributed, producing a characterization of the global exposome that will be catalogued in the Green Deal Data Space (GDDS). Initial harmonized processing is executed for global datasets on tropical cyclones, earthquakes, and riverine and coastal flooding, showing hazard intensity and human exposure have different spatial patterns. For example, populated coastal and riverine regions are substantially exposed to flooding relative to their physical hazard extent, whereas other hazards leave populations unaffected. These emerging contrasts illustrate the importance of harmonized global exposome characterization. The assessment framework lays a foundation for analyses on co-exposures, spatial patterns and equitable public health strategies.

How to cite: Kuipers­­­, E., Schmitz, O., Griffioen, R., Bood, R. J., Oonk, R., Loffredo, L., and Karssenberg, D.: Characterization of the world population’s exposomes , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9715, https://doi.org/10.5194/egusphere-egu26-9715, 2026.

Air pollution is an increasing public health concern in Sri Lanka, driven by rapid urbanization, regional pollutant transport, and continued reliance on solid fuels in rural areas. Exposure to fine particulate matter (PM2.5) is associated with elevated risks of cardiovascular and respiratory diseases, stroke, and premature mortality, contributing to over 20% of total disability-adjusted life years (DALYs) and deaths nationally, according to the most recent iteration of the Global Burden of Diseases Study. However, high-resolution exposure data and short-term health impact assessments remain limited.

In this study, we develop the first high-resolution (1 × 1 km, daily) PM2.5 dataset for Sri Lanka by combining in-situ measurements, satellite retrievals, and reanalysis products using a hybrid modeling framework. We then quantify the acute effects of PM2.5 exposure on respiratory health using daily hospital admission data (eIMMR) for 2020–2023, focusing on acute respiratory infections (ICD-10 codes J00–J06, J09–J18, J20–J22). We apply a Distributed Lag Non-Linear Model (DLNM) to capture non-linear exposure–response relationships and delayed effects, considering lags up to 50 days. Models control relative humidity, temperature, precipitation, carbon monoxide, day of week, and month. Confounding was assessed using a leave-one-out approach, while effect modification was examined through tertile-based stratification and pairwise statistical tests.

Population-weighted PM2.5 concentrations show a rapidly increasing trend, particularly in and around the national capital. We find that PM2.5 effects are strongest on the day of exposure (lag 0) and decrease with increasing lag. A 10-µg m^-3 increase in PM2.5 is associated with a 16.6% (10–22%) increase in hospitalizations for acute respiratory diseases. Relative humidity emerges as a key confounder, while precipitation significantly modifies the PM2.5–hospital admission relationship, with substantially stronger effects on low-precipitation days (RR ≈ 1.40). Children under 15 years’ experience higher risks compared to adults and the elderly.

These findings highlight the growing respiratory health burden of air pollution in Sri Lanka and underscore the need for integrated air quality management and health-informed policy. Future work will incorporate additional pollutants (NO2, O3), socioeconomic factors, and extend analyses to cardiovascular outcomes and joint PM2.5–temperature effects.

How to cite: Aunan, K., Amarakoon, P. M., Jayawardena, R., Diunugala, A., Sandvik, B., Ferkingstad Sandve, G. K., Fleck Fossen, E. I., and Chowdhury, S.: Assessing the health risk of air pollution exposure in Sri Lanka, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9786, https://doi.org/10.5194/egusphere-egu26-9786, 2026.

Climate change is intensifying summertime heat exposure across Europe, with growing implications not only for physical human health but also for population mental well-being. However, heat-health research has focused mainly on the physical outcomes of heat exposure; mental health impacts remain underexplored. In particular, spatially and temporally explicit analyses that capture variation in heat exposure across diverse regions are scarce, limiting the systematic identification and monitoring of vulnerable populations. Switzerland serves as a suitable case study for addressing this gap, given its pronounced warming trends and environmental heterogeneity, while the underlying analytical approach is transferable to other countries and regions.

This study investigates the relationship between heat and mental health outcomes in Switzerland by integrating population health survey data with satellite data-based heat metrics in a spatially and temporally explicit framework. The study is grounded in a heat-mental health risk framework linking thermal hazard, spatiotemporal exposure, and demographic vulnerability. Individual-level mental health data from the Swiss Health Survey (a comprehensive national health survey conducted in 2007, 2012, 2017, and 2022) are combined with high-resolution land surface temperature (LST) derived from MODIS Aqua as the primary heat exposure indicator, alongside gridded near-surface air temperature for comparison and benchmarking. The temperature metrics are designed to represent environmental heat load rather than single-day extremes. Mental health is expressed through multiple standardised indices capturing psychological burden, vitality, depressive symptoms, and anxiety. To account for spatiotemporal dependencies, we apply hierarchical Bayesian ordinal regression models that also serve as predictive models for scenario-analysis.

Results indicate that higher LST is generally associated with poorer mental health outcomes across Switzerland, with the strongest and most credible associations observed during the exceptionally hot summer of 2022. We also found that LST-based models outperform air-temperature-based models, which indicates the added value of thermal remote sensing in heat-health studies across spatially heterogenous areas. Spatial analyses reveal pronounced regional and urban-rural gradients in both heat exposure and baseline mental health, while demographic factors such as age and biological sex exhibit substantial variation in mental health vulnerability but do not significantly modify the heat-mental health relationship itself.

By integrating remote sensing, climatological data, and population health records, this study demonstrates a scalable interdisciplinary approach for assessing climate-sensitive mental health risks across space and time. It provides a foundation for integrating mental health into climate adaption, heat warning systems, and spatially targeted public health planning.

How to cite: Schubiger, E., Adams, J. S., Santos, M. J., Fischer, S., and Naegeli, K.: Remote Sensing of Mental Health: The Burden of Heat Exposure in Switzerland. An Interdisciplinary Study Combining Earth Observation and Epidemiology., EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11659, https://doi.org/10.5194/egusphere-egu26-11659, 2026.

Extreme event attribution is a branch of climate science that aims to quantify the extent to which the frequency and intensity of extreme weather events such as heatwaves, cold spells, droughts and floods can be said to have been influenced by human-caused climate change. Extreme heat is the deadliest type of weather, although heat-related illnesses and deaths are not directly captured in death certificates or hospital records, and the risks are rarely appreciated by the public. In this talk we introduce a recent collaboration between scientists at Imperial College London and the London School of Hygiene and Tropical Medicine that brought together established methods from attribution and epidemiology to estimate in near real time the expected number of heat-related deaths in cities across Europe during the summer of 2025, and the proportion of those deaths that can be attributed to human-caused climate change. Across 854 cities in Europe we found an estimated 24,404 (95% interval: 21,968 - 26,806) excess deaths during the summer months, with almost 70% of those attributable to human-caused climate change, although vulnerability to heat varies across the continent. This work received widespread media attention, showing the importance of timely information for public awareness of both the risks to health and the contribution of climate change as the extreme weather unfolded.

How to cite: Barnes, C., Konstantinoudis, G., Masselot, P., Mistry, M., Gasparrini, A., Vicedo-Cabrera, A. M., Clarke, B., Theokritoff, E., and Otto, F.: Near-real-time attribution of mortality to extreme heat, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12610, https://doi.org/10.5194/egusphere-egu26-12610, 2026.

Heatwaves are among the deadliest climate extremes, with children especially vulnerable due to physiological sensitivity and limited adaptive capacity. Yet cumulative lifetime exposure of children remains poorly quantified, particularly in high-latitude countries such as Canada, where warming is occurring at roughly twice the global average. Here, we present the first national-scale assessment of Lifetime Heatwave Exposure (LHE) for Canadian children under multiple global warming pathways.

We integrate high-resolution temperature observations, downscaled CMIP6 climate projections, and demographic data to estimate the number of severe heatwave events children are expected to experience over their lifetime. Heatwaves are defined using locally relevant thresholds based on exceedance of the 98th percentile of daily maximum temperature, ensuring consistency across Canada’s climate zones. Exposure is evaluated across warming levels from 1 °C to >5 °C at sub-provincial population scales.

Our results demonstrate a clear generational shift in heat exposure. Under 3 °C warming, over 80% of Canadian children are projected to experience unprecedented lifetime heatwave exposure exceeding the historical maximum. Analysis of 45 major historical heat events shows that 70% of reported heat-related deaths occurred during LHE-level events, including the 2021 Pacific Northwest heat dome, when 447 excess deaths were recorded in Vancouver alone. Projections indicate a nationwide transition from rare, once-in-a-lifetime heatwaves to recurrent generational hazards, with western provinces reaching full exposure earliest and eastern and northern regions converging rapidly by mid-century.

By shifting focus from short-term extremes to cumulative lifetime exposure, this study introduces a child-centric, policy-relevant metric for climate risk assessment. The findings highlight growing intergenerational inequity and underscore the urgency of global mitigation alongside targeted local adaptation—such as urban greening, cooling infrastructure, and heat-health early-warning systems—to protect current and future generations of Canadian children.

How to cite: Zaerpour, M., Papalexiou, S. M., and Helldén, D.: Cumulative Lifetime Heatwave Exposure for Canadian Children in a Warming Climate, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12862, https://doi.org/10.5194/egusphere-egu26-12862, 2026.

Climate change is a key determinant of public health, influencing disease patterns and human and environmental well-being. Mosquito-borne diseases such as dengue and West-Nile virus continue to pose significant public health challenges worldwide, particularly in regions where environmental conditions favour mosquito production and spread. In recent years, there has been a resurgence of several vector-borne diseases in Europe, driven by climate change, altered water management, and the expanding distribution of invasive mosquito species. Spain has been increasingly affected by this trend, with repeated outbreaks of West-Nile virus—especially in southern regions—and sporadic locally acquired dengue cases reported since 2018.  

Mosquito population dynamics are largely determined by climatic factors, including temperature and water availability. Therefore, understanding the linkage between climate, local water resources, and mosquito dynamics is crucial for better predicting current and future health risks and informing effective disease control and health management. We investigated how the temporal and spatial distribution of water availability and climatic conditions influenced mosquito populations in the Aiguamolls de l’Empordà, a natural wetland area connected to La Muga and El-Fluvia river basins (Catalonia, Northeast Spain), under current and projected climatic scenarios. To do so, we developed a machine learning based Random Forest (RF) model fed withCulex mosquito abundance data (weekly data from 12 traps), climate (rainfall and temperature), and hydrological simulated data (discharge, actual and potential evapotranspiration, and aridity) from 2001 to 2021.  We use projected daily climate from ensemble projections of climate scenarios of the REMO2015 regional climate model under the RCP2.6 and RCP8.5 scenarios (2031-2060) to project the future abundance of mosquito populations in the study area. Our model comprised 48 environmental predictors and the Culex population as the predictand.

 The Culex mosquito population showed a strong positive correlation with temperature-related variables and a negative relationship with discharge and aridity. The RF model showed reasonably good performance in training (R² = 0.90) and testing (R² = 0.61), showing a well-matched temporal pattern of average condition per trap with observed data. Based on Mean Decrease in Impurity analysis, potential evaporation and temperature were found to be highly important predictors.  According to the climate projection under RCP 8.5, in general, mean annual rainfall over the study area will decrease, while minimum and maximum temperatures will increase in the future (2031-2060) compared to the baseline (1981-2010). Thus, these changes could create more favourable conditions for mosquitoes, resulting in substantial additional risk to public health. These results underscore the mounting risk of mosquito-borne diseases in Europe and the necessity for enhanced surveillance and preventive management. Our results contribute to the project “Infectious Disease Decision-support Tools and Alert systems to build climate Resilience to emerging health Threats (IDAlert)” funded by the European Union.

Keywords: Wetlands, Machine Learning, Health Risk, Climate change, Mosquito-borne diseases

How to cite: Sirisena, J., Rodriguez, J., Bernal, S., Bartumeus, F., Costa, M. M., and Bouwer, L. M.: Simulating Mosquito Populations through the Integration of Climate and Water Resource Modelling, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13289, https://doi.org/10.5194/egusphere-egu26-13289, 2026.

Assessing the efficacy of malaria interventions is increasingly complicated by a changing climate, which can mask or mimic the impacts of public health policies. To robustly attribute changes in disease burden, it is essential to isolate the non-linear impacts of climate trends and variability from intervention effects.

This study introduces the scientific framework of the ACCLIMATISE project (funded by the Wellcome Trust in the ATTRIVERSE program) utilizing the VECTRI dynamical malaria model to simulate transmission under a range of climate counterfactuals. Using latrge ensembles, Our approach filters driving temperature and precipitation data to selectively remove specific modes of variability—ranging from climate change through decadal and multi-year cycles to interannual variability. This experimental setup allows us to disentangle the distinct roles of warming and hydrological variability in driving transmission dynamics across Africa.

We present preliminary results demonstrating how these filtered climate drivers alter simulated malaria baselines, highlighting the sensitivity of the model to specific timescales of climate forcing through temperature and rainfall separately as well as their nonlinear interaction. These simulations establish a "climate-only" reference frame. The ACCLIMATISE project will confront these counterfactual baselines with health observations to attribute the role of climate in this health outcome and separate the signal of malaria interventions from the influence of climate variability and change. 

How to cite: Tompkins, A., DiSera, L., Zornoza, M., Caminade, C., Thiam, M., and Munoz, A.: Multiple timescale climate drivers of malaria: Counterfactual ensembles for climate attribution in health from the ACCLIMATISE Project, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15078, https://doi.org/10.5194/egusphere-egu26-15078, 2026.

Previous modelling studies of mosquitoes in Europe have primarily focused on local and regional climate drivers, while the influence of large-scale atmospheric teleconnection patterns on mosquito populations remains poorly understood. This study examines how major European–North Atlantic (EUNA) teleconnection patterns—including the North Atlantic Oscillation (NAO), Arctic Oscillation (AO), East Atlantic (EA), East Atlantic/Western Russia (EAWR), Scandinavian (SCAND), and Summer East Atlantic (SEA) patterns—influence mosquito abundance across Germany. Using nationwide mosquito surveillance data (2016–2024), we combined rotated temporal-mode principal component analysis (T-mode PCA) of mean sea level pressure fields with spatiotemporal generalized linear mixed models (GLMMs) to quantify regional- and seasonal-specific relationships among circulation modes, local weather anomalies, and mosquito abundance. Results reveal pronounced regional and seasonal variability in the climate-mediated associations between circulation patterns and mosquito abundance. Effects were strongest and predominantly positive in the Continental Dry, Northwest Cool, Warmest, and Coastal regions, particularly from summer to early autumn, whereas responses in Alpine and other mountainous regions were weaker or negative due to cooler, wetter and windier conditions that constrain mosquito activity. Local temperature and humidity anomalies associated with EUNA circulation patterns were consistently linked to increases in mosquito abundance while precipitation and windspeed anomalies showed negative effects. Positive temperature and negative humidity anomalies during EA⁺, EAWR⁺, SCAND⁻, and SEA⁺ phases exhibited the most consistent positive relationships with mosquito abundance. These findings demonstrate that large-scale climate variability plays a significant role in shaping mosquito population dynamics in central Europe and highlight the value of incorporating teleconnection indices into early-warning and forecasting systems of mosquito-borne diseases.

How to cite: Adeleke, E., Merkenschlager, C., Schäfer, M., Lühken, R., Gutjahr, P., Voll, C., and Hertig, E.: Regional and Seasonal Variability in the Impacts of the North Atlantic Oscillation and Other European North Atlantic Teleconnections on Mosquito Populations in Germany, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15200, https://doi.org/10.5194/egusphere-egu26-15200, 2026.

Heat wave (HW) is a hazardous climate extreme that can lead to serious impacts on human health, posing challenges to the UN Sustainable Development Goals #3, #11, and #13. This study examines the heat wave characteristics across South Asia and surrounding regions during a 45-year period (1980-2024) with a particular focus on recent intensification and increasing population exposure. Daily 2-m air temperature data of European Centre for Medium-Range Weather Forecasts (ECMWF) Reanalysis 5 (ERA5) was used for identification of hot days and resultant heat waves. The annual count of hot days (DH) is the number of days the daily maximum temperature surpasses 90^th percentile daily maximum temperature, whereas hot nights (NH) refers to exceedance of 90^th percentile daily minimum temperature, calculated over a 15-day moving window representing long term climatology of the time of the year being considered. A minimum of three consecutive compound hot day and nights was identified as a HW event. Three HW indices were computed annually: the number of HW events (HW_n), the number of HW participating days (HW_p), and HW magnitude (HW_m), which accounts for combined daytime and nighttime temperature departures. For each grid location, these indices were aggregated at decadal scales from the 1980s to the 2020s to examine the evolution of the HW characteristics across the study domain.

The study revealed that globally, NH has increased substantially (266.19%) from 1980s to 2020s leading to more frequent HWs (44,907 events/year in 1980s to 333,424 events/year in 2020s) during the study period. In Peninsular India, HW_n was found to be as high as 98 events and HW_p was as high as 533 days in the 2020s. HW_m was even more than 500 °C² in some locations in Eastern Asia during the same decade, indicating that both day-time and night-time temperatures showed large anomalies with respect to the long-term climatology.

To quantify the impact of rising heatwaves on rising population, gridded population data from WorldPop of University of Southampton was used to determine the change in population exposure to the HW indices over the recent decade (2015–2024). The major cities with marked increase in population exposure to HW occurrences (i.e., HW_n) were identified as Zhengzhou (China), Chengdu (China), Dhaka (Bangladesh), Faridabad (India) and Lahore (Pakistan) with exposure changes ranging from 2.92 × 10⁷ person-events to 6.34 × 10⁷ person-events. The maximum change in population exposure to DH is in Istanbul, Turkey (1.57 × 10⁸ person-days) whereas the same for NH is in Ho Chi Minh City, Vietnam (5.61 × 10⁸ person-days). The rising exposure to NH indicates that many cities are losing the ability of natural night-time cooling and require targeted intervention. Thus, this study offers valuable insights on the spatial and temporal evolution of heat wave characteristics across the most densely populated regions of the world and is expected to be useful for developing policies on climate-resilient urban infrastructure planning.

How to cite: Mandal, N. S., Mandal, N., Das, P., and Chanda, K.: Evolution of Heat wave characteristics across South Asia and identification of the most affected cities in the recent decade, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16288, https://doi.org/10.5194/egusphere-egu26-16288, 2026.

Rapid urbanization has led to increased population density and more impermeable paving and buildings, causing heat to accumulate on the ground and building shells, resulting in a continuous rise in urban temperatures. Studies have shown that environmental meteorological parameters and air pollutants interact, and air pollution concentrations also have a cumulative effect with environmental factors such as temperature, wind field, and rainfall, impacting human health.

This study aimed to investigate the immediate effects of wet-bulb black bulb concentration (WBGT) and temperature on heart rate variability (HRV) and heart rate in participants. A simple, portable device was used to monitor PM_2.5 and other environmental factors. The correlation between temperature and residents' health status was analyzed, with separate analyses conducted for summer and winter-spring seasons to examine the seasonal health impacts.

How to cite: Yu, S.-Y. and Lung, S.-C. C.: Assessment of Immediate health impacts of temperature and PM2.5 in urban residential areas of southern Taiwan., EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17098, https://doi.org/10.5194/egusphere-egu26-17098, 2026.

Research at the intersection of climate, weather, and health is rapidly expanding and inherently interdisciplinary, requiring integration of information across multiple disciplines. This includes comprehensive, accessible, reliable, and harmonized datasets that combine high-quality observational data with bias-corrected and downscaled climate projections from Global Climate Models (GCMs), such as from the sixth phase of the Coupled Model Intercomparison Project (CMIP6). However, despite the availability of numerous gridded observational datasets and pre-processed projections, individual products vary in strengths, limitations, and representations of fine-scale spatiotemporal patterns, which can substantially affect downstream modelling and projection of current and future health outcomes. Moreover, the operational scale of epidemiological analysis is typically defined by administrative units, rather than by regular grids, and therefore often relies on the inclusion of area-level estimates that are additionally weighted by indicators such as human population. Hence, spatially resolved weather and climate data, typically provided in specialized formats (e.g., NetCDF), generally require substantial preprocessing before they can be used for respective analysis.

To address these challenges, we developed a tailored, quasi-global weather and climate dataset designed to support high-resolution infectious disease transmission modelling in tropical settings. Our dataset comprises (1) high-resolution (0.1°) daily climate projections between 60°N and 60°S, and (2) corresponding spatially averaged (population-weighted) area-level estimates at administrative unit levels 0-2 for over 100 countries. We therefore selected and evaluated multiple global observational datasets, including model- and satellite-based products such as ERA5 and CHIRPS, across heterogeneous, disease-relevant tropical study domains. The observational datasets showing the highest performance in our comparative analysis served as reference climatologies for generating high-resolution, bias-corrected climate projections downscaled from six CMIP6 GCMs, focusing on two scenarios from the Shared Socioeconomic Pathways-Representative Concentration Pathway (SSP-RCP) framework: SSP2-4.5 and SSP5-8.5.

For the first time, we hence provide a robust, open-access resource that combines observational datasets and bias-corrected, downscaled climate projections in a coherent manner and translates them into harmonized, spatially aggregated variables suitable for easy use by non-specialists from various disciplines. As an example application, we present the impact of climate change and the sensitivity of administrative-level vector-borne disease transmission risk in South America to the choice of global climate model and emissions scenario. We focus on yellow fever, a vaccine-preventable zoonotic arbovirus endemic to tropical regions of South America and Africa. We anticipate that our unified weather and climate database will be particularly valuable to infectious disease modelers, epidemiologists, and practitioners conducting climate-sensitive health impact assessments.

How to cite: Jahn, S., Fraser, K., Gaythorpe, K. A. M., Dorigatti, I., Winskill, P., Hinsley, W., Wainwright, C. M., Toumi, R., and Ferguson, N. M.: Developing and Applying a Unified Weather and Climate Database to Assess Climate Change Impacts on Tropical Infectious Disease Transmission and Burden, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17736, https://doi.org/10.5194/egusphere-egu26-17736, 2026.

Climate change is increasing Europe’s vulnerability to vector-borne diseases. One example are the increasing number of outbreaks caused by the West Nile Virus.  Whereas in the early 2010s, the virus was only found in south-eastern Europe, local infections are now also detected in more northern and western countries. To inform health care institutions as well as citizens it is necessary to be able to predict where and when the next outbreak will happen. There have been studies that show that land use, climate and weather influence the risk of human West Nile Virus infections, but it is less clear what the relative contributions of land use, climate change and weather are. In particular, it is not determined yet if the northward expansion of WNV can be better explained by the gradual change in climate, or by the occurrence of specific weather conditions that increase the risk of WNV infections. In this study, a random forest model is used to determine what is the best predictor of WNV infections: weather or climate. It shows that on a spatial scale of NUTS3-regions, the climate mean seasonal cycle of the 2m temperature is the best predictor for human WNV outbreaks, and that including the weather in the model does not improve its performance. Moreover, results indicate that WNV risk is higher in areas in which the climate mean seasonal cycle of temperature is in between 20-26 °C for one or more weeks. This can help explain and predict the emergence of human WNV infections in new regions in Europe.

How to cite: Boonekamp, M., De Roode, S., Siebesma, P., Bogaard, T., Van der Schrier, G., Sikkema, R., Schrama, M., Koopmans, M., and Marsboom, C.: Human West Nile Virus infections in Europe: weather or climate?, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18547, https://doi.org/10.5194/egusphere-egu26-18547, 2026.