EGU General Assembly 2023  

In 2004, we painstakingly measured the thinning of a single glacier tongue (the Mer de Glace, Mont-Blanc) from pairs of SPOT (CNES) satellite optical stereo-images. It then took us nearly 20 years before we managed to up-scale such observations to the global scale. In this presentation, I will illustrate the advances (in terms of data availability and processing) and all the collaborative work that led to a spatially-resolved and almost complete estimation of mass changes for the more than 200,000 glaciers on Earth. These new data paint a global picture of accelerating glacier mass loss since 2000 and pave the way toward improved projections of future glacier mass and sea level contribution.

How to cite: Berthier, E., Belart, J., Blazquez, A., Brun, F., Deschamps-Berger, C., Dussaillant, I., Flament, T., and Hugonnet, R.: From the tongue of the Mer de Glace to the world’s glaciers : 20 years of progress in measuring glacier mass changes from space, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-4976, https://doi.org/10.5194/egusphere-egu23-4976, 2023.

Glaciers outside the ice sheets are key contributors to sea-level rise and act as essential fresh-water resources in various regions around the world. Glaciers are also important sources of natural hazards, directly impact biodiversity, and have a significant touristic value. Given these crucial societal and environmental roles, having reliable projections on the evolution of these precious ice bodies under changing climatic conditions is of paramount importance.

In this presentation, I will highlight how modelling glacier changes goes hand in hand with rapidly increasing remotely sensed glacier observations and derived products at the global scale (e.g., glacier outlines, surface elevation, ice thickness, surface velocities, and elevation changes). These new observations, combined with ever-increasing computational capacities and novel numerical tools have recently allowed us to transition from modelling a few individual glaciers to now modelling the evolution of glaciers at regional- to global scales while accounting for ice-dynamical processes. Whereas these recent advances are encouraging, I will also highlight the challenges that we are still facing and that we will need to tackle in the coming years to provide more trustworthy glacier evolution projections.

How to cite: Zekollari, H.: Changing glaciers in a changing climate through changing modelling approaches, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9607, https://doi.org/10.5194/egusphere-egu23-9607, 2023.

We partitioned Antarctic ice mass changes (2003-2020) into the contributions of surface mass balance (SMB) and ice discharge over 27 drainage basins, based on the combined estimates of satellite gravimetry and altimetry observations and a numerical SMB model. Our analysis indicates that the ice discharge has played a dominating role in ongoing ice mass losses and accelerations, especially in the glaciers near Amundsen and Bellingshausen Sea in West Antarctica. In particular, the mass losses in the Thwaites and Pine Island Glaciers have been mostly controlled by ice discharge, while the contribution of SMB has been minor. On the other hand, SMB contributed large portions of ice mass imbalance in East Antarctica, such as glaciers near the Dronning Maud Land and Wilkes Land. An inaccurate GIA model is a potential source of uncertainty in our estimates.

How to cite: Kim, B.-H., Seo, K.-W., Lee, C.-K., Kim, J.-S., and Lee, W. S.: Partitioning the contribution of surface mass balance and ice discharge in Antarctic glacier mass variations (2003-2020), EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-676, https://doi.org/10.5194/egusphere-egu23-676, 2023.

Debris-cover representation is rarely included in regional or global glacier models although it plays a key role in the regulation of melt processes. Debris cover that is more than a few centimeters thick reduces melt by insulating glacier ice. However, mass loss and retreat of debris-covered glaciers are not necessarily slower than those of clear ice. Debris-covered glaciers are widespread in the Northern Caucasus. It is important to reliably quantify their evolution because the contribution of glacial runoff to total discharge is significant in the region.

This study assesses the influence of debris cover on the evolution of glaciers in the basins of the Terek and Kuban rivers in the Northern Caucasus in the 21st century and quantifies its effects on glacier mass balance, ice velocity, ice thinning, changes in glacier area, volume, and position of the glacier fronts. We use the GloGEMflow glacier model and introduce a new debris cover dynamic module. The mass balance is calibrated separately for the explicitly modelled debris cover and for clean-ice glaciers (debris cover is implicit in the degree-day factor calibration). The model is calibrated using newly mapped debris cover outlines and ice thickness data from Rounce et al. (2021). The debris evolution is simulated with a steady deposit model adapted from Verhaegen et al. (2020) and Anderson & Anderson (2016), where debris input onto the glacier surface is generated from a fixed point on the flow line.

We compare spatio-temporal changes in glacier geometry including the evolution of debris cover for the explicit and implicit debris-cover formulation for five SSP scenarios from CMIP6. The debris-cover evolution patterns differ significantly between the Terek and the Kuban basins. In the Kuban basin, glaciers located generally at lower elevations, retreat rapidly and lose ice at the debris-covered glacier tongues. On the contrary, the supraglacial debris of the Terek basin glaciers may, under certain climate scenarios, expand and play an increasingly-important role in glacier evolution with time. However, under the high-end warming scenario SSP5-8.5, the ice loss by 2100 overwhelms the debris-cover effects in both regions.

The maximum difference in glacier length, area and volume depending on the explicit or implicit mode of debris-cover modeling occurs before 2100, but by the end of the century it is eliminated due to the retreat of debris-bearing parts of the glaciers or due to the elevation-stabilization effect. In general, explicitly accounting for debris cover in the projections only has a minor effect on the overall projected regional mass loss, but improves the representation of processes on the intra-glacier scale.

This study was carried out under Governmental Order to Water Problems Institute, Russian Academy of Sciences, subject no. FMWZ- 2022-0001, and was funded by the RSF grant number 22-17-00133.

How to cite: Postnikova, T., Rybak, O., Zekollari, H., Huss, M., and Gubanov, A.: Future evolution of glaciers in the Caucasus: focus on debris-cover evolution., EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-893, https://doi.org/10.5194/egusphere-egu23-893, 2023.

Glacier monitoring has been internationally coordinated since the late 19th century. For the last 25 years (i.e., 1998-2023), the compilation and dissemination of global glacier datasets has been coordinated by the Global Terrestrial Network for Glaciers (GTN-G, https://gtn-g.org). Authorized under the Global Climate Observing System (GCOS) and supported by an international Advisory Board, GTN-G is jointly run by the science officers from the World Glacier Monitoring Service (WGMS, https://wgms.ch), the US National Snow and Ice Data Center (NSIDC, https://nsidc.org), and the Global Land Ice Measurements from Space initiative (GLIMS, https://glims.org), in collaboration with related working groups of the International Association of Cryospheric Sciences (IACS, https://cryosphericsciences.org/).

We present an updated overview of the various observational glacier datasets (https://www.gtn-g.ch/data_catalogue/), including world regions for regional glacier assessments (GTN-G Glacier Regions), glacier attributes (WGI: World Glacier Inventory), centerlines, and outlines (GLIMS, RGI: Randolph Glacier Inventory), ice velocities (ITS_LIVE), ice thickness (GlaThiDa: Glacier Thickness Database), glacier photographs (GPC: Glacier Photograph Collection), and glacier maps (GMC: Glacier Map Collection). For each dataset, we provide insights into the increased amount and richness of available data. We also demonstrate the value of these datasets by presenting selected findings from our own analyses as well as from user applications.

How to cite: Zemp, M., Windnagel, A., Welty, E., Raup, B., Paul, F., Nussbaumer, S., Maussion, F., Hoelzle, M., Hock, R., Gärtner-Roer, I., and Berthier, E.: Insights into and findings from global datasets on glacier distribution and changes, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-2326, https://doi.org/10.5194/egusphere-egu23-2326, 2023.

Peripheral glaciers (PGs, i.e., glaciers that are dynamically decoupled from the ice sheet) play a significant role in the mass balance and freshwater runoff of Greenland's land ice. Their evolution has implications for the sea level, global climate and ocean circulation, as well as for coastal communities and ecosystems. In order to accurately model the contribution of PGs to Greenland's overall mass balance and freshwater supply, it is necessary to consider their characteristics that set them apart from the main ice sheet. In this study, we used the Open Global Glacier Model (OGGM) to conduct large-scale regional modeling of approximately 19,000 PGs and project future mass losses, freshwater runoff, and sea level contributions under different climate scenarios. Our results show that PGs are likely to experience 29 % to 52 % of mass loss compared to their 2020 levels by the end of the 21st century, resulting in 10 mm to 19 mm of sea level rise under SSP126 and SSP585, respectively. Under the high emission scenario, PGs are expected to contribute 184 Gt yr^-1 of liquid freshwater and 3 Gt yr^-1 of calved solid ice to the ocean during 2020-2100, affecting ocean density, circulation, and mixing. Peakwater is projected for the 2080s under SSP585, after which annual freshwater contributions are expected to decline due to reduced glacier area. Regional mass and freshwater balance differences were found to be influenced by local climate, ocean-ice interaction, proportion of marine-terminating glaciers, and initial glacier volume. The central-west, southeast, and central-east subregions will experience the largest mass losses (79 %, 69 %, and 63 % of mass in 2020), with peakwater occurring earlier than for all PGs considered together under SSP585. The northeast subregion is expected to contribute 35 % of total liquid freshwater, 73 % of solid ice calving, and 37 % of sea level rise, despite lower mass loss (wrt 2020) due to regional conditions (i.e., climate, glacier geometries, and surrounding ocean). Submarine melting is likely to impact the mass balance of marine-terminating PGs, but further research is needed to explore this effect at a regional scale. This study highlights the importance of considering the distinct behavior of PGs in modeling Greenland's freshwater balance and understanding the factors that influence their mass changes.

How to cite: Shafeeque, M., Malles, J.-H., Vlug, A., Marzeion, B., Möller, M., and Eis, J.: Regional modeling of peripheral glaciers in Greenland: Implications for mass balance, freshwater runoff, and sea level rise , EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-3329, https://doi.org/10.5194/egusphere-egu23-3329, 2023.

The Tarim River Basin, situated in the Eurasia hinterland, serves as the heart of China’s Silk Road Economic Belt. It covers an area of 1.02 million km² and is surrounded by the Tienshan Mountains to the north, the Kunlun Mountains to the south and the Pamir to the west. The runoff is recharged by glacier melt, snow melt, and rainfall. There are large amount of glaciers distributed in the high mountains of Tienshan and the Kunlun Mountains. In recent decades, the glaciers in the Tienshan Mountains are retreating at a faster rate, while glacier in the Kunlun Mountains show less retreat or even advancing trend (i.e., Karakorum anomaly). These changes in glaciers will definitely alter the future runoff.

This study extended the hydrological model SWAT-Glacier by incorporating a degree-day based glacier melt and accumulation module. The hydrological processes of the headwaters of the Tarim River were simulated using SWAT-Glacier model. The model was calibrated and validated using multiple variables, including glacier mass balance, snow cover area, snow water equivalent, daily streamflow and the balance between snowfall, snow melt and sublimation. The hydrological model was forced by the bias-corrected climate from 6 regional climate models in CORDEX. Results indicated that the runoffs of the headwaters originated from the south Tienshan Mountains (i.e., Kaidu River, Aksu River) demonstrated a slight increase or even decrease trend. For the Kaidu River, there will be a slight decrease in runoff under SSP585, as the contribution of glacier melt water is less than 10%. For the Kumarak River, the runoff showed slightly increase and the glacier melt runoff will reach peak point before 2050s. For the rivers originated from the north Kunlun and Karakorum Mountains, the runoff will increase dramatically.

This study provide a basin-scale runoff changes under multiple constraints in the endoreic Tarim River Basin. However, the glacier accumulation and ablation suffers from great uncertainty as the precipitation observation is rare in the high mountains. More efforts should be taken to utilize more state-of-the-art technology in revealing the meteorological and hydrological processes in these alpine catchments.

How to cite: Fang, G. and Chen, Y.: Future changes in runoffs in the headwaters of the Tarim River, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-3720, https://doi.org/10.5194/egusphere-egu23-3720, 2023.

Retreating and thinning glaciers are icons of climate change and impact the local hazard situation, regional runoff as well as global sea level. For past reports of the Intergovernmental Panel on Climate Change (IPCC), regional glacier change assessments were challenged by the small number and heterogeneous spatio-temporal distribution of in situ measurement series and uncertain representativeness for the respective mountain range as well as by spatial and temporal limitations and technical challenges of geodetic methods. Towards IPCC SROCC and AR6, there have been considerable improvements with respect to available geodetic datasets. Geodetic volume change assessments for entire mountain ranges have become possible thanks to recently available and comparably accurate digital elevation models (e.g., from ASTER or TanDEM-X). At the same time, new spaceborne altimetry (CryoSat-2, IceSat-2) and gravimetry (GRACE-FO) missions are in orbit and about to release data products to the science community. This opens new opportunities for regional evaluations of results from different methods as well as for truly global assessments of glacier mass changes and related contributions to sea-level rise. At the same time, the glacier research and monitoring community is facing new challenges related to the spread of different results as well as new questions with regard to best practises for data processing chains and for related uncertainty assessments.In this presentation, we introduce the Glacier Mass Balance Intercomparison Exercise (GlaMBIE) project of the European Space Agency, which is building on existing activities and the network of the International Association of Cryospheric Sciences (IACS) working group on Regional Assessments of Glacier Mass Change (RAGMAC) to tackle these challenges in a community effort. We will present our approach to develop a common framework for regional-scale glacier mass-change estimates towards a new data-driven consensus estimate of regional and global mass changes from glaciological, DEM-differencing, altimetric, and gravimetric methods.

How to cite: Jakob, L., Zemp, M., Gourmelen, N., Dussaillant, I., Nussbaumer, S. U., Hock, R., Berthier, E., Wouters, B., Gardner, A. S., Moholdt, G., Brun, F., and Braun, M. H.: GlaMBIE – An intercomparison exercise of regional and global glacier mass changes, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-5944, https://doi.org/10.5194/egusphere-egu23-5944, 2023.

It is essential to test the potential impact of processes missing in the global-scale models that are used to project on glacier mass balance in the glacier models, such as the modification of the mass balance through debris-cover. In this study, we evaluate the impact of a parameterization of debris cover on glacier mass change projections using the Open Global Glacier Model (OGGM). The assessment of uncertainties about the potential impact of debris cover on global scale modeling is complicated by the scarcity of suitable data on debris-covered glaciers for validation (e.g., mass balance measurements for individual elevation bins on debris-covered glaciers). To calibrate and validate the mass balance module, we rely on glacier-wide geodetic volume changes. Debris cover can enhance ice melting if less than a few centimeters thick, or decrease ice melting through insulation of the underlying ice by a thick layer of debris. Ice cliffs, supraglacial ponds and streams associated with debris cover may increase the absorption of heat and increase ice melting. In OGGM, the effects of debris cover are parameterized by a simple modification of the mass balance module, through introducing an elevation-dependent temperature sensitivity parameter (“degree-day factor”) and including a debris-related melt correction factor. While debris cover plays only a minor role on glacier mass change on the global scale, this becomes important on regional and individual glacier scales. Our results show the effect of debris cover could improve model performance on the mass balance gradient, not the overall mass balance. To validate our results on the mass balance gradient, we rely on the geodetic mass balance for each elevation band.

How to cite: Mojtabavi, S., Rounce, D., Maussion, F., and Marzeion, B.: Uncertainty assessment of modeling the impact of debris cover on global glacier mass change: challenges and solutions, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9828, https://doi.org/10.5194/egusphere-egu23-9828, 2023.

How to cite: Draeger, C. and Radic, V.: Reconstruction of Glacier Mass Balance with Surface Energy Balance Modeling across Southwestern Canada, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10429, https://doi.org/10.5194/egusphere-egu23-10429, 2023.

A recent large-scale glacier model intercomparison revealed a strong influence of model design choice on glacier projections. Here we examine the influence of various temperature-index mass-balance models and calibration options. With the Open Global Glacier Model (OGGM) framework, we compare the performance and projections of model options such as the use of surface-type dependent degree-day factors as well as varying temporal climate resolution (daily, monthly) and downscaling strategies (temperature lapse rates, temperature and precipitation correction). We focus on 88 glaciers with long term observations of mass-balance profiles and seasonal mass-balance, allowing us to assess the added value of using multiple mass-balance statistics in the calibration process. We find that using interannual mass-balance variability to calibrate otherwise fixed parameters generally leads to an improved representation of the mass-balance gradient, which in turn is a crucial explanatory variable for future glacier evolution. Therefore, we also find a strong influence of the calibrated temperature lapse rates on future glacier volume. Using different degree-day factors for snow, firn, and ice leads to nonlinear sensitivities, where future glacier loss depends on how the accumulation area changes compared to the calibration period. Our study illustrates the strong impact of temperature-index model choice on projected glacier volume and runoff. But it cannot clearly demonstrate the added value of additional model complexity due to the lack of independent observations.

How to cite: Schuster, L., Rounce, D., and Maussion, F.: Glacier projections sensitivity to temperature-index model and climate downscaling parameter calibration choices, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-11983, https://doi.org/10.5194/egusphere-egu23-11983, 2023.

 Glacier meltwater is a significant component of the regional runoff In High Mountain Asia (HMA),. However, the majority of the HMA's glaciers are rapidly losing their mass, putting the long-term viability of meltwater as a component of river flow at risk. It is, hence, crucial to comprehend the long-term glacier response to climate change at the regional scale as well as the impact of non-climatic characteristics like morpho-topographic factors on ice loss. We estimate changes for 445 glaciers in the upper Alaknanda basin and neighboring transboundary glaciers using multi-temporal optical satellite images from 1973 to2020. Our measurements indicate a mean annual area change of −1.14 ± 0.07 m a^–1 and a geodetic glacier mass balance of −0.34 ± 0.08 m w.e.a^–1 for the whole period. Before 2000 (1973-2000), the mean regional glacier mass loss rate was -0.30 ± 0.07 m w.e.a^-1, which increased to -0.43 ± 0.06 m w.e.a^-1 during 2000-2020. The mass loss increased further (-0.68 ± 0.09 m w.e.a^-1) in the recent period (2015-2020) and we observed heterogeneous mass loss both in spatial and temporal scales. Our analysis revealed that the current significant glacier imbalance is probably a result of the rising temperature trend as revealed from the ERA5 Land reanalysis data. An extended ablation season due to the strong seasonal temperature increase has further accelerated glacial mass loss. Steep and higher elevation glaciers were less affected by negative mass budget. This can be explained beside the lower average temperatures at higher elevation by a rapid transfer of snow and ice that helped them to readjust their geometry compared to glaciers at lower elevation, having more gentle slopes and lower dynamics. Such low elevation glaciers are unlikely to recover in coming decades if the current trend of warming continues. We also identified a surging glacier draining onto the Tibetan Plateau that advanced rapidly by around 800 m within three months in Sep-Dec 2019. The advance is still ongoing, though at a much-reduced rate. Our temporally detailed measurements of glacier change provide an in-depth view of glacier evolution in the Alaknanda Basin and will improve the estimation of meltwater run-off component of the hydrological cycle. 

How to cite: Bhattacharya, A., Mukherjee, K., King, O., and Bolch, T.: Examining the impact of climatic and non-climatic attributes on glacier mass budget and surging in Alaknanda Basin, India, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12836, https://doi.org/10.5194/egusphere-egu23-12836, 2023.

Glaciers and ice caps in the Tien Shan mountains play a crucial role in terms of water supply for irrigation, industry and drinking. The retreat of these ice masses can consequently have a major impact on downstream densely populated dry lowland areas. Therefore, it is crucial to understand how ice masses in the Tien Shan are reacting to climate change and how they will evolve in the future. In this study, we model the future evolution of all glaciers and ice caps in the Tien Shan mountains under CMIP6 SSP climate scenarios using the large-scale GloGEMflow model. The model is calibrated to match glacier-specific geodetic mass balances while accounting for debris cover under recent climatic conditions (downscaled climate reanalysis, ERA5). In our modelling framework, we rely on a total of six independent ice thickness datasets, of which the effect on the modelled future glacier evolution is analysed in detail. Our results reveal an exceptionally pronounced retreat of most of the ice masses under all climate scenarios (vs. other regions), which can be related to the specific climate regime. Since most of the precipitation on Tien Shan glaciers occurs in spring and early summer, temperature increases not only increase melt (as is the case in most regions around the world), but additionally strongly influence the precipitation type (solid vs. liquid). In all scenarios, the total runoff for the major river catchments in the Tien Shan is projected to drastically reduce by the end of the 21st century. Further, peak water is modelled to be reached before the middle of the century and the annual runoff peak is anticipated to shift from early summer towards late spring.

How to cite: Van Tricht, L., Zekollari, H., Farinotti, D., Huss, M., Compagno, L., and Huybrechts, P.: Future glacier and runoff evolution in the Tien Shan mountains, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-13056, https://doi.org/10.5194/egusphere-egu23-13056, 2023.

We present Bedmap3, the first comprehensive and openly available compilation of Antarctic Ice Sheet survey datasets, plus the latest gridded mapping products of ice thickness and the surface and bed topography of the whole Antarctic continent and continental shelf. For 60 years, scientists have strived to understand the past, present and future of the ice sheet, a goal that has become ever more urgent as ice loss accelerates. Key to this research has been the mapping of the bed-topography, surface slope and ice-thickness parameters that are crucial for modelling ice flow, and hence for predicting future ice loss and ensuing sea level rise. Supported by the Scientific Committee on Antarctic Research (SCAR) and data contributions from the international survey community, the Bedmap3 Action Group has now produced substantially updated gridded maps of these parameters and, for the first time, has standardized and made available all underlying geophysical survey data points from all Antarctic ice-thickness survey campaigns since the 1950s. Here, we describe the standardization that makes these and future datasets accessible under the ‘Findable, Accessible, Interoperable and Reusable’ (FAIR) data principles, allowing scientists to re-use these data freely for their own analysis. We also describe the results of our new mapping, and how this changes our view of Antarctica’s hidden landscapes and its potential to dominate future sea level rise.

How to cite: Fremand, A., Pritchard, H., Fretwell, P., and Bodart, J.: Bedmap3: new data and gridded products of Antarctic ice thickness, surface and bed topography, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-13665, https://doi.org/10.5194/egusphere-egu23-13665, 2023.

Summer heat waves have a substantial impact on glacier melt as emphasized by the extreme summer of 2022 that caused unprecedented mass losses to the Swiss glaciers. Despite the dramatic impact on glaciers, the summer of 2022 offered a unique opportunity to analyze the implications that such extraordinary events have on glacier melt and related runoff release.

This study presents a novel approach based on computer-vision techniques for automatically determining daily mass balance variations at the local scale. The approach is based on the automated recognition of color-taped ablation stakes from camera images acquired at six sites on three Alpine glaciers in the period 2019-2022. The validation of the method revealed an uncertainty of the automated readings of ±0.81 cm d^-1. By comparing the automatically retrieved mass balances at the six sites with the average mass balance of the last decade derived from seasonal in situ observations, we detect extreme melt events in the summer seasons of 2019-2022.

The in-depth analysis of summer 2022 allows us to assess the impact that the summer heat waves have on glacier melt. With our approach we detect 23 days with extreme melt over the summer, emphasizing the strong correspondence between heat waves and extreme melt events. The Swiss-wide glacier mass loss during the 25 days of heat waves in 2022 is estimated as 1.27 ± 0.10 Gt, corresponding to 35% of the overall glacier mass loss in the summer of 2022. As compared to the 2010-2020 average glacier mass change, days with extreme melt in 2022 correspond to 56% of the mass change during the summer period, thus demonstrating the significance of heat waves for seasonal melt.

How to cite: Cremona, A., Huss, M., Landmann, J. M., Borner, J., and Farinotti, D.: Automated ice ablation readings reveal the significance of summer heat waves for glacier melt, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-1062, https://doi.org/10.5194/egusphere-egu23-1062, 2023.

Glaciers in the Tibetan Plateau are melting at an unprecedented rate in the context of global warming. Hailuogou (HLG) Glacier, a rapidly receding temperate land-terminating glacier in the southeastern Tibetan Plateau, has been observed to lose mass partly through ice frontal mechanical ablation (i.e., ice collapse).

In this study, we present analysis from Uncrewed Aerial Vehicles (UAV) surveys conducted over nine field campaigns to the HLG Glacier, providing evidence of glacier change and frontal ice collapse between 2017 and 2021. Structure from Motion with Multi-View Stereo was applied to produce multi-temporal Digital Surface Models (DEMs) and orthophoto mosaics, from which geomorphological maps and DEMs of Difference were derived to quantify the changes of the glacier snout and the ice loss from frontal ice collapse. Based on that, a linear correlation of Area-Volume for frontal ice collapse was subsequently built. Planet images were used to identify additional ice collapse events (i.e., 2017 to 2021) and to extract time-sequenced glacier extents. ASTER-derived DEMs generated by NASA Ames Stereo Pipeline (ASP) were then differenced to calculate the ice volume changes in the period. Combined with frontal ice collapse events identified from Planet, the contribution of that to the glacier mass balance can be estimated from the established Area-Volume correlation.

These analyses reveal that at the margins of the glacier terminus retreated 132.1 m over the period of analysis, and that in the area specifically affected by collapsing (i.e., the glacier collapsed terminus), it retreated 236.4 m. Overall the volume lost in the terminal area was of the order of 184.61 ± 10.32 x 10⁴ m³, within which the volume change due to observed collapsing events comprises approximately 28%. We show that ice volume changes at the terminus due to a single ice collapse event may exceed the interannual level of volume change, and the daily volume of ice loss due to ice calving exceeds the seasonal and interannual level by a factor of ~ 2.5 and 4. The contribution to the mass balance change of the entire glacier that is attributed to frontal ice collapse is limited (i.e., ranges from 0.48% to 1.12% from 2017 to 2021). However, the mechanical ablation (e.g., frontal ice collapse and subglacial/englacial conduit’s roof collapse) has probably changed the way of losing ice mass to some extent.

Our results suggest that the evolution of the HLG Glacier terminus is dominantly controlled by the frontal ice collapse. The projection of the recession rate of the HLG Glacier may well be underestimated if based on surface mass balance alone, as the frontal ice collapsing might be more frequent and larger under the context of warming. If the future evolution of glaciers such as HLG Glacier is to be robustly predicted, the contribution of mechanical ablation should be accounted for by numerical models.

How to cite: Xu, S., Fu, P., Quincey, D., Feng, M., Marsh, S., Liu, Q., and Jia, T.: The Changes of Hailuogou Glacier in the Southeastern Tibetan Plateau and the Impacts on Glacier Dynamics from the Mechanical Ablation, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-2555, https://doi.org/10.5194/egusphere-egu23-2555, 2023.

Similar to the Patagonian Icefields, the Cordillera Darwin Icefield in Tierra del Fuego experienced important ice loss during the last decades. The difficult accessibility and the harsh weather conditions in that area result in scarce in-situ observations of climatic conditions and glacier mass balances. Under these challenging conditions, this study investigates calibration strategies of surface mass balance models in the Monte Sarmiento Massif, western Cordillera Darwin, with the goal to achieve realistic simulations of the regional surface mass balance in the period 2000-2022.

We apply three calibration strategies ranging from a local single-glacier calibration to a regional calibration with and without the inclusion of a snowdrift parametrization. Furthermore, we apply four models of different complexity ranging from an empirical degree-day model to a fully-fledged surface energy balance model. This way, we examine the model transferability in space, the benefit of including regional mass change observations as calibration constraint and the advantage of increasing the model complexity regarding included processes. In-situ measurements comprise ablation stakes, ice thickness surveys and weather station records at Schiaparelli Glacier as well as elevation changes and flow velocity from satellite data for the entire study site. Performance of simulated surface mass balance is validated against geodetic mass changes and stake observations of surface melting.

Results show that transferring mass balance models in space is a challenge, and common practices can produce distinctly biased estimates. The use of remotely sensed regional observations can significantly improve model performance. Increasing the complexity level of the model does not result in a clear improvement in our case where all four models perform similarly. Including the process of snowdrift, however, significantly increases the agreement with geodetic mass balances. This highlights the important role of snowdrift for the surface mass balance in the Cordillera Darwin, where strong and consistent westerly winds prevail.

How to cite: Temme, F., Farías-Barahona, D., Seehaus, T., Jaña, R., Arigony-Neto, J., Gonzalez, I., Arndt, A., Sauter, T., Schneider, C., and Fürst, J. J.: Calibrating Surface Mass Balance Models at the Monte Sarmiento Massif, Tierra del Fuego, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-2824, https://doi.org/10.5194/egusphere-egu23-2824, 2023.

The Ladakh region in the western Himalaya relies directly on snow-glacier-fed first order streams for domestic and agricultural needs of the population. Despite the significant contribution of glacier meltwater towards community livelihood and its vulnerability in a warmer climate, glaciers in the Ladakh region have not been studied comprehensively. Previous studies, mostly at a poor spatial (~90 m) and temporal (10-15 years) scales, focused only on geodetic glacier mass balance estimation, hence their controlling climatic drivers remain unknown. In this study, we estimate the geodetic mass balance of the glaciers of the Ladakh region, using multiple digital elevation models of 30 m resolution acquired between 2000 (Shuttle Radar Topography Mission; SRTM) and 2021 (Advanced Spaceborne Thermal Emission and Reflection Radiometer; ASTER). Due to the large aerial coverage, we divided the whole Ladakh region into two sub-regions namely the eastern and western Ladakh. The primary climatic drivers of glacier mass balances were examined using the long-term ERA5-Land reanalysis data, complemented by available in-situ meteorological data. The role of non-climatic (morphological) variables on glacier mass balances was also investigated in detail. The results reveal a negative glacier mass balance over the Ladakh region during the last two decades, with significant spatial variability. Glaciers in western Ladakh lost higher mass (-0.35 ± 0.07 to -0.37 ± 0.07 m w.e. a^-1) compared to eastern Ladakh (-0.21 ± 0.07 to -0.33 ± 0.05 m w.e. a^-1). Although the widespread mass loss in Ladakh is primarily caused by warming, the variations in spatial mass loss are primarily caused by the morphological settings of the glaciers. The eastern Ladakh glaciers are located at higher elevations and small sized, whereas western Ladakh glaciers are large sized and their tongues are situated at lower elevations (low-elevation-hypsometry), therefore, the impact of temperature is much higher in them, leading to higher mass loss. The non-climatic factors (morphological control) exhibit a dominant role than climatic factors in governing the glacier mass balances, particularly in the EL. The comparison between ASTER-based and the Ice, Cloud and land Elevation Satellite (ICESat)-2 laser altimetry-based mass balances shows a good agreement, reaffirming the robustness of regional mass balance estimates. Overall, glaciers of the Ladakh region are losing mass and the western Ladakh glaciers are potentially more susceptible to warming climate compared to the eastern Ladakh.

How to cite: Mandal, A., Vishwakarma, B. D., Angchuk, T., Azam, M. F., Garg, P. K., and Soheb, M.: Glacier mass balance and its climatic and non-climatic drivers in the Ladakh region during 2000-2021 from remote sensing data, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6319, https://doi.org/10.5194/egusphere-egu23-6319, 2023.

Kersten Glacier, a slope glacier on the southern flank of Kilimanjaro, has been observed to shrink for many decades. Quantitatively, the glacier’s mass balance has been studied by Mölg et al. (2009) with a distributed physically based mass balance model. In this study, the research question is revisited using the open-source COupled Snowpack and Ice surface energy and mass balance model in PYthon (COSIPY; Sauter et al., 2020). A spatially distributed simulation of the surface energy and mass balance of Kersten Glacier is performed for February 2005 to January 2008. The model is driven by hourly observations from an automated weather station in the top region of the glacier at 5873 m MSL. In this contribution, we present findings such as the different components of the mass and energy balance, their temporal variation and elevational characteristics, and compare them to the results obtained from the 2009 analysis. This should allow a first assessment of COSIPY’s skill as a future tool for simulating snow and ice variability in equatorial latitudes.

How to cite: Schramm, M. and Mölg, T.: Comparative study of the surface energy and mass balance of Kersten Glacier on Mt. Kilimanjaro: COSIPY versus previous modeling, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6623, https://doi.org/10.5194/egusphere-egu23-6623, 2023.

Debris-covered glacier tongues are widespread in high-relief mountains and are characterised by highly undulated surfaces with supraglacial ponds and circular cliffs in hummocky topography. These features are known to strongly enhance mass loss on debris-covered surfaces and have been widely mapped, but their formation mechanisms and underlying controls have not been studied in detail.

Here we aim to investigate the role of supraglacial streams on the morphological development of debris-covered glacier surfaces and related supraglacial features such as ice cliffs based on high-resolution DEMs and orthophotos (Pleiades and UAV) from two debris-covered glaciers of contrasting spatial scales: the Satopanth Glacier located in the Indian Himalaya and the Zmuttgletscher in the European Alps. We systematically analyse the morphological development of the debris-covered surface along the glacier from the onset zone of debris cover and supraglacial channels down to the hummocky and sunken tongue surfaces. We perform this using a semi-automated approach that includes meltwater flow routing, the extraction of surface roughness, profiles and extents of supraglacial channel-influenced valleys, as well as the mapping of ice cliffs.

Based on this analysis, we find a clear and coherent succession of morphological developments along both glaciers that seems initiated through erosion from supraglacial streams. On the initially smooth debris-covered surface, locally incised and meandering channels initiate ice cliffs that progressively backwaste, creating a downstream-widening supraglacial valley with an undulated surface. This ‘mobility area’ is advected downstream even beyond the moulins where the supraglacial channels drain to the bed. Further downstream, neighbouring ‘mobility area’ valleys laterally merge and create the quasi-chaotic highly undulated surfaces typically observed on tongues of debris-covered glaciers. We integrate these interpretations into a conceptual model that links the downstream morphological development of debris-covered surfaces and explains the genesis of related features such as ice cliffs.

How to cite: Ouvry, B., Kneib, M., Purves, R. S., and Vieli, A.: Development of supraglacial meltwater streams and their influence on the morphology of debris-covered glacier surfaces., EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-7645, https://doi.org/10.5194/egusphere-egu23-7645, 2023.

The search for satellite-derived proxies of the surface mass balance (SMB) of glaciers is of crucial importance for an updated estimation of the SMB worldwide. The minimum mean  glacier albedo (αmin, calculated as the minimum mean albedo over the whole glacier) attained along a season has proved to be a good proxy. In this work we demonstrate that SMB estimations can be improved by adding albedo decay parameters as predicting variables. The SMB of Hurd glacier (Livingston Island, Antarctica) has been continuously monitored since 2001, with available values of annual, summer and winter SMB. MODIS MOD10A1 daily snow albedo product with a spatial resolution of 500 m over the glacier was downloaded using the Google Earth Engine Application Programming Interface. Data from 2000-2001 to 2020-2021 season are considered in this work. MOD10A1 data were filtered using a maximum filter followed by a first-order Butterworth filter. Each season extends from September to March of two consecutive years. The seasonal albedo was fitted to an exponential decay α=αm+Aexp(-βt), and parameters αm, A and β as well as the albedo decay duration (D) were calculated for all pixels. Mean values of αm, β, A and D were calculated over the whole glacier. Monthly and seasonal mean αmin were also estimated from MOD10A1 data. Simple linear regressions show that the minimum albedo in the period January-February explains the summer SMB, while the minimum albedo in the period December-January explains the annual SMB. Multiple linear regression models including snow albedo decay parameters improve the quality of the models, increasing the value of the coefficient of determination and decreasing the root mean square difference between measured and predicted SMB. These results show that the SMB is determined not only by αmin but also by how fast and for how long the albedo decay takes place, and by the difference between the minimum albedo and the surface albedo at the beginning of the season. On the other hand, most of the snow albedo decay takes place in the period from late September to December. Previous investigations of mass balance over the Hurd Peninsula have established that the snow melt in Hurd Peninsula takes place mostly from December to March. The fact that snow melting lags behind albedo decay can be explained if we consider that some surface snow metamorphic processes occur prior to melting and that melting continues after surface snow has attained its maximum degree of metamorphism.

How to cite: Calleja, J., Muñiz, R., Navarro, F., Otero, J., and Fernández, S.: Improvements in the estimation of glacier surface mass balance taking into account albedo decay parameters, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-7890, https://doi.org/10.5194/egusphere-egu23-7890, 2023.

Western Canada contains 72% of the glaciers within the Randolph Glacier Inventory (RGI) region 2 (Western Canada and USA), and these glaciers constitute 95% of the region’s total ice cover. Recent studies exploiting stereoscopic imagery from NASA’s Terra satellite (ASTER) have reduced biases in the number, type and distribution of glaciers used to assess regional glacier mass change. The elevation uncertainty of ASTER digital terrain models, in addition to infrequent sampling, confounds its use to detect trends in elevation change at seasonal scales or for small glaciers (< 1km²). Since summer 2014, the Hakai-UNBC Airborne Coastal Observatory (ACO) routinely acquires laser altimetric data at the end of the accumulation (late-April to early May) and ablation (early to late September) over many of western Canada’s glaciers and icefields using an aircraft equipped with an 1064-nm laser scanner and dedicated positional hardware. Post-processed uncertainties of repeated, co-registered elevational data over stable terrain yield uncertainties that are typically below ±0.3 m (±1s). Since 2020, our bi-annual acquisitions sample over 800 glaciers (about 2,000 km²), which constitutes about 15% of the total areal extent of ice in RGI-02. While the area-altitude distribution of ACO sampled glaciers largely accord with those of RGI-02, our sampling program captures fewer glaciers that exist at highest elevations, and the average glacier size surveyed by us is about three times larger than the average glacier size within RGI-02 (0.77 km²). Our archive reveals important aspects of glacier elevation change that cannot be obtained from existing publicly available sources of digital terrain data such as the magnitude of seasonal to inter-annual changes during the accumulation and ablation seasons, short-term horizontal transfers of mass, changes in volume and extent of transient late-lying snow, and the effects of short-term meteorological events (e.g. heat waves or forest fires) on regional melt events for a given year. We plan to release this archive to the public so it can be used to validate in-situ mass balance measurement programs, improve melt models and provide insight into physical factors that drive glacier change.

How to cite: Menounos, B., Heathfield, D., Beffort, S., Viner, N., Gonzalez Arriola, S., and White, R.: A nine-year, airborne laser scanning archive of glacier change, western Canada., EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9820, https://doi.org/10.5194/egusphere-egu23-9820, 2023.

Information on glacier velocity is imperative to understand glacier mass, ice volume, topography, surge events of the glacier and response to climate change. Therefore, inter-annual surface ice velocity (SIV) of the Panchi Nala glacier has been calculated in the current study between the first two decades of the twenty-first century. To do so, the SIV has been computed by the feature tracking technique using the Co-registration of Optically Sensed Images and Correlation (COSI-Corr) method applied on the multi-temporal Landsat (TM and OLI) and sentinel -2 MSI images acquired between 2000 and 2021. The results of the study show that the mean velocity of the debris-covered tongue of the Panchi Nala Glacier is ∼10.60 ± 5.56 m/y during the study period. Additionally, the highest average glacier velocity is 13.77 ± 4.64 m/y, whereas the lowest is 8.92 ± 2.78 m/y, respectively, observed in 2005 and 2015. Also, the 95% confidence interval of the mean annual velocity lies between 9.76 and 11.43 m/y during the entire study period. The annual heterogeneity is linked with the variation of summer precipitation. Statistically, a 100 mm increment of summer precipitation can reduce the velocity around 1.3 m/y. The main reason behind this is the Panchi Nala glacier is located in high-elevation where the climate is much colder and during the summer precipitation, the lower temperatures cause the precipitation to take the form of snow, which freezes and accumulates on the glacier. This reduces the process of basal sliding. Further, detailed investigations with additional parameters need to be carried out to elucidate the comprehensive causes for inter-annual fluctuations in surface velocity. In this perspective, future research maybe directed towards higher temporal and spatial scale remote sensing-based investigations and validation of glacier surface velocity using field measurements, to better understand the glacier dynamics.

Keywords: Glacier surface ice velocity; debris cover; climate change; western Himalayas.

How to cite: Prajapati, M., Garg, P. K., Shukla, A., and Guha, S.: Dynamics of the Panchi Nala glacier, western Himalaya: trends and controlling factors., EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10964, https://doi.org/10.5194/egusphere-egu23-10964, 2023.

Glacier fronts are retreating across the globe in response to climate warming, revealing valleys, fiords, and proglacial lakes. The piedmont lobe of San Quintín, the largest glacier of the Northern Patagonia Icefield, in southern Chile, has recently entered a catastrophic phase of frontal retreat, where its terminus is rapidly disintegrating into large tabular icebergs calving into a new proglacial lake. We present results of a unique airborne GPR survey of the terminus of this large Patagonian glacier (763 km² in 2017), complemented with an analysis of ice flow velocity, satellite imagery, and ice elevation change to show that the ongoing retreat is caused by recent detachment of a floating terminus from the glacier bed and may shortly lead to the disappearance of the last existing piedmont lobe in Patagonia. Finally, we discuss how the observations of San Quintín’s ongoing collapse may give insights into processes governing frontal retreat of fast-flowing temperate glaciers and the quasi-stability of the floating termini.

How to cite: Pętlicki, M., Rivera, A., Oberreuter, J., Uribe, J. A., Reinthaler, J., and Bown, F.: Frontal collapse of San Quintín glacier (Northern Patagonia Icefield), the last piedmont glacier lobe in the Andes, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10966, https://doi.org/10.5194/egusphere-egu23-10966, 2023.

Glaciers distinct from the ice sheets melt quickly around the world. Rounce et al., 2023, recently projected (28±9) % glacier mass will be lost in 2100 compared to 2015 in the low-emission scenario SSP1-2.6. In our study, we focus on the observation of recent glacier mass change on Novaya Zemlya, Russian Arctic, using the SARIn altimeter CryoSat-2. Taking advantage of CryoSat-2's spatiotemporal resolution and a novel processor optimized for mountain glaciers, we show a prompt response to the surface temperatures modeled by the regional atmosphere model MAR. Prominent warm and high-melt years were 2013, 2016, and 2020, while 2014 featured lower surface temperature and melt than average. We discuss the potential drivers for those extreme years and furthermore exploit surface mass balance estimates to analyze the relative contribution of ice dynamics and surface processes to the observed mass loss acceleration, and regional variations therein.

How to cite: Haacker, J. and Wouters, B.: Prompt response of glaciers on Novaya Zemlya to recent warm summers shown by CryoSat swath data, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-11912, https://doi.org/10.5194/egusphere-egu23-11912, 2023.

A developed boundary layer can decouple a glacier's response to the ambient meteorological conditions, though glacier retreat can limit this boundary layer development and increase a glacier’s sensitivity to climate change. We explore six years of distributed meteorological data on a small Swiss glacier in the period 2001-2022 to highlight its changing response to local conditions. We find an increased sensitivity (ratio) of on-glacier to off-glacier temperature changes as the glacier has retreated and its debris-cover area expanded. The glacier lost ~60% of area since 1994, coinciding with notable frontal retreat post 2005 and an observed switch from down-glacier to up-glacier winds in the upper ablation zone from 2001-2022. Increased sensitivity to external temperature changes is thus driven by a combination of increased up-glacier winds and the larger extent of ice exposed to warm air at a retreating, debris-covered glacier terminus. Calculated sensible heat fluxes on the glacier are therefore increasingly determined by the conditions occurring outside the boundary layer of the glacier, highlighting the expected negative feedback of smaller Alpine glaciers as the climate continues to warm and experience an increased frequency of extreme summers.

How to cite: Shaw, T., Buri, P., McCarthy, M., Miles, E., Ayala, Á., and Pellicciotti, F.: Meteorological Feedbacks on a Decaying Alpine Glacier, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12248, https://doi.org/10.5194/egusphere-egu23-12248, 2023.

Iceland enjoys a much warmer climate than the average for its latitude. A major reason for this is the warm ocean currents in the Atlantic south of Iceland. There is a large heat flux from the ocean to the atmosphere and the air temperature therefore depends to a high degree on the ocean temperature. During the last roughly two decades, glaciers in Iceland have generally been retreating as well as having a negative mass balance due to a warmer climate, whereas during three decades prior to that, most of the glaciers in Iceland were advancing. The air temperature in Iceland south of the largest Icelandic glacier, Vatnajökull, showed a rise in temperature of about 1°C from 1995 to the early 2000’s and since then it has mostly remained at this high level. Often this warmer climate is attributed entirely to global warming. However, the temperature in the warm and saline Atlantic water south of Iceland also increased by about 1°C during the same period. This rise in ocean temperature was accompanied by an increase in salinity which indicates that the temperature rise was mostly due to a change in the ocean circulation, resulting in advection of warmer and saltier water to the area. In the period from 1995 to the early 2000’s the ocean heat flux with the Atlantic water across the Greenland-Scotland ridge increased by 21 TW, partly through Denmark Strait towards the continental shelf north of Iceland. The increased heat flux was attributed to a rising temperature as well as increased flow of Atlantic water. Only about 0.5% of this heat flux increase is needed to explain the recent melting of Icelandic glaciers. With a relatively sudden 1°C rise in temperature the glaciers will take decades to reach equilibrium with this new temperature and if the temperature does not decrease, the glaciers will continue to lose mass. There are records of advancing and retreating Icelandic glaciers from 1930 and they show a good correspondence with the Atlantic Multidecadal Oscillation (AMO), that reflects temperature variations in the North Atlantic Ocean.

How to cite: Jónsson, S.: The role of the ocean circulation in melting the glaciers in Iceland, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12667, https://doi.org/10.5194/egusphere-egu23-12667, 2023.

With around 46 km², Nevado Coropuna (NC, 15°32’S, 72°39’W; 6377 m) is the largest tropical icecap in the world. NC is situated on a stratovolcano structure with six peaks over 6000 meters, in the arid border of the Andean plateau, Southern Peru. NC is a vital source of freshwater for the communities within the Majes valley and the vast irrigation plans located in the same valley and on the arid coastal strip. Our MOTICE project will model the evolution of NC until 2100 CE in response to climate change.

We present initial results on the modelling of NC, which will be used to tune the glaciological parameters for the projections under different RCP scenarios. Mass balance was modelled using the COupled Snowpack and Ice surface energy and MAss balance model in Python (COSIPY). This was forced with climate data for the 1950-2020 period from the ERA5-Land reanalysis, which provided surface pressure, cloud cover, incoming shortwave and longwave radiation, wind speed, 2-meter air temperature and relative humidity fields. This was combined with the RAIN4PE gridded product outputs for daily precipitation during the 1981-2015 period. The climate dataset was downscaled and validated with observed temperature and relative humidity from a weather station located on the Cavalca glacier (5800 m above sea level), at the northern part of NC. Glacier mass balance results were validated with measured mass balance in the same glacier for the 2014-2019 period. The mass-balance outputs from COSIPY were used for glacial flow modelling, using the Parallel Ice Sheet Model (PISM).

Subglacial topography was modelled using the Volume and Topography Automation (VOLTA) tool in a DEM that had been previously corrected with 70 differential GPR points measured “in situ”. The subglacial topography was also tuned and validated against in-situ GPR measurements in four glaciers of NC. Both GPR and GPS measurements were conducted during the 2022 fieldwork campaign, in which large areas of debris-covered ice were also located, mapped and measured. However, debris covered ice has not been considered in this initial model run.

Our preliminary results were compared to the actual 1955-2020 glacier surface evolution, which was retrieved from aerial photography and topographic maps for the initial stage in 1955 and from satellite images from 1975 onwards. This work highlighted the difficulty of modelling tropical glaciers, especially accounting for processes important to tropical ice, such as sublimation, and short-lived meteorological events.

How to cite: Pellitero, R., Bonshoms, M., Ely, J. C., and Liguori, G.: Modelling the future of Nevado Coropuna (Peru), the world’s largest tropical ice cap., EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12767, https://doi.org/10.5194/egusphere-egu23-12767, 2023.

Glaciers in Norway are retreating, following the global trend caused by climate change. Future mass loss is projected to increase and cause the majority of Norwegian glaciers to disappear by the end of the century. This alters runoff and downstream hydrology, thus affecting available water resources for local communities. Quantifying this future change is essential to aid societal climate adaptation.

In this work, our aim is to assess both the rate of change and how trustworthy these estimates are given uncertainty in future climate and simulation tools. We organize the study around the Folgefonna ice cap in Western Norway, a prime example of a maritime ice cap. For the people living in its vicinity, the ice cap is important as it acts as a reservoir for both hydropower generation and drinking water. Despite this, little is known about how it will change in the future.

Using the model BESSI (The Bergen Snow Simulator), we simulate the surface mass balance of Folgefonna at a high spatial resolution. Recent developments of BESSI have made it a suitable option for small-scale glacier studies. These model alterations are presented here as well as results of past and future surface mass balance for the ice cap. We quantify how sensitive Folgefonna is to climate change and discuss limitations to the tools available for future glacier projections.

How to cite: Frøystad, R. and Born, A.: Sensitivity of Folgefonna ice cap to anthropogenic climate change, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12975, https://doi.org/10.5194/egusphere-egu23-12975, 2023.

Recent glacier mass changes are very heterogeneous in High Mountain Asia, owing to climatic variability and the mass balance sensitivity to climate, which may differ from one region to another. Mera glacier in the Everest region is one of the longest field-based monitored and well-studied glaciers of the Central Himalaya. In this study, we examine the sensitivity of Mera glacier mass balance to climate variables using the COupled Snowpack and Ice surface energy and mass balance model in PYthon (COSIPY), using 4 years (2016-2020) of in-situ meteorological data recorded at different elevations in the ablation and accumulation zones of the glacier. This shows that the net short-wave radiation is the main energy input at the surface, and in turn albedo is a key parameter controlling the glacier mass balance. As a result, at 5360 m asl, in the ablation zone, surface melt accounts for 90% of mass loss whereas sublimation and subsurface melt account for less than 10%. This analysis is performed at point scale at 5360 and 5770 m asl, in the ablation and accumulation zones respectively, as well as in a distributed way. We produce and analyze 88 distinct climatic scenarios, varying from dry and warm to wet and cold conditions. Dry conditions, primary during the pre-monsoon and secondary during the monsoon, strongly decrease the glacier mass balance, revealing that the annual amount and the seasonal distribution of snowfalls primary drives the glacier-wide mass balance of Mera Glacier.   

How to cite: Khadka, A., Wagnon, P., and Brun, F.: Energy and mass balance of Mera glacier (Everest region, Central Himalaya) and its sensitivity to climate, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-13874, https://doi.org/10.5194/egusphere-egu23-13874, 2023.

Debris-covered glaciers can react differently to external forcings than clean-surface glaciers. Depending on its thickness, a supraglacial debris layer impacts the glacier mass balance by either enhancing the surface melt or protecting the underlying ice. Based on previous works focusing on simple flowline geometries, we extended the model setup to three-dimensional complex geometries. The framework is implemented using the Ice-sheet and Sea-level System Model (ISSM) and applied to a typical alpine glacier geometry. Ice dynamics are solved on high-resolution with full-Stokes and coupled to the surface debris transport equation. The employed surface mass balance (SMB) model is capable of describing the melt rate for all debris thicknesses by including turbulent fluxes within the upper debris cover. This SMB formulation resolves the enhanced melt rates for a thin debris cover as well as the decreasing melt rates for thickening debris. To test the sensitivity of future projections of alpine glaciers on the debris layer, simulations are forced with high-resolution regional climate model (RCM) data from the EURO-CORDEX ensemble (RCP2.6 and RCP8.5).

How to cite: Rückamp, M., Morlighem, M., and Mayer, C.: Modelling the future evolution of an alpine debris-covered glacier, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-14348, https://doi.org/10.5194/egusphere-egu23-14348, 2023.

The importance of glaciers in High Mountain Asia (HMA) is significant in sustaining mountain hydrology and the runoff of several river systems that originate from these glaciers. The availability of numerous remote sensing products provides an opportunity to assess the current regional glacier mass balance comprehensively. We updated and presented recent glacier mass loss from the HMA and regional variability using Ice, Cloud, and Land Elevation Satellite (ICESat-2) data from October 2018 to December 2021. The HMA experienced accelerated mass loss in recent years of -62.62 ± 11.81 Gt a^-1, (-17.11 ± 2.89 m w.e a^-1). All 22 regions of HMA lost mass during the study period, regional mass loss range between -0.28 ± 0.47 m w.e. a^-1 in the Western Kunlun Shan and -1.71 ± 0.22 m w.e. a^-1 in the Hengduan Shan. The highest mass loss occurred at Hengduan Shan, East Tibetan Mountains, and Nyainqentanglha. Glaciers within the altitude range up to 4000 m a.s.l experienced the most negative mass balance. In recent years glacier mass loss increased by more than two-fold compared to previous studies even in the regions where glaciers were previously in balance or less negative mass balance such as Western Kunlun Shan (-0.28 ± 0.47 m w.e. a^-1), Eastern Pamir (-0.47 ± 0.374 m w.e. a^-1), and Karakoram (-0.61 ± 0.40 m w.e. a^-1). The complex regional pattern of variable glacial mass balance and recent mass loss can be attributed to heterogeneous climate change signals and changes in meteorological conditions over the regions.

How to cite: Hassan, J., Colgan, W., and Abbas Khan, S.: Rates of High Mountain Asian Glacier Ice Loss from ICESat-2 Observations, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-16195, https://doi.org/10.5194/egusphere-egu23-16195, 2023.

Glaciers are an important contributor to the freshwater supply in the Central Asian region. Their response to climate change has profound consequences for the land-use applications, and is thus essential to understand. The collapse of the Soviet Union has interrupted the vast majority of the conducted glacier mass balance observations, which began to re-establish in 2010. The existing data gap, limited spatial resolution of glaciological measurements, and the high heterogeneity of the region limits the use of in-situ data. Mass balance models rely on observation-based calibration and validation data, such as transient snowlines (TSLs), a transition between snow and ice-covered surfaces on a glacier at a given point in time. At the end of the ablation season TSL approximates the equilibrium line. From TSL we can calculate the snow-covered area fraction (SCAF), the area on the glacier surface that is snow covered in relation to the total glacier area. The TSL and SCAF can be mapped from satellite imagery due to the distinctive spectral and structural signature of snow over time. Our approach presented in this contribution is based on the MODIS time-series, harnessing the advantage of long and close-to-daily observations records for the period before high-resolution satellites became available. To resolve the issue of MODIS coarse spatial resolution, we retrieved SCAF from multispectral Sentinel-2 and cloud-independent Sentinel-1 SAR imagery using established workflow. The automatic classification and calculation of SCAF is performed using the cloud computing service of the Google Earth Engine, which makes the entire approach easily applicable on a large number of remote glaciers worldwide. We validated the results independently with Landsat data over selected glaciers in Central Asia. From the  SCAF time-series we analysed changes in various parameters indicative for the atmospheric conditions and its changes  (amongst others the length of ablation period, the minimum SCAF, and the seasonal SCAF changes ) as well as their 20-year trends. Our results provide a unique time series of temporally and spatially high-resolved SCAF estimates giving observation-based information on the heterogeneity of the region’s climatic setting as well as its changes on subseasonal scale. 

How to cite: Kim, D., Callegari, M., Ullmann, T., and Barandun, M.: Sub-seasonal snowline dynamics of glaciers in Central Asia from multi-sensor satellite observations, 2000-2021, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-225, https://doi.org/10.5194/egusphere-egu23-225, 2023.

Understanding water resources' long-term availability under varying climatic conditions is vital for water security and adaptation strategies. Due to increased temperature and changes in precipitation patterns, the Himalayan glaciers undergo rapid mass loss. The extent of glaciation has varied considerably since the global Last Glacier Maxima (LGM), ~18-24 ka. Moraine studies provide a better understanding of glaciation’s paleo-extent and timing. This work investigates the extent of glaciation and volume in the Baspa basin during the global LGM. We reconstruct the extent and volume using different remote sensing techniques, including the laminar flow method (HIGHTIM), Topo to Raster tool, Volume – Area scaling, and polygon area method. These methods are applied on 61 glaciers in the basin, considering glacial geomorphology and topographic parameters of the terrain. The glacier boundary and moraines were delineated using Landsat 8 satellite image and high resolution google earth images. The current spatial distribution of ice thickness and volume was estimated using the HIGHTHIM model. Further, the ice thickness was extrapolated to the moraine area using the Topo to Raster tool of ArcGIS, with the help of Cartosat DEM to estimate paleo glacier volume. On smaller glaciers (area<1km²), the current and paleo volume was estimated using the volume-area scaling equation. The glaciated area covers about 161.3 ± 8 km² and the moraine area is calculated as 49.3 ± 0.46 km². The estimate suggests ice volume at LGM is 18.71 km³ and the current volume is 8.51 km³, i.e., about a 52% loss in ice volume from LGM. We propose estimating the mass loss in the current decades, which will help understand the acceleration of mass loss under global warming conditions.

How to cite: Jose, N., Kulkarni, A., Sk, S., and Krishnan, S.: Estimation of paleo-extent and volume of glaciers in the Baspa basin, India, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-629, https://doi.org/10.5194/egusphere-egu23-629, 2023.

This work aims to contribute to progress in the detection of hidden or transient hydrological events. Passive seismic methods offer high temporal resolution and the ability to monitor seismic sources hidden from direct view, making it an ideal candidate to complement other in-situ and satellite methods in these cases. The dynamics of a glacier can be greatly affected by its hydrological system, whether this be through water mediated ice fracturing, or the influence the water has on friction at the ice-bed interface. Effective detection of moving meltwater is therefore of great interest for anticipating future glacier changes and sensitivities.

To effectively infer any hidden process from the observed seismic waveforms, we require a physically rigorous modelling framework. Our work therefore combines hydrodynamic models depicting meltwater flow with seismic wave propagation methods to produce synthetic seismograms. The hydrodynamic model of choice is smoothed particle hydrodynamics (SPH). This is a full, three-dimensional computational fluid dynamics method, meaning we can make minimal assumptions on the exact seismogenic mechanism prior to simulation. SPH has the capacity to capture a broad range of signal-generating processes that may prove to be of interest for modelling meltwater flow, such as fluid-solid impact events, free-surface behaviour (e.g., wave breaks), and some forms of turbulence. Beyond the modelling of complex flow, SPH also allows a simple implementation of arbitrarily shaped solid boundaries and the computation of force of the water on these boundaries; a necessary output for waveform simulation.

We propose a correspondence between different types of meltwater flow and the attributes of the waveforms they produce, as a step towards better detection and characterisation of hidden and short-lived events. Across a diverse set of model geometries and flow types, we anticipate the collection of synthetically generated signals will be useful for categorising archived and real-time signals according to a mechanistic process using unsupervised machine learning methods in ongoing work.

How to cite: Magyar, J., Reading, A. M., Turner, R. J., and Cook, S.: Simulation of water-induced seismic waveforms in glaciers through hydrodynamic modelling, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-898, https://doi.org/10.5194/egusphere-egu23-898, 2023.

We present an analytic framework to model seismic body waves due to supraglacial, englacial or subglacial flows in solid ice based on a smoothed particle hydrodynamic (SPH) simulation. Consisting of two parts, i) hydrodynamic modelling and ii) seismic wave propagation, the flexible framework allows for a pre-existing fluid simulation to be supplied to generate synthetic seismic signals. The field of glacier-related seismology has seen rapid development in recent years, with an expanded availability of passive seismic datasets that contain records of seismic disturbances generated by glacier processes. Some of these processes, such as basal slip and crevasse propagation, have mechanisms with plate tectonic deformation counterparts, however, many glacier signals are generated by moving melt water. This contribution aims to inform the interpretation of such signals.

Our approach tracks the trajectories of fluid particles near the water-ice interface, as recorded in standard simulation outputs, to create a catalogue describing the energetics of each collision. We illustrate the capability of this framework using four end-member cases of water flow along surface channels with different geometries. Seismic signals are simulated at a variety of locations around the channel based on the impulse of the database of simulated collisions. We consider the change in character of the seismic waveforms by modelling frequency-dependent attenuation and weak dispersion in the glacial ice, in addition to the standard geometric spreading. The acceleration time series produced in this work are invariant to the temporal and spatial resolution of the hydrodynamic simulation, provided more than some minimum resolution is used. These time series may be converted to velocity or displacement for comparison with observed seismic signals.

Investigating the seismic waves generated for our four channel geometries, we find distinct waveform envelope shapes with different first and later amplitude peaks matching initial and subsequent collisions of the melt water surge with the supraglacial channel walls. The change in waveform character with distance is also captured such that the character attributes due to the process and the those due to the propagation effects may be understood. The flexibility inherent in the model framework will allow for the generation of the seismic signals from simulations of a variety of different water flow geometries including simple 3D channels into and through a glacier. We make the code available as an open source resource for the polar geophysics community with the aim of adding to the toolbox of available approaches to inform the potential future seismic monitoring of melt water movement and related glacier processes.

How to cite: Turner, R. J., Magyar, J., Cook, S., and Reading, A. M.: Analytical framework to model seismic signals from fluid particle collisions in hydrodynamic simulations of glacier melt water, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-913, https://doi.org/10.5194/egusphere-egu23-913, 2023.

Snow and ice albedo play a crucial role in the mass losses from the NW Antarctic Peninsula and the South Shetlands Islands since absorption of solar radiation is the largest energy source for surface melt in the cryosphere. Snow albedo exhibits a large variation at different time scales (hours, days, seasonal, long-term trend). It has been established that the thickness of the snow/ice layer that affects the albedo varies from 20 cm to 50 cm.   The surface geomorphology of this layer exerts a strong control in the snow albedo evolution over a day because they affect the amount of solar energy received, and the exposure to prevailing winds. In order to discover patterns of variation between snow albedo and surface geomorphology over Hurd Peninsula glaciers at metric scales, we used four DEMs (1957,2000, 2013 and 2019) of 1 m of spatial resolution. In the four DEMs topographical variables (altitude, slope, plan and profile curvature, aspect) and indexes (diurnal differential heating, wind exposition, roughness index) were calculated using QGis_Gdal_SAGA tools.  Because of the high spatiotemporal variability of the snow cover, the albedo data obtained from fixed stations provide a partial picture of the actual field behaviour.  In order to obtain spatially distributed albedo measurements over Hurd Peninsula glaciers, we designed a portable albedometer. The device consists of two pyranometers, one facing the sky and another facing the ground at 1,20m above the ground, working together with a GNSS. The dataloggers of each pyranometer were set up to take a measurement every 5 seconds. Using this equipment, in January of 2018 and 2019 we surveyed the glaciers of Hurd Peninsula along the same tracks but under different weather conditions (fog, clear sky, clouds and clearings). The population of albedo data obtained with this method was about 800 measured points per track per day.  Using QGis we obtained the values of topographical variables and indexes for all the albedo points. Lineal correlations between albedo and topographical variables and indexes were explored. The R² was especially high in the tracks performed during open sky days. There are not significant correlations between DTMs variables and albedo data in tracks performed in foggy days. Moreover, we built a Linear Regression Model (forward stepwise) of open_sky day’s albedo with Adjusted R²= 0.86 and Std. Error of estimate: 0.00428 with diurnal differential heating_2019, altitudes (1957, 2000, 2013), slope_2019 and convexity_2019 as predictive variables. To estimate the surface albedo all across Hurd Peninsula extending linear models using QGis. Also, we calculated the DTM of 1957-2019 altitude changes in meters (Minimum=-32.694, Maximum=53.188, Mean=7.959, StdDev=10.526), which shows strong correlation with albedo of all open sky tracks.  We interpreted these results in relation to the density, structure and state of metamorphism of the snow cover that could represent the layer of snow that affects the albedo in Hurd Peninsula glaciers. The preliminary results seem to indicate that the surface melting intra-annual variability of the Hurd Peninsula glaciers, registered in the glacier surfaces geomorphology, exerts a strong influence on the current albedo.

How to cite: Fernandez Menendez, S. C., Fernandez Calleja, J., Muñiz Sanchez, R., Otero Garcia, J., and Navarro Valero, F.: On-site albedo data and their relationship to the long-term evolution of surface morphology in glaciers of Hurd Peninsula, South Shetland Islands, Antarctica, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-3941, https://doi.org/10.5194/egusphere-egu23-3941, 2023.

Surface melting of alpine glaciers is spatially and temporally heterogeneous and is strongly influenced by climatic and non-climatic variables. Especially supra-glacial debris causes significant uncertainty on the ice melting rate with its physical property and thickness. A thick debris layer decrease ice melting rate, whereas a thin layer increase it with its low albedo. Therefore, a scientific method for spatial quantification of debris influences on the melting rate should be established to assess future projection of glacier shrinkage corresponding to the climate change.

Estimating  thermal resistance (TR), which quantifies how hard the ground heat flux (G) travels to ice-debris interface, with remote-sensing data has been attempted in several studies. However, uncertainties caused by the linear temperature gradient have not been resolved, as well as non-negligible underestimation of TR values. Therefore, this study aims to assess TR calculations on multi-spatial and multi-temporal conditions in detail, and discusses the current state of TR estimation and the potential for further improvement.

A study site is defined in Satopanth glacier [30.77°N; 79.40°E], a debris-covered glacier located in Garhwal region, India. Sampling domains of 12 circles with 100-m radius are put with 1-km intervals through the flow line of supra-glacial debris. In addition, four circles of the same size were placed outside the glacier in the lower reaches.

To calculate TR, first, broadband albedo and surface temperature (Ts) are calculated from all available Landsat-8 data in an orbit path [Path 145; Row 39] acquired from 2013 to the present (N= 81). These have a revisit cycle of 16 days. Cloud-covered pixels and frozen pixels (Ts < 0°C) are removed. Second, downward shortwave/longwave radiations and sensible/latent heat flux are collected from a 1-hour resolution product of ERA5. Combining these inputs, considering surface energy budget, G is calculated as a residue of energy flux, and then TR is calculated as Ts (°C) divided by G.

Our result shows a positive correlation that higher Ts leads higher values of G and TR. This trend have no significant difference between debris-covered surface and off-glacier terrains. It suggests that, in most sample areas with relatively thick debris, G does not reach the ice-debris interface. High-gradient inclinations of G versus Ts increase is identified in a lower part of Ts range (0-5°C). It may be caused by heat absorption because of ice mass under relatively thin debris layer, but such inclination is not reflected in TR’s gradient. In the lower Ts surfaces (0-5°C), slightly lower R is estimated for thinner-debris domains. For the thinner debris layer TR might reflects debris thickness. These features in other multiple glaciers will be shown and compared in the presentation.

Our result suggested that maximum debris thickness of G transfer (DTmax) may be defined. Smaller than the DTmax, G transfer may be estimated, whereas larger than the DTmax, G might be zero. In such domain, spatial distributions of ice cliff and supra-glacial pond are more dominant for melting projection. Further assessments might derive a perspective of multiple models for TR estimation according to debris thickness.

How to cite: Nagai, H., Nuimura, T., Takigawa, M., Patel, L., Laha, S., Pratap, B., Konya, K., Sharma, P., Fujita, K., Sato, Y., and Sakai, A.: Toward improvement of satellite-derived thermal resistance for supra-glacial debris, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-4768, https://doi.org/10.5194/egusphere-egu23-4768, 2023.

In situ monitoring of glacier mass balance – through a network of reference sites – is essential to improve process understanding and detect changes, as well as provide calibration and validation for local and large-scale modeling and remote sensing studies. Still, the interpretation and representativeness of measured mass balances can be affected by unsteady glacier flow dynamics, from minor pulsations to extreme surges. Such events can alter mass balance gradients and apparent trends; englacial water storage associated with glacier surges can lead to biases in high‑resolution geodetic assessments, and in situ measurements can be disrupted by surface changes during rapid ice movement. At the same time, mass balance and its perturbations can exert multiple influences on glacier dynamics, from an indirect control on surge frequency to the direct triggering of instabilities and propagating waves. So far, a very small number of glaciers worldwide has seen combined long-term observations of mass balance and unsteady ice flow, limiting understanding of their interaction.

Here, we investigate the flow dynamics of the Abramov glacier (Pamir-Alay, Kyrgyzstan), whose measured mass balance series is one of the longest in Central Asia. We use multi-sensor optical remote sensing, including the recently released SPOT World Heritage archive as well as ASTER, IRS-1C/D and RapidEye data, to augment the Landsat record over the past 25 years. Through a dense series of digital elevation models and orthoimages, we quantify a front advance over 2000‑2005 at sub‑seasonal resolution. While the event was not observed in situ, negative mass balances in the preceding decades support an interpretation in terms of unstable ice dynamics. Moreover, asynchronous front advances on several neighboring glaciers reveal a high prevalence of unsteady ice flow in the region. We compare the Abramov advance to a previous surge of 1972/73, which was well documented with extensive in situ measurements but was virtually unknown outside Soviet glaciology. Both events do not fit well within the traditional surge models of hydrological or thermal switch, similar to previous observations in the Pamir-Karakoram.

At present, the west branch of the Abramov glacier is undergoing a slow, regular and widespread inter‑annual speed-up. In the light of continued surface thinning and front retreat, the current evolution likely represents the build-up phase to the next episode of dynamic instability.

How to cite: Mattea, E., Machguth, H., Berthier, E., and Hoelzle, M.: Minor pulsations of Abramov glacier (Kyrgyzstan) observed with multi-sensor optical remote sensing, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6296, https://doi.org/10.5194/egusphere-egu23-6296, 2023.

The free and open data policy of the Landsat legacy and the Sentinel-2 satellite constellation have enhanced our knowledge considerably. This enrichment is due to an increase in spatial spread; from glacier change at targeted glaciers, to a coverage at regional or global scale. Procedures for snow cover mapping and glacier outline product generation from such multi-spectral data are well matured. Though more potential is present for glacier specific information in such data sources.

Recently several studies presented the capabilities of elevation change extraction via shadow cast detection. Though these are at the proof of concept stage, as many steps are based on heuristics and manual adjustments/measurements.

Here we present a fully automatic pipeline which is based on an open source library, which we developed. The emphasis of this code base, which we named dhdt, is focused on large scale processing. Hence, all steps are fully automatic and structured so distributed processing is possible. It generates spatial temporal elevation changes, over small mountain glaciers. We hope, this will advance the glaciological community forward, giving a new instrument in its toolbox.

How to cite: Altena, B. and Nattino, F.: dhdt: a Python library to transform shifting shadows to glacier elevation change, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-8711, https://doi.org/10.5194/egusphere-egu23-8711, 2023.

Quantifying glacier elevation and volume changes is critical to understanding the response of glaciers to climate change and related impacts, such as regional runoff and global sea-level rise. Spaceborne remote sensing techniques enable the quantification of spatially distributed glacier elevation changes at regional and global scales using multi-temporal digital elevation models (DEMs). A growing number of spaceborne studies exist to assess glacier elevation changes but they show widespread differences often beyond the error bars. Here, we present the results of a community-based inter-comparison experiment using spaceborne optical (ASTER) and radar (TanDEM-X) sensors to assess elevation changes for selected individual glaciers and regional glacier samples. Using a predefined set of DEMs, participating groups provided their own estimates using various processing strategies.  

For the selected individual glaciers, the results were validated using airborne data. The validation shows that the median of the spaceborne ensemble is biased by a few decimetres per year with a standard deviation of about half a meter per year. An interesting finding is that no sensor and no processing strategy perform significantly better for all experiment sites. At the regional scale, we find that the co-registration of DEMs is the most relevant processing step for an accurate assessment of elevation change. Other corrections such as gap filling, filtering, and radar penetration have less impact in general but can be essential for individual cases. Temporal corrections (i.e. seasonal and annual) can have a great impact; however, they are not yet well resolved by the remote sensing community.

Our study confirms that the currently available spaceborne geodetic assessments result in relatively widespread glacier elevation changes. Therefore, we recommend an ensemble approach of observations from multiple observational sources. Furthermore, there is a need to establish best practices for related uncertainty estimates.

How to cite: Piermattei, L., Brun, F., Sommer, C., Braun, M. H., and Zemp, M.: Observing glacier elevation changes from spaceborne optical and radar sensors, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9215, https://doi.org/10.5194/egusphere-egu23-9215, 2023.

Time-lapse photogrammetry in conjunction with high-resolution digital elevation models is used to quantify the surficial velocity field and the ablation of the Pasterze, a rapidly retreating alpine valley glacier in Austria. Three automatic time-lapse cameras were installed along the orographic left valley flank, c. 150 m above the glacier’s surface, to monitor retreat and ablation from July to September 2020. Although time-lapse photogrammetry offers spatial and temporal high-resolution data, the various processing steps to calculate the glacier’s velocity field are challenging. Digital image correlation of the time-lapse series photos is achieved using a published Python library (How et al. (2020) PyTrx: a Python-based monoscopic terrestrial photogrammetry toolset for glaciology. Frontiers in Earth Science 8:21, doi:10.3389/feart.2020.00021), which was adapted for the Pasterze data set. Typical time intervals for the digital image correlation were two to ten days. Factors that hamper computing flow velocities from time-lapse series photos alone are low contrast of the glacier’s surface and ablation rates exceeding the horizontal flow velocity. The latter problem is solved with the aid of two high-resolution digital elevation models (DEMs), which were calculated using 790 drone images from flight missions on July 13 and September 15 2020. The photogrammetric (Structure from Motion) software Agisoft Metashape (v. 1.8.4) is used to calculate the two DEMs from dense point clouds with a resolution of 5 cm/pixel. Linear interpolation of the glacier’s elevation between the July and September DEM was used to provide an approximate surface for a given date within the monitoring period. With this additional processing step, we can project tracked pixels from the digital image correlation process on an adapted DEM, which reflects the absolute height on a certain date best. Our workflow illustrates that time-lapse series photos taken obliquely to the glacier’s surface can be used to compute the surficial velocity field with the aid of digital elevation models that yield the glacier’s surface at the beginning and at the end of the monitoring period. In fact, the glacier’s velocity field computed in that fashion is consistent with direct measurements from a sparse network of stakes and mapped structures.

How to cite: Mayrhofer, F., Grasemann, B., and Schöpfer, M.: Quantification of glacier flow velocities using time-lapse photogrammetry in conjunction with high-resolution digital elevation models, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-11141, https://doi.org/10.5194/egusphere-egu23-11141, 2023.

In Peru, there are 20 glacial mountain ranges that in almost 60 years have lost 54% of their glacial coverage. The accelerated glacial recession due to climate change raises the need to know the levels of retraction of the glacier surface efficiently and reliably. During the preparation of the last inventory by the National Institute for Research in Glaciers and Mountain Ecosystems (INAIGEM), it was identified that carrying out the inventory faces challenges, such as differentiating between temporary snow and glacier ice, especially in areas with complex meteorological characteristics. like the Cordillera Blanca (Peru). Due to the Peruvian Andes’ geographical and climatic complexity, satellite images usually show cloudiness and temporary snow, even in the dry season (April to November). In this way, the objective is to obtain an ideal image with annual glacier coverage with minimal snow reflecting the current glacier surface. For this, the script "Normalized Differentiated Snow Index - minimum NDSI" was developed, which analyzed the Sentinel-2A image catalog of the year 2020 and delimited the glaciers of the Huascarán and Huandoy systems in the Cordillera Blanca.

The proposed methodology aims to evaluate the techniques for generating glacier cover, which allows the proposed objective to get the minimum glacier area. Three glacier cover generation techniques were evaluated: mosaic, medium, and minimum. For the mosaic and average reductions, the algorithm applied a cloud filter to the Sentinel-2A image set and calculated the NDSI for the month that was lowest in the historical average from Landsat 5, 7, and 8 images (1990-2020), applying a threshold of 0.4 and exporting the results with mean and mosaic reduction, respectively. While the minimum NDSI was calculated annually (2020) from Sentinel-2A images, with a cloud filter to which the reduction by minimums is applied, within the same area of interest, applying the threshold of 0.4 and exporting the results in raster format. Finally, the three results were evaluated in terms of the percentage of overestimation concerning glacier coverage in 2016.

The results reveal that the Huascarán and Huandoy glacier systems present a lower NDSI value during August and October, with standard deviations of 0.12 and 0.14, respectively. The glacier cover generated by the minimum NDSI was compared in the percentage of overestimation (m2) concerning 2016 with the average and mosaic NDSI, finding as a result that the minimum glacier cover, for the year 2020, evidences a lower percentage of temporal snow in the Huascarán system between 157.12% and 76.84% less than the filtering methods: average and mosaic. Likewise, in the Huandoy system, there is evidence of a lower percentage of temporary snow between 205.91% and 191.65% less than the average and mosaic methods. Finally, it is necessary to indicate that the developed algorithm has improved the obtaining of glacier coverage from the inventory developed by INAIGEM and reduces the overestimation of glacier coverage due to temporary snow.

How to cite: Fernandez, J. D. D., Fajardo, B., Curo, Y., Mejia, M., Celmi, G., Robles, D., and Castañeda, A.: Proposed algorithm for the identification of glacier cover from Sentinel-2A images, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-11201, https://doi.org/10.5194/egusphere-egu23-11201, 2023.

Glaciers are crucial sources of freshwater in particular for the arid lowlands surrounding High Mountain Asia. In order to better constrain glacio-hydrological models, annual, or even better, seasonal information about glacier mass changes is highly beneficial. In this study, we test the suitability of very high-resolution Pleiades DEMs to measure glacier-wide mass balance at annual and seasonal scales in two regions of High Mountain Asia (Muztagh Ata in Eastern Pamir and parts of Western Nyainqêntanglha, South-central Tibetan Plateau), where recent estimates have shown contrasting glacier behaviour. We find that the average annual mass balance in Muztagh Ata between 2020 and 2022 was -0.11 ±0.21 m w.e. a-1, suggesting the continuation of a recent phase of slight mass loss following a prolonged period of balanced mass budgets previously observed. The mean annual mass balance in Western Nyainqêntanglha for the same period was highly negative (-0.60 ±0.15 m w.e. a-1 on average), suggesting increased mass loss rates. The 2022 winter (+0.21 ±0.24 m w.e.) and summer (-0.31 ±0.15 m w.e.) mass budgets in Muztag Ata and Western Nyainqêntanglha (-0.04 ±0.27 m w.e. [winter]; -0.66 ±0.07 m w.e. [summer]) suggest winter and summer accumulation-type regimes, respectively. We support our findings by implementing a Sentinel-1–based Glacier Index to identify the firn and wet snow areas on glaciers and characterize accumulation type and demonstrate the potential of very high-resolution Pleiades data to monitor mass balance at short time scales and to improve our understanding of glacier accumulation regimes across High Mountain Asia.

How to cite: Bolch, T., Falaschi, D., Bhattacharya, A., Huang, L., and King, O.: Annual to seasonal glacier mass balance in High Mountain Asia derived from Pléiades stereo images: examples from the Pamir and the Tibetan Plateau, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12612, https://doi.org/10.5194/egusphere-egu23-12612, 2023.

Glaciers in the tropical Andes of Peru and Bolivia are important but rapidly declining water resources. Precise knowledge of their extent is thus mandatory for calculation of their volume, area changes and mass balance. Due to wrongly mapped seasonal snow, the glacier outlines currently available from the widely used RGIv6 are often too large, resulting in errors for change assessment and volume estimation. Apart from snow cover, also frequent cloud cover and shadows cast by the steep terrain make glacier mapping in this region challenging.

For this study, we have mapped all glaciers in Peru and Bolivia using cloud-free Landsat TM scenes from 1998 and Sentinel-2 scenes from 2020. In both years seasonal snow off glaciers was largely absent. Glacier extents were mapped with a standard band ratio (red/SWIR) and a scene specific threshold value. Wrongly classified lakes and missing debris cover were manually corrected, the latter also by using the very high-resolution satellite images available in the ESRI Basemap. The Copernicus DEM GLO-30 was used to derive new drainage divides and topographic information for each glacier.

In total, we mapped 3586 glaciers larger than 0.01 km² covering an area of 1747 km² in 1998. This is 419 km² or 20% less than the 2166 km² in RGIv6. As glacier outlines in RGIv6 are from 2000 to 2009 and most area change studies found continuous and strong area decrease over this period, a ‘back-calculation’ of all glacier areas to the year 1998 with an annual shrinkage rate of 1% gives a mean area overestimation of 27% for 1998. This value varies regionally and is smaller in the Cordillera Blanca and much larger (>50%) in other regions. The related modelled glacier volumes for these regions are thus also overestimated. From 1998 to 2020 glaciers have lost 23% of their area (-1% per year).

How to cite: Paul, F. and Rastner, P.: Glacier extents in Peru and Bolivia are overestimated in RGIv6 by 27%, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12724, https://doi.org/10.5194/egusphere-egu23-12724, 2023.

Glaciers are one of the most important indicators of climate change. Monitoring their evolution is, therefore, crucial for safeguarding the Earth’s ecosystem. In this study, exploiting photogrammetric optical satellite processing techniques, we used image pairs from Ikonos-2 and Plèiades-HR satellites to generate Digital Surface Models (DSMs) of Forni Glacier (Ortles–Cevedale group, Italy) and compute its morphological variations between 2009 and 2016. In addition, we used DSMs generated from Unmanned Aerial Vehicle (UAV) acquisitions collected during summer campaigns from 2014 to 2021 for comparison with very high-resolution DSMs on the terminal portion of the glacier including its tongue, which is also the area more affected by morphological changes. To evaluate the glacier height loss, DSMs co-registration was applied to remove DSM biases due to inconsistencies in the georeferencing of the different satellite image pairs. For this purpose, we used the 2016 UAV DSM as a reference, and we co-registered all the optical DSMs to the 2016 UAV DSM using the Nuth and Kaab algorithm. The DSM height differences after co-registration highlighted a final accuracy of one meter. Since optical satellite data have the advantage of providing information on very large areas, we analysed glacier change not only on small areas of Forni Glacier tongue but also on larger regions including parts of the entire glacial apparatus to depict the evolution of the glacier at different altitudes. Results from optical DSMs were consistent with the average annual variation of the glacier suggested by UAV DSMs analysis, confirming an average 5.00 m/y loss on the Forni tongue during 2014-2016.  Furthermore, based on both UAV and optical data, melting trends have highlighted how climate change is causing an acceleration in the melting process, with values averaging 3.3 m/y in the period 2009-2013, 3.8 m/y in 2009-2016 and 4.7 m/y in 2009-2021. With reference to the optical data only, we observed that the intensity of melting varied at different altitudes, with 10 m of maximum variation above 3000 m, and 30 m between 2600-3000 m during 2009-2016. Our results suggested that despite the limitations related to weather conditions (e.g. cloud coverage) and time revisit, high-resolution optical satellite imagery can certainly be used to estimate relevant morphological variations of glaciers in the order of meter/years, offering the opportunity of monitoring large-scale areas.

How to cite: Ranaldi, L., Belloni, V., Fugazza, D., and Crespi, M.: Glacier change monitoring using optical satellite imagery: the case of Forni Glacier, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-13794, https://doi.org/10.5194/egusphere-egu23-13794, 2023.

Glacier displacement can in principle be measured at the large-scale by cross-correlation of satellite images. At weekly to monthly scales, the expected displacement is often of the same order as the noise for the commonly used satellite images, complicating the retrieval of accurate glacier velocity. Assessments of velocity changes on short time scales and over complex areas such as mountain ranges are therefore still lacking, but are essential to better understand how glacier dynamics are driven by internal and external factors. In this study, we take advantage of the wide availability and redundancy of satellite imagery over the Western Pamir to retrieve glacier velocity changes over 10 days for 7 years for a wide range of glacier geometry and dynamics. Our results reveal strong seasonal trends. In spring/summer, we observe velocity increases of up to 300% compared to a slow winter period. These accelerations clearly migrate upglacier throughout the melt-season, which we link to changes in subglacial hydrology efficiency. In autumn, we observe glacier accelerations that have rarely been observed before. These episodes are primarily confined to the upper ablation zone with a clear downglacier migration. We suggest that they result from glacier instabilities caused by sudden subglacial pressurization in response to (1) supraglacial pond drainage and/or (2) gradual closure of the hydrological system. Our 10-day resolved measurements allow us to characterize the short-term response of glacier to changing meteorological and climatic conditions.

How to cite: Nanni, U., Scherler, D., Ayoub, F., Millan, R., Herman, F., and Avouac, J.-P.: Climatic control on seasonal variations of moutain glacier surface velocity, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-13863, https://doi.org/10.5194/egusphere-egu23-13863, 2023.

With temperatures in the Arctic rising rapidly at a rate of 3-4 times the global mean, monitoring the state of the Greenland ice sheet has never been more relevant.

In-situ observations from the Arctic, in particular from the Greenland ice sheet, are scarce due to the cost and difficulty of maintaining instrumentation in the harsh and remote environment. Yet, the fate of the ice sheet concerns 100s of millions of people living in coastal zones worldwide. To gain understanding of the ice sheet processes leading to sea level rise and increase our ability to capture those in climate models, there is an urgent need to collect in-situ observations from the ice sheet surface. Similarly, ground-truthing observations are necessary for validation and calibration of satellite-derived estimates of ice sheet change.

The Geological Survey of Denmark and Greenland (GEUS), along with partner institutions Asiaq and DTU Space, currently operates a combined network of 40 automatic weather stations (AWS) on ice in Greenland, mainly through the Programme for Monitoring of the Greenland Ice Sheet (PROMICE) and the Greenland Climate Network (GC-Net).

GEUS has implemented a new pipeline providing near-real-time hourly weather observations from the PROMICE and GC-Net stations to its users and the World Meteorological Organization (WMO) for use in numerical weather prediction. The satellite-transmitted AWS data is processed and submitted to the WMO via the Danish Meteorological Institute with a latency of 7 minutes after observations are recorded.

Here, we present the recent advances in our AWS instrumentation, data processing and database solution to invite discussion on how we can best meet the community needs for in-situ observations from the ice sheet.

How to cite: Ahlstrøm, A. P., Fausto, R. S., Box, J. E., Karlsson, N. B., How, P. R., Wright, P. J., Vandecrux, B., Rutishauser, A., Mankoff, K. D., Colgan, W. T., Citterio, M., Messerli, A., Solgaard, A. M., Larsen, S. H., Korsgaard, N. J., Kjeldsen, K. K., Nielsen, R. B., Houtz, D., and Andersen, S. B. and the GEUS GlacioLab Team: Recent advances in monitoring surface mass balance of the Greenland ice sheet, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-13870, https://doi.org/10.5194/egusphere-egu23-13870, 2023.

Summer 2022, in less than a week, two glaciers collapsed. First in Italy on July 3 (Marmolada) then in Kyrgyzstan on July 9 (Juuku pass). The collapse of the Marmolada glacier caused eleven fatalities. In both cases, we immediately requested the tasking of Pléiades satellites to estimate the collapse volumes by photogrammetry. In both cases, the images were acquired three days after the event, and less than one day later we had the first estimates. We found that the Marmolada glacier lost 65,000 ± 10,000 m3. The Jukuu pass glacier lost a volume almost twenty times greater (1,145,000 m3), which remains however much lower than the volumes involved during the collapse of the Aru glaciers on the Tibetan plateau in 2016 (68 and 83 million m3) or the Kolka Glacier in 2002 (130 million m3). From the Marmolada elevation model and orthoimages we could also estimate that the rupture was approximately 80 m wide and 25 m deep.  In the case of the Juuku glacier, the tongue of the glacier collapsed entirely over a width of almost 300 m and the maximum elevation drop reached 50 m.

How to cite: Gascoin, S. and Berthier, E.: Remote sensing analysis of Marmolada (Italy) and Juuku pass (Kyrgyzstan) glacier collapses, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-14011, https://doi.org/10.5194/egusphere-egu23-14011, 2023.

In situ glaciological monitoring of A. P. Olsen Ice Cap in NE Greenland has been ongoing since 2008. The monitoring effort is part of the Greenland Ecosystem Monitoring (GEM) programme at Zackenberg research station. The monitoring includes: Three automatic weather and ablation stations along a transect, manual stake observations upon every visit in April, ground based radar profiles of last year snow accumulation and geodetic mass balance from satellite based observations. Here we present an overview of all glaciological monitoring data, and use the dataset to estimate the mass balance of A. P. Olsen over the past 13 years.

How to cite: Hillerup Larsen, S., Rutishauser, A., Binder, D., Korsgaard, N., Hynek, B., and Citterio, M.: Glaciological monitoring at A. P. Olsen Ice Cap in NE Greenland, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-15647, https://doi.org/10.5194/egusphere-egu23-15647, 2023.

Digital elevation models (DEMs) from the spaceborne interferometric radar mission TanDEM-X hold a large potential for glacier elevation change assessments and monitoring. However, a bias is potentially introduced through the penetration of the X-band signal into snow and firn that can be substantial. The magnitude of this bias has been analysed in some glaciarized regions of the world; still, the knowledge about X-band penetration of TanDEM-X in the European Alps is limited.

In this study, we investigated the unique situation of almost synchronous acquisition of TanDEM-X and Pléiades DEMs over the Grosser Aletschgletscher, complemented with in-situ observations (ground penetrating radar, snow cores, snow pits), all within a four-day period in late winter 2021. The comparison of the TanDEM-X and Pléiades DEM revealed an elevation bias due to radar penetration of up to 8 m above 3400 m. Further, the concurrent in-situ measurements reveal that the signal is not obstructed by the last summer horizon but reaches into perennial firn.

Our study improves our understanding about the magnitude of X-band penetration of TanDEM-X in the Alps and the underlying process with a relevance for glaciology, snow science, remote sensing and the wider geoscience community.

How to cite: Bannwart, J., Piermattei, L., Dussaillant, I., Krieger, L., Floricioiu, D., Berthier, E., Roeoesli, C., Machguth, H., and Zemp, M.: Elevation bias due to penetration of spaceborne radar signal on Grosser Aletschgletscher, Switzerland, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-15661, https://doi.org/10.5194/egusphere-egu23-15661, 2023.

A national glacier outline inventory for several different epochs since the end of the Little Ice Age (LIA) in Iceland has been created with input from several research groups and institutions, and has been submitted to the GLIMS (Global Land Ice Measurements from Space, nsidc.org/glims) database, where it is openly available. The glacier outlines have been revised and updated for consistency and the most representative outline chosen. The maximum glacier extent during the LIA was not reached simultaneously in Iceland, but many glaciers started retreating from their outermost LIA moraines around 1890. The total area of glaciers in Iceland in 2021 was ~10,300 km². The total glacier area has decreased by ~2300 km² since the end of the 19th century and by ~830 km² since ca. 2000. During the first two decades of the 21st century, the decrease rate has on average been ~40 km² a^–1. In this period, some tens of small glaciers have disappeared entirely. Temporal glacier inventories are important for climate change studies, for calibration of glacier models, for studies of glacier surges and glacier dynamics, and they are essential for better understanding of the state of glaciers. Although surges, volcanic eruptions and jökulhlaups influence the position of some glacier termini, glacier variations have been rather synchronous in Iceland, largely following climatic variations since the end of the 19th century.  The glacier outlines are also available on a new glacier web portal (www.islenskirjoklar.is), together with measurements of frontal positions and mass balance and numerous photographs of glaciers at different times. The photographic glacier archive will be updated through systematic photographic surveys, including rephotography of historical photos, based on a collaboration of the Iceland Glaciological Society and Extreme Ice Survey. This website, which is intended for scientists, students and lay people alike, is a joint effort of institutions involved in glacier research in Iceland and the Iceland Glaciological Society. It will serve as a powerful tool for outreach on glacier and climate change in Iceland. The glacier inventory is planned to be updated every other year in the future as part of regular monitoring of glacier changes in Iceland. Furthermore, the larger ice caps will be divided into ice-flow basins along the ice divides of individual outlet glaciers determined from ice-surface DEMs, which will allow for more detailed analysis of area variations with time.

How to cite: Hannesdóttir, H., Sigurðsson, O., Guðmundsson, S., Belart, J. M. C., Þrastarson, R. H., Pálsson, F., Magnússon, E., and Jóhannesson, T.: Glacier changes in Iceland since the Little Ice Age maximum - glacier outlines, terminus measurements and photographic evidence, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-15715, https://doi.org/10.5194/egusphere-egu23-15715, 2023.

The geodetic method has become a popular tool to measure glacier elevation changes over large glacierized regions with high accuracy for multiannual/decadal time periods. In contrast, the glaciological method provides annually to seasonally resolved information on glacier mass balance, but only for a small sample of the world’s glaciers (less than 1%). Various methods have been proposed to bridge the gap regarding the spatio-temporal coverage of glacier change observations and provide annually resolved glacier mass balances using the geodetic sample as calibration. Thanks to a new globally near-complete (96% of the world’s glaciers) dataset of geodetic mass balance between 2000 and 2020, a global-scale assessment of annual mass changes at glacier-specific level has now become feasible. Inspired by previous methodological frameworks, we developed a new approach to combine the glacier outlines from the globally complete Randolph Glacier Inventory with the mass balance and elevation change observations from the Fluctuation of Glaciers database of the World Glacier Monitoring Service (WGMS). Our results provide a global assessment of annual glacier mass change and related uncertainties for every individual glacier since 1976. The glacier-specific time series can then be integrated into an annually-resolved gridded global glacier change product at any user-requested spatial resolution, useful for comparison with, for example, gravity-based products, calibration or validation of glacier mass balance models operating at a global scale and to improve assessments of glacier contribution to regional hydrology and global sea-level rise. These developments additionally open a new door of opportunity to keep on pushing the frontiers of glacier change observations towards future assessments of global glacier mass changes at increased temporal resolutions.

How to cite: Dussaillant, I., Hugonnet, R., Huss, M., Berthier, E., and Zemp, M.: An annual mass balance estimate for each of the world’s glaciers based on observations., EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-16174, https://doi.org/10.5194/egusphere-egu23-16174, 2023.

We present the Programme for Monitoring of the Greenland Ice Sheet (PROMICE) ice velocity product, which is a time series of ice velocity mosaics derived using offset tracking on Sentinel-1 SAR data.  The time series starting in January 2016 is continuously updated with a new mosaic every 12 days and is posted at 500 m grid resolution. Within PROMICE, the ice velocity product is used directly as input to estimate the solid ice discharge from the Greenland Ice Sheet as well as to study ice dynamic processes on seasonal and multi-annual time scales. Recently, we have made changes to the processing chain due to spurious cases of slow down detected in a few glaciers in Southeast Greenland. In this contribution, we discuss how this was resolved as well as other recent improvements to the product.

How to cite: Solgaard, A., Kusk, A., Hillerup Larsen, S., Mankoff, K., and Fausto, R.: Monitoring the flow of the Greenland Ice Sheet: The PROMICE ice velocity product and recent updates, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-16473, https://doi.org/10.5194/egusphere-egu23-16473, 2023.

Livingston Island is one of the eleven islands of the South Shetland Archipelago which is separated from the Antarctic Peninsula by Bransfield Strait and from South America by the Drake Passage. The South Shetland Islands, where the Bulgarian Antarctic Base is located, is characterized by complex geodynamics, including: subduction zones, zones of splitting of the old crust and arising of a new crust, numerous volcanoes. The most of the territory of the Island is coated by permanent snow cover and glaciers. In recent decades, a number of scientific institutions have been working on projects related to the study of various aspects of the seismic regime and the structure of the Еarth's crust in the region of the South Shetland Islands and Antarctica in general, as well as the state and dynamics of the ice sheet. However, this region of the Earth remains still unexplored. Harsh climatic conditions are a serious obstacle for conducting long-term research. Bulgarian scientists from Sofia University "Sv. Kliment Ohridski" and the National Institute of Geophysics, Geodesy and Geography of the Bulgarian Academy of Sciences have been studying the seismicity of the Livingston Island region and the behavior of the glaciers near the Bulgarian Antarctic Base within the framework of three scientific projects since 2014. The projects are funded by the Bulgarian Scientific Fund and National Center for Polar Studies.

One of the aim of the projects is to study the activity of the glaciers during different seasons by combining seismic and GNSS measurements. Seismic registration was carried out by the seismic station LIVV installed previously as temporary station and later developed as a year-round one. GNSS measurements were carried out at certain points on the surface of the glacier Balkan. The resulting estimates of the speed of glacier movement at these points were interpreted in conjunction with the seismic data, thus making an initial attempt to determine the nature of the recorded icequakes. The relationship between seismic activity in the glacier and the change in the temperature of the environment during the astral summer and the astral winter was studied.

The study of the seismicity and ground structure of Livingston Island and the surrounding area have been carried out using the data recorded by the Broadband seismic equipment of the LIVV station, applying a software code developed for this purpose.

Investigating seasonal variations in seismic noise is another goal of the projects. The recorded seismic noise provided information on the condition and behavior of the seismic equipment throughout the recording period, as well as on the sources of seismic noise and their influence on the recording capabilities of the station.

How to cite: Dimitrova, L., Georgieva, G., and Gourev, V.: Exploring seismicity and ice cover of Livingston Island - research projects, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-16573, https://doi.org/10.5194/egusphere-egu23-16573, 2023.

The Arctic is warming more rapidly than the rest of the world. This warming has had an especially profound impact on Greenland’s ice cover. Only 4% of Greenland’s ice cover are small peripheral glaciers that are distinct from the ice sheet proper. Despite comprising this relatively small area, these small peripheral gIaciers are responsible for 11% of the ice loss associated with Greenland’s recent sea-level rise contribution. Using the satellite laser platforms ICESat and ICESat-2, we estimate that ice loss from these Greenland glaciers increased from 27±6 Gt/yr (2003–2009) to 42±6 Gt/yr (2018–2021). We find that the largest acceleration in ice loss is in North Greenland, where we observe ice loss to increase by a factor of four between 2003 and 2021. In some areas, it appears that recent increases in snowfall at high altitudes have partially counteracted recent increases in melt at low altitudes. While many recent Greenland ice loss assessments have focused on only the ice sheet, the recent sharp increase in ice loss from small peripheral glaciers highlights the importance of accurately monitoring Greenland’s small peripheral glaciers. These small peripheral glaciers appear poised to play an outsized role in Greenland ice loss for decades to come.

How to cite: Khan, S. A., Colgan, W., Neumann, T., van den Broeke, M., Brunt, K., Noël, B., Bamber, J., Hassan, J., and Bjørk, A.: Accelerating ice loss from peripheral glaciers in North Greenland, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-16855, https://doi.org/10.5194/egusphere-egu23-16855, 2023.

We use cross-correlations of ambient seismic noise data between pairs of 9 broadband three component seismometers to investigate variations in velocity structure and anisotropy in the vicinity of the EastGRIP camp along and across flow of the Northeast Greenland Ice Stream (NEGIS).

From the 9-component correlation tensors associated with all station pairs we derive dispersion curves of Rayleigh and Love wave group velocities between station pairs at frequencies from 1 to 25 Hz. The distributions of the Rayleigh and Love group velocities exhibit anisotropy variations for the along and across flow component. To better assess those variations, we invert the dispersions curves to shear wave velocities in the horizontal (Vsh) and vertical (Vsv) direction for the top 300 m of the NEGIS using a Markov Chain Monte Carlo approach.

The reconstructed 1-D shear velocity model revels radial anisotropy in the NEGIS. Along and across flow vertical shear wave velocities (Vsv) identify comparable velocity profiles for all depths. However, horizontal shear wave velocities (Vsh) are faster by approximately 250 m/s in the along flow direction below a depth of 100 m, i.e. below the firn-ice transition.

This type of anisotropy seems to arise from the alignment of a crystallographic preferred orientation, due to deformation associated with shear zones. The role of anisotropy as e.g. created by air bubbles in the firn and ice matrix, is yet unclear.

Faster Vsh velocities in the along flow direction support that the NEGIS has crystal orientation alignment normal to the plane of shear compression (i.e. ice crystals orientated across flow) within the upper 300 m of the ice stream and are in alignment with the results from other methods. We demonstrate that simple, short duration (2-3 weeks), passive seismic deployment and environmental noise-based analysis can be used to determine the anisotropy of the upper part of ice masses.

How to cite: Pearce, E., Zigone, D., Schoonman, C., Franke, S., Eisen, O., and Rimpot, J.: Investigating firn and ice anisotropy around the EastGRIP Camp, North East Greenland Ice Stream, from ambient noise surface waves, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-40, https://doi.org/10.5194/egusphere-egu23-40, 2023.

When modelling ice sheet and glacier dynamics, a consideration of basal conditions is essential. Bed topography, hydrology and materials provide important controls on ice flow; however, the materials underlying large sections of the polar ice sheets are unknown. Seismic amplitude-versus-offset (AVO) analysis provides a means of inferring glacier bed properties, namely acoustic impedance and Poisson's ratio, by measuring bed reflectivity as a function of incidence angle.

However, existing methods of applying AVO to glaciology only consider the compressional-wave component of the wavefield and solutions suffer from non-uniqueness. This can be addressed using multi-component seismic datasets, in which a strong converted-wave arrival (downgoing compressional-wave energy converted to shear-wave energy upon reflection at the glacier bed) is often present. We present a method of jointly inverting compressional (PP) and converted-wave (PS) seismic data to improve constraint of glacier bed properties.

Using synthetic data, we demonstrate that for typical survey geometries, joint inversion of PP- and PS-wave AVO data delivers better-constrained bed acoustic impedance and Poisson’s ratio estimates compared with PP-only inversion. Furthermore, joint inversion can produce comparably constrained results to PP inversion when using input data with a smaller range of incidence angles/offsets (0-30 degree incidence for joint inversion, versus 0-60 degrees for PP- only). This could simplify future field acquisitions on very thick ice, where obtaining data at large incidence angles is difficult.

Joint AVO inversion therefore has the potential to improve identification of glacier bed materials and simplify field acquisitions of glacial AVO data. We also present preliminary results from Korff Ice Rise, West Antarctica, where better constraints on bed conditions can help improve our knowledge of ice sheet history in the Weddell Sea sector. Routine measurements of this kind will help constrain ice-sheet model inputs and reduce uncertainty in predictions of sea-level rise.

How to cite: Agnew, R., Booth, A., Brisbourne, A., Clark, R., and Smith, A.: Improving identification of glacier bed materials using converted-wave seismics, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-92, https://doi.org/10.5194/egusphere-egu23-92, 2023.

Seismic surveys are widely used to characterise the properties of glaciers, their basal material and conditions, and ice dynamics. The emerging technology of Distributed Acoustic Sensing (DAS) uses fibre optic cables as seismic sensors, allowing observations to be made at higher spatial resolution than possible using traditional geophone deployments. Passive DAS surveys generate large data volumes from which the rate of occurrence and failure mechanism of ice quakes can be constrained, but such large datasets are computationally expensive and time consuming to analyse. Machine learning tools can provide an effective means of automatically identifying seismic events within the data set, avoiding a bottleneck in the data analysis process.

Here, we present a novel approach to machine learning for a borehole-deployed DAS system on Store Glacier, West Greenland. Data were acquired in July 2019, as part of the RESPONDER project, using a Silixa iDAS interrogator and a BRUsens fibre optic cable installed in a 1043 m-deep borehole. The data set includes controlled-source vertical seismic profiles (VSPs) and a 3-day passive record of cryoseismicity.  To identify seismic events in this record, we used a convolutional neural network (CNN). A CNN is a deep learning algorithm and a powerful classification tool, widely applied to the analysis of images and time series data, i.e. to recognise seismic phases for long-range earthquake detection.

For the Store Glacier data set, a CNN was trained on hand-labelled, uniformly-sized time-windows of data, focusing initially on the high-signal-to-noise-ratio seismic arrivals in the VSPs. The trained CNN achieved an accuracy of 90% in recognising seismic energy in new windows. However, the computational time taken for training proved impractical. Training a CNN instead to identify events in the frequency-wavenumber (f-k) domain both reduced the size of each data sample by a factor of 340, yet still provided accurate classification. This decrease in input data volume yields a dramatic decrease in the time required for detection. The CNN required only 1.2 s, with an additional 5.6 s to implement the f-k transform, to process 30 s of data, compared with 129 s to process the same data in the time domain. This suggests that f-k approaches have potential for real-time DAS applications.

Continuing analysis will assess the temporal distribution of passively recorded seismicity over the 3 days of data. Beyond this current phase of work, estimated source locations and focal mechanisms of detected events could be used to provide information on basal conditions, internal deformation and crevasse formation. These new seismic observations will help further constrain the ice dynamics and hydrological properties of Store Glacier that have been observed in previous studies of the area.

The efficiency of training a CNN for event identification in the f-k domain allows detailed insight to be made into the origins and style of glacier seismicity, facilitating further development to passive DAS instrumentation and its applications.

How to cite: Pretorius, A., Booth, A., Smith, E., Nowacki, A., de Ridder, S., Christoffersen, P., and Hubbard, B.: Efficient neural network-based detection of seismicity in fibre optic data from Store Glacier, West Greenland, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-93, https://doi.org/10.5194/egusphere-egu23-93, 2023.

The ice mass balance of Antarctica has been steadily and strongly decreasing over recent decades, with major ramifications for global sea levels. Satellite remote sensing offers global, daily coverage of ice mass changes, which is essential for understanding land ice changes and their effects on global climate. However, we need to correct for processes including firn densification, glacial isostatic adjustment, elastic compensation of the Earth’s surface, ocean tides, and inverse barometer effect. Of these corrections, understanding the changes to the firn layer constitutes one of the largest uncertainties in making estimates of the surface mass balance from space. Furthermore, the development of firn models that aid our understanding of firn densification processes is hampered by a lack of observations.

Radar sounder reflections contain information about the roughness and permittivity of the reflecting interface, allowing us to map the spatial variability of the ice surface characteristics. In 2022, a helicopter-mounted ice-penetrating radar system developed by the University of Texas Institute for Geophysics collected high-quality radar observations over the Dotson Ice Shelf, West Antarctica. These surveys obtained clearly defined surface and bed reflections. We derived near-surface density along these survey flight lines using the radar statistical reconnaissance method developed by Grima, 2014. We calibrated our estimates with contemporary observations, including ground penetrating radar, a shallow ice core, an Autonomous phase-sensitive Radio Echo-sounder (ApRES), and radar soundings of well-defined surfaces from a calibration flight.

How to cite: Eayrs, C., Beem, L., Lee, C.-K., Lee, W. S., Chung, J., Pierce, C., Stutz, J., and Holland, D.: Variability of surface density at Dotson Ice Shelf, West Antarctica, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-952, https://doi.org/10.5194/egusphere-egu23-952, 2023.

Although rock glaciers represent a common periglacial landform in the alpine environment, and have a significant contribution to the hydrological regime of the related areas, their hydrodynamic is relatively less defined if compared to moraines, talus, and hillslope deposits. So far, the hydraulic behaviour of frozen layers that may be found inside rock glaciers has been investigated only with geochemical analysis of their spring water. These previous studies observed that the frozen layer acts as an aquiclude (or aquitard) and separates a supra-permafrost flow component, originating from snow-ice melting and rainwater, and a deeper aquifer at the bottom of the rock glacier systems.

In this work we verified, for the first time with a geophysical monitoring method, the low-permeability hydraulic behaviour associated to the frozen layer of mountain permafrost subsoils. In the inactive rock glacier of Sadole Valley (Southern Alps, Trento Province, Italy) we performed an infiltration experiment combined with 2D electrical resistivity tomography (ERT) measurements in time-lapse configuration. Considering the same ERT transect, a time zero dataset (t₀) has been collected before the water injection, subsequently about 800 liters of salt water have been spilled (approximately in a point) on the surface of the rock glacier in the middle of the electrodes array, and 10 ERT datasets have been collected periodically in the following 24 hours. To highlight the variations of electrical resistivity in the frozen subsoil, related to the injected salt water flow, only the inverted resistivity model derived from t₀ dataset has been represented in terms of absolute resistivities, while the other time steps results have been evaluated in terms of percentage changes of resistivity with respect to the t₀ initial model.

Our results clearly agree with the assumption that a frozen layer acts as an aquiclude (or aquitard) in a mountain permafrost aquifer, since during the infiltration experiment the injected salt water was not able to infiltrate into the underlying permafrost layer. The positive outcome of this test, fronting impervious environment and logistic constraints, opens up interesting future scenarios regarding the application of this geophysical monitoring method for the hydraulic characterization of rock glaciers. The experiment, used in this work to evaluate the permeability of the frozen layer, could be adapted in future to evaluate (in a quantitative way) the hydraulic conductivity of the active layer in rock glacier aquifers.

How to cite: Pavoni, M., Boaga, J., Carrera, A., Zuecco, G., Carturan, L., and Zumiani, M.: Hydraulic behaviour of a mountain permafrost subsoil revealed by an infiltration experiment and ERT time-lapse measurements, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-1137, https://doi.org/10.5194/egusphere-egu23-1137, 2023.

On the Greenland Ice Sheet, the firn layer holds the potential to trap and refreeze surface meltwater within its pore space. Acting as a buffer, it prevents meltwater from leaving the ice sheet. However, several meter-thick ice slabs have developed in the firn during the last two decades, reducing subsurface permeability and inhibiting vertical meltwater percolation. Ice slabs are located above the long-term equilibrium line along the west, north and northeast coasts of the ice sheet. Through time, ice slabs have thickened while new ones have developed at higher elevations. Concomitantly, the area of the ice sheet drained by surface rivers has increased by 29% from 1985 to 2020. Nowadays, 5-10% of surface losses through meltwater runoff originates from these newly drained areas, which correspond strongly with where ice slabs are located.

Here, we demonstrate that the highest elevation which is drained by surface rivers – termed the maximum visible runoff limit – is controlled by the ice content in the subsurface firn. Using ice slab thickness derived from the accumulation radar and annual maximum visible limit retrievals from Landsat imagery from 2002 to 2018, we show that a sub-surface ice content threshold triggers the shift from a ‘firn deep percolation regime’ to a ‘firn runoff regime’. Although ice slabs act as an aquitard, vertical meltwater percolation can still take place where visible meltwater ponds at the surface. We show that once the firn runoff regime is underway, ice slabs are thicker in locations with active surface hydrology compared to locations where no meltwater is visible at the surface. Spatial heterogeneity in ice slab thickness is therefore predominantly controlled by surface hydrology features.

How to cite: Jullien, N., Tedstone, A., and Machguth, H.: Ice slab thickening drives surface runoff expansion from the Greenland Ice Sheet’s percolation zone - and vice versa, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-2250, https://doi.org/10.5194/egusphere-egu23-2250, 2023.

The processing of Synthetic Aperture Radar (SAR) images is based on the coherent integration of Doppler frequencies. The associated Doppler spectrum is generated from the variation of the relative location between the radar and the scatterer. In geometries where the moving radar-platform follows a straight trajectory at constant velocity, the Doppler frequency depends on the angle of elevation from the radar to the scatterer, according to the electromagnetic (EM) propagation. In ice-sounding by airborne SAR, the EM path depends on the air-ice interface and the firn ice properties. For any of the scatterers under test, after integrating the received radar echoes from the multiple radar locations into a single pixel, the resulting amplitude image forgets which is the backscattering angle from each of the radar locations. However, this information is still within the Doppler spectrum of the image. We decompose the Doppler spectrum of the SAR image into three non-overlapping sub-bands; assign to each sub-band one of the primary colours red, green or blue, forming three sub-images; and finally merge the sub-images into a single one. Rather than a single full-beamwidth averaged amplitude value, the new composition now includes angular backscattering information, coded by one of the primary colours. Blue colour is assigned to scattering received from forwards, when the scatterer is ahead of the radar (positive Doppler frequencies); green approximately from the vertical (near zero-Doppler geometries); and red to scattering received from backwards (negative Doppler). Thus, heterogeneous scattering will be represented by one or two colours, whereas homogeneous scattering will be grey, with all the primary colours uniformly weighted. Features like internal layering, crevasses, SAR focussing quality and discrimination of multiple reflections from surface and bottom, can now be better interpreted. We present and discuss the results from the British Antarctic Survey (BAS) airborne radar PASIN2 for deep-ice sounding, in Recovery and Rutford ice streams, respectively in East and West Antarctica during seasons 2016/17 and 2019/20.

How to cite: Arenas-Pingarrón, Á., Corr, H. F. J., Brennan, P. V., Robinson, C., Jordan, T. A., Brisbourne, A., and Martín, C.: New Representation of Synthetic Aperture Radar Images for Enhanced Ice-Sounding Interpretation, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-2856, https://doi.org/10.5194/egusphere-egu23-2856, 2023.

In many regions of the Northern Hemisphere, permafrost is thawing due to climate change. In steep terrain, this permafrost degradation can affect slope stability. In one of Iceland's eastern fjords, Seyðisfjörður, nine major landslide cycles have occurred in the last century, originating from the lower parts (< 500 m a.s.l.) of the Strandartindur slopes, with the largest landslide event ever recorded in Iceland occurring in December 2020. Its triggering mechanism is being intensively studied and its development is being monitored. In addition to these instabilities, slow movements are also observed in the upper part (> 500 - 1010 m a.s.l.) of these slopes. In these upper areas, it is not known whether permafrost is present in the subsurface or what is causing it to creep downward. To further investigate the stability of these slopes, it is important to know and map the distribution and condition of possible permafrost layers. Therefore, electrical resistivity tomography (ERT) and ground penetrating radar (GPR) measurements were performed to study the presence and distribution of permafrost in the mountain, Strandartindur, above Seyðisfjörður. A combination of measurements is used as ERT responds primarily to the electrical resistivity of the subsurface, but this can depend strongly on other factors such as porosity, water content, etc., and GPR can help map the presence of different interfaces in the soil determined by their different physical properties, such as relative electrical permittivity, but also conductivity, which is the reciprocal of resistivity. Combining the two methods allows us to get a clearer picture of the subsurface. As a benchmark for ERT measurements in the field, a laboratory setup was performed with soil and rock samples at different temperatures and water saturations to study the behavior of frozen and non-frozen conditions in our geologic environment. With all of these measurements, we aim to answer the questions of whether permafrost is present in the selected area, what the distribution of permafrost is, whether we can use laboratory ERT to establish reference resistivity values, and if these methods are appropriate for this area.

How to cite: von der Esch, A., Piispa, E. J., and Sæmundsson, Þ.: Electrical Resistivity Tomography and Ground-Penetrating Radar Measurements for Permafrost Detection on a Mountain Slope at Strandartindur, Seyðisfjörður - East Iceland, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-4135, https://doi.org/10.5194/egusphere-egu23-4135, 2023.

Cold regions are increasingly subjected to higher air temperatures, causing warming of permafrost and a deepening of the active layer. This activates hydrogeological groundwater flow and new groundwater pathways to emerge. Monitoring of the active layer depth occurs mainly with the use of temperature observations, but a more flexible and non-invasive method to study transient subsurface processes is with the use of Electrical Resistivity Tomography (ERT) observations. 

Automated time-laps ERT arrays are used to monitor the frozen ground evolution during various seasons, observing resistivity variations during freezing and thawing. Similarly, the leaching of meltwater into the ground under freezing/thawing conditions can be observed. Not only geophysical changes such as fluctuations in water content and water table, but also temperature variations affect the electrical resistivity field. In order to track the development of permafrost active-layer freeze-thaw fronts using ERT observations, it is thus essential that the effect of temperature on the resistivity is clearly defined at realistic scales representing field conditions. Our aim is to determine fluid resistivity at various stages during freezing and thawing and validate current temperature–resistivity relations for partly frozen soils.

This study used a soil column (0.4 m diameter, 1 m heigh) equipped with 96 stainless steel electrodes placed at 8 horizontal rings of 12 electrodes each at various heights around the circumference of the column alongside with temperature sensors. The column was fully insulated on the sides and top except for the bottom, creating a 1D heat transfer system. The soil column was filled with quartzite sand with a D50 of 350 (μm) and organic matter content of 5 (wgt %). The experimental setup was placed within a climate chamber where the column was frozen to -4 °C and thawed to 3 °C over a 3-month period. During the freezing and thawing phase, a full 3D resistivity image was collected using the ERT at a weekly interval. Initial results show that the setup is capable of simulating permafrost freezing and thawing dynamics and ongoing work focuses on the relation between the temperature and time lapse ERT resistivity observations.

How to cite: de Bruin, J., Bense, V., and van der Ploeg, M.: Detecting permafrost freeze-thaw front propagation using time-laps ERT observations in a large column experiment, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-4895, https://doi.org/10.5194/egusphere-egu23-4895, 2023.

We present maps of the sub-ice platelet layer (SIPL) thickness and ice volume fraction beneath the land-fast sea ice in Atka Bay adjacent to the Ekström Ice Shelf (southeastern Weddell Sea, Antarctica). The widespread SIPL beneath Antarctic fast ice is indicative of basal melt of nearby ice shelves, contributes to the sea ice mass balance and provides a unique ecological habitat. Where plumes of supercooled Ice Shelf Water (ISW) rise to the surface rapid formation of platelet ice can lead to the presence of a semi-consolidated SIPL beneath consolidated fast ice.

Here we present data from extensive electromagnetic (EM) induction surveying with the multi-frequency EM sounder GEM-2 between May and December, 2022. It includes monthly survey data along a fixed transect line across Atka Bay between May and October, as well as comprehensive mapping across the entire bay in November and December. The GEM-2 surveys were supplemented by drill hole thickness measurements, ice coring and CTD profiles. A new data processing and inversion scheme was successfully applied to over 1000 km of EM profiles with a horizontal resolution of one meter. We obtained layer thicknesses of the consolidated ice plus snow layer, the SIPL, and the respective layer conductivities. The latter were used to derive SIPL ice volume fraction and an indicator for flooding at the snow-ice interface. The robustness of the method was validated by drill hole transects and CTD profiles.

Our results support conclusions about the spatial variability of the ocean heat flux linked to outflow of ISW from beneath the ice shelf cavity. Temporally, we found that the end of SIPL growth and the onset of its thinning in summer can be linked to the disappearance of supercooled water in the upper water column.

How to cite: Neudert, M., Arndt, S., Schulze, M., Hendricks, S., and Haas, C.: High resolution maps of the sub-ice platelet layer in Atka Bay from electromagnetic induction sounding, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-5545, https://doi.org/10.5194/egusphere-egu23-5545, 2023.

Since 2012 we monitor avalanche activity using distributed acoustic and seismic fiber optic sensing at our avalanche test area at Lech am Arlberg, Austria. The method is based on an optical time domain reflectometer system that detects seismic vibrations and acoustic signals on a fiber optic cable that can have a length of up to 30 km in 80 cm resolution. While in the first years we focused on successfully developing an operational avalanche detection system that is able to tell in real time reliably when an avalanche was triggered and what the size of the avalanche is, we now present our investigations of the seismic signals to measure snow properties such as snow depth and avalanche properties such as flow behavior. Our test in winter 2022 recorded by blasting triggered avalanches and during data post processing we extracted seismic guided waves. We discuss methods for extracting information from guided waves for measuring snow depth, which was verified against spatial snow depth measurements from terrestrial laser scanning. Analyzing the seismic signals of avalanches with run-out distances ranging from a few metres to approximately 250 m allows us to differentiate between wet and snow avalanches, which is discussed in the context of avalanche dynamics.

How to cite: Prokop, A., Agostinetti, N. P., and Graseman, B.: Measuring snow and avalanche properties using acoustic and seismic distributed fiber optic sensing, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-5851, https://doi.org/10.5194/egusphere-egu23-5851, 2023.

The short observational record is one of the main obstacles to improving the present understanding of the future of the Polar ice sheets. While the quantity and quality of observations presently are increasing observations from before the 1990s are scarce. Here, we present the first results from a newly digitized ice-penetrating radar dataset acquired over the Greenland Ice Sheet in the 1970s. The data consist of more than 170,000 km of radar flight lines. While the ice thickness information from the data has been digitized by previous studies, the data itself (notably the z-scopes) were until recently only available as 35-mm films, microfiche copies of the films, and enlarged positives: Formats that are not usable for digital analysis.

In 2019, the film rolls were scanned by a digital scanner and subsequently, a large effort has been directed at carrying out quality control of the data with the view of making them publicly available.  Here we present the first results from this digitization. The overall data quality is good, and we are able to retrieve valuable information on layer stratigraphy and ice-flow dynamics.

How to cite: Karlsson, N. B., Schroeder, D., Sørensen, L. S., Chu, W., Teisberg, T., Tarzano, A., Skou, N., and Dall, J.: A new view of a 1970s radar dataset from Greenland, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6935, https://doi.org/10.5194/egusphere-egu23-6935, 2023.

The climatic conditions over ice sheets at the time of snow deposition and compaction imprint distinctive crystallographic properties to the resulting ice. As it gets buried, its macroscopic structure evolves due to vertical compression but retains traces of the climatic imprint that generate distinctive mechanical, thermal and optical properties. Because climate alternates between glacial periods, that are colder and dustier, and interglacial periods, the ice sheets are composed from layers with alternating mechanical properties. Here we compare ice core dust content, crystal orientation fabrics and englacial vertical strain-rates, measured with a phase-sensitive radar (ApRES), at the South Pole and EPICA Dome C ice cores. In agreement with previous observations, we show that ice deposited during glacial periods develops stronger crystal orientation fabrics. In addition, we show that ice deposited during glacial periods is harder in vertical compression and horizontal extension, up to about three times, but softer in shear. These variations in mechanical properties are ignored in ice-flow models but they could be critical for the interpretation of ice core records. Also, we show that the changes in crystal orientation fabrics due to transitions from interglacial to glacial conditions can be detected by radar. This information can be used to constrain age-depth at future ice-core locations.

How to cite: Martin, C., Mulvaney, R., Conway, H., Koutnik, M., Stevens, C. M., Corr, H., Ritz, C., Nicholls, K., Drews, R., and Ershadi, M. R.: Climatic imprint in the mechanical properties of ice sheets and its effect on ice flow: Observations fromSouth Pole and EPICA Dome C ice cores, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-7198, https://doi.org/10.5194/egusphere-egu23-7198, 2023.

Anisotropic crystal fabrics in ice sheets develop as a consequence of deformation and hence record information of past ice flow. Simultaneously, the fabric affects the present-day bulk mechanical properties of glacier ice because the susceptibility of ice crystals to deformation is highly anisotropic. This is particularly relevant in dynamic areas such as fast-flowing glaciers and ice streams, where the formation of strong fabrics might play a critical role in facilitating ice flow. Anisotropy is ignored in most state-of-the-art ice sheet models, and while its importance has long been recognized, accounting for fabric evolution and its impact on the ice viscosity has only recently become feasible. Both the application of such models to ice streams and their verification through in-situ observations are still rare. We present an extensive dataset of fabric anisotropy derived from ground-based and air-borne radar data, covering approximately 24,000 km² of the Northeast Greenland Ice Stream onset region. Our methods yield the horizontal anisotropy and are based on travel time anisotropy as well as birefringence-induced power modulation of radar signals. These methods complement each other and show good agreement. We compare the in-situ observations with the results obtained from a fabric-evolution model employed along flow line bundles in the ice stream onset to discuss the fabric in light of past flow history and its significance for the current flow mechanics of the ice stream.

How to cite: Gerber, T. and Eisen, O. and the NEGIS community: Spatial variation of ice crystal fabric and implications of anisotropic flow in the Northeast Greenland Ice Stream, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-7695, https://doi.org/10.5194/egusphere-egu23-7695, 2023.

Promontory ice rises are locally grounded features adjacent to ice shelves that are still connected to the ice sheet. Ice rises are an archive for the atmospheric and ice dynamic history of the respective outflow regions where the presence, absence, or migration of Raymond arches in radar stratigraphy represents a memory of the ice-rise evolution. However, ice rises and their inferred dynamic history are not yet used to constrain large-scale ice flow model spin-ups because matching the arch amplitudes includes many unknown parameters, e.g., those pertaining to ice rheology. In particular, anisotropic ice flow models predict gradients in ice fabric anisotropy on either side of an ice divide. However, this has thus far not been validated with observations.

The ground-based phase-sensitive Radio Echo Sounder (pRES) has previously been used to infer ice fabric types for various flow regimes using the co-polarized polarimetric coherence phase as a metric to extract information from the birefringent radar backscatter. Here, we apply this technique using quad-polarimetric radar data along a 5 km transect across a ridge near the triple junction of Hammarryggen Ice Rise at the Princess Ragnhild Coast. A comparison with ice core data collected at the dome shows that the magnitude of ice fabric anisotropy can reliably be reconstructed from the quad-polarimetric data. We use the combined dataset also to infer the spatial variation of ice fabric orientations in the vicinity of the triple junction. The observations are integrated with airborne radar profiles and strain rates based on the shallow ice approximation. We then discuss whether estimated anisotropy from radar polarimetry on ice rises, in general, can be another observational constraint to better ice rises as an archive of ice dynamics.

How to cite: Ershadi, M. R., Drews, R., Tsibulskaya, V., Sun, S., Henry, C., Oraschewski, F., Koch, I., Martin, C., Tison, J.-L., Wauthy, S., Bons, P., Eisen, O., and Pattyn, F.: Ice rise evolution derived from radar investigations at a promontory triple junction, Dronning Maud Land, East Antarctica, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-8197, https://doi.org/10.5194/egusphere-egu23-8197, 2023.

The internal, isochronous layering of glaciers is shaped by accumulation and ice deformation. Information about these processes can be inferred from observing the layers using radar sounding. The reflectivity of the layers depends on density (permittivity) and acidity (conductivity) contrasts which tend to decrease with depth. At places like alpine glaciers where logistic limitations often only allow the deployment of lightweight and power-constrained ground-penetrating radar systems, it can therefore be challenging to illuminate the deeper radio-stratigraphy.

The phase-sensitive Radio Echo Sounder (pRES) is a lightweight frequency modulated continuous wave radar which allows the use of coherent Synthetic Aperture Radar (SAR) processing techniques to improve the signal-to-noise ratio of internal reflection horizons. Using a mobile pRES we collected a radar profile on an alpine glacier (Colle Gnifetti, Italy/Switzerland). Here, we demonstrate how to apply layer-optimized SAR techniques to make deep internal layers visible, which could not be seen by a conventional pulsed radar. We evaluate the requirements on spatial resolution and positioning accuracy during data acquisition, necessary for applying layer-optimized SAR processing, as they constrain the feasibility of the method. We further discuss implications on how density and acidity contribute to decreasing dielectric contrasts.

How to cite: Oraschewski, F. M., Koch, I., Ershadi, M., Hawkins, J., Eisen, O., and Drews, R.: Layer-optimized SAR processing with a mobile pRES to illuminate the internal layering of an alpine glacier, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9273, https://doi.org/10.5194/egusphere-egu23-9273, 2023.

We have developed a drone-based GPR system at the University of Lausanne that allows for the safe and efficient acquisition of large, high-density, 3D and 4D datasets over alpine glaciers. The system is able to record approximately 4 line-km of high-quality GPR data per set of drone batteries in less than 30 minutes of operation which, combined with multiple sets of batteries and/or the possibility of charging at the field site, means that 3D datasets over a large area can be acquired with unprecedented efficiency. The latter performance is possible thanks to (i) a custom-made real-time-sampling GPR controller that has been specifically designed for glaciers studies, (ii) minimization of the total payload weight using custom-built antennas and carbon-fiber components, and (iii) development of an optimized survey methodology. Further, because the drone is equipped with real-time kinematic GPS positioning, survey paths can be repeated with great precision, which opens new opportunities in term of 4D data acquisitions.

In the summer of 2022, we acquired both 3D and 4D data over two Swiss glaciers. On the Otemma glacier, we surveyed a grid of 462 profiles representing a total length of 112 line-km of data in only four days. After 3D binning, the trace spacing intervals in the in-line and crossline directions were respectively 0.4 m and 1 m, making this arguably the largest 3D GPR dataset of such density ever recorded over ice. The interface between the ice and the bedrock, visible on all profiles, extends to 1000 ns which translates into a depth of approximately 80 m. In addition, internal englacial and subglacial 3D structures are clearly detectable.

In parallel, we visited the Rhône glacier on a monthly basis between June and September 2022. A collapse feature, identified by the presence of large circular crevasses, had formed and was evolving close to the snout of the glacier. This represented a great opportunity to test the 4D acquisition capabilities of our system. We collected four high-density 3D datasets on the same survey grid. The repeatability of the trajectories was excellent as the paths differ only by a few centimeters between occurrences. Clear variations in the internal structure of the glacier are visible which will be investigated in the upcoming months.

How to cite: Ruols, B., Baron, L., and Irving, J.: High-density 3D and 4D GPR data acquisitions over alpine glaciers using a newly developed drone-based system., EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9619, https://doi.org/10.5194/egusphere-egu23-9619, 2023.

The Whillans Ice Stream (WIS) is a major outlet of the West Antarctic Ice Sheet. Significantly, the downstream portion of the WIS is presently decelerating, possibly stagnating by the end of this century. Additionally, this downstream region of WIS is unique in that it moves primarily by stick-slip motion. However, both the rate of deceleration as well as the percent of motion accommodated by stick-slip motion is spatially variable. Such spatial variability is potentially linked to associated variability in basal conditions. Active source seismic measurement are capable of providing high-resolution insights into basal conditions, however, they are time-consuming to collect, limiting the spatial extent over which they can be acquired. In this presentation, we will use passive seismic measurements collected at over 50 seismic stations to map sediment thickness and ice-bed conditions across the region. This will be done using the receiver function method which images the depth and physical properties of sediments by modeling the arrival times and amplitudes of seismic waves that interact with subglacial sedimentary structures. We will first map conditions at the ice-bed interface by using relatively high-frequency waveforms (> 2 Hz) as they are sensitive to the physical properties of the shallow (< 20m ) subglacial sediments layers. Across the entirety of the study region, we find that this uppermost layer of sediments is characterized by relatively high porosity sediments.  Second, we will utilize lower frequencies (< 2 Hz) to map the depth basement, finding that the entire region is underlain 100’s of meters of sediments (Gustafson et al., Science, 2022). We will use our maps of sediment properties and thickness to investigate potential mechanisms for the observed variability in deceleration and stick-slip behavior of the WIS.  

How to cite: Winberry, J. P.: Basal Conditions and Sedimentary Structure of the Whillans Ice Stream., EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9621, https://doi.org/10.5194/egusphere-egu23-9621, 2023.

Englacial temperature and water content play critical roles in glacier dynamics, both within ice sheets and mountain glaciers. As radio wave attenuation is sensitive to both of these properties, radio-echo sounding (RES) serves as a useful tool for mapping out their distributions within glaciers. Ground-based bistatic surveys, in which multi-offset measurements are taken, provide a large diversity in bed incidence angles and travel-path lengths. Provided the anomaly of interest is sufficiently sampled, these measurements can be exploited to perform attenuation tomography, thereby recovering the distribution of englacial radio wave attenuation from which englacial temperature can be estimated. Extensive RES surveys have been carried out over Antarctica using airborne radar; however, due to the monostatic geometry, methods for estimation of englacial radio wave attenuation and basal roughness have relied primarily on nadir returns. These estimates are often derived from 2D spatial correlation of basal return power and ice thickness or by employing layer-tracking methods. These techniques are limited in that the former uses echoes from a large spatial footprint, preventing the detection of small-scale anomalies, while the latter assumes a known, spatially invariant reflectivity for tracked layers. However, by considering returns from off-nadir in airborne surveys, techniques from multi-offset surface surveys can be modified and extended to perform airborne attenuation tomography. While not reaching the range of path diversity achievable in surface-based surveys due to limitations imposed by total internal reflection at the ice-air interface, airborne off-nadir returns contain valuable information about subglacial and englacial conditions that is often ignored. Thus, we propose a method for estimating englacial attenuation and basal roughness using the drop in power from the peak to tail of hyperbolic scattering events in unfocussed radargrams associated with the rough bed surface. The travel-paths of the bed returns across a given hyperbolic event vary in both length and bed incidence angle. Thus, the drop in return power across a hyperbolic event gives insight into both the integrated attenuation along a travel-path, as well as the scattering function at the bed. Specular reflections from internal layers with varying dips similarly provide diversity in travel-path lengths, allowing the derivation of a relationship between path length and return power without the complications brought about by diffuse scattering at rough surfaces. Using the diverse path lengths and angles through the ice, a tomographic inversion to map the spatial distribution englacial attenuation anomalies can be implemented. This technique is applied to synthetic data, as well as data collected using the British Antarctic Survey’s Polarimetric-radar Airborne Science Instrument (PASIN), specifically to lines collected over the Eastern Shear Margin of Thwaites Glacier. This location was chosen as constraining bed conditions and identifying expected englacial thermal anomalies are critical to understanding the history and modelling the future of Thwaites.

How to cite: May, D., Schroeder, D., and Young, T. J.: Radar Attenuation Tomography for Mapping Englacial Temperature Distributions Using Off-Nadir Airborne Radio-Echo Sounding, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9833, https://doi.org/10.5194/egusphere-egu23-9833, 2023.

Monitoring of the thickness and elastic parameters of floating ice on lakes and the sea is of interest in understanding the climate change impact on Alpine and Arctic environments, assessing ice safety for recreational and engineering purposes, studying ice shelves as well as exploring possibilities for the future exploration of the icy crusts of ocean worlds in our solar system. A multitude of geophysical methods exist today to monitor sea and lake ice thickness as well as elastic parameters. Mostly, seismic and radar measurements are used. Both methods have in common that they come with significant logistical effort and expensive equipment. In this study, we present a novel low cost approach using acoustic sensors for ice monitoring.

We explored the possibility of using microphones deployed on frozen lakes in the Swiss Alps to monitor the lake ice-thickness using acoustic signals originating from frequently occurring ice quakes. Data were obtained during a three-month-long field campaign at Lake St. Moritz in Switzerland in winter 2021/2022. Three microphone stations were placed on the lake in addition to five conventional seismometers. These seismometers were used to compare the acoustic signals with the seismic ice quake recordings. Additionally, also active-source experiments were conducted using hammer strokes as source, which were used to constrain elastic parameters of the ice.

The acoustic recordings of ice quakes allowed us to exploit the unique characteristics of so-called air-coupled waves to determine time-dependent ice thickness curves of Lake St. Moritz for winter 2021/2022 using acoustic data only. Furthermore, the acoustic data allowed us to gain new insights into the ice/air coupling of seismic waves in ice. 

How to cite: Wetter, C., Schmelzbach, C., and Stähler, S. C.: Monitoring lake ice with acoustic sensors, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10127, https://doi.org/10.5194/egusphere-egu23-10127, 2023.

The snowpack is a key component in several fields like climatology, hydrology, or natural hazards research and mitigation, not least in mountainous regions. One of the most considerable snowpack features is the snow water equivalent (SWE), representing the mass of water stored in the snowpack and – in another perspective – the weight straining objects the snow is settling on (snow load). In comparison to snow depth, measuring SWE is rather complex and prone to errors. Consecutive observations of SWE do not have a long tradition in many regions.

Despite various recent developments in measuring SWE by means of remote sensing or other noninvasive methods, e.g. with scales, GNSS reflectometry, signal attenuation and time delay techniques, cosmic-ray neutron sensing, etc. the standard measuring technique still are snow tubes or gauging cylinders, often in combination with digging pits. Tubing-technique is commonly used as reference for the validation of named modern methods, although studies addressing its accuracy, precision and repeatability are very rare.

This contribution provides results from comparing different types of SWE measurement tubes with reference standard oberservation. Several field tests were executed at different sites in the Austrian Alps covering a great variety of snow conditions (e.g. dry and wet), snow depths and SWEs. For the reference observation 3x4 m rectangular fields were dug snow-free and the respective snow masses have been weighted stepwise using ca. 50-liter-buckets. Due to the large total mass of snow of typically around two tons per rectangular, relative uncertainties are extremely small and the results highly accurate. Additionally, different snow tubes were compared to each other. The cylinder or tube designs vary a lot: from meters long metal coring tubes of typical inner diameters of ca. 4-7 cm (without the need of pits) or PVC cylinders with typical lengths of 0.5 to 1.5 m and diameters ranging from about 5-20 cm to small aluminum tubes holding a maximum of 0.5 liter of snow.  

Many statistical measures like variance and bias vary quite a lot primarily depending on the equipment used, but also on the different snow conditions. A synopsis on the suitability of the various methods depending on the questioning or objective of the observation is provided.

How to cite: Radlherr, A. and Winkler, M.: Validating manual measurements of snow water equivalent against a reference standard, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-11787, https://doi.org/10.5194/egusphere-egu23-11787, 2023.

Many of the gauged snow depth measurements in the European Alps began in the late 19th and early 20th centuries. We leverage this reasonably long period of record to investigate the historical sensitivity of snow depths as a function of precipitation, mean temperature, and elevation. By controlling for changes in precipitation, we can isolate the influence that different temperature changes have on snow depths at varying elevation bands. This simple, yet effective, approach to defining our historical sensitivity can provide a robust observational framework to evaluate the impact that a range of different future warming scenarios would have on snow depths across the Alps. As a result, adaptation and mitigation measures can be put in place for a variety of end users, such as ski tourism and water resource management. Furthermore, this provides an observational reference by which to evaluate the performance of climate model simulations.

How to cite: Switanek, M., Schöner, W., and Resch, G.: Snow depth sensitivity to mean temperature and elevation in the European Alps, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-13127, https://doi.org/10.5194/egusphere-egu23-13127, 2023.

While climate change driven increase in air temperature has been correctly modeled in recent decades, the extent of its consequences is still uncertain. In high alpine environments, especially in steep rock walls, permafrost degradation reduces slope stability with critical consequences for people and infrastructures: to properly assess the risk, the rate of these changes must be monitored. In the last decades, electrical resistivity tomography (ERT) has been used in more than hundred studies to detect permafrost, but there are only limited long-term monitoring cases that mostly do not provide quantitative information. 

Here we compare ERT measurements from two alpine landforms with different altitude and lithology: Steintälli ridge (3160m asl, CH) and Mt. Zugspitze rock wall (2750 m asl, DE/AT). Standard procedures and permanently installed electrodes allow the collection of a unique dataset of consistent measurements since 2006. Supporting information like resistivity-temperature calibration from former studies, rock surface and borehole temperatures as well as active seismic refraction measurements enable an advanced quantitative interpretation of the results. 

Permafrost at both sites is close to disappearing and in both cases resistivity changes are evident and in good agreement with air temperature increase, although with different magnitudes according to the landform. The yearly 3D measurements of the Steintälli ridge show a sudden and conspicuous degradation (~40% of the volume in 15 years), while the monthly 2D monitoring of the north face of Mt. Zugspitze shows slow constant decrease in summer (~15% of the surface in 15 years) and a strong variation in winter in correlation with snow-height. 

For the first time we provide a quantification of alpine permafrost degradation rates in different landforms over 15 years. These datasets help to better understand the different characteristics of the thermal responses to the climate change induced stress on alpine permafrost environments.

How to cite: Scandroglio, R., Offer, M., and Krautblatter, M.: Long-time permafrost evolution in alpine bedrock: quantifying climate change effects with geoelectrical monitoring, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-14390, https://doi.org/10.5194/egusphere-egu23-14390, 2023.

How to cite: Smith, E. C., Karplus, M., Walter, J., Nakata, N., Booth, A. D., Gonzalez, L., Pretorius, A., Agnew, R., Veitch, S., Dawson, E. J., May, D., Summers, P., Young, T. J., Christoffersen, P., and Tulaczyk, S.: Thwaites Glacier Eastern Shear Margin: Insights from two broadband seismic arrays, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-16308, https://doi.org/10.5194/egusphere-egu23-16308, 2023.

Snow microstructure defines the physical, mechanical and electromagnetic properties of snow. Accurate information of snow structure is needed by many applications, including avalanche forecasting (Hirashima et al., 2008) and numerical weather prediction (de Rosnay et al., 2014). The interaction of electromagnetic waves with snow properties can be applied in satellite remote sensing to retrieve, for example, global information of snow mass (Pulliainen et al., 2020). Objective in-situ observations of snow microstructure are needed to validate and develop both physical models and satellite snow retrieval algorithms. Conventional measurements of snow grain size are unsatisfactory in this regard, as the parameter is difficult to measure objectively, and even its definition is ambiguous (Mätzler, 2002). Hence, recent efforts have focused on developing forward models of microwave interactions and snow specific surface area (SSA), which can be objectively measured in field and laboratory conditions using various methods. A recently proposed approach links SSA to microwave scattering properties through another physically defined parameter (Picard et al., 2022).

In the SnowAPP project, three field campaigns were carried out at the Finnish Meteorological Institute Arctic Research Centre in Sodankylä, with the goal of collecting data on snow microstructural properties and establishing the relation of microstructure to both optical reflectance and microwave emission and scattering from snow.  During the spring 2019 campaign, six different methods were used for measuring SSA; and several methods were used for measuring snow density, another important factor affecting especially the extinction of microwave energy. Furthermore, multi-frequency radiometry and a wide-band, high resolution spectrometer were used to measure microwave emission and reflectance. In this study, we compare objectively the SSA and density values obtained by the different methods in a round-robin exercise. The relation of measured snow microstructures to measured spectral properties of snow are discussed.

SnowAPP was funded by the Academy of Finland, with contributions from WSL Institute for Snow and Avalanche Research SLF, Centre Tecnològic de Telecomunicacions de Catalunya, Beijing Normal University, National Institute for Polar Research, Meteorological Research Institute (Japan), and Université Grenoble Alpes.

de Rosnay, P., Balsamo, G., Albergel, C., Muñoz-Sabater, J., & Isaksen, L. (2014). Initialisation of land surface variables for numerical weather prediction. Surveys in Geophysics, 35(3), 607–621. https://doi.org/10.1007/s10712-012-9207-x

Hirashima, H., Nishimura, K., Yamaguchi, S., Sato, A., & Lehning, M. (2008). Avalanche forecasting in a heavy snowfall area using the snowpack model. Cold Regions Science and Technology, 51(2–3), 191–203. https://doi.org/10.1016/j.coldregions.2007.05.013

Mätzler, C., 2002. Relation between grain-size and correlation length of snow. J. Glaciol., (48)162: 461-466.

Picard, G., Löwe, H., Domine, F., Arnaud, L., Larue, F., Favier, V., & Meur, E. le. (2022). The Microwave Snow Grain Size: A New Concept to Predict Satellite Observations Over Snow-Covered Regions. https://doi.org/10.1029/2021AV000630

Pulliainen, J., Luojus, K., Derksen, C., Mudryk, L., Lemmetyinen, J., Salminen, M., Ikonen, J., Takala, M., Cohen, J., Smolander, T., & Norberg, J. (2020). Patterns and trends of Northern Hemisphere snow mass from 1980 to 2018. Nature, 581(7808), 294–298. https://doi.org/10.1038/s41586-020-2258-0

How to cite: Kontu, A., Leppänen, L., Pirazzini, R., Hannula, H.-R., Lemmetyinen, J., Räisänen, P., McFarlane, A., Espin Lopez, P., Anttila, K., Rimali, A., Suokanerva, H., Yang, J., Aoki, T., Niwano, M., Picard, G., Ollivier, I., Arnaud, L., Matzl, M., Merkouriad, I., and Schneebeli, M.: Intercomparison of quantification methods for snow microstructure during the SnowAPP experiment, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-16342, https://doi.org/10.5194/egusphere-egu23-16342, 2023.

Permafrost peatlands are highly vulnerable ecosystems in a warming climate; their thaw greatly impacts carbon storage capacity and endangers existing landscape morphology. Due to their remoteness and, in some cases, protected status, it is difficult to characterise and monitor the subsurface using invasive methods. Geophysical investigations are useful in such cases allowing relatively rapid and extensive subsurface mapping. We focus here on the emerging high-frequency induced polarisation (HFIP) method, which can be effective in permafrost hydrology research as the geoelectrical properties of frozen water display a characteristic frequency-dependence between ranges of 100 Hz and 100 kHz.

HFIP field measurements were conducted using the Chameleon-II equipment (Radic Research) on two peat permafrost sites located in Abisko, Northern Sweden: Storflaket mire and Heliport mire. The sites have been subject to routine permafrost monitoring since 1978 and are known to have an upper peat layer underlain by a silt-rich subsoil. We present the results of 2D surveys measuring frequencies ranging from 1 Hz to 57 kHz, which capture a high-frequency phase shift peak. Field data are inverted for each measured frequency separately with ResIPy, using an appropriate data error quantification model. The spectral data analysis captures heterogeneity within the subsurface, i.e., layered medium, permafrost mire boundary and ice-rich versus ice-poor regions. Identification of spectrally distinct regions allows the application of an appropriate relaxation model. For this study, we apply a two-component mixture model for ice-content estimation. Our results extend the existing knowledge at this site by quantifying ice content in a 2D plane, thus improving the foundation for further modelling studies.

How to cite: Sugand, M. G., Hördt, A., and Binley, A.: Resolving ice content heterogeneity within permafrost peatlands using high-frequency induced polarisation., EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-16522, https://doi.org/10.5194/egusphere-egu23-16522, 2023.

The Dome Fuji (DF) region in Antarctica is a potential site for holding an ice record older than one million years. Here, we combine the internal airborne radar stratigraphy with a 1-D inverse model to reconstruct the age field of ice in the DF region. As part of the Beyond EPICA - Oldest Ice reconnaissance (OIR), the region around DF was surveyed with a total of 19000 km of radar lines in the 2016/17 Antarctic summer. Internal stratigraphy in this region has now been traced. Through these tracked radar isochrones, we transfer the age-depth scale from DF ice core to the adjacent 500 km² region. A 1-D inverse model has been applied at each point of the survey to extend the age estimates to deeper regions of the ice sheet where no direct or continuous link of internal stratigraphy to the ice cores is possible, and to construct basal thermal state and accumulation rates. Through the reliability index of each model, we can evaluate the reliability of the 1-D assumption. Mapped age of basal ice and age density imply there might exist promising sites with ice older than 1.5 million years in the DF region. Moreover, the deduced basal state, i.e., melting rates and stagnant ice provide constraints for finding old-ice sites with a cold base. The accumulation rate ranges from 0.014 to 0.038 m a^-1 (in ice equivalent) in the DF region, which is also an important criterion for potential old ice.

How to cite: Wang, Z., Eisen, O., Chung, A., Steinhage, D., Parrenin, F., and Freitag, J.: Mapping stagnant ice and age in the Dome Fuji region, Antarctica, by combining radar internal layer stratigraphy and flow modeling, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-787, https://doi.org/10.5194/egusphere-egu23-787, 2023.

Ice cores remain the highest resolution proxy for measuring past surface mass balance (SMB) that can be used for model-data comparison. However, there is a clear difference in the spatial resolution of the ice cores, with a surface sample on the order of cm², and the spatial resolution of models, with at best a surface footprint on the order of a few km². Comparing ice core SMB records and model SMB outputs directly is therefore not a one-to-one comparison. In addition, it is well known that ice cores, as point measurements, sample very local SMB conditions which can be affected by local wind redistribution of the SMB at the surface.

We set out to answer the question: how representative are ice-cores of regional SMB? For this, we use several ground-penetrating radar (GPR) surveys in East Antarctica, which have co-located ice core drill sites. Most of our sites share a relatively similar climatology, as they are all coastal ice promontories/rises along the Dronning Maud Land coast, with the exception of the Dome Fuji survey on the high plateau in the interior of the continent.

We will show that the comparison of the SMB signals of the GPR and the ice core records allows us to estimate the spatial footprint of the ice cores, and that this spatial footprint varies widely from site to site. We will provide a summary of the spatial and temporal characteristics for each location.

How to cite: Cavitte, M. G. P., Goosse, H., Matsuoka, K., Wauthy, S., Dey, R., Goel, V., Tison, J.-L., Van Liefferinge, B., and Meloth, T.: What to watch out for when assimilating ice-cores as regional SMB proxies?, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-792, https://doi.org/10.5194/egusphere-egu23-792, 2023.

Glaciers and ice streams account for the majority of ice mass discharge to the ocean from the Antarctic Ice Sheet, and are bounded by intense bands of shear that separate fast-flowing from slow or stagnant ice, called shear margins. The anisotropy of glacier ice (i.e. a preferred crystal orientation) stemming from high rates of shear at these margins can greatly facilitate fast streaming ice flow, however it is still poorly understood due to a lack of in-situ measurements. If anisotropy is incorporated into numerical ice sheet models at all, it is usually as a simple scalar enhancement factor that represents the "flow law" that governs the model's rheology. Ground-based and airborne radar observations along two transects fully crossing the Eastern Shear Margin of Thwaites Glacier reveal rapid development of highly anisotropic fabric tightly concentrated around a lateral maximum in surface shear strain. These measurements of fabric strength at the centre of the shear margin are indicative of a horizontal pole configuration, which potentially represents ice that is “softened” to shearing in some directions and hardened in others. The resulting flow enhancement revealed by our results suggest that the viscosity of ice is highly variable and regime-dependent, and supports the importance of considering anisotropic flow laws to model the rheology of ice sheets.

How to cite: Young, T. J., Martin, C., Jordan, T., Zeising, O., Eisen, O., Christoffersen, P., Lilien, D., and Rathmann, N.: Radar-derived ice fabric anisotropy and implications on flow enhancement along the Thwaites Glacier Eastern Shear Margin, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-2178, https://doi.org/10.5194/egusphere-egu23-2178, 2023.

Permafrost, the frozen layer beneath a freezing and thawing active layer, is an impermeable frozen soil that persists for multiple years. The gradual thawing of permafrost and thickening of the active layer allows a glimpse into the evolution of the hydraulic processes that shape the periglacial landscape. One question in understanding the governing mechanics within the rapidly evolving periglacial landscape is how water retains within or segregates through the active layer to eventually feed rivers. 

In this exploratory study, we analyze data from multiple periglacial hydraulic catchments over time and characterize their hydraulic response rate to stressors. We test whether deconvolution and demixing of noisy time series can isolate precipitation from thawing permafrost signals in river discharge. We use the Ensemble Rainfall-Runoff (ERRA) script, which is effective in inferring nonstationary and nonlinear responses to precipitation using Runoff Response Distribution (RRD), to further test temperature signatures. Using this tool, we measure the RRD for the same catchments both over the years and over the summer months. We hypothesize that an increase in active layer thickness over years and over summer months will delay the RRD due to an increase in water storage.

By analyzing the parameters that change the RRD of periglacial systems with time, soil moisture content, average seasonal and yearly temperatures, and precipitation, we can begin a systematic understanding of how the active layer modulates hydraulic responses and how the responses may be different from other hydraulic systems.

How to cite: Culha, C. and Kirchner, J.: Characterizing melt water properties in the periglacial active layer through seasonal and yearly variations in catchment hydrology., EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-4291, https://doi.org/10.5194/egusphere-egu23-4291, 2023.

While being one of the most important variables for predicting the future of the ice sheets, observations of ice thickness are only available along flight tracks, separated by a few to a few tens of kilometres. For many applications, these observations need to be interpolated on grids at a much higher resolution than the actual average spacing between tracks.

The mass conservation method is an inverse method that combines the sparse ice thickness data with high resolution surface velocity observations to obtain a high-resolution map of ice thickness that conserves mass and minimizes the departure from observations.  As with any inverse method, the problem is ill-posed and requires some regularisation. The classical approach is to use a Tikhonov regularisation that penalizes the spatial derivatives of the ice thickness and therefore favours smooth solutions with implicit spatial correlation structures. In a Bayesian framework, regularization can be seen as an implicit assumption for the prior probability distribution of the inverted parameter. Other popular geostatistical interpolation algorithms, such as kriging, usually require to parameterize the spatial correlation of the interpolated field using standard correlation functions (e.g., gaussian, exponential, Matèrn).

Here we replace the Tikhonov regularisation term in the mass conservation method  by a term that penalises the departure from a prior, where the error statistics are parametrized with the same standard correlation functions. This makes the regularisation independent from grid spacing and regularisation weights do not need to be adjusted. We present and discuss the sensitivity of the mass conservation method to the regularisation scheme using a suite of synthetic and “true” bed from deglaciated areas and show that prescribing the correct regularisation always provides the most accurate solution.

How to cite: Gillet-Chaulet, F., Jager, E., and Morlighem, M.: Sensitivity of the mass conservation method to the regularisation scheme, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-5504, https://doi.org/10.5194/egusphere-egu23-5504, 2023.

How to cite: Eisen, O., Franke, S., Bons, P. D., Westhoff, J., Weikusat, I., Binder, T., Streng, K., Steinhage, D., Helm, V., Paden, J. D., Eagles, G., and Jansen, D.: Greenland ice-stream dynamics: short-lived and agile?, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6770, https://doi.org/10.5194/egusphere-egu23-6770, 2023.

The ice shelves buttressing the Antarctic ice sheet determine its stability. Over half of all mass loss in Antarctica occurs due to ice melting at the water-ice boundary at the base of ice shelves. Different contemporary methods of estimating the spatial distribution of the melting rates do not produce consistent results, and provide no information about decadal to centennial timescales. We explore a new method to infer the spatial distribution of the basal mass balance (BMB) using the internal stratigraphy which may contain additional information not present in other sources such as ice thickness and surface velocities alone. The method estimates the Bayesian posterior distribution of the BMB,  and provides us with a principled measure of uncertainty in our estimates. 

Our inference procedure is based on simulation-based inference (SBI) [1], a novel machine learning inference method. SBI utilizes artificial neural networks to approximate probability distributions which characterize those parameters that yield data-compatible simulations, without the need of an explicit likelihood function. We demonstrate the validity of our method on a synthetic ice shelf example, and then apply it to Ekström ice shelf, East Antarctica, where we have radar measurements of the internal stratigraphy. The inference procedure relies on a simulator of the dynamics of the ice shelves. For this we use the Shallow Shelf Approximation (SSA) implemented in the Python library Icepack [2], and a time-discretized layer tracing scheme [3].  These detailed simulations, along with available stratigraphic data and the SBI methodology, allows us to compute a spatially-varying posterior distribution of the melting rate. This distribution corroborates existing estimates and extends upon them by quantifying the uncertainty in our inference. This uncertainty should be incorporated in future forecasting of ice shelf dynamics and stability analysis.

[1] Lueckmann et al.: Benchmarking simulation-based inference (2020).

[2] Shapero et al.: icepack: a new glacier flow modeling package in Python, version 1.0. (2021).

[3] Born: Tracer transport in an isochronal ice-sheet model (2017).

How to cite: Moss, G., Višnjević, V., Schröder, C., Macke, J., and Drews, R.: Determining Basal Mass Balance of Ice Shelves Using Simulation-Based Inference, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6900, https://doi.org/10.5194/egusphere-egu23-6900, 2023.

RUC land surface model (LSM) was designed for short-range weather predictions with an emphasis on severe weather. The model has been operational at NCEP since 1998. Currently it is utilized in the operational WRF-based Rapid Refresh (RAP) and High-Resolution Rapid Refresh (HRRR) regional models. Being available to the world WRF community, RUC LSM is also used as a land-surface component in operational weather prediction models in Austria, New Zealand and Switzerland.

At present time, RUC LSM is being tested in the regional application of the UFS-based Rapid Refresh FV3 Standalone (RRFS) model to replace operational RAP and HRRR at NCEP.

RUC LSM has improved and matured over the years. The unique feature of this land-surface model is continuous evolution of soil/snow states within moderately coupled land data assimilation (MCLDA). To avoid possible drifts, this feature requires high skill from RUC LSM as well as accurate atmospheric forcing. Continuous snow cycling includes the following snow state variables: snow cover fraction, snow depth, snow water equivalent and snow temperature. To avoid possible inaccuracies in the location of cycled snow on the ground, snow depth is corrected daily using 4-km IMS snow cover information. Work is also underway to further improve RUC snow model for better surface predictions over snow-covered areas. RUC snow model uses “mosaic” approach to account for subgrid variability of snow cover. Within this approach, snow-covered and snow-free portions of the grid cells are treated separately in the solution of energy and moisture budgets. Thus, snow cover fraction becomes a critical parameter, and modifications to its computation have been developed and tested in the RRFS retrospective experiments. Results from these validation experiments will be presented at the meeting.

How to cite: Smirnova, T., Kliewer, A., He, S., and Benjamin, S.: Advancements in RUC Snow Model for Implementation in the Regional Application of the Unified Forecasting System (UFS), EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-8775, https://doi.org/10.5194/egusphere-egu23-8775, 2023.

The Greenland ice sheet mass loss is one of the main sources of contemporary sea-level rise. The mass loss is primarily caused by surface melt and the resulting runoff. During the melt season, the ice sheet’s surface receives energy from sunlight absorption and sensible heating, which subsequently heats the subsurface snow and ice. The energy from the previous melt season can also enhance melting in the following summer as less heating is required to bring the snow and ice to the melting point. Subsurface temperatures are therefore both a result and a driver of the timing and magnitude of surface melt on the ice sheet. We present a dataset of more than 3900 measurements of ice, snow and firn temperature at 10 m depth across the Greenland ice sheet spanning the years from 1912 to 2022. We construct an artificial neural network (ANN) model that takes as input the ERA5 reanalysis monthly near-surface air temperature and snowfall for the 1954-2022 period and train it on our compilation of observed 10-meter temperature. We use our dataset and the ANN to evaluate three broadly used regional climate models (RACMO, MAR and HIRHAM). Our ANN model provides an unprecedented and observation-based description of the recent warming of the ice sheet’s near-surface and our evaluation of the three climate models highlights future development for the models. Overall, these findings improve our understanding of the ice sheet’s response to recent atmospheric warming and will help reduce uncertainties of ice sheet surface mass balance estimates.

How to cite: Vandecrux, B., Fausto, R. S., Box, J. E., Covi, F., Hock, R., Rennermalm, A., Heilig, A., Abermann, J., Van As, D., Løkkegaard, A., Fettweis, X., Smeets, P. C. J. P., Kuipers Munneke, P., Van Den Broeke, M., Brils, M., Langen, P. L., Mottram, R., and Ahlstrøm, A. P.: Historical snow and ice temperature compilation documents the recent warming of the Greenland ice sheet, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9080, https://doi.org/10.5194/egusphere-egu23-9080, 2023.

Projections of the future development of the Antarctic Ice Sheet still exhibit a large degree of uncertainty due to difficulties in constraining parameters of ice-flow models such as basal boundary conditions. Deriving better estimates of these parameters from radargrams could greatly improve model accuracy, but integration of inferred radar attributes into ice-flow models is not yet widespread.

Here, we develop a radar forward modeling framework that is geared to train a machine learning workflow (likely simulation-based inference) to extract radar attributes such as the internal stratigraphy and basal boundary conditions (e.g., frozen vs. wet) from radar data. The workflow starts with ice-dynamic forward models predicting physically sound stratigraphies and internal/basal temperatures for synthetic flow settings using shallow ice, shallow shelf and higher order ice-flow models. This is then used as input to the radar simulator (here gprMax), which calculates the radar image produced by such a stratigraphy. To do so, we match the synthetic permittivities of the modeled stratigraphy with statistical properties known from ice-core logging data and prescribe temperature dependent attenuation via an Arrhenius relation. gprMax is optimized for acceleration using GPUs which can be efficiently employed when solving the FDTD discretized Maxwell equations. Currently, 200 m wide and 500 m deep sections can be simulated on a single NVIDIA GeForce RTX 2070 Super graphics card within 390 minutes. The runtime can be substantially improved in a HPC environment. In order to obtain radar simulations comparable with observations, we also add system specific noise and contributions from volume scattering with variable surface roughness. Here, we focus on 50 MHz pulse radar for which we have many observational counterparts. However, the workflow is designed to encompass multiple ice-dynamic settings and different radar frequencies.

The application of physical forward models will result in physically meaningful radargrams which are indistinguishable from observations. This provides a tool to create datasets for training machine learning workflows for inference without the limitations of hand-labeled data.

How to cite: Muhle, L. S., Moss, G., Henry, A. C. J., and Drews, R.: Radar forward modelling as a precursor for statistical inference, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12495, https://doi.org/10.5194/egusphere-egu23-12495, 2023.

Fast ice stream flow at speeds of hundreds to thousands of meters per year is sustained by sliding at the ice sheet base, whereas slow flow outside of ice streams is characterized by limited-to-no basal sliding. In this sense, the transition from no sliding to significant sliding exerts a key control on ice stream flow. The detailed physical processes that enable the onset of basal sliding are somewhat debated, but laboratory experiments, recent theoretical work, and a handful of direct observations support the notion of sliding initiating below the melting point as a result of regelation and premelting. 

In this contribution we describe a recently funded glacier-scale field experiment that has been designed to advance the understanding of sliding onset physics by testing the hypothesis that sliding starts below the melting point. The experiment will take place at the Grenzgletscher (Swiss Alps), which is known to have a cold-based accumulation region and a temperate-based ablation region. Our work will involve extensive surface geophysics (radio echo sounding, terrestrial radar interferometry, radar thermal tomography) aimed at identifying the sliding onset region. This work will guide the site selection for a subsequent borehole study of englacial deformation that is meant illuminate the relation between sliding velocity and basal temperature. The borehole work will allow us to test systematically the hypothesis that sliding starts below the melting point through an extended region of temperature-dependent sliding, and possibly to advance the formulation of temperature-dependent friction laws that are used to describe the onset of sliding in ice flow models.

The focus of this contribution will be specifically on the experimental design - how it is informed by existing theory and observations, and how it will support theoretical and ice flow modeling advances, at the glacier scale and beyond.

How to cite: Mantelli, E., Drews, R., Eisen, O., Farinotti, D., Luethi, M., Mingo, L., Schroeder, D., and Vieli, A.: At the bottom of ice streams: unraveling the physics of sliding onset through a glacier-scale field experiment, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12553, https://doi.org/10.5194/egusphere-egu23-12553, 2023.

Regional Climate Models (RCM) are the primary source of climate data available for impact studies over Antarctica. These climate-models experience significant, large-scale biases over Antarctica for variables such as snowfall, surface temperature and melt. Correcting for these biases is desirable for impact models being driven by meteorological data that aim to produce optimal estimates of for example surface run-off and ice discharge. Typical approaches to bias correction often neglect the handling of uncertainties in parameter estimates and don’t account for the different supports of climate-model and observed data. Here a fully Bayesian approach using latent Gaussian processes is proposed for bias correction, where parameter uncertainties are propagated through the model. Advantages of this approach are demonstrated by bias-correcting RCM output for near-surface air temperature over Antarctica.

How to cite: Carter, J., Chacón Montalván, E., and Leeson, A.: Bias correction of climate models using observations over Antarctica., EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-14292, https://doi.org/10.5194/egusphere-egu23-14292, 2023.

The stability of the Greenland ice sheet is poorly constrained even for benchmark periods such as the mid-Holocene or Eemian. Since ice stratigraphy holds a record of both surface mass balance (SMB) and ice dynamics, dated radiostratigraphy offer a potential route to improved reconstructions. Here we explicitly simulate isochrones and employ inverse methods to optimize the solution. The Englacial Layer Simulation Architecture (ELSA) coupled with a thermomechanical ice sheet model computes the isochrones or ice layers, which enable the direct comparison with the radiostratigraphy data. The accumulation rates force ELSA, and are adjusted until the model reproduces the observations within their uncertainties. We deploy the Ensemble Kalman Smoother for the data assimilation. This results in not only the reconstruction of the SMB; an optimized simulation of the ice sheet is obtained by solving the corresponding forward problem. Hence, the contribution to sea level change by Greenland over the same period can also be constrained.

Here we present our initial approach and preliminary results of SMB reconstruction. Future plans and expansions of the work are also presented, involving the study of several model parameters such as basal traction.

How to cite: Voigt, P. I. and Born, A.: Reconstructing accumulation rates of the Greenland ice sheet using dated radiostratigraphy, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-15374, https://doi.org/10.5194/egusphere-egu23-15374, 2023.

Whereas most of the continent of Antarctica is covered by snow, in some areas, blue-colored ice emerges to the surface. In these blue ice areas (BIAs), mass is removed at the surface through ablative processes. This mass removal exposes deeper layers of ice that are normally located closer to the underlying bedrock. As a result, we can find old ice at the surface of BIAs, as well as the material contained within the ice, such as meteorites and terrestrial rocks. BIAs are unique locations for sampling old ice for palaeoclimatic purposes and collecting meteorites (about ⅔ of all meteorites ever retrieved on Earth come from Antarctica BIAs). Hence, a high-quality BIA map is essential for meteorite searches, the quest for the oldest ice, and surface mass balance modeling.

Prior efforts to map BIAs across the Antarctic continent using remote sensing are single-sensor based, introducing biases related to temporary snow coverage of the exposed ice, and sensor-dependent conditions such as solar illumination angles, anisotropic reflectance, or cloud coverage. To overcome these challenges, we opt for using multi-sensor observations in a deep learning framework to create a new BIA map. The observations we use are (i) radar backscatter, (ii) surface morphology, (iii) elevation, and (iv) multi-spectral reflectance. The deep learning algorithm consists of the well-established convolutional neural network U-Net, which allows for an efficient training process and inclusion of spatial context. The algorithm outputs a pixel-level prediction of blue ice presence. Moreover, by training multiple, randomly initialized models and rotating and flipping data, we obtain multiple predictions for each pixel. Thanks to this data augmentation at test time, we estimate the variation in the predictions, which we then use as an indication of uncertainty. 

We use an existing dataset of BIA outlines as reference for training the model. It is known that these existing labels are noisy due to i) large uncertainties related to the use of a single sensor, and ii) biases as a result of applying a threshold that is based on local observations over the entire continent. However, convolutional neural networks, combined with regularization methods like weight decay and batch normalization, can learn from underlying ‘clean’ patterns of noisy labels during initial epochs of training (i.e., at the start of the training process). Here, we demonstrate this noise-eliminating property by assessing the algorithm's performance on noisy pixels that are used for training, where we see that over 80% of these noisy instances are attributed correctly. Furthermore, we optimize the performance of the neural network based on a reduced set of "noise-free", hand-labeled validation data. Last,  we test the performance of our model on hand-labeled test data, therefore having a realistic estimate of the model performance on precise, so far unused data. These tests indicate that it is possible for the neural net to learn how to map blue ice from the noisy data, leading to an improved map of BIAs in Antarctica.

How to cite: Tollenaar, V., Zekollari, H., Tuia, D., Rußwurm, M., Kellenberger, B., Lhermitte, S., and Pattyn, F.: A new blue ice area map of Antarctica, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-88, https://doi.org/10.5194/egusphere-egu23-88, 2023.

Advanced environmental technologies have made available an increasing amount of data from remote sensing satellites, and more sophisticated ground data. Their assimilation into dynamic models is progressively becoming the most frequent, and conceivably the most successful, solution to estimate snow water resources. Models reliability is therefore bounded to data quality, which is often low in mountain, high-elevation, and unattended settings. To add new value to snow-depth sensor measurements, we developed a machine-learning algorithm to automatize the QA/QC procedure of near-surface snow depth observations collected through ground stations data. Starting from a consolidated manual classification, based on the expert knowledge of hydrologists in Valle D'Aosta, a Random Forest classifier was developed to discriminate snow cover from grass or bare ground data and detect random errors (e.g., spikes). The model was trained and tested on Valle d’Aosta data and then validated on 3 years of data from 30 stations on the Italian territory. The F1 score was used as scoring metric, being it most suited to describe the performances of a model in case of a multiclass imbalanced classification problem. The model proved to be robust and reliable in the classification of snow cover and grass/bare ground discrimination (F1 values above 90%), yet less reliable in random error detection, mostly due to the dataset imbalance. No clear correlation with single year meteorology was found in the training domain, and the promising results from the generalization to a larger domain corroborates the model robustness and reliability.This machine learning application of data quality assessment provides more reliable snow ground data, enhancing the quality of snow models.

How to cite: Blandini, G., Avanzi, F., Gabellani, S., Ponziani, D., Stevenin, H., Ratto, S., and Ferraris, L.: A Random Forest approach to quality-chacking automatic snow-depth sensor measurements, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-180, https://doi.org/10.5194/egusphere-egu23-180, 2023.

The European Copernicus Sentinel-1 SAR mission offers a unique chance to compare and analyse long time series of freely accessible SAR images with frequent coverage in the northern polar areas. In our case, during the ExtremeEarth project (H2020 grant agreement No 825258), we concentrated on a two-year analysis of multi-season ice cover categories around Belgica Bank in Greenland where we can easily use typical examples of SAR image targets ranging from snow-covered ice to melting ice surfaces as well as open sea scenes with ships and icebergs.

Our primary goal was to search for most powerful ice type classification algorithms exploiting the well-known characteristics of the Sentinel-1 satellites for SAR imaging in polar areas, both taken from ascending and descending orbit branches with C-band transmission and an incidence angle of about 39°, a resulting ground sampling distance of 10 m or more, HH or HV polarization, and recorded in wide-swath or high-resolution modes as provided and distributed routinely by ESA´s level-1 processing system as amplitude or complex-valued data.

In order to be compatible with established international ice type standards we used the Canadian MANICE semantic labelling system providing up to 10 different polar ice and polar target types.

Our algorithms are based on a patch-based classification approach, where we assigned the most probable primary label for each given square image patch with a size of 256×256 pixels. This prevented us from creating many noise-related single-pixel categories.

Within the ExtremeEarth project, were generated semantic classification maps, topic representations, change maps, or physical scattering representations. A library of algorithms was created, among these algorithms we mention the following ones: classification based on Gabor filtering and SVMs, classification based on compression rates, variational auto-encoders for SAR feature learning, topic representations based on LDA, physical scattering representations based on LDA and CNNs, etc.

When the attempted image content classification based on current machine learning approaches, it turned out that we had to consider several important parameters such as typical applications, main semantic goals to be reached, applied processing algorithms, common types of data, available datasets and already predefined categories to be used, pixel-based versus patch-based data processing, single- and multi-labelling of image patches, confidence calculations and annotations, as well as attainable runtimes, implementation effort and risk - all depending on the target area characteristics. When it came to time series of target area images, we also had to consider the chances offered by short and long data sequences.

It turned out that this large number of aspects can be grouped together depending on the applied human expert supervision approach for semantic classification, namely unsupervised, self-supervised, semi-supervised, and supervised algorithms together with their individual training and testing strategies. In future, we want provide some justifications for next-generation remote sensing applications that require (near) real-time capabilities.

How to cite: Dumitru, O., Schwarz, G., Karmakar, C., and Datcu, M.: Polar Ice Coverage Classified by Various Machine Learning Algorithms, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-452, https://doi.org/10.5194/egusphere-egu23-452, 2023.

Climate change and the resulting accelerating melt on the Greenland and Antarctic ice sheets are causing dramatic and irreversible changes at a global scale, significantly contributing to sea-level rise. In this scenario, monitoring the evolution of diagenetic snow facies can provide valuable insights to better comprehend climate-related variables and trends. Previous studies of the Greenland ice sheet led to the definition of four main snow facies, depending on the amount of snow melt and on the properties of the snow pack itself: the inner dry snow zone, where melt does not occur; the percolation zone, where a limited amount of melt per year occurs, leading to the generation of larger snow grains and the formation of small ice structures; the wet snow zone, where a substantial part of the snow melt drains off during summer and is characterized by the presence of multiple ice layers; and the outer ablation zone, where the previous year accumulation completely melts during summer, resulting in a surface of bare ice and surface moraine. By exploiting X-band TanDEM-X interferometric synthetic aperture radar (InSAR) acquisitions, previous works explored the idea of classifying different snow facies of the Greenland ice sheet utilizing an unsupervised machine learning clustering approach. The analysis was performed using data acquired in winter 2010/2011 only, under the assumption of stable climatic conditions and similar acquisition geometries. In this paper, we further investigate the evolution of the snow facies of Greenland throughout the last decade of TanDEM-X observations, proposing unsupervised machine learning strategies for snow facies characterization by using InSAR features such as backscatter, volume decorrelation, the incidence angle and height of ambiguity. We use TanDEM-X data acquired during the winter of 2010/2011, 2015/2016, 2016/2017, 2020/2021, and 2021/2022, where full or partial coverage of the Greenland ice sheet is available. The challenges and caveats of such approaches for different image acquisition geometries will be presented. Finally, the potential of TanDEM-X for investigating large-scale interannual changes in the dry snow zone over Greenland will be investigated as well.  

How to cite: Becker Campos, A., Rizzoli, P., Gonzalez, C., Bueso-Bello, J.-L., and Braun, M.: The Evolution of the Snow Facies on the Greenland Ice Sheet Observed by the Last Decade of TanDEM-X Interferometric SAR Data, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-2096, https://doi.org/10.5194/egusphere-egu23-2096, 2023.

Monitoring glacier change processes leads to a better understanding of how glaciers respond to various external forcings. For sub-annual remote sensing, Synthetic Aperture Radar (SAR) imagery is indispensable, as polar night and cloud cover limit the temporal continuity of optical imagery. Especially with the launch of the Sentinel-1 mission, the availability of suitable SAR imagery has increased substantially. This high amount of data leads to another challenge: Manual inspection is no longer feasible. Therefore, in recent years, many studies applied deep learning to automate the segmentation of calving fronts in satellite imagery. To ensure comparability between different deep learning models, they must be trained and tested on the same data with a predefined test set and evaluated with the same metrics. A dataset intended to be used for this purpose is called a benchmark dataset and needs to provide the ready-to-use satellite imagery and the corresponding calving front labels. Gourmelon et al. [1] provide such a dataset for SAR imagery of calving fronts called CaFFe (https://doi.pangaea.de/10.1594/PANGAEA.940950). CaFFe includes multi-mission data (ERS-1/2, RADARSAT 1, Envisat, ALOS, Sentinel-1A/B, TerraSAR-X, and TanDEM-X), providing a spatial resolution between 6 and 20 meters and covering the period from 1996 to 2020. It contains images of seven glaciers from Antarctica to Greenland and Alaska. For each of the 681 images contained in the benchmark, two labels are provided: One displaying the calving front versus background and the other showing different zones (ocean, rock outcrops, glacier area, and no information available). CaFFe is split into a train set and a predefined out-of-sample test set, which comprises all images from two of the seven glaciers. A split of the train set into training and validation is not specified, as different approaches like cross-validation shall be possible. The benchmark covers a wide variety of different conditions to capture the variability of SAR calving front images. For example, images with open oceans and images with ice-melange-covered oceans are included in the dataset. Especially including images featuring ice-melange is of great importance, as deep learning models have shown difficulties in accurately segmenting calving fronts under this condition. Including images with ice-melange in the train set helps models to learn accurate predictions even under these circumstances.  Adding such images to the test set ensures that evaluated models are able to cope with ice melange. The test set of CaFFe is specifically designed to be challenging, such that the generalizability of models to different conditions and spatial transferability even to other continents can be verified. Gourmelon et al. provide baselines (one for each of the available labels) complementing the benchmark dataset. Current collaborative work aims to evaluate recently published deep learning techniques for calving front extraction on CaFFe and compare it with the baselines.

[1] N. Gourmelon, T. Seehaus, M. Braun, A. Maier, and V. Christlein: "Calving Fronts and Where to Find Them: A Benchmark Dataset and Methodology for Automatic Glacier Calving Front Extraction from SAR Imagery," Earth System Science Data, vol. 14, no. 9, pp. 4287-4313, 2022, doi: 10.5194/essd-14-4287-2022.

How to cite: Gourmelon, N., Seehaus, T., Braun, M., Maier, A., and Christlein, V.: Comparability of Deep Learning Techniques for Calving Front Segmentation in SAR Imagery, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-5069, https://doi.org/10.5194/egusphere-egu23-5069, 2023.

How to cite: Bolibar, J., Sapienza, F., Maussion, F., Lguensat, R., Pérez, F., and Wouters, B.: Functional Inversion of Glacier Rheology from Ice Velocities using ODINN.jl, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-5137, https://doi.org/10.5194/egusphere-egu23-5137, 2023.

The Arctic has been warming four times faster than the global mean over the last forty years. In response, glaciers across the Arctic have been retreating and losing mass at accelerated rates in recent years, including Greenland, Alaska, Canadian Arctic, Iceland, Svalbard and Russian Arctic. To predict their evolution with confidence, it is important to understand the mechanisms driving mass loss across the Arctic, especially the interconnected relationships between glacier retreat, ice dynamics, and mass imbalance. Over the past several decades, satellite remote sensing has been used to image glaciers over large spatial scales and at high temporal resolution. The volume of data produced, however, has challenged the traditional manual-based approaches to quantify glacier calving dynamics at a sub-annual scale across the whole Arctic. To address this limitation, we use a fully automated deep learning approach to generate a new calving front dataset for pan-Arctic glaciers at a high temporal resolution, by harmonizing multiple satellite missions that are available from the 1970s onwards, including optical missions such as Landsat, ASTER and Sentinel-2, and various SAR missions such as ERS-1/2, Envisat, RADARSAT-1, TerraSAR-X and Sentinel-1. We first present a new training dataset for the Arctic glaciers. We then present a new deep-learning framework for mapping the pan-Arctic glacier calving fronts. We show the interannual and seasonal variability of glacier termini positions by applying this method at scale and investigate the responses of Arctic glaciers to climate change.

How to cite: Li, T., Heidler, K., Mou, L., Igneczi, A., Zhu, X., and Bamber, J.: Towards pan-Arctic glacier calving front variability with deep learning, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-5656, https://doi.org/10.5194/egusphere-egu23-5656, 2023.

Accurate estimates of iceberg populations, disintegration rates and iceberg movements are essential to understand ice sheet contributions to global sea level change and freshwater and heat balances. Knowledge and prediction of iceberg distributions is also important for the safety and efficiency of shipping operations in polar seas. The dynamics, persistence, fragmentation rates, melt rates and dispersal of icebergs are, however, poorly understood due to a lack of automated approaches for monitoring them. Better monitoring of icebergs would help parameterise the locations and quantities of freshwater and nutrient inputs within hydrographic and ecological models respectively and help mitigate collision hazards for navigation.

Here we present a combination of Bayesian approaches to the identification of icebergs in synthetic aperture radar imagery and their subsequent tracking across multiple years. For detection we use a Dirichlet Process Mixture Model, while for tracking we adapt Bayesian Tracker, a probabilistic multi-object tracking algorithm originally developed for cell microscopy applications. We are able to reconstruct iceberg paths and lineages, which we validate against synthetic data and manual annotations. We demonstrate that icebergs across the size distribution can be tracked successfully from their point of calving in dense fields of objects, through dispersal and fragmentation, to distal locations.

How to cite: Evans, B., Hosking, S., Fleming, A., and Lowe, A.: Probabilistic detection and tracking of icebergs in the Amundsen Sea embayment, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6495, https://doi.org/10.5194/egusphere-egu23-6495, 2023.

Snow cover can be measured from multi-spectral optical images through the Normalized Difference Snow Index (NDSI). However, cloud cover affects the acquisition frequency. Radar snow detection offers independence from weather conditions. Despite existing method to detect snow melt [1] or measure snow depth [2], no existing method offers the possibility to detect any type of snow using only SAR images. To assess the different evolutions of the SAR signal during a winter season, we use a deep learning auto-encoding approach for the 2018-2019 winter over the Guil Basin in the French Hautes-Alpes using SLC Sentinel-1 images for three relative orbits: D66 (Descending), A88 (Ascending) and D139 (Descending). The images were geocoded using the French IGN DEM (BDALTI). On top of displaying the most representative temporal SAR signal profils on this area, this study help us to assess the spatial stationnarity of the SAR signal.

All the temporal profils were auto-encoded in three embedding following the framework detailed in [3]. The network was trained five times for each orbit. The chosen embeddings were the ones exhibiting the smaller correlation, leading to absolute value correlation between 0.10 and 0.20. The correlation between these embedding and the geographical features (latitude, longitude, altitude and incidence angle) is also below 0.30. 

Then these embbedings were used to group the pixels in six clusters using a kmeans framework. The mean temporal profile was estimated for each cluster, as well as the histogram of the elevation distribution. Behaviours appear consistently for the three orbits. One cluster correspond to shadow or dark areas pixels, with a constitent low backscattering over the year and a spread elevation histogram. Another cluster correspond to pixels in high altitude areas which exhibit an increase in backscattering between October and March that we attribute to snow fall. The third  cluster includes also high altitude pixels with a short drop of backscattering in October and May, certainly related to snow melt. The spatial pattern of the clusters for the A88 orbit shows a east-west shift in the class repartition while for the D66 and D139 the class repartition is more impacted by the altitude and follow the southward mountain arc. In further work, a  train/validation/test dataset with no dataset shift will be design using the stationnarity of this cluster, as well as a second test set introducing a geographical dataset shift that can take in account both the ascending/descending differences and the topographic and climatic variation of snow cover.  

[1] T. Nagler, et al.  "Advancements for Snowmelt Monitoring by Means of Sentinel-1 SAR". Remote Sens. 2016. https://doi.org/10.3390/rs8040348 
[2] H. Lievens, et al. "Snow depth variability in the Northern Hemisphere mountains observed from space". Nat Commun 2019. https://doi.org/10.1038/s41467-019-12566-y
[3] T. Di Martino et al. "Beets or Cotton? Blind Extraction of Fine Agricultural Classes Using a Convolutional Autoencoder Applied to Temporal SAR Signatures".IEEE TGRS 2022, 10.1109/TGRS.2021.3100637.

Acknoledgment: This work is part of the AI4GEO project. The authors would like to thank Thomas Di Martino for his precious advices.

How to cite: Weissgerber, F. and Montginoux, M.: Temporal and spatial stationnarity of the snow regime assessed through deep-learning auto-encodding of SAR-image stacks, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6858, https://doi.org/10.5194/egusphere-egu23-6858, 2023.

The emulator used for projections of the land ice contribution to sea level by 2100 in the Intergovernmental Panel on Climate Change Sixth Assessment Report (Edwards et al., 2021) used a novel approach for incorporating structural uncertainty from the underlying computer models. This statistical (Gaussian Process) emulator represented entire multi-model ensembles at once -- for the Greenland and Antarctic ice sheets, and the world’s glaciers, under the model intercomparison projects ISMIP6 and GlacierMIP, respectively -- by using a noise (nugget) term to allow for multiple estimates of sea level contribution for a given set of model input values, analogous to kriging of spatial data.

However, the emulator was rather simple in other respects: in particular, the sea level projection for each year from 2015-2100, and from each region of land ice (splitting Antarctica into 3 regions, and the glaciers into 19), were modelled independently, so temporal correlations emerged only upon smoothing the projections. The emulator was also not formally evaluated with observations, because the underlying simulations were only driven with meaningful climate forcings from 2015. These limitations have presented difficulties for users, who often need continuous time series projections and prefer, of course, these to be assessed with observations.

Here the IPCC land ice emulator is improved for interpretation and use by decision-makers by estimating spatio-temporal correlations directly from the underlying simulations (Rougier, 2008; Rougier et al., 2009), to produce meaningful trajectories of sea level contribution from each land ice source. The extent to which the land ice emulator can be evaluated with data, now and in future, is also discussed.

References:

Edwards et al. (2021) Projected land ice contributions to twenty-first-century sea level rise, Nature, 593, 74–82.

Rougier, J. (2008), Efficient Emulators for Multivariate Deterministic Functions, Journal of Computational and Graphical Statistics, 17(4):827–843.

Rougier, J.C. et al. (2009), Expert Knowledge and Multivariate Emulation: The Thermosphere-Ionosphere Electrodynamics General Circulation Model (TIE-GCM), Technometrics, 51(4), 414-424.

How to cite: Edwards, T., Rougier, J., and Turner, F.: Improving and evaluating the IPCC land ice emulator, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6961, https://doi.org/10.5194/egusphere-egu23-6961, 2023.

In order to improve the forecasting of hydraulic production, EDF uses optical satellite images to evaluate the snow cover [1]. These images are acquired daily by the MODIS instrument of the Terra satellite and provide a snow product through the Normalized Difference Snow Index (NDSI). However, part of the information on the snow cover is lost due to clouds. To complete those gaps, radar satellite images can be interesting because it does not depend on weather conditions.

Dry snow and wet snow have different SAR signature. Wet snow can be detected since its backscatter decreases [2]. Dry snow detection is more challenging. It may be performed with a polarimetric approach [3], and the snow depth (SD) can be estimated using optical images as auxiliary inputs [4]. In this work, wet snow was detected and SD was estimated over the Guil basin in the Alps (420 km²) for the years 2018 and 2019 on three relative orbits of Sentinel-1: the D66 (descending, 87 images), A88 (ascending, 119 images), and D139 (descending, 90 images). The results show an accumulation of snow in autumn on the SD and a peak of snowmelt in spring on the detection of wet snow.

Then we propose to detect the snow from SAR images using a convolutional neural network trained with optical images from MODIS as labels. For the dataset, a smaller area is chosen around Abriès (of approximately 59km²) and we select 36 images for each of the three orbits to study the winter 2018-2019. A binary semantic segmentation is computed from two SAR inputs: R_wet from [2], and R_dry a polarimetric ratio inspired from [3]. The trained model, called SESAR U-net, gives a snow detection with an overall accuracy of 80% for our test set. This low accuracy result can be explained by the fact that MODIS images have a resolution 25 to 100 times coarser than the SAR images, which hinder both the training and the evaluation of the model. Further works will consider the uncertainty of the MODIS label in the loss computation to improve the training.

[1] M. Le Lay et al., “Use of snow data in a hydrological distributed model: different approaches for improving model realism,” in EGU General Assembly Conference Abstracts, EGU General Assembly Conference Abstracts, p. 14545, Apr. 2018.
[2] T. Nagler et al., “Advancements for snowmelt monitoring by means of sentinel-1 sar,” Remote Sensing, vol. 8, 04 2016.
[3] A. Reppucci et al., “Estimation of snow-pack characteristics by means of polarimetric SAR data,” in Remote Sensing for Agriculture, Ecosystems, and Hydrology XIV (C. M. U. Neale and A. Maltese, eds.), vol. 8531, p. 85310Z, International Society for Optics and Photonics, SPIE, 2012.
[4] H. Lievens et al., “Snow depth variability in the Northern Hemisphere mountains observed from space,” Nature Communications, vol. 10, p. 4629, Dec. 2019.

How to cite: Montginoux, M., Weissgerber, F., Monteil, C., and Girard, A.: Snow cover estimation by deep-learning segmentic segmentation of radar images based on optical image references, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-7240, https://doi.org/10.5194/egusphere-egu23-7240, 2023.

Accurate estimation of ice thickness is essential for understanding and predicting the behaviors of glaciers, since the ice thickness provides valuable information about the glacier state and helps anticipate the evolution of whole glaciers systems. This latter knowledge is particularly important given the fact that glaciers act as natural reservoirs in the global water cycle. They are also indicators of the current and past state of global and local climate, as the properties and dynamics of glaciers reveal insights into the environmental conditions they have experienced.

Current methods for estimating ice thickness face issues : field measurements such as the Ground Penetrating Range method can be costly and may not provide dense, continuous, renewable coverage. On the other hand, methods based on physical modeling can be computationally intensive and are dependent on assumptions about model parameters that may be unreliable in a poorly understood context. Results may thus be sensitive to the choice of prior information and prone to bias when working with limited or noisy data.

Deep learning methods provide a promising solution for ice thickness prediction. One key advantage is their ability to handle large, multi-dimensional datasets and to learn directly from raw data without prior knowledge. Additionally, deep learning models are able to exploit non-linear relationships between datasets. Using such models also allows simultaneous training of other tasks, such as terrain classification to identify the presence of glaciers when it is not provided or outdated. In this study, we propose the use of such an approach relying on convolutional models based on VGGNet, ResNet and U-Net.

Our goal is to obtain an accurate estimation of the glacier thickness distribution. We propose the use of neural networks in order to 1) be free from statistical/physical assumptions, 2) leverage deep relationships between observed data and physical parameters to be estimated, 3) overcome inaccuracies in collected data, and 4) accurately represent complex patterns such as non-linear thickness variations within the glacier. Additionally, it is important that these models should not be prone to common issues of deep learning such as overfitting and lack of explainability.

We conduct our study on Alpine glaciers in Switzerland. The input data for our neural network models includes: 1) average ice velocity fields calculated from correlation of Sentinel-2 images with a resolution of 50 metres, and 2) altitudes and slopes derived from the Swiss digital elevation model with a resolution of 10 metres. To verify the accuracy of the predicted ice thickness values, we use ground truth data obtained from GPR surveys conducted in profile form, from 2012 to 2021.

In addition to estimating the ice thickness, we also perform direct classification of glaciers vs. non-glacier areas. Results demonstrate the feasibility of quickly training a neural network model with limited training data and producing stable, high-quality ice thickness estimates for different glaciers in the study region.

How to cite: Lopez Uroz, L., Yan, Y., Benoit, A., Rabatel, A., Dehecq, A., and Giffard-Roisin, S.: Improving ice thickness estimation of glaciers using deep learning methods : a case study in the Swiss Alps, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-8339, https://doi.org/10.5194/egusphere-egu23-8339, 2023.

ESA has developed a series of seven TEPs on different subjects to provide insight into how our oceans, atmosphere, land and ice operate and interact as part of an interconnected earth system by exploiting the unprecedented flow of high-quality global data on the state of our planet, combined with long-term EO archives, in-situ networks and models. The Polar Thematic Exploitation Platform (Polar TEP) was developed to address the particular needs of the polar community.

Polar TEP provides a complete working environment where users can access algorithms and data remotely to obtain computing resources and tools that they might not otherwise have and avoid the need to download and manage large volumes of data. This new approach removes the need to transfer large Earth Observation data sets around the world, while increasing the analytical power available to researchers and operational service providers. Polar TEP provides new ways to exploit EO and other large datasets for research scientists, industry, operational service providers, regional authorities, and policy analysts. Polar TEP provides:

• Data Discovery - Polar TEP makes satellite and other polar data easily accessible for browsing or analysis within the cloud or within the user’s own environment. The infrastructure takes care of the complexity of handling satellite imagery archives and makes the data available via web services. Users can instantly access petabytes of Sentinel, Landsat, and other Earth observation imagery, both historic and the latest acquisitions.

• Interactive Development Environment - Polar TEP offers a managed JupyterLab instance with curated base images. The platform provides different flavors of computational resources and a network file system for persistent data storage. Headless notebook execution is supported.

• Machine Learning - Polar TEP has implemented the MLflow platform to support machine learning activities. MLflow manages all stages of the machine learning lifecycle, including experimentation, reproducibility, deployment, and a central model registry.

• Execution Environment - Docker containers are used to provide processors with a separate custom environment having minimal execution overhead. The computing resources used by the execution environment are scaled to the current demand.

• Application Hosting Environment - Users can host their own applications on a VM within the Polar TEP environment.

• Story Telling - Polar TEP provides tools to communicate analysis results to other researchers or the public.

Polar TEP is an integral part of the wider polar data ecosystem, contributing to data interoperability and fostering the use of information about the polar regions to support environmental protection, safety, and sustainable economic development.

This presentation will illustrate how the power of Polar TEP to process massive amounts of data is being applied to topics such as machine learning for operational sea ice charting, daily calculations of Greenland ice sheet albedo, and providing information to support traditional ways of life in the Arctic.

How to cite: Arthurs, D.: Polar TEP – A Platform for Polar Big Data Analytics and Machine Learning, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-8706, https://doi.org/10.5194/egusphere-egu23-8706, 2023.

While a majority of mass loss from the Greenland Ice Shelf is attributed to glacial terminus retreat via calving, the superimposed force factors of the ice-ocean interface create a challenge for physically modeling terminus change. Here we use time series of environmental and glacial data, input as features into a machine learning regression model, to forecast terminus retreat for marine-terminating glaciers in Greenland. We then identify the critical features that most impact a glacier’s likelihood of retreat using feature importance analysis. We further analyze the heterogeneous outcomes for individual glaciers to classify them by their terminus change profile.  By better understanding the parameters impacting glacial retreat, we inform physical models to reduce uncertainty in mass change projections.

How to cite: Shionalyn, K., Catania, G., Trugman, D., Felikson, D., and Stearns, L.: Predicting Glacier Terminus Retreat Using Machine Learning, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9122, https://doi.org/10.5194/egusphere-egu23-9122, 2023.

Polar ice sheets Greenland and Antarctica are integral parts of the climate system. Understanding their history, dynamics and past accumulation rates determines projections of sea level change. Ice englacial stratigraphy is used to assign ages, taken from ice cores, to radar reflections and subsequently connect these known layers over large areas. One of the main methods to investigate these characteristics is radar reflections. Ground-penetrating radar (GPR) has been used as the primary technique to detect internal ice architecture.

Mapping the internal reflection horizons in order to study and investigate the features, accumulation rates, and ice streams is an important step, which is conventionally done through a semi-automatic process. Such methods are prone to shortcomings in terms of continuity and layer geometry. Moreover, it is highly time-consuming to map an entire profile, the abundance of unmapped radar profiles especially from antarctic ice sheet is an evidence for this. Thus, there is the need for more comprehensive and efficient methods.

The use of machine learning to perform this task automatically will make a significant difference for internal layer detection in terms of efficiency and accuracy. Such machine and deep learning methods would be a suitable fit for radar surveys with different properties, such as center frequencies, making them appropriate for both ice, firn and snow data. In this project, apart from classical computer vision methods and image processing, deep learning methods are used to map the internal reflection horizons (IRH). Convolutional Neural Networks (CNN) are a powerful tool to learn features and track the IRHs continuously.

In this talk, the implemented classical computer vision methods are enumerated, and the machine learning methods that have been used (the specific pre-processing methods unique to this project, labeling method, architecture and hyperparameters) are explained. The results from some more promising architectures such as U-net are presented and compared to the results from image processing methods. The main challenges in this project are lack of complete training data, unknown number of IRHs in a profile, and abundance of features in a single radargram. These challenges are shown and possible solutions are presented.

How to cite: Moqadam, H., Steinhage, D., Eisen, O., and Wilhelm, A.: Tracing Extended Internal Stratigraphy in Ice Sheets using Computer Vision Approaches, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9532, https://doi.org/10.5194/egusphere-egu23-9532, 2023.

There is growing evidence that ice mélange, the granular mixture of sea ice and icebergs at the termini of tidewater glaciers, impacts ice sheet discharge through physical buttressing forces and alterations to fjord circulation via iceberg melting. However, ice mélange is a highly dynamic, fragmented and mobile phenomenon which varies over a range of timescales (e.g. hours, days, weeks) and hence is difficult to monitor using traditional ground-based and spaceborne sensors. In this contribution, we utilise high spatio-temporal satellite imagery acquired from the ICEYE small satellite constellation to assess correlations between ice mélange characteristics and tidewater glacier dynamics. ICEYE is a growing constellation of 20+ small satellites each equipped with an X-band Synthetic Aperture Radar (SAR) and capable of mapping the entire globe at least once a day with fine spatial resolution (1-3 m). We utilised the ICEYE SAR imagery to study the perennial mélange matrix at the terminus of Helheim Glacier in southeast Greenland. ICEYE SAR imagery was acquired during summer and winter to assess how seasonal ice mélange conditions impact tidewater glacier dynamics. Sentinel-1 SAR imagery and ground-based TLS 3D data from two autonomous terrestrial laser scanners (ATLAS) were used to validate remote sensing analysis and provide additional data sources for interpretation of the glaciological processes. We will report on the following: (1) a spatial texture analysis (e.g. Grey Level Co-occurrence Matrix (GLCM), Gabor Transforms) of ice mélange at the terminus of Helheim Glacier using high resolution ICEYE SAR imagery; (2) results of hierarchical and random forest classifiers to map icebergs, sea ice and open water within the ice mélange matrix; (3) quantification of glacier and mélange flow variability at daily to weekly timescales; and (4) the development of observational models correlating ice mélange texture, iceberg distributions, mélange/glacier flow rates, and tidewater glacier stability. Our case study at Helheim Glacier aims to demonstrate a new approach to rapidly monitor ice mélange conditions and tidewater glacier stability using high resolution SAR imagery. In particular, this study pushes forward our Earth Observation capabilities and will help us better understand the complex processes operating at the ice-ocean interface which is critical for improved predictions of how the Greenland Ice Sheet will evolve under a warming climate.

How to cite: Harcourt, W. D., Stearns, L. A., Shahin, M. G., and Shankar, S.: Assessment of ice mélange impacts on tidewater glacier dynamics using high resolution ICEYE imagery, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9812, https://doi.org/10.5194/egusphere-egu23-9812, 2023.

Sea ice plays a vital role in Earth’s human-climate system. It regulates the Earth’s overall energy balance by seasonally increasing surface albedo and reflecting solar radiation; it governs thermodynamic exchanges between the ocean and atmosphere and thereby impacts mid-latitude weather patterns; it buttresses key continental ice shelves in Greenland and Antarctica; it provides an ecosystem in which land, marine, and airborne species thrive; it enables the livelihoods of indigenous populations across the Arctic; it poses a major obstacle to global shipping logistics; and it serves as a key indicator of climate change given the sensitivity of the polar regions to anthropogenically-induced warming. Regular and automated monitoring of sea ice concentration and type may therefore prove valuable to a broad and diverse set of parties. Conventional approaches in sea ice monitoring involve the use of remotely sensed microwave radiometer data with low resolution of 6-25 km and high instrumental sensitivities to environmental factors such as atmospheric water vapour, near-surface brightness temperature, and wind-induced surface roughening. Dual-polarity synthetic aperture radar (SAR) imagery offers a higher resolution alternative, which can also distinguish between sea ice and open water year-round independent of weather conditions. However, manual interpretation of such imagery is time-consuming. In this work, we develop a deep learning system to automatically generate high-resolution maps of sea ice concentration and type using 40m-resolution SAR imagery obtained from the Sentinel-1 mission between 2017 and 2021. Focusing on the East Weddell Sea, a region where compacted sea ice is renowned for inhibiting ship navigation and an active area of iceberg calving, we train the system against reference sea ice charts produced through manual interpretation by experts. We identify strengths and weaknesses of the system and discuss implications for future research at the intersection of machine learning and polar science.

How to cite: McDonald, A., Dimasaka, J., Plumridge, M., Torry, J., Zúñiga González, A. C., van Zeeland, L., Rogers, M., and Hosking, S.: Classifying sea ice in high-resolution SAR imagery using deep learning, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9816, https://doi.org/10.5194/egusphere-egu23-9816, 2023.

Atmospheric errors in interferometric synthetic aperture radar (InSAR)-derived estimates of surface deformation often obscure real signals, especially in mountainous terrain. By taking advantage of the differing spatial characteristics of periglacial landforms and atmospheric noise, we trained a deep convolutional neural network (CNN) to remove atmospheric noise from individual interferograms. Unlike existing corrections, which rely on coarse climate reanalysis or radiometer data, this computer vision correction is applied at the spatial and temporal resolution of the interferogram. We processed Sentinel 1 interferograms of the Colorado Rocky Mountains using the Alaska Satellite Facility's Hybrid Pluggable Processing Pipeline (ASF HyP3) and used the Miami INsar Time-series software in PYthon (MintPy) package to generate low-noise line-of-sight (LOS) velocity maps containing primarily rock glacier and hillslope motion. These maps were combined with noisy short temporal-baseline interferograms to contrive a training dataset. Model performance was assessed using the structural similarity index measure (SSIM) and compared to that of other widely used corrections. We find that our CNN significantly outperforms standard corrections and that previously hidden intraseasonal kinematic behavior is revealed in Colorado rock glaciers. We suggest that insights from external validation against GNSS data and sensitivity analysis could be used to further improve model performance and assess model scalability and transferability. 

How to cite: Brencher, G., Henderson, S., and Shean, D.: Removing Atmospheric Noise from Interferograms in Mountainous Regions with a Deep Convolutional Neural Network, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10088, https://doi.org/10.5194/egusphere-egu23-10088, 2023.

Global mean sea level rise has been accelerating significantly over the past decades, a substantial part of which is attributed to increased surface melting from the Greenland ice sheet (GrIS). Climate models project the GrIS to contribute 9-18 cm to global mean sea level rise by 2100 for the Shared Socioeconomic Pathway SSP5-8.5. The significant uncertainty in this projection prevents accurate mitigation of the effects of sea level rise. The uncertainty stems from a not-comprehensive understanding of the physical processes controlling surface melting. In particular, we lack understanding of ice albedo evolution/variability, a crucial factor in surface melt processes. Ice albedo is a complex and highly variable property of the ice surface that is not well represented in climate model projections, leading to imprecise predictions of sea level rise. The high complexity and number of drivers and feedbacks responsible for ice albedo variability prevent us from building a comprehensive predictive ice albedo model that accurately incorporates all these processes.

From this point of view, we adopt a machine learning-based approach to predict ice albedo variability on the GrIS. We use daily regional climate model output of atmospheric, radiative, and glaciological variables from the Modèle Atmosphérique Régional (MAR) as input data and daily broadband albedo data from the Moderate Resolution Imaging Spectroradiometer (MODIS) as output data. From these data, we construct a Convolutional Long Short-Term Memory (CNN-LSTM) network that models daily ice albedo variability on 6.5 km spatial resolution. A CNN is a neural network that works particularly well for extracting patterns from spatial data. An LSTM is a special kind of recurrent neural network (RNN) that is well-suited for finding patterns and trends in temporal data on a much longer time scale than classic RNNs. Preliminary results show a significant improvement of the correlation between observed and simulated bare ice albedo, with the CNN-LSTM outperforming MAR. Besides the predictive ability of this physically-based machine learning ice albedo model and its suitability for implementation in climate models, it also allows us to gain understanding of what variables drive ice albedo variability on the GrIS, now and in the future.

How to cite: Antwerpen, R., Tedesco, M., Gentine, P., Alexander, P., and van de Berg, W. J.: Predicting Greenland Ice Albedo Using A Physically-Based Convolutional Long Short-Term Memory Network, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10377, https://doi.org/10.5194/egusphere-egu23-10377, 2023.

Surface melt occurs on most ice shelves in Antarctica each summer, with potential impacts on their strength and stability and thus on the ice sheet's contribution to global sea level rise. However, many questions remain regarding the spatiotemporal variability of surface melt and the processes driving it, particularly in East Antarctica where few in situ observations exist. Previous work in this field has largely relied on remote sensing observations to monitor the occurrence and extent of surface melt, often using metrics such as the onset and freeze-up dates of melt each summer, the number of melt days, or the cumulative melting area. Whilst such metrics are often necessary to handle the sheer volume of data produced by satellite observations, much of the information contained within the datasets is lost, hindering attempts to build a more complete picture of melt variability at different spatial and temporal scales, and thus of disentangling the different processes driving melt.

To help address this problem, we use the machine learning approach of a Self-Organising Map (SOM) and nearly two decades (2002/03–2020/21) of daily observations from the AMSR-E and AMSR-2 passive microwave sensors, gridded at a spatial resolution of 12.5 km. Here, we present results focused on the Shackleton Ice Shelf in East Antarctica, but our code, implemented in the R programming language, is openly available and can be applied to any Antarctic ice shelf, or adapted for use with other melt datasets.

Our results show that the daily distribution of surface melt on the Shackleton Ice Shelf can be described by nine representative spatial patterns of melt. These patterns demonstrate the potential for heterogeneous melt behaviour across the shelf, and thus provide insight into the influence of surface topography, katabatic winds, and surface albedo in driving surface melt. A sensitivity analysis of the SOM algorithm shows that the same general spatial patterns are returned repeatedly regardless of the parameter values used, strengthening confidence in our results and interpretation, and demonstrating the suitability of our approach. We further examine the temporal variability of the nine melt patterns, both within and across melt seasons, finding that there are no significant trends in any of the patterns. Instead, our analysis identifies a number of summers with unusual melt behaviour and also reveals correlations with shelf-wide, summer-averaged surface air temperatures, highlighting that both local and large-scale controls are important for driving surface melt in Antarctica.

How to cite: Saunderson, D., Mackintosh, A., McCormack, F., Jones, R. S., and Picard, G.: Self-organising maps and surface melt on East Antarctic ice shelves, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10430, https://doi.org/10.5194/egusphere-egu23-10430, 2023.

Safety related applications like avalanche warning or risk management depend on timely information on avalanche occurrences. Today, this is gathered in a non-systematic way by observers in the field, even though remote sensing already proved capable of providing spatially continuous information on avalanche occurrences over large regions. Satellite imagery has the big advantage of large coverage, however the information is available only on selected dates. Depending on the application, a better temporal resolution is necessary. Webcams are ubiquitous and capture numerous avalanche prone slopes several times a day. The cameras mounted in a stable position may even be georeferenced to allow for an exact transfer of the location from the image to a map. To complement the knowledge about avalanche occurrences with more precise release time information, we propose making use of this webcam imagery for avalanche mapping.

For humans, avalanches are relatively easy to identify in imagery, but the manual mapping of their outlines is cumbersome and time intensive. To counter this, we propose automating the process with deep learning. Relying on interactive object segmentation we want to extract the avalanche outlines from those images in a time efficient manner with feedback from human experts (in the form of few corrective clicks on an image). We test existing models, searching for the best fit for avalanche outline segmentation. By adapting the best model where necessary we are aiming for outlines of good quality with a low number of clicks. For imagery we rely on current and archive data from our 14 webcams covering the Dischma valley near Davos, Switzerland with imagery available every 30 minutes during the day. Since the images are georeferenced, we may import identified avalanches directly into designated databases and therefore make them available for the relevant stakeholders.

On a more long-term perspective, the resulting avalanche outlines will enlarge the webcam training, test and validation dataset and consequently help to fully automate the avalanche outline identification from webcam imagery with object segmentation.

How to cite: Hafner, E., Oberson, L., Kontogianni, T., Daudt, R. C., Wegner, J. D., Schindler, K., and Bühler, Y.: Using interactive object segmentation to derive avalanche outlines from webcam imagery, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10867, https://doi.org/10.5194/egusphere-egu23-10867, 2023.

Greenland and other polar regions are highly sensitive to global warming. The impact of climate change on Greenland's glaciers can be seen through an increase in calving events. To better understand this impact, it is important to quantify and document calving activity. However, direct observations are difficult to perform repeatedly over long periods of time due to the hostile climatic conditions, the lack of human witnesses and of the possibility to install in-situ sensors in these remote areas.

With the installation of a regional seismological network in Greenland in the 2000’ ,and the densification of the one in north-eastern Canada, seismic signals caused by large volume calving events, known as glacial earthquakes, were recorded at distances of hundreds of km from the source. These signals have a wide range of frequencies, making it hard to distinguish them from tectonic events, anthropogenic noise, and other natural noise. Using two catalogs of known events - one of 444 glacial earthquakes that occurred between 1993 and 2013, and one for 400 earthquakes that occurred during the same time period selected from USGC - we trained and tested a detection algorithm based on the STA/LTA method to extract event signals from continuous data. We then trained a supervised machine learning algorithm (Random Forest) to automatically classify these signals into two different classes : glacial earthquakes and earthquakes, with a probability of belonging to each class.

With a workflow designed to limit the false alarm rate based on the probability scores of each events, we finally analyzed over 800 days of data from the Greenland regional seismic network and identified almost 1500 new glacial earthquake events using the trained machine learning model. Our detection methods makes it possible to detect four times more ice-quake than the original catalogue.

How to cite: Pirot, E., Hibert, C., and Mangeney, A.: Seismological Monitoring of Calving Events in Greenland: A Machine Learning Approach, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-11891, https://doi.org/10.5194/egusphere-egu23-11891, 2023.

The Antarctic Ice Sheet is an important indicator of climate change and a major contributor to sea level rise. Hence, precise, long-term observations of surface elevation change are required to assess changes and their contribution to sea level rise. Satellite altimetry has been used by various missions to measure surface elevation change since 1992. It has been shown that, next to the surface slope and complex topography, one of the most challenging issues is the spatial and temporal variability of radar pulse penetration into the snowpack, especially over the vast East Antarctic plateau. This results in an inaccurate measurement of the true surface elevation and consequently affects surface elevation change (SEC) estimates.

To increase the accuracy and correct the SEC, we developed a deep convolutional neural network (CNN) architecture. The CNN was trained by a simulated waveform data set containing more than 3.6 million waveforms, considering different surface slopes, topography, and attenuation. The CNN follows standard architectural design choices. The successfully trained network is finally applied as a CNN-retracker to the full time series of CryoSat-2 low resolution mode (LRM) waveforms over the Antarctic ice sheet. We will show the CNN retrieved SEC and compare it to estimates of conventional retrackers like OCOG or ICE2. Our preliminary results show reduced uncertainty and a strongly reduced time variable radar penetration, making backscatter or leading edge corrections typically applied in SEC processing obsolete. This technique provides new opportunities to utilize convolutional neural networks in altimetry, waveform retracking, and processing altimetry data, which can be applied to historical, recent, and future altimetry missions.

How to cite: Dehghanpour, A., Helm, V., Humburt, A., Hänsch, R., and Horwath, M.: Waveform retracking based on a Convolutional Neural Network applied to Cryosat-2 altimeter data, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-11954, https://doi.org/10.5194/egusphere-egu23-11954, 2023.

How to cite: Turner, F., Edwards, T., and Rougier, J.: The use of multivariate Gaussian process emulation in making projections of land ice contributions to sea level rise, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12361, https://doi.org/10.5194/egusphere-egu23-12361, 2023.

Snow avalanches pose a significant danger to the population and infrastructure in the Austrian Alps. Although rigorous prevention and mitigation mechanisms are in place in Austria, accidents cannot be prevented, and victims are mourned every year. A comprehensive mapping of avalanches would be desirable to support the work of local avalanche commissions to improve future avalanche predictions. In recent years, mapping of avalanches from satellite images has been proven to be a promising and fast approach to monitor the avalanche activity. The Copernicus Sentinel-1 mission provides weather independent synthetic aperture radar data, free of charge since 2014, that has been shown to be suitable for avalanche mapping in a test region in Norway. Several recent approaches of avalanche detection make use of deep learning-based algorithms to improve the detection rate compared to conventional segmentation algorithms.

          Building upon the success of these deep learning-based approaches, we are setting up a modular data pipeline to map previous avalanche cycles in Sentinel-1 imagery in the Austrian Alps. As segmentation algorithm we make use of a common U-Net approach as a baseline and compare it to mapping results from an additional algorithm that has originally been applied to an autonomous driving problem. As a first test case, the extensive labelled training dataset of around 25 000 avalanche outlines from Switzerland will be used to train the U-Net; further test cases will include the training dataset of around 3 000 avalanches in Norway and around 800 avalanches in Greenland. To obtain training data of avalanches in Austria we tested an approach by manually mapping avalanches from Sentinel-2 satellite imagery and aerial photos.

          In a new approach, we will introduce high-resolution weather data, e.g., weather station data, to the learning-based algorithm to improve the detection performance. The avalanches detected with the algorithm will be quantitatively evaluated against held-out test sets and ground-truth data where available. Detection results in Austria will additionally be validated with in situ measurements from the MOLISENS lidar system and the RIEGL VZ-6000 laser scanner. Moreover, we will assess the possibilities of learning-based approaches in the context of avalanche forecasting.

How to cite: Kapper, K. L., Goelles, T., Muckenhuber, S., Trügler, A., Abermann, J., Schlager, B., Gaisberger, C., Grahn, J., Malnes, E., Prokop, A., and Schöner, W.: Next steps to a modular machine learning-based data pipeline for automated snow avalanche detection in the Austrian Alps, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12774, https://doi.org/10.5194/egusphere-egu23-12774, 2023.

A novel generation of sea-ice models with Elasto-Brittle rheologies can represent the drift and deformation of sea-ice with an unprecedented resolution and accuracy. To speed-up these computationally heavy simulations and to facilitate subgrid-scale parameterizations, we investigate supervised deep learning techniques for surrogate modelling of large-scale, Arctic-wide, neXtSIM Lagrangian simulations. We tailor convolutional neural networks to emulate the sea-ice thickness for 12 hours in advance. In our most successful approach, the U-Net learns to make beneficially use of information from multiple temporal and spatial scales, an important feature of the neural network for sea-ice prediction. Consequently, cycling the neural network performs in average 36% better than persistence on a daily timescale and up to 43 % on a monthly timescale. These promising results therefore demonstrate a way towards surrogate modelling of Arctic-wide simulations. 

How to cite: Durand, C., Finn, T., Farchi, A., Bocquet, M., and Olason, E.: Deep learning for surrogate modelling of neXtSIM, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12810, https://doi.org/10.5194/egusphere-egu23-12810, 2023.

Small-scale, subgrid processes on the ice sheets, such as localized surface melt, remain unnoticed by current coarse-resolution Regional Climate Models (RCMs), leading to uncertainties in climate reanalyses and projections. Deep learning allows us to enhance the spatial resolution of RCMs but requires sophisticated model development. Earlier studies have shown that rudimental techniques, such as single-image super-resolution, have failed to capture Antarctic surface melt patterns accurately, because the spatial transferability of these models is low. In this study, we add remote sensing data to a super-resolution model: daily observations of surface albedo from MODIS are used to guide the downscaling of low-resolution surface melt (RACMO2, 27 km) to a high-resolution version (RACMO2, 5.5 km) for a 20-year period, between 2001-2019. We extend a conventional SRResNet and add the MODIS data in different configurations (i.e., spatial-channel communication, content communication, and empirical-physical activation). The models are trained over the Antarctic Peninsula, for which RACMO2 simulations are available at 5.5 km resolution (Van Wessem et al., 2016). We verify the performance of the models with three independent datasets to inspect (1) the overall performance (using QuickSCAT); (2) spatial patterns (using Sentinel-1); and (3) temporal patterns (using automatic weather stations). Our work shows the potential of adding remote sensing data to deep learning-based downscaling models, leading to improved spatial transferability compared to single-image downscaling models.

How to cite: de Roda Husman, S., Hu, Z., Kuipers Munneke, P., van Tiggelen, M., Lhermitte, S., and Wouters, B.: The Added Value of Remote Sensing Data in Downscaling Regional Climate Models, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12927, https://doi.org/10.5194/egusphere-egu23-12927, 2023.

The AutoICE Competition, launched on ESA’s AI4EO platform, brings together AI and Earth Observation practitioners to address the challenge of “automated sea ice mapping” from Sentinel-1 SAR data. Traversing the polar waters safely and efficiently requires up-to-date maps of the constantly moving and changing sea ice conditions showing the current sea ice extent, local concentration, and auxiliary descriptions of the ice conditions. For several decades, sea ice charts have been manually produced by visually inspecting and analysing satellite imagery.

The objective of the AutoICE challenge is to advance the state-of-the-art for automatic sea ice parameter retrieval from SAR data to derive more robust and accurate sea ice maps. The challenge design and evaluation criteria have been created with input from machine learning experts and members of the International Ice Charting Working Group (IICWG). In this competition, participants are tasked to build machine learning models using the available state-of-the-art challenge dataset and to submit their model results for each of the three sea ice parameters: sea ice concentration, stage of development and floe size. The dataset made available in this challenge contains Sentinel-1 active microwave (SAR) data and corresponding Microwave Radiometer (MWR) data from the AMSR2 satellite sensor to enable challenge participants to exploit the advantages of both instruments and to create data fusion models. Label data in the challenge datasets are ice charts produced by both the Greenland ice service at the Danish Meteorological Institute (DMI) and the Canadian Ice Service (CIS). The challenge datasets also contain other auxiliary data such as the distance to land and numerical weather prediction model data. Two versions of the challenge dataset are available, a raw dataset and a ready-to-train dataset. The datasets each consist of the same 513 training and 20 test (without label data) scenes, however, the ready-to-train version has been further prepared for model training. In addition, a number of tools are made available to help the participants get started quickly, including access to machine learning computing resources on the ESA Polar Thematic Exploitation Platform (Polar TEP). The competition was initiated on the 23rd of  November 2022 and is expected to conclude on the 17th of April 2023.

Here, we present the overall challenge, the underlying objective, the available state-of-the-art dataset and resources, the progress of the challenge and its results, as well as a sneak peek of our upcoming ASID-v3 dataset. 

How to cite: Stokholm, A., Buus-Hinkler, J., Wulf, T., Korosov, A., Saldo, R., Arthurs, D., Solberg, R., Longépé, N., and Kreiner, M.: The AutoICE Competition: Automatically Mapping Sea Ice in the Arctic, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-13038, https://doi.org/10.5194/egusphere-egu23-13038, 2023.

Understanding how the presence of fractured ice alters the dynamics, hydrology and energy balance of glaciers and ice shelves is important in determining the future evolution of the Antarctic Ice Sheet (AIS). However, these processes are not all well understood, and large-scale quantitative observations of fractures are sparse. Fortunately, the large amount of satellite data covering Antarctica gives us the opportunity to change this.

The Sentinel-1 satellite cluster, from the European Space Agency's Copernicus programme, has acquired synthetic aperture radar (SAR) data over the AIS with a repeat period of 6-12 days for the last 8 years. A broad range of crevasse types are visible in this imagery: rifts, surface crevasses and some basal crevasses on ice shelves, and fine surface crevasses on grounded ice streams - even those bridged by snow or pixel-scale in width.

In this study, we use machine learning to automatically map crevasses in this imagery; producing monthly composite maps over the AIS at 50m resolution. We developed algorithms to partition crevasses into those on grounded and floating ice, and extract these features in parallel using a mixture of convolutional neural networks, trained in a weakly supervised way, and more traditional computer vision techniques.

Having developed parallelisable routines for the large-scale batch processing of SAR data, we have processed every Sentinel-1 acquisition over the Antarctic Ice Sheet. The resulting dense timeseries of fracture maps allows us to assess the evolution of crevasses during the Sentinel-1 acquisition period. By measuring the density of fractures we develop a method to quantify structural change on ice shelves, and investigate those of the Amundsen Sea Embayment. We show an increase in crevassing in buttressing regions of the Pine Island and Thwaites ice shelves over the last 8 years, with observed changes elsewhere in the Amundsen Sea dominated by the advection of existing crevasses.

Finally, we develop methods demonstrating how our fracture data can be assimilated into numerical modelling experiments aiming to quantify the impact of ice shelf fracture on glacier dynamics.

How to cite: Surawy-Stepney, T., Hogg, A. E., Cornford, S. L., and Hogg, D.: Mapping Antarctic Crevasses at High Spatiotemporal Resolution with Deep Learning applied to Synthetic Apertur Radar Data, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-14207, https://doi.org/10.5194/egusphere-egu23-14207, 2023.

Runoff from Arctic rivers has a direct influence on the sea ice dynamics in the Arctic Ocean, producing significant effects from local to global scales. Despite their key role, the knowledge of the processes that influence the Arctic rivers streamflow is still limited, and their behavior is not fully understood.

In the literature, these analyses are usually performed adopting classical statistical methods and simple linear models, which are probably unable to fully capture underlying nonlinearities and redundancy of candidate drivers.

In this study, we use automatic feature selection techniques to detect the main drivers of the five major Arctic Rivers’ runoff (Ob, Yenisei, and Lena in Asia, Mackenzie and Yukon in North America). Daily time series of temperature and precipitation recorded by several stations spread across the Arctic region, and the average snow cover of each basin are used as candidate input variables.

The feature selection analysis is carried out with two algorithms: Wrapper for Quasi Equally Informative Subset Selection (W-QEISS) and Iterative Input Section (IIS). W-QEISS adopts neural predictive models to select alternative sets of drivers providing similar in terms of accuracy, but with different relevance, redundancy, and cardinality. Conversely, IIS directly produces a ranking of the input variables relying on tree-based models and combining computational efficiency and scalability to high input dimensionality.

The two algorithms achieve noticeably consistent results, with minor differences that can be explained by numerical factors typical of machine learning. Results also show that autoregressive terms have a crucial role in all the hydrological basins, while the importance of the other drivers is different for each river.

This preliminary research opens the floor for further analysis to broaden the knowledge of Arctic hydro-meteorological dynamics.

How to cite: Zeno, M., Sangiorgio, M., and Castelletti, A.: Detection of Arctic rivers streamflow drivers through automatic feature selection, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-14210, https://doi.org/10.5194/egusphere-egu23-14210, 2023.

Icebergs account for half of all ice loss from Antarctica. Their melting affects the surrounding ocean properties through the intrusion of cold, fresh meltwater and the release of terrigenious nutrients. This in turn influences the local ocean circulation, sea ice formation and biological production. To locate and quantify the fresh water flux from Antarctic icebergs, we need to track them and monitor changes in their area and thickness. While the locations of large icebergs are tracked operationally by manual inspection, delineation of iceberg extent requires detailed analysis – either also manually or through automated segmentation of high resolution satellite imagery.

In this study, we apply three machine learning techniques to 191 Sentinel-1 images between 2014 and 2020 and assess their skill to segment seven giant Antarctic icebergs between 54 and 1052 km² in size. Most previous studies to detect icebergs have focused on smaller bergs. In contrast, we aim to segment selected giant icebergs with the goal to automate the calculation of their changing area, volume, and freshwater input. Two of our techniques are standard segmentation techniques (k-means and Otsu thresholding) and the third one is a deep neural network (U-net). It is the first study to apply a deep learning algorithm to iceberg detection.

We analyse the strengths and weaknesses of the different machine learning approaches across a range of challenging environmental conditions: These include scenes where the iceberg is surrounded by deformed sea ice, when other big bergs are present and when berg fragments are close to the main iceberg. We also cover cases when the iceberg drifts close to the coast and summer images with surface thawing conditions, which invert the backscatter contrast between iceberg and ocean.

How to cite: Braakmann-Folgmann, A., Shepherd, A., Hogg, D., and Redmond, E.: Delineating giant Antarctic icebergs with Deep Learning, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-14307, https://doi.org/10.5194/egusphere-egu23-14307, 2023.

The urgency in reducing uncertainties of near-term sea level rise relies on improved modeling of ice sheet response to climate change. Predicting future ice sheet change requires a tremendous effort across a range of disciplines in ice sheet science, including expertise in observational data, paleoglaciology, numerical ice sheet modeling, and the widespread use of emerging methodologies for learning from the data. However, significant knowledge and disciplinary barriers make collaboration between data and model groups the exception rather than the norm. We seek to improve the efficiency in collaboration among traditionally disparate approaches to this problem. We present Ghub, a community-building scientific and educational cyberinfrastructure framework that includes models and data processing tools, online simulation, and collaboration support, available for use at theghub.org. Ghub enables collaboration between ice sheet scientific communities and acts as a host for the open-source tools that are becoming more common in the field of ice sheet science. We provide an overview of the Ghub framework, with examples of tools, tutorials, and educational content that are ready to use, and visions for extending these and other upcoming developments. These tools target a wide range of audiences, ranging from ice sheet modeling community efforts such as the Ice Sheet Model Intercomparison Project for CMIP6 (ISMIP6) to more specialized process-orientated investigations. We also outline the process for scientists to host their data and tools on the platform.

How to cite: Goliber, S., Briner, J., Nowicki, S., Csatho, B., Jones-Ivey, R., Lipscomb, W., Patra, A., Poinar, K., Quinn, J., Schenk, A., and Thayer-Calder, K.: Ghub: A new community-driven data-model resource for ice-sheet scientists, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-17128, https://doi.org/10.5194/egusphere-egu23-17128, 2023.

How to cite: Chen, W., Tsamados, M., Willatt, R., Takao, S., Nelson, C., Lawrence, I., Lee, S., Brockley, D., Landy, J., De Rijke-Thomas, C., Shirazi, D., Stroeve, J., and Francis, A.: Discrimination of sea ice leads and floes using Deep Learning applied to Sentinel-3 Ocean and Land Colour Instrument (OLCI) imaging spectrometer, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-17318, https://doi.org/10.5194/egusphere-egu23-17318, 2023.

Recent work has demonstrated how Gaussian Process Regression (GPR) can be used to interpolate Pan-Arctic radar freeboard of sea ice as measured by satellites. Sea ice freeboard is crucial to measuring sea ice thickness, and thus sea ice volume, which can play an important role in climate models. Similarly sea surface heights from altimetry are essential to determine the geostrophic currents from space. Using GPR can be computationally burdensome for modest dataset sizes and prohibitive for large datasets. To avoid having to deal with a large dataset the raw satellite observations were binned (averaged) onto a regularly spaced grid. We look at how these calculations can be reduced in terms of run time by utilising a Graphical Processing Unit (GPU), a dedicated GPR python package and by making practical adjustments to the methodology. We find by adopting these changes the overall run time of a single day’s interpolation can be greatly reduced by a factor of over 60, making it practical to run such calculations on an environment with a GPU. We then extend the method to use raw satellite observation data (no binning), which greatly increases the number of training points, requiring the use a sparse method for GPR. We conclude with recommendations for further work on this subject as it has the potential for widespread use in remote sensing applications.

How to cite: MacEachern, R., Tsamados, M., Gregory, W., Lawrence, I., and Takao, S.: Fast interpolation of satellite altimetry data with probabilistic machine learning and GPU, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-17323, https://doi.org/10.5194/egusphere-egu23-17323, 2023.

Ice shelves, the floating extensions of glaciers and ice sheets, create a safety band around Antarctica. They control the flow of ice that drains into the ocean by buttressing the upstream grounded ice. Loss of ice shelf stability and integrity results in reduced buttressing and leads to increased discharge contributing to global sea level rise. Therefore, it is important to monitor ice shelf dynamics to accurately estimate future sea level rise.

So far, the potential of SAR data has not yet been full exhausted as data of early SAR satellites has only been used to a very limited extent for calving front monitoring. To fill this research gap, we made use of the entire ERS and Envisat archive within West Antarctic Pine Island Bay, a region that requires particular attention due to drastic ongoing changes. A 20-year time series (1992-2011) of ice shelf front dynamics was derived based on a deep neural network architecture that combines segmentation and edge detection. By testing different data preparation, training and post-processing configurations we identified the best performing model for ERS and Envisat data. This includes transfer learning based on a model originally trained on Sentinel-1 data and post-processing with filtering and temporal compositing to remove artefacts from geolocation errors and limited data availability.

The resulting product of yearly, half-year and monthly ice shelf front positions reveals individual dynamic patterns for all five investigated ice shelves. The most considerable fluctuations were found for Pine Island Ice Shelf in terms of frequency of calving events (multiple cycles of calving and re-advance) and Thwaites ice tongue in terms of size of break-up (80 km retreat in early 2002). Despite different change rates and magnitudes, most ice shelves show similar signs of destabilisation. This manifests through retreating front positions and changing ice shelf geometries. Signs of weakening appear in the form of fracturing, disintegration events and loss of connection to lateral confinements.

How to cite: Wagner, L., Baumhoer, C., Dietz, A., Kuenzer, C., and Ullmann, T.: Analysis of ice shelf front dynamics in Pine Island Bay (Antarctica) based on long-term SAR time series and deep learning, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-17372, https://doi.org/10.5194/egusphere-egu23-17372, 2023.

The terrestrial carbon cycle is one of the largest sources of uncertainty in climate projections. The terrestrial carbon sink which removes a quarter of anthropogenic CO2 emissions; is highly variable in time and space depending on climate. Previous studies have found that data-driven models such as random forest, artificial neural networks and long short-term memory networks can be used to accurately model Net Ecosystem Exchange (NEE) and Gross Primary Productivity (GPP) accurately, which are two important metrics to quantify the direction and magnitude of CO2 transfer between the land surface and the atmosphere. Recently, a new class of machine learning models called transformers have gained widespread attention in natural language processing tasks due to their ability to learn from large volumes of sequential data. In this work, we use Transformers to model NEE and GPP from 1996-2022 at 39 Flux stations in the ICOS Europe network using ERA5 reanalysis data. We can compare our results with traditional machine learning approaches to evaluate the generalisability and predictive performance of transformers for carbon flux modelling.

How to cite: Mazumder, S. and Prasad, A.: Modeling the Variability of Terrestrial Carbon Fluxes using Transformers, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-582, https://doi.org/10.5194/egusphere-egu23-582, 2023.

Weather forecasts today are typically issued in the form of ensemble simulations based on multiple runs of numerical weather prediction models with different perturbations in the initial states and the model physics. In light of the continuously increasing spatial resolutions of operational weather models, this results in large, high-dimensional datasets that nonetheless contain relevant spatial and temporal structure, as well as information about the predictive uncertainty. We propose invariant variational autoencoder (iVAE) models based on convolutional neural network architectures to learn low-dimensional representations of the spatial forecast fields. We specifically aim to account for the ensemble character of the input data and discuss methodological questions about the optimal design of suitable dimensionality reduction methods in this setting. Thereby, our iVAE models extend previous work where low-dimensional representations of single, deterministic forecast fields were learned and utilized for incorporating spatial information into localized ensemble post-processing methods based on neural networks [1], which were able to improve upon model utilizing location-specific inputs only [2]. By additionally incorporating the ensemble dimension and learning representation for probability distributions of spatial fields, we aim to enable a more flexible modeling of relevant predictive information contained in the full forecast ensemble. Additional potential applications include data compression and the generation of forecast ensembles of arbitrary size.

We illustrate our methodological developments based on a 10-year dataset of gridded ensemble forecasts from the European Centre for Medium-Range Weather Forecasts of several meteorological variables over Europe. Specifically, we investigate alternative model architectures and highlight the importance of tailoring the loss function to the specific problem at hand.

References:

[1] Lerch, S. & Polsterer, K.L. (2022). Convolutional autoencoders for spatially-informed ensemble post-processing. ICLR 2022 AI for Earth and Space Science Workshop, https://arxiv.org/abs/2204.05102.

[2] Rasp, S. & Lerch, S. (2018). Neural networks for post-processing ensemble weather forecasts. Monthly Weather Review, 146, 3885-3900.

How to cite: Chen, J., Höhlein, K., and Lerch, S.: Spatial representation learning for ensemble weather simulations using invariant variational autoencoders, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-1825, https://doi.org/10.5194/egusphere-egu23-1825, 2023.

The AtmoRep project asks if one can train one neural network that represents and describes all atmospheric dynamics. AtmoRep’s ambition is hence to demonstrate that the concept of large-scale representation learning, whose principle feasibility and potential was established by large language models such as GPT-3, is also applicable to scientific data and in particular to atmospheric dynamics. The project is enabled by the large amounts of atmospheric observations that have been made in the past as well as advances on neural network architectures and self-supervised learning that allow for effective training on petabytes of data. Eventually, we aim to train on all of the ERA5 reanalysis and, furthermore, fine tune on observational data such as satellite measurements to move beyond the limits of reanalyses.

We will present the theoretical formulation of AtmoRep as an approximate representation for the atmosphere as a stochastic dynamical system. We will also detail our transformer-based network architecture and the training protocol for self-supervised learning so that unlabelled data such as reanalyses, simulation outputs and observations can be employed for training and re-fining the network. Results will be presented for the performance of AtmoRep for downscaling, precipitation forecasting, the prediction of tropical convection initialization, and for model correction. Furthermore, we also demonstrate that AtmoRep has substantial zero-short skill, i.e., it is capable to perform well on tasks it was not trained for. Zero- and few-shot performance (or in context learning) is one of the hallmarks of large-scale representation learning and to our knowledge has never been demonstrated in the geosciences.

How to cite: Lessig, C., Luise, I., and Schultz, M.: AtmoRep: Large Scale Representation Learning for Atmospheric Data, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-3117, https://doi.org/10.5194/egusphere-egu23-3117, 2023.

Numerical Earth system models (ESMs) are our primary tool for projecting future climate scenarios. Their simulation output is used by impact models that assess the effect of anthropogenic global warming, e.g., on flood events, vegetation changes or crop yields. Precipitation, an atmospheric variable with arguably one of the largest socio-economic impacts, involves various processes on a wide range of spatial-temporal scales. However, these cannot be completely resolved in ESMs due to the limited discretization of the numerical model. 
This can lead to biases in the ESM output that need to be corrected in a post-processing step prior to feeding ESM output into impact models, which are calibrated with observations [1]. While established post-processing methods successfully improve the modelled temporal statistics for each grid cell individually, unrealistic spatial features that require a larger spatial context are not addressed.
Here, we apply a cycle-consistent generative adversarial network (CycleGAN) [2] that is physically constrained to the precipitation output from Coupled Model Intercomparison Project phase 6 (CMIP6)  ESMs to correct both temporal distributions and spatial patterns. The CycleGAN can be naturally trained on daily ESM and reanalysis fields that are unpaired due to the deviating trajectories of the ESM and observation-based ground truth. 
We evaluate our method against a state-of-the-art bias adjustment framework (ISIMIP3BASD) [3] and find that it outperforms it in correcting spatial patterns and achieves comparable results on temporal distributions. We further discuss the representation of extreme events and suitable metrics for quantifying the realisticness of unpaired precipitation fields.

 [1] Cannon, A.J., et al. "Bias correction of GCM precipitation by quantile mapping: How well do methods preserve changes in quantiles and extremes?." Journal of Climate 28.17 (2015): 6938-6959.

[2] Zhu, J.-Y., et al. "Unpaired image-to-image translation using cycle-consistent adversarial networks." Proceedings of the IEEE international conference on computer vision. 2017.

[3] Lange, S. "Trend-preserving bias adjustment and statistical downscaling with ISIMIP3BASD (v1.0)." Geoscientific Model Development 12.7 (2019): 3055-3070.

How to cite: Hess, P., Lange, S., and Boers, N.: Improving global CMIP6 Earth system model precipitation output with generative adversarial networks for unpaired image-to-image translation, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-3128, https://doi.org/10.5194/egusphere-egu23-3128, 2023.

Machine learning, and particularly neural networks, have been touted as a valuable accelerator for physical processes. By training on data generated from an existing algorithm a network may theoretically learn a more efficient representation and accelerate the computations via emulation. For many parameterized physical processes in weather and climate models this being actively pursued. Here, we examine the value of this approach for radiative transfer within the IFS, an operational numerical weather prediction model where both accuracy and speed are vital. By designing custom, physics-informed, neural networks we achieve outstanding offline accuracy for both longwave and shortwave processes. In coupled testing we find minimal changes to forecast scores at near operational resolutions. We carry out coupled inference on GPUs to maximise the speed benefits from the emulator approach.

How to cite: Chantry, M., Ukkonen, P., Hogan, R., and Dueben, P.: Emulating radiative transfer in a numerical weather prediction model, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-3256, https://doi.org/10.5194/egusphere-egu23-3256, 2023.

Dynamical models used in climate prediction often have systematic errors that can bias the predictions. In this study, we utilized machine learning to address this issue. Machine learning was applied to learn the error corrected by data assimilation and thus build a data-driven model to emulate the dynamical model error. A hybrid model was constructed by combining the dynamical and data-driven models. We tested the hybrid model using synthetic observations generated by a simplified high-resolution coupled ocean-atmosphere model (MAOOAM, De Cruz et al., 2016) and compared its performance to that of a low-resolution version of the same model used as a standalone dynamical model.

To evaluate the forecast skill of the hybrid model, we produced ensemble predictions based on initial conditions determined through data assimilation. The results show that the hybrid model significantly improves the forecast skill for both atmospheric and oceanic variables compared to the dynamical model alone. To explore what affects short-term forecast skills and long-term forecast skills, we built two other hybrid models by correcting errors either only atmospheric or only oceanic variables. For short-term atmospheric forecasts, the results show that correcting only oceanic errors has no effect on atmosphere variables forecasts but correcting only atmospheric variables shows similar forecast skill to correcting both atmospheric and oceanic errors. For the long-term forecast of oceanic variables, correcting the oceanic error can improve the forecast skill, but correcting both atmospheric and oceanic errors can obtain the best forecast skill. The results indicate that for the long-term forecast of oceanic variables, bias correction of both oceanic and atmospheric components can have a significant effect.

How to cite: He, Z., Brajard, J., Wang, Y., Wang, X., and Shen, Z.: Using machine learning to improve dynamical predictions in a coupled model, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-3321, https://doi.org/10.5194/egusphere-egu23-3321, 2023.

Data-driven models are becoming increasingly competent at tasks fundamental to weather and climate prediction. Relative to machine learning (ML) based atmospheric models, which have shown promise in short-term forecasting, ML-based ocean forecasting remains somewhat unexplored. In this work, we present a data-driven emulator of an ocean GCM and show that performance over a single predictive step is skilful across all variables under consideration. Iterating such data-driven models poses additional challenges, with many models suffering from over-smoothing of fields or instabilities in the predictions. We compare a variety of methods for iterating our data-driven emulator and assess them by looking at how well they agree with the underlying GCM in the very short term and how realistic the fields remain for longer-term forecasts. Due to the chaotic nature of the system being forecast, we would not expect any model to agree with the GCM accurately over long time periods, but instead we expect fields to continue to exhibit physically realistic behaviour at ever increasing lead times. Specifically, we expect well-represented fields to remain stable whilst also maintaining the presence and sharpness of features seen in both reality and in GCM predictions, with reduced emphasis on accurately representing the location and timing of these features. This nuanced and temporally changing definition of what constitutes a ‘good’ forecast at increasing lead times generates questions over both (1) how one defines suitable metrics for assessing data-driven models, and perhaps more importantly, (2) identifying the most promising loss functions to use to optimise these models.

How to cite: Furner, R., Haynes, P., Jones, D., Munday, D., Paige, B., and Shuckburgh, E.: An iterative data-driven emulator of an ocean general circulation model, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-3340, https://doi.org/10.5194/egusphere-egu23-3340, 2023.

While efforts have been made to curb CO₂ emissions from aviation, the more uncertain non-CO₂ effects that contribute about two-thirds to the warming in terms of radiative forcing (RF), still require attention. The most important non-CO₂ effects include persistent line-shaped contrails, contrail-induced cirrus clouds and nitrogen oxide (NO_x) emissions that alter the ozone (O₃) and methane (CH₄) concentrations, both of which are greenhouse gases, and the emission of water vapour (H₂O). The climate impact of these non-CO₂ effects depends on emission location and prevailing weather situation; thus, it can potentially be reduced by advantageous re-routing of flights using Climate Change Functions (CCFs), which are a measure for the climate effect of a locally confined aviation emission. CCFs are calculated using a modelling chain starting from the instantaneous RF (iRF) measured at the tropopause that results from aviation emissions. However, the iRF is a product of computationally intensive chemistry-climate model (EMAC) simulations and is currently restricted to a limited number of days and only to the North Atlantic Flight Corridor. This makes it impossible to run EMAC on an operational basis for global flight planning. A step in this direction lead to a surrogate model called algorithmic Climate Change Functions (aCCFs), derived by regressing CCFs (training data) against 2 or 3 local atmospheric variables at the time of emission (features) with simple regression techniques and are applicable only in parts of the Northern hemisphere. It was found that in the specific case of O₃ aCCFs, which provide a reasonable first estimate for the short-term impact of aviation NO_x on O₃ warming using temperature and geopotential as features, can be vastly improved [1]. There is aleatoric uncertainty in the full-order model (EMAC), stemming from unknown sources (missing features) and randomness in the known features, which can introduce heteroscedasticity in the data. Deterministic surrogates (e.g. aCCFs) only predict point estimates of the conditional average, thereby providing an incomplete picture of the stochastic response. Thus, the goal of this research is to build a new surrogate model for iRF, which is achieved by :

1. Expanding the geographical coverage of iRF (training data) by running EMAC simulations in more regions (North & South America, Eurasia, Africa and Australasia) at multiple cruise flight altitudes,

2. Following an objective approach to selecting atmospheric variables (feature selection) and considering the importance of local as well as non-local effects,

3. Regressing the iRF against selected atmospheric variables using supervised machine learning techniques such as homoscedastic and heteroscedastic Gaussian process regression.

We present a new surrogate model that predicts iRF of aviation NO_x-O₃ effects on a regular basis with confidence levels, which not only improves our scientific understanding of NO_x-O₃ effects, but also increases the potential of global climate-optimised flight planning.

References

[1] Rao, P.; et al. Case Study for Testing the Validity of NO_x-Ozone Algorithmic Climate Change Functions for Optimising Flight Trajectories. Aerospace 2022, 9, 231. https://doi.org/10.3390/aerospace9050231

How to cite: Rao, P., Dwight, R., Singh, D., Maruhashi, J., Dedoussi, I., Grewe, V., and Frömming, C.: Towards a new surrogate model for predicting short-term NOx-O3 effects from aviation using Gaussian processes, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-4337, https://doi.org/10.5194/egusphere-egu23-4337, 2023.

Time transfer functions describe the change of state variables over time in geoscientific numerical simulation models. The identification of these functions is an essential but challenging step in model building. While traditional methods rely on qualitative understanding or first order principles, the availability of large spatio-temporal data sets from direct measurements or extremely detailed physical-based system modelling has enabled the use of machine learning methods to discover the time transfer function directly from data. In this study we explore the feasibility of this data driven approach for numerical simulation of the co-evolution of soil, hydrology, vegetation, and grazing on landscape scale, at geological timescales. From empirical observation and hyper resolution (1 m, 1 week) modelling (Karssenberg et al, 2017) it has been shown that a hillslope system shows complex behaviour with two stable states, respectively high biomass on deep soils (healthy state) and low biomass on thin soils (degraded or desertic state). A catastrophic shift from healthy to degraded state occurs under changes of external forcing (climate, grazing pressure), with a transient between states that is rapid or slow depending on system characteristics. To identify and use the time transfer functions of this system at hillslope scale we follow four procedural steps. First, an extremely large data set of hillslope average soil and vegetation state is generated by a mechanistic hyper resolution (1 m, 1 week) system model, forcing it with different variations in grazing pressure over time. Secondly, a machine learning model predicting the rate of change in soil and vegetation as function of soil, vegetation, and grazing pressure, is trained on this data set. In the third step, we explore the ability of this trained machine learning model to predict the rate of system change (soil and vegetation) on untrained data. Finally, in the fourth step, we use the trained machine learning model as time transfer function in a forward numerical simulation of a hillslope to determine whether it is capable of representing the known complex behaviour of the system. Our findings are that the approach is in principle feasible. We compared the use of a deep neural network and a random forest. Both can achieve great fitting precision, although the latter performs much faster and requires less training data. Even though the machine learning based time transfer function shows differences in the rates of change in system state from those calculated using expert knowledge in Karssenberg et al. (2017), forward simulation appeared to be possible with system behaviour generally in line with that observed in the data from the hyper resolution model. Our findings indicate that discovery of time transfer functions from data is possible. Next steps need to involve the use observational data (e.g., from remote sensing) to test the approach using data from real-world systems.

Karssenberg, D., Bierkens, M.F.P., Rietkerk, M., Catastrophic Shifts in Semiarid Vegetation-Soil Systems May Unfold Rapidly or Slowly. The American Naturalist 2017. Vol. 190, pp. E145–E155.

How to cite: Pomarol Moya, O. and Karssenberg, D.: Machine learning for data driven discovery of time transfer functions in numerical modelling: simulating catastrophic shifts in vegetation-soil systems, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-4454, https://doi.org/10.5194/egusphere-egu23-4454, 2023.

Physical parameterization is one of the major components of Numerical Weather Prediction system. In Korean Integrated Model (KIM), physical parameterizations account for about 30 % of the total computation time. There are many studies of developing neural network based emulators to replace and accelerate physics based parameterization. In this study, we develop a planetary boundary layer(PBL) emulator which is based on Shin-Hong (Hong et al., 2006, 2010; Shin and Hong, 2013, 2015) scheme that computes the parameterized effects of vertical turbulent eddy diffusion of momentum, water vapor, and sensible heat fluxes. We compare the emulator performance with Multi-Layer Perceptron (MLP) based architectures: simple MLP, MLP application version, and MLP-mixer(Tolstikhin et al., 2021). MLP application version divides data into several vertical groups for better approximation of each vertical group layers. MLP-mixer is MLP based architecture that performs well in computer vision without using convolution and self-attention. We evaluate the resulting MLP based emulator performance. MLP application version and MLP-mixer showed significant performance improvement over simple MLP.

How to cite: Jang, J., Oh, T.-J., An, S., Park, W., Na, I., and Kim, J.: Development of PBL Parameterization Emulator using Neural Networks, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-4695, https://doi.org/10.5194/egusphere-egu23-4695, 2023.

Numerical weather prediction models contain parameters that are inherently uncertain and cannot be determined exactly. Traditionally, the parameter tuning has been done manually, which can be an extremely labourious task. Tuning the entire model usually requires adjusting a relatively large amount of parameters. In case of manual tuning, the need to balance a number of requirements at the same time can lead the tuning process being a maze of subjective choices. It is, therefore, desirable to have reliable objective approaches for estimation of optimal values and uncertainties of these parameters. In this presentation we present how to optimise 20 key physical parameters having a strong impact on forecast quality. These parameters belong to the Stochastically Perturbed Parameters Scheme in the atmospheric model Open Integrated Forecasting System.

The results show that simultaneous optimisation of O(20) parameters is possible with O(100) algorithm steps using an ensemble of O(20) members, and that the optimised parameters lead to substantial enhancement of predictive skill. The enhanced predictive skill can be attributed to reduced biases in low-level winds and upper-tropospheric humidity in the optimised model. We find that the optimisation process is dependent on the starting values of the parameters that are optimised (starting from better suited values results in a better model). The results also show that the applicability of the tuned parameter values across different model resolutions is somewhat questionable since the model biases seem to be resolution-specific. Moreover, our optimisation algorithm tends to treat the parameter covariances poorly limiting its ability to converge to the global optimum.

How to cite: Tuppi, L., Ekblom, M., Ollinaho, P., and Järvinen, H.: Algorithmic optimisation of key parameters of OpenIFS, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-4817, https://doi.org/10.5194/egusphere-egu23-4817, 2023.

In the preparation of the global kilometre-resolution coupled ICON climate model, it is necessary to calibrate cloud microphysical parameters. Here we explore the avenue towards optimally calibrating such parameters using machine learning. The emulator developed by Watson-Parris et al. (2021) is employed in combination with a perturbed-parameter ensemble of limited-area atmosphere-only ICON simulations for the North Atlantic ocean. In a first step, the autoconversion scaling parameter is calibrated, using satellite-retrieved top-of-atmosphere and bottom-of-atmosphere radiation fluxes. For this purpose, limited area simulations of the north atlantic are performed with ICON. In which different cloud microphysical parameters are changed, in order to evaluate possible influences on the output of radiation fluxes.

How to cite: Eichholz, H. M., Kretzschmar, J., Watson-Parris, D., Umlauft, J., and Quaas, J.: Towards machine-learning calibration of cloud parameters in the kilometre-resolution ICON atmosphere model, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-5003, https://doi.org/10.5194/egusphere-egu23-5003, 2023.

Machine learning approaches have been widely used for improving the representation of subgrid scale parameterizations in Earth System Models. In our study we target the Cloud Microphysics parameterization, in particular the two-moment bulk scheme of the ICON (Icosahedral Non-hydrostatic) Model. 

Cloud microphysics parameterization schemes suffer from an accuracy/speed tradeoff. The simplest schemes, often heavy with assumptions (such as the bulk moment schemes) are most common in operational weather prediction models. Conversely, the more complex schemes with fewer assumptions –e.g. Lagrangian schemes such as the super-droplet method (SDM)– are computationally expensive and used only within research and development. SDM allows easy representation of complex scenarios with multiple hydrometeors and can also be used for simulating cloud-aerosol interactions. To bridge this gap and to make the use of more complex microphysical schemes feasible within operational models, we use a data-driven approach. 

Here we train a neural network to mimic the behavior of SDM simulations in a warm-rain scenario in a dimensionless control volume. The network behaves like a dynamical system that converts cloud droplets to rain droplets–represented as bulk moments–with only the current system state as the input. We use a multi-step training loss to stabilize the network over long integration periods, especially in cases with extremely low cloud water to start with. We find that the network is stable across various initial conditions and in many cases, emulates the SDM simulations better than the traditional bulk moment schemes. Our network also performs better than any previous ML-based attempts to learn from SDM. This opens the possibility of using the trained network as a proxy for imitating the computationally expensive SDM within operational weather prediction models with minimum computational overhead. 

How to cite: Sharma, S. and Greenberg, D.: Machine Learning Parameterization for Super-droplet Cloud Microphysics Scheme, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-5149, https://doi.org/10.5194/egusphere-egu23-5149, 2023.

Machine learning represents a potential method to cope with the gray zone problem of representing motions in dynamical systems on scales comparable to the model resolution. Here we explore the possibility of using a neural network to directly learn the error caused by unresolved scales. We use a modified shallow water model which includes highly nonlinear processes mimicking atmospheric convection. To create the training dataset, we run the model in a high- and a low-resolution setup and compare the difference after one low-resolution time step, starting from the same initial conditions, thereby obtaining an exact target. The neural network is able to learn a large portion of the difference when evaluated on single time step predictions on a validation dataset. When coupled to the low-resolution model, we find large forecast improvements up to 1 d on average. After this, the accumulated error due to the mass conservation violation of the neural network starts to dominate and deteriorates the forecast. This deterioration can effectively be delayed by adding a penalty term to the loss function used to train the ANN to conserve mass in a weak sense. This study reinforces the need to include physical constraints in neural network parameterizations.

How to cite: Ruckstuhl, Y., Kriegmair, R., Rasp, S., and Craig, G.: Using weak constrained neural networks to improve simulations in the gray zone, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-5523, https://doi.org/10.5194/egusphere-egu23-5523, 2023.

In recent years, machine learning (ML) based parameterizations have become increasingly common in Earth System Models (ESM). Sub-grid scale physical processes that would be computationally too expensive, e.g., atmospheric chemistry and cloud microphysics, can be emulated by ML algorithms such as neural networks.

Neural networks are trained first on simulations of the sub-grid scale process that is to be emulated. They are then used in so-called inference mode to make predictions during the ESM run, replacing the original parameterization. Training usually requires GPUs, while inference may be done on CPU architectures.

At first, neural networks are evaluated offline, i.e., independently of the ESM on appropriate datasets. However, their performance can ultimately only be evaluated in an online setting, where the ML algorithm is coupled to the ESM, including nonlinear interactions.

We want to shorten the time spent in neural network development and offline testing and move quickly to online evaluation of ML components in our ESM of choice, ICON (Icosahedral Nonhydrostatic Weather and Climate Model). Since ICON is written in Fortran, and modern ML algorithms are developed in the Python ecosystem, this requires efficient bridges between the two programming languages. The Fortran-Python bridge must be flexible to allow for iterative development of the neural network. Changes to the ESM codebase should be as few as possible, and the runtime overhead should not limit development.

In our contribution we explore three strategies to call the neural network inference from within Fortran using (i) embedded Python code compiled in a dynamic library, (ii) pipes, and (iii) MPI using the ICON coupler YAC. We provide quantitative benchmarks for the proposed Fortran-Python bridges and assess their overall suitability in a qualitative way to derive best practices. The Fortran-Python bridge enables scientists and developers to evaluate ML components in an online setting, and can be extended to other parameterizations and ESMs.

How to cite: Arnold, C., Sharma, S., Weigel, T., and Greenberg, D.: Best Practices for Fortran-Python Bridges to Integrate Neural Networks in Earth System Models, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-5766, https://doi.org/10.5194/egusphere-egu23-5766, 2023.

Accurate forecasts of the atmosphere demand large-scale simulations with high spatio-temporal resolution. Atmospheric chemistry modeling, for example, usually requires solving a system of hundreds of coupled ordinary partial differential equations. Due to the computational complexity, large high performance computing resources are required, which is a challenge as the spatio-temporal resolution increases. Machine learning methods and specially deep learning can offer an approximation of the simulations with some factor of speed-up while using less compute resources. The goal of this study is to investigate the feasibility, opportunities but also challenges and pitfalls of replacing the compute-intensive chemistry of a state-of-the-art atmospheric chemistry model with a trained neural network model to forecast the concentration of trace gases at each grid cell and to reduce the computational complexity of the simulation. In this work, we introduce a neural network model (ICONET) to forecast trace gas concentrations without executing the traditional compute-intensive atmospheric simulations. ICONET is equipped with a multifeature Long Short Term Memory (LSTM) model to forecast atmospheric chemicals iteratively in time. We generated the training and test dataset, our ground truth for ICONET, by execution of an atmospheric chemistry simulation in ICON-ART. Applying the ICONET trained model to forecast a test dataset results in a good fit of the forecast values compared to our ground truth dataset. We discuss appropriate metrics to evaluate the quality of models and present the quality of the ICONET forecasts with RMSE and KGE metrics. The variety in the nature of trace gases limits the model's learning and forecast skills according to the variable. In addition to the quality of the ICONET forecasts, we described the computational efficiency of ICONET as its run time speed-up in comparison to the run time of the ICON-ART simulation. The ICONET forecast showed a speed-up factor of 3.1 over the run time of the atmospheric chemistry simulation of ICON-ART, which is a significant achievement, especially when considering the importance of ensemble simulations.

How to cite: Azmi, E., Meyer, J., Strobl, M., Weimer, M., and Streit, A.: Approximation and Optimization of Atmospheric Simulations in High Spatio-Temporal Resolution with Neural Networks, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6287, https://doi.org/10.5194/egusphere-egu23-6287, 2023.

One of the largest sources of uncertainty when projecting the Antarctic contribution to sea-level rise is the ocean-induced melt at the base of Antarctic ice shelves. This is because resolving the ocean circulation and the ice-ocean interactions occurring in the cavity below the ice shelves is computationally expensive.

Instead, for large ensembles and long-term projections of the ice-sheet evolution, ice-sheet models currently rely on parameterisations to link the ocean temperature and salinity in front of ice shelves to the melt at their base. However, current physics-based parameterisations struggle to accurately simulate basal melt patterns.

As an alternative approach, we explore the potential use of a deep feedforward neural network as a basal melt parameterisation. To do so, we train a neural network to emulate basal melt rates simulated by highly-resolved circum-Antarctic ocean simulations. We explore the influence of different input variables and show that the neural network struggles to generalise to ice-shelf geometries unseen during training, while it generalises better on timesteps unseen during training. We also test the parameterisation on separate coupled ocean-ice simulations to assess the neural network’s performance on independent data.  

How to cite: Burgard, C., Jourdain, N. C., Mathiot, P., and Smith, R.: Parameterising melt at the base of Antarctic ice shelves with a feedforward neural network, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6836, https://doi.org/10.5194/egusphere-egu23-6836, 2023.

Retrieving ozone from limb measurements necessitates the modeling of scattered light through the atmosphere.  However, accurately modeling multiple scattering (MS) during retrieval requires excessive computational resources; consequently, operational retrieval models employ approximations in lieu of the full MS calculation.  Here we consider an alternative MS approximation method, where we use radiative transfer (RT) simulations to train neural network models to predict the MS radiances.  We present our findings regarding the best-performing network hyperparameters, normalization schemes, and input/output data structures.  Using RT calculations based on measurements by the Ozone Mapping and Profiling Suite's Limb Profiler (OMPS/LP), we compare the accuracy of these neural-network models with both the full MS calculation as well as the current MS approximation methods utilized during OMPS/LP retrievals.

How to cite: Himes, M., Kramarova, N., Zhu, T., Mok, J., Bandel, M., Fasnacht, Z., and Loughman, R.: Neural network surrogate models for multiple scattering: Application to OMPS LP simulations, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-7281, https://doi.org/10.5194/egusphere-egu23-7281, 2023.

Some results from the DEEB (Differential Equation Estimation Benchmark) are presented. In DEEB, we compare different machine learning approaches and statistical methods for estimating nonlinear dynamics from data. Such methods constitute an important building block for purely data-driven earth system models as well as hybrid models which combine physical knowledge with past observations.

Specifically, we examine approaches for solving the following problem: Given time-state-observations of a deterministic ordinary differential equation (ODE) with measurement noise in the state, predict the future evolution of the system. Of particular interest are systems with chaotic behavior - like Lorenz 63 - and nonparametric settings, in which the functional form of the ODE is completely unknown (in particular, not restricted to a polynomial of low order). To create a fair comparison of methods, a benchmark database was created which includes datasets of simulated observations from different dynamical systems with different complexity and varying noise levels. The list of methods we compare includes: echo state networks, Gaussian processes, Neural ODEs, SINDy, thin plate splines, and more.

Although some methods consistently perform better than others throughout different datasets, there seems to be no silver bullet.

How to cite: Schötz, C.: Comparison of Methods for Learning Differential Equations from Data, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-7368, https://doi.org/10.5194/egusphere-egu23-7368, 2023.

How to cite: Brolly, M.: Learning fluid dynamical statistics using stochastic neural networks, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-7391, https://doi.org/10.5194/egusphere-egu23-7391, 2023.

Monitoring the state of the sea is a fundamental task for economic activities in the coastal zone, such as transport, tourism and infrastructure design. In recent years, regular wave height monitoring for marine risk assessment and mitigation has become unavoidable as global warming impacts in more intense and frequent swells.
In particular, the Mediterranean Sea has been considered as one of the most responsive regions to global warming, which may promote the intensification of hazardous natural phenomena as strong winds, heavy precipitation and high sea waves. Because of the high density population along the Mediterranean coastlines, heavy swells could have major socio-economic consequences. To reduce the impacts of such scenarios, the development of more advanced monitoring systems of the sea state becomes necessary.
In the last decade, it has been demonstrated how seismometers can be used to measure sea conditions by exploiting the characteristics of a part of the seismic signal called microseism. Microseism is the continuous seismic signal recorded in the frequency band of 0.05 and 0.4 Hz that is likely generated by interactions of sea waves together and with seafloor or shorelines.
In this work, in the framework of i-WaveNET INTERREG project, we performed a regression analysis to develop a model capable of predicting the sea state in the Sicily Channel (Italy) using microseism, acquired by onshore instruments installed in Sicily and Malta. Considering the complexity of the relationship between spatial sea wave height data and seismic data measured at individual stations, we used supervised machine learning (ML) techniques to develop the prediction model. As input data we used the hourly Root Mean Squared (RMS) amplitude of the seismic signal recorded by 14 broadband stations, along the three components, and in different frequency bands, during 2018 - 2021. These stations, belonging to the permanent seismic networks managed by the National Institute of Geophysics and Volcanology INGV and the Department of Geosciences of the University of Malta, consist of three-component broadband seismometers that record at a sampling frequency of 100 Hz.
As for the target, the significant sea wave height data from Copernicus Marine Environment Monitoring Service (CMEMS) for the same period were used. Such data is the hindcast product of the Mediterranean Sea Waves forecasting system, with hourly temporal resolution and 1/24° spatial resolution. After a feature selection step, we compared three different kinds of ML algorithms for regression: K-Nearest-Neighbors (KNN), Random Forest (RF) and Light Gradient Boosting (LGB). The hyperparameters were tuned by using a grid-search algorithm, and the best models were selected by cross-validation.  Different metrics, such as MAE, R² and RMSE, were considered to evaluate the generalization capabilities of the models and special attention was paid to evaluate the predictive ability of the models for extreme wave height values.
Results show model predictive capabilities good enough to develop a sea monitoring system to complement the systems currently in use.

How to cite: Cannavo', F., Minio, V., Saitta, S., Alparone, S., Borzì, A. M., Cannata, A., Ciraolo, G., Contrafatto, D., D’Amico, S., Di Grazia, G., and Larocca, G.: Machine Learning and Microseism as a Tool for Sea Wave Monitoring, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-7492, https://doi.org/10.5194/egusphere-egu23-7492, 2023.

A course of action to combat the emission of greenhouse gasses (GHG) in a Danish context is to re-wet previously drained peatlands and thereby return them to their natural hydrological state acting as GHG sinks. GHG emissions from peatlands are known to be closely coupled to the hydrological dynamics through the groundwater table depth (WTD). To understand the effect of a changing and variable climate on the spatio-temporal dynamics of hydrological processes and the associated uncertainties, we aim to produce a high-resolution local-scale climate projection ensemble from the global-scale CMIP6 projections.

With focus on hydrological impacts, uncertainties and possible extreme endmembers, this study aims to span the full ensemble of local-scale climate projections in the Danish geographical area corresponding to the CMIP6-ensemble of Global Climate Models (GCMs). Deep learning founded statistical downscaling methods are applied bridge the gap from GCMs to local-scale climate change and variability, which in turn will be used in field-scale hydrological modeling. The approach is developed to specifically accommodate the resolutions, event types and conditions relevant for assessing the impacts on peatland GHG emissions through their relationship with WTD dynamics by applying stacked conditional generative adversarial networks (CGANs) to best downscale precipitation, temperature, and evaporation. In the future, the approach is anticipated to be extended to directly assess the impacts of climate change and ensemble uncertainty on peatland hydrology variability and extremes.

How to cite: Quistgaard, T., Langen, P. L., Denager, T., Schneider, R., and Stisen, S.: Deep Learning guided statistical downscaling of climate projections for use in hydrological impact modeling in Danish peatlands, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-7561, https://doi.org/10.5194/egusphere-egu23-7561, 2023.

In oceanography, altimetry products are used to measure the height of the ocean surface, and ocean modeling is used to understand and predict the behavior of the ocean. There are two main types of gridded altimetry products: operational sea level products, such as DUACS, which are used for forecasting and reconstruction, and ocean model reanalyses, such as Glorys 12, which are used to forecast seasonal trends and assess physical characteristics. However, advances in ocean modeling do not always directly benefit operational forecast or reconstruction products.

In this study, we investigate the potential for deep learning methods, which have been successfully applied in simulated setups, to leverage ocean modeling efforts for improving operational altimetry products. Specifically, we ask under what conditions the knowledge learned from ocean simulations can be applied to real-world operational altimetry mapping. We consider the impact of simulation grid resolution, observation data reanalysis, and physical processes modeled on the performance of a deep learning model.

Our results show that the deep learning model outperforms current operational methods on a regional domain around the Gulfstream, with a 50km improvement in resolved scale. This improvement has the potential to enhance the accuracy of operational altimetry products, which are used for a range of important applications, such as climate monitoring and understanding mesoscale ocean dynamics.

How to cite: Febvre, Q., Fablet, R., Le Sommer, J., Ubelmann, C., and Benaïchouche, S.: Learning operational altimetry mapping from ocean models, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-8288, https://doi.org/10.5194/egusphere-egu23-8288, 2023.

Neural Differential Equations (NDEs) provide a powerful framework for hybrid modeling. Unfortunately, the flexibility of the neural network component of the model comes at the expense of potentially violating known physical invariants, such as conservation laws, during inference. This shortcoming is especially critical for applications requiring long simulations, such as climate modeling, where significant deviations from the physical invariants can develop over time. It is hoped that enforcing physical invariants will help address two of the main barriers to adoption for hybrid models in climate modeling: (1) long-term numerical stability, and (2) generalization to out-of-sample conditions unseen during training, such as climate change scenarios. We introduce Stabilized Neural Differential Equations, which augment an NDE model with compensating terms that ensure physical invariants remain approximately satisfied during numerical simulations. We apply Stabilized NDEs to the double pendulum and Hénon–Heiles systems, both of which are conservative, chaotic dynamical systems possessing a time-independent Hamiltonian. We evaluate Stabilized NDEs using both short-term and long-term prediction tasks, analogous to weather and climate prediction, respectively. Stabilized NDEs perform at least as well as unstabilized models at the “weather prediction” task, that is, predicting the exact near-term state of the system given initial conditions. On the other hand, Stabilized NDEs significantly outperform unstabilized models at the “climate prediction” task, that is, predicting long-term statistical properties of the system. In particular, Stabilized NDEs conserve energy during long simulations and consequently reproduce the long-term dynamics of the target system with far higher accuracy than non-energy conserving models. Stabilized NDEs also remain numerically stable for significantly longer than unstabilized models. As well as providing a new and lightweight method for combining physical invariants with NDEs, our results highlight the relevance of enforcing conservation laws for the long-term numerical stability and physical accuracy of hybrid models.

How to cite: White, A. and Boers, N.: Stabilized Neural Differential Equations for Hybrid Modeling with Conservation Laws, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9285, https://doi.org/10.5194/egusphere-egu23-9285, 2023.

Permafrost, i.e. ground material that remains perennially frozen, plays a key role in Arctic ecosystems. Monitoring the response of permafrost to rapid climate change remains difficult due to the sparse availability of long-term, high quality measurements of the subsurface. Numerical models are therefore an indispensable tool for understanding the evolution of Arctic permafrost. However, large scale simulation of the hydrothermal processes affecting permafrost is challenging due to the highly nonlinear effects of phase change in porous media. The resulting computational cost of such simulations is especially prohibitive for sensitivity analysis and parameter estimation tasks where a large number of simulations may be necessary for robust inference of quantities such as temperature, water fluxes, and soil properties. In this work, we explore the applicability of recently developed physics-informed machine learning (PIML) methods for accelerating numerical models of permafrost hydrothermal dynamics. We present a preliminary assessment of two possible applications of PIML in this context: (1) linearization of the nonlinear PDE system according to Koopman operator theory in order to reduce the computational burden of large scale simulations, and (2) efficient parameterization of the surface energy balance and snow dynamics on the subsurface hydrothermal regime. By combining the predictive power of machine learning with the underlying conservation laws, PIML can potentially enable researchers and practitioners interested in permafrost to explore complex process interactions at larger spatiotemporal scales.

How to cite: Groenke, B., Langer, M., Gallego, G., and Boike, J.: Exploring physics-informed machine learning for accelerated simulation of permafrost processes, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10135, https://doi.org/10.5194/egusphere-egu23-10135, 2023.

Foehn winds are accelerated, warm and dry winds that can have significant environmental impacts as they descend into the lee of a mountain range. For example, in the McMurdo Dry Valleys in Antarctica, foehn events can cause ice and glacial melt and destabilise ice shelves, which if lost, resulting in a rise in sea level. Consequently, there is a strong interest in a deeper understanding of foehn winds and their meteorological signatures. Most current automatic detection methods rely on rule-based methodologies that require static thresholds of meteorological parameters. However, the patterns of foehn winds are hard to define and differ between alpine valleys around the world. Consequently, data-driven solutions might help create more accurate detection and prediction methodologies. 

State-of-the-art machine learning approaches to this problem have shown promising results but follow a supervised learning paradigm. As such, these approaches require accurate labels, which for the most part, are being created by imprecise static rule-based algorithms. Consequently, the resulting machine-learning models are trained to recognise the same static definitions of the foehn wind signatures. 

In this paper, we introduce and compare the first unsupervised machine-learning approaches for detecting foehn wind events. We focus on data from the Mc Murdo Dry Valleys as an example, however, due to the unsupervised nature of these approaches, our solutions can recognise a more dynamic definition of foehn wind events and are therefore, independent of the location. The first approach is based on multivariate time-series clustering, while the second utilises a deep autoencoder-based anomaly detection method to identify foehn wind events. Our best model achieves an f1-score of 88%, matching or surpassing previous machine-learning methods while providing a more flexible and inclusive definition of foehn events. 

How to cite: Milz, T., Hofsteenge, M., Katurji, M., and Vetrova, V.: Foehn Wind Analysis using Unsupervised Deep Anomaly Detection, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10256, https://doi.org/10.5194/egusphere-egu23-10256, 2023.

Data assimilation is often viewed as a framework for correcting short-term error growth in dynamical climate model forecasts. When viewed on the time scales of climate however, these short-term corrections, or analysis increments, closely mirror the systematic bias patterns of the dynamical model. In this work, we show that Convolutional Neural Networks (CNNs) can be used to learn a mapping from model state variables to analysis increments, thus promoting the feasibility of a data-driven model parameterization which predicts state-dependent model errors. We showcase this problem using an ice-ocean data assimilation system within the fully coupled Seamless system for Prediction and EArth system Research (SPEAR) model at the Geophysical Fluid Dynamics Laboratory (GFDL), which assimilates satellite observations of sea ice concentration. The CNN then takes inputs of data assimilation forecast states and tendencies, and makes predictions of the corresponding sea ice concentration increments. Specifically, the inputs are sea ice concentration, sea-surface temperature, ice velocities, ice thickness, net shortwave radiation, ice-surface skin temperature, and sea-surface salinity. We show that the CNN is able to make skilful predictions of the increments, particularly between December and February in both the Arctic and Antarctic, with average daily spatial pattern correlations of 0.72 and 0.79, respectively. Initial investigation of implementation of the CNN into the fully coupled SPEAR model shows that the CNN can reduce biases in retrospective seasonal sea ice forecasts by emulating a data assimilation system, further suggesting that systematic sea ice biases could be reduced in a free-running climate simulation.

How to cite: Gregory, W., Bushuk, M., Adcroft, A., and Zhang, Y.: Deep learning of systematic sea ice model errors from data assimilation increments, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10351, https://doi.org/10.5194/egusphere-egu23-10351, 2023.

Current numerical weather prediction models contain significant systematic errors, due in part to indeterminate ground forcing (GF). This study considers an optimal virtual GF (GF_o) derived by training observed and simulated datasets of 10-m wind speeds (WS₁₀) for summer and winter. The GF_o is added to an offline surface multilayer model (SMM) to revise predictions of WS₁₀ in China by the Weather Research and Forecasting model (WRF). This revision is a data-based optimization under physical constraints. It reduces WS₁₀ errors and offers wide applicability. The resulting model outperforms two purely physical forecasts (the original WRF forecast and the SMM with physical GF parameterized using urban, vegetation, and subgrid topography) and two purely data-based revisions (i.e., multilinear regression and multilayer perceptron). Compared with original WRF forecasting, using the GF_o scheme reduces the Root Mean Square Error (RMSE) in WS₁₀ across China by 25% in summer and 32% in winter. The frontal area index of GF_o indicates that it includes both the effects of indeterminate GF and other possible complex physical processes associated with WS₁₀.

How to cite: Feng, J.: Mitigate forecast error in surface wind speed using an offline single-column model with optimal ground forcing, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10394, https://doi.org/10.5194/egusphere-egu23-10394, 2023.

The lack of ground level observation stations outside of settlements makes monitoring and forecasting local weather and permafrost challenging in the Arctic. Such predictive pieces of information are essential to help prepare for potentially hazardous weather conditions, especially during winter. In this study, we aim at enhancing predictive analytics in Alaska of permafrost and temperature by using a hybrid forecasting technique. In particular, we propose VMD-WT-InceptionTime model for short-term air temperature forecasting.

This proposed technique incorporates data preprocessing techniques and deep learning to enhance the accuracy of the next seven days air temperature forecasts. Initially, the Spearman correlation coefficient is utilized to examine the relationship between different inputs and the forecast target temperature. Following this, Variational Mode Decomposition (VMD) is used to decompose the most output-correlated input variables (i.e., temperature and relative humidity) to extract intrinsic and non-stationary time-frequency features from the original sequences. The Wavelet Transform (WT) is then employed to further extract intrinsic multi-resolution patterns from these decomposed input variables. Finally, a deep InceptionTime model is used for multi-step air temperature forecasting using these processed sequences. This forecasting technique was developed using an open dataset holding 20+ years of data from three locations in Alaska: North Slope, Alaska, Arctic National Wildlife Refuge, Alaska, and Diomede Island region, Bering Strait. Model performance has been rigorously evaluated of metrics including RMSE, MAPE and error.

Results highlight the effectiveness of the proposed hybrid model in providing more accurate short-term forecasts than several baselines (GBDT, SVR, ExtraTrees, RF, ARIMA, LSTM, GRU, and Transformer). More specifically, this technique reported RMSE and MAPE average increase rates amounting to 11.21% and 16.13% in North Slope, 30.01% and 34.97% in Arctic National Wildlife Refuge, and 16.39%, 23.46% in Diomede Island region. In addition, the proposed technique produces forecasts over all seven horizons with a maximum error of <1.5K, a minimum error of >-1.2K, and an average error lower than 0.18K for North Slope. For Arctic National Wildlife Refuge, a maximum error of <1K, a minimum error of >-0.9K, and an average of < 0.1K. While a maximum error of <0.9K, a minimum error of >-0.8K, and an average of <0.13K, for Diomede Island region. However, the worst performances achieved were errors of around 6K in the third horizon (i.e., 3rd day) for North Slope and the Arctic National Wildlife Refuge and the last horizon (i.e., 7th day) for the Diomede Islands region. Most of the worst performances of the proposed technique in all three locations can be attributed to having to produce forecasts of higher variations and wider temperature ranges than their averages.

Overall, this research highlights the potential of the decomposition techniques and deep learning to: 1) reveal and effectively learn the underlying cyclicity of air temperatures at varying resolutions that allows for accurate predictions without any knowledge of the governing physics, 2) produce accurate multi-step temperature forecasts in Arctic climates.

How to cite: Putkonen, J., Ahajjam, M. A., Pasch, T., and Chance, R.: A hybrid VMD-WT-InceptionTime model for multi-horizon short-term air temperature forecasting in Alaska, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10726, https://doi.org/10.5194/egusphere-egu23-10726, 2023.

Adjoints have become a staple of the oceanic and atmospheric numerical modeling community over the past couple of decades as they are useful for tuning of dynamical models, sensitivity analyses, and data assimilation. One such application is generation of reanalysis datasets, which provide an optimal record of our past weather, climate, and ocean. For example, the state-of-the-art ocean-ice renanalysis dataset, ECCO, is created by optimally combining a numerical ocean model with heterogeneous observations through a technique called data assimilation. Data assimilation in ECCO minimizes the distance between model and observations by calculating adjoints, i.e., gradients of the loss w.r.t. simulation forcing fields (wind and surface heat fluxes). The forcing fields are iteratively updated and the model is rerun until the loss is minimized to ensure that the numerical model does not drastically deviate from the observations. Calculating adjoints, however, either requires  disproportionately high computational resources  or rewriting the dynamical model code to be autodifferentiable. 

Therefore, we ask if deep learning-based emulators can provide fast and accurate adjoints. Ocean data is smooth, high-dimensional, and has complex spatiotemporal correlations. Therefore, as an initial foray into ocean emulators, we leverage a combination of neural operators and transformers. Specifically, we have adapted the FourCastNet architecture, which has successfully emulated ERA5 weather data in seconds rather than hours, to emulate an idealized ocean simulation.

We generated a ground-truth dataset by simulating a double-gyre, an idealized representation of the North Atlantic Ocean, using MITgcm, a state-of-the-art dynamical model. The model was forced by zonal wind at the surface and relaxation to a meridional profile of temperature — warm/cold temperatures at low/high latitudes. This simulation produced turbulent western boundary currents embedded in the large-scale gyre circulation. We performed 4 additional simulations by modifying the magnitude of SST relaxation and wind forcing to introduce diversity in the dataset. From these simulations, we used 4 state variables (meridional and zonal surface velocities, pressure, and temperature) as well as the forcing fields (zonal wind velocity and relaxation SST profile) sampled in 10-day steps. The dataset was split into training, validation, and test datasets such that validation and test datasets were unseen during training. These datasets provide an ideal testbed for evaluating and comparing the performance of data-driven ocean emulators.

We used this data to train and evaluate Oceanfourcast. Our initial results in the following figure show that our model, Oceanfourcast, can successfully predict the streamfunction and pressure for a lead time of 1 month. 

We are currently working on generating adjoints from Oceanfourcast.  We expect the adjoint calculation to require significantly less compute time than that from a full-scale dynamical model like MITgcm.  Our work shows a promising path towards deep-learning augmented data assimilation and uncertainty quantification.

How to cite: Bire, S., Lütjens, B., Newman, D., and Hill, C.: Oceanfourcast: Emulating Ocean Models with Transformers for Adjoint-based Data Assimilation, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10810, https://doi.org/10.5194/egusphere-egu23-10810, 2023.

Recent works have demonstrated the viability of employing data-driven / machine learning 
methods for the purposes of learning more about ocean turbulence, with applications to turbulence parameterisations in ocean general circulation models. Focusing on mesoscale geostrophic turbulence in the ocean context, works thus far have mostly focused on the choice of algorithms and testing of trained up models. Here we focus instead on the choice of eddy flux data to learn from. We argue that, for mesoscale geostrophic turbulence, it might be beneficial from a theoretical as well as practical point of view to learn from eddy fluxes with dynamically inert rotational fluxes removed (ideally in a gauge invariant fashion), instead of the divergence of the eddy fluxes as has been considered thus far. Outlooks for physically constrained and interpretable machine learning will be given in light of the results. 

How to cite: Yan, F., Mak, J., and Wang, Y.: On the choice of turbulence eddy fluxes to learn from in data-driven methods, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10904, https://doi.org/10.5194/egusphere-egu23-10904, 2023.

Small-scale fluctuations in vertical wind velocity, unresolved by climate and weather forecast models play a particularly important role in determining vapor and tracer fluxes, turbulence and cloud formation. Fluctuations in vertical wind velocity are challenging to represent since they depend on orography, large scale circulation features, convection and wind shear. Parameterizations developed using data retrieved at specific locations typically lack generalization and may introduce error when applied on a wide range of different conditions. Retrievals of vertical wind velocity are also difficult and subject to large uncertainty. This work develops a new data-driven, neural network representation of subgrid scale variability in vertical wind velocity. Using a novel deep learning technique, the new parameterization merges data from high-resolution global cloud resolving model simulations with high frequency Radar and Lidar retrievals.  Our method aims to reproduce observed statistics rather than fitting individual measurements. Hence it is resilient to experimental uncertainty and robust to generalization. The neural network parameterization can be driven by weather forecast and reanalysis products to make real time estimations. It is shown that the new parameterization generalizes well outside of the training data and reproduces much better the statistics of vertical wind velocity than purely data-driven models.

How to cite: Barahona, D., Breen, K., and Kalesse-Los, H.: Deep learning parameterization of small-scale vertical velocity variability for atmospheric models, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10959, https://doi.org/10.5194/egusphere-egu23-10959, 2023.

Driven in large part by policy initiatives designed to increase food security and realized via the construction of thousands of center-pivot irrigation fields since the 1970s, agriculture development in Saudi Arabia has undergone tremendous changes. However, little is known about the accurate number, acreage, and the changing dynamics of the fields. To bridge the knowledge gap between the political drivers and in-field response, we leveraged a hybrid machine learning framework by implementing Density-Based Spatial Clustering of Applications with Noise, Convolutional Neural Networks, and Spectral Clustering in a stepwise manner to delineate the center-pivot fields on a national scale in Saudi Arabia using historical Landsat imagery since 1990. The framework achieved producer's and user's accuracies larger than  83.7% and 90.2%, respectively, when assessed against 28,000 manually delineated fields collected from different regions and periods. We explored multi-decadal dynamics of the agricultural development in Saudi Arabia by quantifying the number, acreage, and size distribution of center-pivot fields, along with the first and last detection year of the fields since 1990. The agricultural development in Saudi Arabia experienced four stages, including an initialization stage before 1990, a contraction stage from 1990 to 2010, an expansion stage from 2010 to 2016, and an ongoing contraction stage since 2016. Most of the fields predated 1990, representing over 8,800 km² in that year, as a result of the policy initiatives to stimulate wheat production, promoting Saudi Arabia as the sixth largest exporter of wheat in the 1980s. A decreasing trend was observed from 1990 to 2010, with an average of 8,011 km² of fields detected during those two decades, which was a response to the policy initiative implemented to phase-out wheat after 1990. As a consequence of planting fodder crops to promote the dairy industry, the number and extent of fields increased rapidly from 2010 to 2015 and reached its peak in 2016, with 33,961 fields representing 9,400 km². Agricultural extent has seen a continuous decline since 2016 to a level lower than 1990 values in 2020. This decline has been related to sustainable policy initiatives implemented for the Saudi Vision 2030. There is some evidence of an uptick in 2021 — also observed in an ongoing analysis for 2022 — which might be in response to global influences, such as the COVID-19 pandemic and the more recent conflict in the Ukraine, which has disrupted the international supply of agricultural products. The results provide a historical account of agricultural activity throughout the Kingdom and provide a basis for informed decision-making on sustainable irrigation and agricultural practices, helping to better protect and manage the nation's threatened groundwater resources, and providing insights into the resilience and elasticity of the Saudi Arabian food system to global perturbations.

How to cite: Li, T., López Valencia, O., Johansen, K., and McCabe, M.: National scale agricultural development dynamics under socio-political drivers in Saudi Arabia since 1990, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-11293, https://doi.org/10.5194/egusphere-egu23-11293, 2023.

Determining the number of existing water masses and defining their boundaries is subject to ongoing discussion in physical oceanography. Traditionally, water masses are defined manually by experts setting constraints based on experience and previous knowledge about the hydrographic properties describing them. In recent years, clustering, an unsupervised machine learning approach, has been introduced as a tool to determine clusters, i.e., volumes, with similar hydrographic properties without explicitly defining their hydrographic constraints. However, the exact number of clusters to be looked for is set manually by an expert up until now. 

We propose a method that determines a fitting number of clusters for hydrographic clusters in a data driven way. In a first step, the method averages the data in different-sized slices along the time or depth axis as the structure of the hydrographic space changes strongly either in time or depth. Then the method applies clustering algorithms on the averaged data and calculates off-the-shelf evaluation scores (Davies-Bouldin, Calinski-Harabasz, Silhouette Coefficient) for several predefined numbers of clusters. In the last step, the optimal number of clusters is determined by analyzing the cluster evaluation scores across different numbers of clusters for optima or relevant changes in trend. 

For validation we applied this method to the output for the subpolar North Atlantic between 1993 and 1997 of the high-resolution Atlantic Ocean model VIKING20X, in direct exchange with domain experts to discuss the resulting clusters. Due to the change from strong to weak deep convection in these years, the hydrographic properties vary strongly in the time and depth dimension, providing a specific challenge to our methodology. 

Our findings suggest that it is possible to identify an optimal number of clusters using the off-the-shelf cluster evaluation scores that catch the underlying structure of the hydrographic space. The optimal number of clusters identified by our data-driven method agrees with the optimal number of clusters found by expert interviews. These findings contribute to aiding and objectifying water mass definitions across multiple expert decisions, and demonstrate the benefit of introducing data science methods to analyses in physical oceanography.

How to cite: Trahms, C., Wölker, Y., and Biastoch, A.: Objectively Determining the Number of Similar Hydrographic Clusters with Unsupervised Machine Learning, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-11687, https://doi.org/10.5194/egusphere-egu23-11687, 2023.

Ecosystems are affected by extreme climate conditions such as droughts worldwide but we still lack understanding of the involved dynamics. Which factors render an ecosystem more resilient, and on which temporal scales do weather patterns affect vegetation state and physiology? Traditional approaches to tackle such questions involve assumption-based land surface modeling or inversions. Machine learning (ML) methods can provide a complementary perspective on how ecosystems respond to climate in a more data-driven and assumption-free manner. However, ML depends heavily on data, and commonly used observations of vegetation at best contain one observation per day, but most products are provided at 16-daily to monthly temporal resolution. This masks important processes at sub-monthly time scales. In addition, ML models are inherently difficult to interpret, which still limits their applicability for process understanding.

In the present study, we combine modern deep learning models in the time domain with observations from the geostationary Meteosat Second Generation (MSG) satellite, centered over Africa. We model fractional vegetation cover (representing vegetation state) and land surface temperature (as a proxy for water stress) from MSG as a function of meteorology and static geofactors. MSG collects observations at sub-daily frequency, rendering it into an excellent tool to study short- to mid-term land surface processes. Furthermore, we use methods from explainable ML for post-hoc model interpretation to identify meteorological drivers of vegetation dynamics and their interaction with key geofactors.

From the analysis, we expect to gather novel insights into ecosystem response to droughts with high temporal fidelity. Drought response of vegetation can be highly diverse and complex especially in arid to semi-arid regions prevalent in Africa. Also, we assess the potential of explainable machine learning to discover new linkages and knowledge and discuss potential pitfalls of the approach. Explainable machine learning, combined with potent deep learning approaches and modern Earth observation products offers the opportunity to complement assumption-based modeling to predict and understand ecosystem response to extreme climate.

How to cite: Kraft, B., Duveiller, G., Reichstein, M., and Jung, M.: Untapping the potential of geostationary EO data to understand drought impacts with XAI, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-11906, https://doi.org/10.5194/egusphere-egu23-11906, 2023.

Soil temperature and soil moisture variations have large effects on ecological processes in the soil. To investigate and understand these processes, high-resolution data of soil temperature and soil moisture are required. Here, we present an approach to generate data of soil temperature and soil moisture continuously in space, depth, and time for a 400 km² study area in the Fichtel Mountains (Germany). As reference data, measurements with 1 m long soil probes were taken. To cover many different locations, the available 15 soil probes were shifted regularly in the course of one year. With this approach, around 250 different locations in forest sites, on meadows and on agricultural fields were captured under a variety of meteorological conditions. These measurements are combined with readily available meteorological data, satellite data and soil maps in a machine learning approach to learn the complex relations between these variables. We aim for a model which can predict the soil temperature and soil moisture continuously for our study area in the Fichtel Mountains, with a spatial resolution of 10 m x 10 m, down to 1 m depth with segments of 10 cm each and in an hourly resolution in time. Here, we present the results of our pilot study where we focus on the temperature and moisture change within the depth down to 1 m at one single location. To take temporal lags into account, we construct a Long Short-Term Memory network based on meteorological data as predictors to make temperature and moisture predictions in time and depth. The results indicate a high ability of the model to reproduce the time series of the single location and highlight the potential of the approach for the space-time-depth mapping of soil temperature and soil moisture.

How to cite: Baumberger, M., Adorf, L., Haas, B., Meyer, N., and Meyer, H.: Modelling Soil Temperature and Soil Moisture in Space, Depth, and Time with Machine Learning Techniques, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-11958, https://doi.org/10.5194/egusphere-egu23-11958, 2023.

The aircraft-based observation is one of the important anchor data used in the numerical weather prediction (NWP) models. Nevertheless, the bias has been noted in the temperature observation through several previous studies. As the performance on the hybrid four-dimensional ensemble variational (hybrid-4DEnVar) data assimilation (DA) system of the Korean Integrated Model (KIM) ⸺ the operational model in the Korea Meteorological Administration (KMA) ⸺ has been advanced, the need for the aircraft temperature bias correction (BC) has been confirmed. Accordingly, as a preliminary study on the BC, the static BC method based on the linear regression was applied to the KIM Package for Observation Processing (KPOP) system. However, the results showed there were limitations of a spatial discontinuity and a dependency on the calculation period of BC coefficients.

In this study, we tried to develop the machine learning-based bias estimation model to overcome these limitations. The MultiLayer Perceptron (MLP) based learning was performed to consider the vertical, spatial and temporal characteristics of each observation by flight IDs and phases, and at the same time to consider the correlation among observation variables. As a result of removing the predicted bias from the bias estimation model, the mean of the background innovation (O-B) decreases from 0.2217 K to 0.0136 K in a given test period. Afterwards, in order to verify the analysis field impact for BC, the bias estimation model will be grafted onto the KPOP system and then several DA cycle experiments will be conducted in the KIM.

How to cite: Kwon, H., Jeon, H., Kang, J., Kwon, I., and Park, S. K.: Bias correction of aircraft temperature observations in the Korean Integrated Model based on a deep learning approach, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12218, https://doi.org/10.5194/egusphere-egu23-12218, 2023.

Machine learning is increasingly being applied to ocean wave modelling. Surrogate modelling has the potential to reduce or bypass the large computational requirements, creating a low computational-cost model that offers a high level of accuracy. One approach integrates in-situ measurements and historical model runs to achieve the spatial coverage of the model and the accuracy of the in-situ measurements. Once operational, such a system requires very little computational power, meaning that it could be deployed to a mobile phone, operational vessel, or autonomous vessel to give continuous data. As such, it makes a significant change to the availability of met-ocean data with potential to revolutionise data provision and use in marine and coastal settings.

This presentation explores the impact that an underlying physics-based model can have in such a machine learning driven framework; comparing training the system on a bespoke regional SWAN wave model developed for wave energy developments in the South West of the UK against training using the larger North-West European Shelf long term hindcast wave model run by the UK Met Office. The presentation discusses the differences in the underlying NWP models, and the impacts that these have on the surrogate wave models’ accuracy in both nowcasting and forecasting wave conditions at areas of interest for renewable energy developments. The results identify the importance in having a high quality, validated, NWP model for training such a system and the way in which the machine learning methods can propagate and exaggerate the underlying model uncertainties.

How to cite: Pillai, A., Ashton, I., Chen, J., and Steele, E.: Comparison of NWP Models Used in Training Surrogate Wave Models, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12355, https://doi.org/10.5194/egusphere-egu23-12355, 2023.

When predicting complex systems such as parts of the Earth system, one typically relies on differential equations which often can be incomplete, missing unknown influences or include errors through their discretization. To remedy those effects, we present PseudoSpectralNet (PSN): a hybrid model that incorporates both a knowledge-based part of an atmosphere model and a data-driven part, an artificial neural network (ANN). PSN is a neural differential equation (NDE): it defines the right-hand side of a differential equation, combining a physical model with ANNs and is able to train its parameters inside this NDE. Similar to the approach of many atmosphere models, part of the model is computed in the spherical harmonics domain, and other parts in the grid domain. The model consists of ANN layers in each domain, information about derivatives, and parameters such as the orography. We demonstrate the capabilities of PSN on the well-studied Marshall Molteni Quasigeostrophic Model.

How to cite: Gelbrecht, M. and Boers, N.: PseudoSpectralNet: A hybrid neural differential equation for atmosphere models, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12403, https://doi.org/10.5194/egusphere-egu23-12403, 2023.

Data Assimilation (DA) is a challenging and expensive computational problem targetting hidden variables in high-dimensional spaces. 4DVar methods are widely used in weather forecasting to fit simulations to sparse observations by optimization over numerical model input. The complexity of this inverse problem and the sequential nature of common 4DVar approaches lead to long computation times with limited opportunity for parallelization. Here we propose using machine learning (ML) algorithms to replace the entire 4DVar optimization problem with a single forward pass through a neural network that maps from noisy and incomplete observations at multiple time points to a complete system state estimate at a single time point. We train the neural network using a loss function derived from the weak-constraint 4DVar objective, including terms incorporating errors in both model and data. In contrast to standard 4DVar approaches, our method amortizes the computational investment of training to avoid solving optimization problems for each assimilation window, and its non-sequential nature allows for easy parallelization along the time axis for both training and inference. In contrast to most previous ML-based data assimilation methods, our approach does not require access to complete, noise-free simulations for supervised learning or gradient-free approximations such as Ensemble Kalman filtering. To demonstrate the potential of our approach, we show a proof-of-concept on the chaotic Lorenz'96 system, using a novel "1.5D Unet" architecture combining 1D and 2D convolutions.

How to cite: Zinchenko, V. and Greenberg, D.: Training Deep Data Assimilation Networks on Sparse and Noisy Observations, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12458, https://doi.org/10.5194/egusphere-egu23-12458, 2023.

Since PM_2.5 (particulate matter with an aerodynamic diameter of less than 2.5 µm) directly threatens public health, in order to take appropriate measures(prevention) in advance, the Korea Ministry of Environment(MOE) has been implementing PM₁₀ forecast nationwide since February 2014. PM_2.5 forecasts have been implemented nationwide since January 2015. The currently implemented PM forecast by the MOE subdivides the country into 19 regions, and forecasts the level of PM in 4 stages of “Good”, “Moderate”, “Unhealthy”, and “Very unhealthy”.

Currently PM air quality forecasting system operated by the MOE is based on a numerical forecast model along with a weather and emission model. Numerical forecasting model has fundamental limitations such as the uncertainty of input data such as emissions and meteorological data, and the numerical model itself. Recently, many studies on predicting PM using artificial intelligence such as DNN, RNN, LSTM, and CNN have been conducted to overcome the limitations of numerical models.

In this study, in order to improve the prediction performance of the numerical model, past observational data (air quality and meteorological data) and numerical forecasting model data (chemical transport model) are used as input data. The machine learning model consists of DNN and Seq2Seq, and predicts 3 days (D+0, D+1, D+2) using 6-hour and 1-hour average input data, respectively. The PM_2.5 concentrations predicted by the machine learning model and the numerical model were compared with the PM_2.5 measurements.

The machine learning models were trained for input data from 2015 to 2020, and their PM forecasting performance was tested for 2021. Compared to the numerical model, the machine learning model tended to increase ACC and be similar or lower to FAR and POD.

Time series trend was showed machine learning PM forecasting trend is more similar to PM measurements compared with numerical model. Especially, machine learning forecasting model can appropriately predict PM low and high concentrations that numerical model is used to overestimate.

Machine learning forecasting model with DNN and Seq2Seq can found improvement of PM forecasting performance compared with numerical forecasting model. However, the machine learning model has limitations that the model can not consider external inflow effects.

In order to overcome the drawback, the models should be updated and added some other machine learning module such as CNN with spatial features of PM concentrations.

How to cite: Wang, K.-H., Lee, C.-Y., Lee, J.-Y., Jung, M.-W., Kim, D.-G., Han, S.-H., Choi, D.-R., and Yun, H.: Comparison of PM2.5 concentrations prediction model performance using Artificial Intelligence, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12566, https://doi.org/10.5194/egusphere-egu23-12566, 2023.

Clouds of all kinds play a large role in many atmospheric processes including, e.g. radiation and moisture transport, and their type allows an insight into the dynamics going on in the atmosphere. Hence, the observation of clouds from Earth's surface has always been important to analyse the current weather and its evolution during the day. However, cloud observations by human observers are labour-intensive and hence also costy. In addition to this, cloud classifications done by human observers are always subjective to some extent. Finding an efficient method for automated observations would solve both problems. Although clouds have already been operationally observed using satellites for decades, observations from the surface shed a light on a different set of characteristics. Moreover, the WMO also defined their cloud classification standards according to visual cloud properties when observations are done at the Earth’s surface. Thus, in this work a utilization of machine learning methods to classify clouds from RGB pictures taken at the surface is proposed. Explicitly, a conditional Generative Adversarial Network (cGAN) is trained to discriminate between 30 different categories, 10 for each cloud level - low, medium and high; Besides showing robust results in different image classification problems, an additional advantage of using a GAN instead of a classical convolutional neural network is that its output can also artificially enhance the size of the training data set. This is especially useful if the number of available pictures is unevenly distributed among the different classes. Additional background observations like cloud cover and cloud base height can also be used to further improve the performance of the cGAN. Together with a cloud camera, a properly trained cGAN can observe and classify clouds with a high temporal resolution of the order of seconds, which can be used, e.g. for model verification or to efficiently monitor the current status of the weather as well as its short-time evolution. First results will also be presented.

How to cite: Rosenberger, M., Dorninger, M., and Weißmann, M.: Using cGAN for cloud classification from RGB pictures, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-13013, https://doi.org/10.5194/egusphere-egu23-13013, 2023.

With fast parallel computing hardware, particularly GPUs, becoming more accessible in the geosciences the now efficiently running deep learning techniques are ready to handle larger amounts of recorded observation and satellite derived data and are able to learn complex structures across time-series. Thus, a suitable deep learning setup is able to generate highly-resolved weather forecasts in real-time and on demand. Forecasts of irradiance and radiation can be challenging in machine learning as they embrace a high degree of diurnal and seasonal variation.

Continuously extended PV/solar power production grows into one of our most important fossil-fuel free energy sources. Unlike the just recently emerging PV power observations, solar irradiance offers long time-series from automized weather station networks. Being directly linked to PV outputs, forecasting highly resolved solar irradiance from nowcasting to short-range plays a crucial role in decision support and managing PV.

In this study, we investigate the suitability of several deep learning techniques adopted and developed to a set of heterogeneous data sources on selected locations. We compare the forecast results to traditional – however computationally expensive - numerical weather prediction models (NWP) and rapid update cycle models. Relevant input features include 3D-fields from NWP models (e.g.: AROME), satellite data and products (e.g.: CAMS), radiation time series from remote sensing, and observation time time-series (site observations and close sites). The amount of time-series data can be extended by a synthetic data generator, a part of our deep learning framework. Our main models investigated includes a sequence-to-sequence LSTM (long-short-term-memory) model using a climatological background model or NWP for post-processing, a Graph NN model, and an analogs based deep learning method. Furthermore, a novel neural network model based on two other ideas, the IrradianceNet and the PhyDNet, was developed. IrradPhyDNet combines the skills of IrradianceNet and PhyDNet and showed improved performance in comparison to the original models.

Results obtained by the developed methods yield, in general, high forecast-skills. For selected case studies of extreme events (e.g. Saharan dust) all novel methods could outperform the traditional methods.  Different combinations of inputs and processing-steps are part of the analysis.

How to cite: Papazek, P., Schicker, I., and Gfähler, P.: Comparison of LSTM, GraphNN, and IrradPhyDNet based Approaches for High-resolution Solar Irradiance Nowcasting, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-13143, https://doi.org/10.5194/egusphere-egu23-13143, 2023.

The ambient noise tomography (ANT) method is widely adopted to reconstruct shear-wave velocity anomalies and to generate high-resolution images of the crust and upper-mantle. A critical step in this process is the extraction of surface-wave dispersion curves from cross-correlation functions of continuous ambient noise recordings, which is traditionally performed manually on the dispersion spectrograms through human-machine interfaces. Picking of dispersion curves is sometimes prone to bias due to human interpretation. Furthermore, it is a laborious and time-consuming task that needs to be resolved in an automatized manner, especially when dealing with dense seismic network of nodal geophones where the large amount of generated data severely hinders manual picking approaches. In the last decade, several studies successfully employed machine learning methods in Earth Sciences and across many seismological applications. Early studies have shown versatile and reliable solutions by treating dispersion curve extraction as a visual recognition problem. 

We review and adapt a specific machine learning approach, deep convolutional neural networks, for use on dispersion spectrograms generated with the usual frequency-time analysis (FTAN) processing on ambient noise cross-correlations. To train and calibrate the algorithm we use several available datasets acquired from previous experiments across different geological settings. The main dataset consists of records acquired with a dense local geophone network (150 short period stations sampling at 250 Hz) deployed for one month in October 2021. The dataset has been acquired during the volcanic unrest of the Vulcano-Lipari complex, Italy. The network also accounts for additional 17 permanent broadband stations (sampling at 100 Hz) maintained by the National Institute of Geophysics and Volcanology (INGV) in Italy. We evaluate the performance of the dispersion curves extraction algorithm. The automatically-picked dispersion curves will be used to construct a shear-wave velocity model of the Vulcano-Lipari magmatic plumbing system and the surrounding area of the Aeolian archipelago.

How to cite: Stumpp, D., Amir Jiwani-Brown, E., Barat, C., Lupi, M., Muñoz, F., Planes, T., and Savard, G.: Nodal Ambient Noise Tomography and automatic picking of dispersion curves with convolutional neural network: case study at Vulcano-Lipari, Italy, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-13322, https://doi.org/10.5194/egusphere-egu23-13322, 2023.

Mesoscale eddies are gyrating currents in the ocean and have horizontal scales from 10 km up to 100 km and above. They transport water mass, heat, and nutrients and therefore are of interest among others to marine biologists, oceanographers, and geodesists. Usually, gridded sea level anomaly maps, processed from several radar altimetry missions, are used to detect eddies. However, operational processors create multi-mission (processing level 4) SLA grid maps with an effective spatiotemporal resolution far lower than their grid spacing and temporal resolution. 

This drawback leads to erroneous eddy detection. We, therefore, investigate if the higher-resolution along-track data could be used instead to solve the problem of classifying the SLA observations into cyclonic, anticyclonic, or no eddies in a more accurate way than using processed SLA grid map products. With our framework, we aim to infer a daily two-dimensional segmentation map of classified eddies. Due to repeat cycles between 10 and 35 days and cross-track spacing of a few 10 km to a few 100 km, ocean eddies are clearly visible in altimeter observations but are typically covered only by a few ground tracks where the spatiotemporal context within the input data is highly variable each day. However conventional convolutional neural networks (CNNs) rely on data without varying gaps or jumps in time and space in order to use the intrinsic spatial or temporal context of the observations. Therefore, this is a challenge that needs to be addressed with a deep neural network that on the one hand utilizes the spatiotemporal context information within the modality of along-track data and on the other hand is able to output a two-dimensional segmentation map from data of varying sparsity. Our approach with our architecture Teddy is to use a transformer module to encode and process the spatiotemporal information along with the ground track's sea level anomaly data that produces a sparse feature map. This will then be fed into a sparsity invariant convolutional neural network in order to infer a two-dimensional segmentation map of classified eddies. Reference data that is used to train Teddy is produced by an open-source geometry-based approach (py-eddy-tracker [1]). 

The focus of this presentation is on how we implemented this approach in order to derive two-dimensional segmentation maps of classified eddies with our deep neural network architecture Teddy from along-track altimetry. We show results and limitations for the classification of eddies using only along-track SLA data from the multi-mission level 3 product of the Copernicus Marine Environment Monitoring Service (CMEMS) within the 2017 - 2019 period for the Gulf Stream region. We find that using our methodology, we can create two-dimensional maps of classified eddies from along-track data without using preprocessed SLA grid maps.

[1] Evan Mason, Ananda Pascual, and James C. McWilliams, “A new sea surface height–based code for oceanic mesoscale eddy tracking,” Journal of Atmospheric and Oceanic Technology, vol. 31, no. 5, pp. 1181–1188, 2014.

How to cite: Bolmer, E., Abulaitijiang, A., Fenoglio-Marc, L., Kusche, J., and Roscher, R.: Framework for creating daily semantic segmentation maps of classified eddies using SLA along-track altimetry data, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-13367, https://doi.org/10.5194/egusphere-egu23-13367, 2023.

Clouds play an important role in determining the Earth’s radiation budget. Despite their complex and three-dimensional (3D) structures, their interactions with radiation in models are often simplified to one-dimensional (1D), considering the time required to compute radiative transfer. Such a simplification ignores cloud Inhomogeneity and horizontal photon transport in radiative processes, which may be an acceptable approximation for low-resolution models, but can lead to significant errors and impact cloud evolution predictions in high-resolution simulations. Since model developments and operations are heading toward a higher resolution that is more susceptible to radiation errors, a fast and accurate 3D radiative transfer scheme becomes important and necessary. To address the need, we develop a machine-learning-based 3D radiative transfer emulator to provide surface radiation, shortwave fluxes at all layers, and heating rate profiles. The emulators are trained for highly heterogeneous shallow cumulus under different solar positions. We will discuss the performance of the emulators in accuracy and efficiency and discuss their potential applications.

How to cite: Chiu, J.-Y. C., Yang, C.-K. K., Gristey, J. J., Feingold, G., and Gustafson, W. I.: Machine Learning Emulation of 3D Shortwave Radiative Transfer for Shallow Cumulus Cloud Fields, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-13771, https://doi.org/10.5194/egusphere-egu23-13771, 2023.

Over the last decades, space oceanography missions, particularly altimeter missions, have greatly advanced our ability to observe sea surface dynamics. However, they still struggle to resolve spatial scales below ~ 100 km. On a global scale, sea surface current are derived from sea surface height by a geostrophical assumption. While future altimeter missions should improve the observation of sea surface height, the observation of sea surface current using altimetry techniques would remains indirect. In the other hands, recent works have considered the use of AIS (automated identification system) as a new mean to reconstruct sea surface current : AIS data streams provide an indirect observational models of total currents including ageostrophic phenomenas. In this work we consider the use of the supervised learning framework 4DVARNet, a supervised data driven approach that allow us to perform multi-modal experiments : We focus on an Observing System Simulation Experiment (OSSE) in a region of the Gulf-Stream and we show that the joint use of AIS and sea surface height (SSH) measurement could improve the reconstruction of sea surface current with respect to product derived solely from AIS or SSH observations in terms of physical and time scale resolved. 

How to cite: Benaïchouche, S., Le Goff, C., Boussidi, B., Rousseau, F., and Fablet, R.: Multi-modal data assimilation of sea surface currents from AIS data streams and satellite altimetry using 4DVARNet, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-14051, https://doi.org/10.5194/egusphere-egu23-14051, 2023.

The Norwegian Meteorological Institute has for many years applied a CFD model to downscale operational NWP forecasts to 100-200m spatial resolution for wind and turbulence forecasting for about 20 Norwegian airports. Due to high computational costs, however, the CFD model can only be run twice per day, each time producing a 12-hour forecast. An approximate approach requiring far less compute resources using deep learning has therefore been developed. In this, the relation between relevant NWP forecast variables at grids of 2.5 km spatial resolution and wind and turbulence from the CFD model has been approximated using neural networks with basic convolutional and dense layers. The deep learning models have been trained on approximately two year of the data separately for each airport. The results show that the models are to a large extent able to capture the characteristics of their corresponding CDF simulations, and the method is in due time intended to fully replace the current operational solution. 

How to cite: Loe, M. K. and Bremnes, J. B.: Deep learning approximations of a CFD model for operational wind and turbulence forecasting, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-15183, https://doi.org/10.5194/egusphere-egu23-15183, 2023.

Mass accumulation rates of sediments[g/cm²/yr] or sedimentation rates[cm/yr] on the seafloor are important to understand various benthic properties, like the rate of carbon sequestration in the seafloor and seafloor geomechanical stability. Several machine learning models, such as random forests, and k-Nearest Neighbours have been proposed for the prediction of geospatial data in marine geosciences, but face significant challenges such as the limited amount of labels for training purposes, skewed data distribution, a large number of features etc. Previous model predictions show deviation in the global sediment budget, a parameter used to determine a model's predicitve validity, revealing the lack of accurate representation of sedimentation rate by the state of the art models. 

Here we present a semi-supervised deep learning methodology to improve the prediction of sedimentation rates, making use of around 9x10⁶  unlabelled data points. The semi-supervised neural network implementation has two parts: an unsupervised pretraining using an encoder-decoder network. The encoder with the optimized weights from the unsupervised training is then taken out and fitted with layers that lead to the target dimension. This network is then fine-tuned with 2782 labelled data points, which are observed sedimentation rates from peer-reviewed sources. The fine-tuned model then predicts the rate and quantity of sediment accumulating on the ocean floor, globally.

The developed semi-supervised neural network provide better predictions than supervised models trained only on labelled data. The predictions from the semi-supervised neural network are compared with that of the supervised neural network with and without dimensionality reduction(using Principle Component Analysis).

How to cite: Parameswaran, N., Gonzalez, E., Bur­wicz-Ga­ler­ne, E., Greenberg, D., Wallmann, K., and Braack, M.: Semi-supervised feature-based learning for prediction of Mass Accumulation Rate of sediments, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-15684, https://doi.org/10.5194/egusphere-egu23-15684, 2023.

The next generation of numerical weather prediction model (so-called digital twin engines) will reach hectometric scale, for which the existing physiography databases are insufficient. Our work leverages machine learning and open-access data to produce a more accurate and higher resolution physiography database. One component to improve is the land cover map. The reference data gathers multiple high-resolution thematic maps thanks to an agreement-based decision tree. The input data are taken from the Sentinel-2 satellite. Then, the land cover map generation is made with image segmentation. This work implements and compares several algorithms of different families to study their suitability to the land cover classification problem. The sensitivity to the data quality will also be studied. Compared to existing work, this work is innovative in the reference map construction (both leveraging existing maps and fit for end-user purpose) and the diversity of algorithms to produce our land cover map comparison.

How to cite: Rieutord, T., Bessardon, G., and Gleeson, E.: Physiography improvements in numerical weather prediction digital twin engines, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-15756, https://doi.org/10.5194/egusphere-egu23-15756, 2023.

Lagrangian particle dispersion models (LPDMs) have been used extensively to calculate source-receptor relationships (“footprints”) for use in greenhouse gas (GHG) flux inversions. However, because a backward-running model simulation is required for each data point, LPDMs do not scale well to very large datasets, which makes them unsuitable for use in GHG inversions using high-resolution satellite instruments such as TROPOMI. In this work, we demonstrate how Machine Learning (ML) can be used to accelerate footprint production, by first presenting a proof-of-concept emulator for ground-based site observations, and then discussing work in progress to create an emulator suitable to satellite observations. In Fillola et al (2023), we presented a ML emulator for NAME, the Met Office’s LPDM, which outputs footprints for a small region around an observation point using purely meteorological variables as inputs. The footprint magnitude at each grid cell in the domain is modelled independently using gradient-boosted regression trees. The model is evaluated for seven sites, producing a footprint in 10ms, compared to around 10 minutes for the 3D simulator, and achieving R2 values between 0.6 and 0.8 for CH4 concentrations simulated at the sites when compared to the timeseries generated by NAME. Following on from this work, we demonstrate how this same emulator can be applied to satellite data to reproduce footprints immediately around any measurement point in the domain, evaluating this application with data for Brazil and North Africa and obtaining R2 values of around 0.5 for simulated CH4 concentrations. Furthermore, we propose new emulator architectures for LPDMs applied to satellite observations. These new architectures should tackle some of the weaknesses in the existing approach, for example, by propagating information more flexibly in space and time, potentially improving accuracy of the derived footprints and extending the prediction capabilities to bigger domains.

How to cite: Fillola, E., Santos-Rodriguez, R., and Rigby, M.: Towards emulated Lagrangian particle dispersion model footprints for satellite observations, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-15892, https://doi.org/10.5194/egusphere-egu23-15892, 2023.

The spatio-temporal reconstruction of a dynamical process from some observationaldata is at the core of a wide range of applications in geosciences. This is particularly true for weather forecasting, operational oceanography and climate studies. However, the re35 construction of a given dynamic and the prediction of future states must take into ac36 count the uncertainties that affect the system. Thus, the available observational measurements are only provided with a limited accuracy. Besides, the encoded physical equa38 tions that model the evolution of the system do not capture the full complexity of the real system. Finally, the numerical approximation generates a non-negligible error. For these reasons, it seems relevant to calculate a probability distribution of the state system rather than the most probable state. Using recent advances in machine learning techniques for inverse problems, we propose an algorithm that jointly learns a parametric distribution of the state, the dynamics governing the evolution of the parameters, and a solver. Experiments conducted on synthetic reference datasets, as well as on datasets describing environmental systems, validate our approach.

How to cite: Lafon, N., Naveau, P., and Fablet, R.: Uncertainty quantification in variational data assimilation with deep learning, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-15994, https://doi.org/10.5194/egusphere-egu23-15994, 2023.

Forests are a significant carbon sink of the total carbon dioxide (CO₂) emitted by humans. Climate change is expected to impact forest systems, and their role in the terrestrial carbon cycle in several ways – for example, the fertilization effect of increased atmospheric CO₂, and the lengthening of the growing season in northern temperate and boreal areas may increase forest productivity, while more frequent extreme climate events such as storms and windthrows or drought spells, as well as wildfires might reduce disturbances return period, hence increasing forest land loss and reduction of the carbon stored in the vegetation and soils. In addition, forest management in response to an increased demand for wood products and fuel can affect the carbon storage in ecosystems and wood products. State-of-the-art Dynamic Global Vegetation Models (DGVMs) simulate the forest responses to environmental and human processes, however running these models globally for many climate and management scenarios becomes challenging due to computational restraints. Integration of process-based models and machine learning methods through emulation allows us to speed up computationally expensive simulations. In this work, we explore the use of machine learning to surrogate the LPJ-GUESS DGVM. This emulator is spatially-aware to represent forests across the globe in a flexible spatial resolution, and consider past climate and forest management practices to account for legacy effects. The training data for the emulator is derived from dedicated runs of the DGVM sampled across four dimensions relevant to forest carbon and yield: atmospheric CO₂ concentration, air Temperature, Precipitation, and forest Management (CTPM). The emulator can capture relevant forest responses to climate and management in a lightweight form, and will support the development of the coupled socio-economic/ecologic model of the land system, namely LandSyMM (landsymm.earth). Other relevant scientific applications include the analysis of optimal forestry protocols under climate change, and the forest potential in climate change mitigation.

How to cite: Natel de Moura, C., Belda, D. M., Antoni, P., and Arneth, A.: A machine learning emulator for forest carbon stocks and fluxes, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-16287, https://doi.org/10.5194/egusphere-egu23-16287, 2023.

One of the many ways in which anthropogenic climate change impacts our planet is
rising sea levels. The rate of sea level rise (SLR) across the oceans is,
however, not uniform in space or time and is influenced by a complex interplay
of ocean dynamics, heat uptake, and surface forcing. As a consequence,
short-term (years to a decade) regional SLR patterns are difficult to model
using conventional deterministic approaches. For example, the latest climate
model projections (called CMIP6) show some agreement in the globally integrated
rate of SLR but poor agreement when it comes to spatially-resolved
patterns. However, such forecasts are valuable for adaptation planning in
coastal areas and for protecting low lying assets.
Rather than a deterministic modeling approach, here we explore the possibility
of exploiting the high quality satellite altimeter derived record of sea surface
height variations, which cover the global oceans outside of ice-infested waters
over a period of 30 years. Alongside this rich and unique satellite record,
several data-driven models have shown tremendous potential for various
applications in Earth System science. We explore several data-driven deep
learning approaches for sea surface height forecasts over multi-annual to
decadal time frames. A limitation of some machine learning approaches is the
lack of any kind of uncertainty quantification, which is problematic for
applications where actionable evidence is sought. As a consequence, we equip
our models with a rigorous measure of uncertainty, namely conformal prediction which
is a model and dataset agnostic method that provides calibrated predictive
uncertainty with proven coverage guarantees. Based on a 30-year satellite
altimetry record and auxiliary climate forcing data from reanalysis such as
ERA5, we demonstrate that our methodology is a viable and attractive alternative
for decadal sea surface height forecasts.

How to cite: Lehmann, N., Bamber, J., and Zhu, X.: Global Decadal Sea Surface Height Forecast with Conformal Prediction, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-16597, https://doi.org/10.5194/egusphere-egu23-16597, 2023.

Sea surface temperature (SST) is a critical parameter in the global climate system and plays a vital role in many marine processes, including ocean circulation, evaporation, and the exchange of heat and moisture between the ocean and atmosphere. As such, understanding the variability of SST is important for a range of applications, including weather and climate prediction, ocean circulation modeling, and marine resource management.

The dynamics of SST is the compound of multiple degrees of freedom that interact across a continuum of Spatio-temporal scales. A first-order approximation of such a system was initially introduced by Hasselmann. In his pioneering work, Hasselmann (1976) discussed the interest in using a two-scale stochastic model to represent the interactions between slow and fast variables of the global ocean, climate, and atmosphere system. In this paper, we examine the potential of machine learning techniques to derive relevant dynamical models of Sea Surface Temperature Anomaly (SSTA) data in the Mediterranean Sea. We focus on the seasonal modulation of the SSTA and aim to understand the factors that influence the temporal variability of SSTA extremes. Our analysis shows that the variability of the SSTA can indeed well be decomposed into slow and fast components. The dynamics of the slow variables are associated with the seasonal cycle, while the dynamics of the fast variables are linked to the SSTA response to rapid underlying processes such as the local wind variability. Based on these observations, we approximate the probability density function of the SSTA data using a stochastic differential equation parameterized by a neural network. In this model, the drift function represents the seasonal cycle and the diffusion function represents the envelope of the fast SSTA response.

How to cite: Ouala, S., Chapron, B., Collard, F., Gaultier, L., and Fablet, R.: Analysis of marine heat waves using machine learning, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-16936, https://doi.org/10.5194/egusphere-egu23-16936, 2023.

Snow plays a crucial role in the hydrological cycle as it serves as an intermediate storage of winter precipitation and renews groundwater reserves. It is therefore of central importance for, among others, our drinking water supply and agriculture. Snow interacts with its environment in many ways, is constantly changing with time, and thus has a highly heterogeneous spatial and temporal distribution. Therefore, modelling snow variability is difficult, especially when additional components such as forests add complexity. To increase our understanding of the spatiotemporal variability of snow as well as to validate snow models, we need reliable validation data. For these purposes, airborne LiDAR surveys or time series derived from snow sensors on the point scale are commonly used. However, these are disadvantageously limited to one point either in space or in time. In this study, we profited from current advances in LiDAR and drone technology, as well as machine learning, to bridge this gap. We present a new dataset on snow variability in forests for the Alptal, a sub-alpine, forested valley in the pre-alps, Switzerland. The core dataset consists of a dense sensor network, repeated UAV-based LiDAR flights and manual snow height and density measurements. Using modern machine learning algorithms, we determine four clusters of similar spatiotemporal behaviour regarding their snowheight. These clusters are characterized and further used to derive daily snow depth and snow water equivalent maps. By using the latter, we obtain spatially continuous key hydrological variables. The results suggest that snow occurs in clusters that reoccur in space. These clusters underline the relation between canopy cover and spatial snow accumulation patterns and (the much more complex) spatial ablation patterns. The presented dataset and derived products are the first to our knowledge that provide daily, high-resolution snow height and hydrologic variables based on field data. The results of this study can therefore improve our understanding of the spatiotemporal variability of snow in forested environments. Moreover, they are ideally suited for the validation of modern snow models.

How to cite: Geissler, J., Rathmann, L., and Weiler, M.: Machine Learning and LiDAR Snowheight Maps from UAVs Reveal Clusters of Snow Variability in a Sub-Alpine Forest., EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-32, https://doi.org/10.5194/egusphere-egu23-32, 2023.

Seasonal snow in the northern regions plays an important role providing water resources for both consumption and hydropower generation. Moreover, the snow depth in the northern Finland during winter exceeds 1 m, impacting local agriculture, vegetation, tourism and recreational activities. The objective of this study is to estimate snow depth using an empirical methodology applied to synthetic aperture radar (SAR) images and compare with in situ measurements collected by automatic weather stations (AWS) and snow courses in northern Finland. Snow depth estimates with high spatiotemporal resolution can improve our understand of seasonal snow mass in complex access areas. Here, we use an adapted version of the empirical methodology developed by Lievens et al. (2019) to estimate snow depth using Sentinel-1 constellation (C-band). The algorithm utilizes changes in the cross-polarized backscatter measurements of SAR images repeatedly acquired on the same orbit to avoid geometry distortions. We use SNAP toolbox, combined with the Copernicus digital elevation model (DEM), posted at 30 meters, in the pre-processing stage.  The snow retrievals between 2019 and 2022 are compared to three automatic weather stations and four snow courses measurements collected over the same period. The ongoing Sentinel-1 snow depth retrievals during the winter 2021/2022 demonstrate a correlation of 0.76, when compared to in situ measurements, supporting the potential ability to derive snow changes in regions where in situ measurements of snow are currently lacking. Despite the good agreement between the empirical algorithm and the collected datasets on land, further investigation is still necessary to better understand the backscatter response over frozen lake areas. Thanks to the effort of international space agencies, we have available currently, and in a near future, global coverage at high resolution SAR imagery and, combined with installed automatic weather stations, opens the possibility of a wide spatial monitoring of snow variations.

How to cite: Lemos, A. and Riihelä, A.: Snow Depth derived from Sentinel-1 compared to in-situ observations in northern Finland, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-998, https://doi.org/10.5194/egusphere-egu23-998, 2023.

The Arctic region has been experiencing significant climate change, with the loss of snow and ice accelerating at an alarming rate. Accurate monitoring of the spatiotemporal dynamics of snow and ice is essential for understanding and predicting the impacts of climate change on Arctic ecosystems and their feedback on global climate. In this paper, we use the Visible Infrared Imaging Radiometer Suite (VIIRS) on the Suomi National Polar-orbiting Partnership (SNPP) satellite to monitor the spatiotemporal dynamics of snow and ice in polar regions. The VIIRS/Day/Night Band (DNB) is a unique instrument that can provide high-resolution imagery of the Earth's surface at night, with a spatial resolution of 750 m and a sensitivity of 0.01 nW/cm2/sr. This enables the detection of faint moonlight and artificial light and allows for mapping snow and ice in polar winter when no sunlight is available for months.  Our aims demonstrate the potential of moonlight remote sensing for continuous monitoring of snow/ice in the Arctic region and analyse the importance of continuous monitoring and research on the impacts of climate change on the Arctic ecosystem and the potential for Arctic seaway.

How to cite: Liu, D. and Zhang, Q.: Monitoring the Spatiotemporal Dynamics of Arctic Winter Snow/Ice with MoonlightRemote Sensing, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-1920, https://doi.org/10.5194/egusphere-egu23-1920, 2023.

The melting snow layer, as a composition of ice, liquid water, and air, supplies meltwater in the runoff phase inducing the melt pond formation. These melting processes of Arctic sea ice alter the surface reflection properties and thereby affect the energy budget. Such sea ice surface reflection properties were surveyed by airborne hyperspectral imagery within the framework of an Arctic field campaign performed in May/June 2017. A retrieval approach based on different absorption indices of pure ice and liquid water in the near infrared spectral range is applied to the campaign data retrieving the spatial distribution of snow layer liquid water fraction and effective radius of snow grains. For the same sceneries the melt pond depth was retrieved based on an existing approach that isolates the dependence of a melt pond reflectance spectrum on the pond depth by eliminating the reflection contribution of the pond ice bottom. The presented retrieval methods show the potential of airborne hyperspectral imagery to map the transition phase of the Arctic sea ice surface examining the snow layer composition and melt pond bathymetry.

How to cite: Rosenburg, S., Lange, C., Jäkel, E., Schäfer, M., and Wendisch, M.: Retrieval of snow layer and melt pond properties based on airborne hyperspectral imagery, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-3304, https://doi.org/10.5194/egusphere-egu23-3304, 2023.

Arctic snow cover dynamics exhibit modification in terms of extent and duration due to recent changes in climate, i.e. increasing temperatures and changing precipitation patterns, i.e. winter rain-on-snow events (WROS). Remote sensing methods based on active radar images (SAR) have demonstrated a significant advantage for snow monitoring, (i) capturing physical and dielectric properties, (ii) overcoming the weakness of optical images limited by cloud cover and polar night.

The aim of this study is dedicated to the analysis of the spatial and temporal variability of snow cover in the Ny-Ålesund area on the BrØgger peninsula, Svalbard (N 78°55’ / E 11° 55’). In-situ snow measurements from two automated weather stations (Ny-Ålesund, and Bayelva), regular snowpits around the village and in spring on the Austre Loéven, were compared with the spaceborne dataset.

The RADARSAT Constellation Mission (RCM) is comprised of three satellites into closely coordinated orbits operating in C-band (5.4 GHz, 5.5 cm). The high temporal (4-day repeat cycle) and spatial resolution of the sensors in Quad-Pol mode (9-m) or Compact-Pol mode (5-m) provide a helpful performance for detecting the spatial variability of snow properties. RCM data are also compared to images of the TerraSAR-X satellite (DLR, Germany) operated in X-band (9.6 GHz, 3.1 cm) at 5-m spatial resolution. Both RCM QP mode and TSX data were acquired with medium incidence angles (33° to 39°) providing better snow penetration for volume backscattering. The RCM CP data were only available under low (23°) and high (53°) beam angles.

The following two snow properties were analyzed:

WROS detection: the focus was set on the 16-17 March 2022 event (+ 5.5 °C, 62 mm). RCM data at cross-polarization VH or HV can clearly detect the impact of rain on snow, indicating an intensity drop of -10 dB, even on the glacier at high elevation.

Snow depth retrieval: the study covers spring 2021 (March-June) and the complete winter season 2021-2022 (November-June).

- Concerning QP mode, better correlation between snow depth and SAR backscattering is observed by the cross-pol VH component, retrieving more volume backscattering information than co-pol configuration or total backscattering power (Span). We observe also that descending orbit images (06 :30 AM) provide a better correlation with snow depth than ascending orbit (15 :30 PM) data.

- Concerning CP mode and Span (RH+RV), the low incidence images (23°) do not match the snow depth observations due to main surface backscattering, contrariwise the high incidence images correlate better with in-situ observations. The analyses of the Stokes vector elements showed a satisfying correlation for the g3 element and the Relative Phase polarimetric decomposition.

Finally, a comparison of Span temporal values between RCM at C-band and TSX at X-band indicates similar time profiles, but clearly lower values of -5 to -10 dB at the C-band.

How to cite: Dedieu, J.-P., Van Noort, J., Montpetit, B., Levistre, M., Vauclare, S., Wendleder, A., Boike, J., Bernard, E., Gallet, J.-C., and Jacobi, H.-W.: Snow cover monitoring in the Arctic (Svalbard) with RADARSAT Constellation Mission (RCM). Comparison with in-situ measurements and TerraSAR-X data, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-4196, https://doi.org/10.5194/egusphere-egu23-4196, 2023.

How to cite: Jans, J.-F., Beernaert, E., Lievens, H., and Verhoest, N.: Dynamics in mountain SNOW water resources by MODEs of climate variability assessed from satellite observations, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6176, https://doi.org/10.5194/egusphere-egu23-6176, 2023.

Seasonal snow is an essential source of water, especially in mountainous regions. However, accurate satellite observations of the snow water equivalent (SWE), i.e., snow depth multiplied by the snow density, are still lacking. Therefore, new and robust remote sensing techniques are urgently needed. This study presents a novel method for SWE retrieval in mountainous regions at sub-weekly temporal and 500-m spatial resolution, based on snow depth observations from the ESA and Copernicus Sentinel-1 (S1) satellite mission and model simulations of snow density. The snow depth observations rely on a change detection algorithm which translates the temporal changes in the S1 radar backscatter measurements into the accumulation or ablation of snow. The snow density estimates are obtained from different modeling approaches, including empirical methods (e.g., based on the day of the year, the snow depth, snow climate class, etc.) and a physics-based mass and energy balance model. The performance of the different snow density modeling approaches is here compared, both with respect to their ability to accurately simulate in situ measurements of snow density, as well as their ability to accurately simulate in situ measurements of SWE after combination with the S1 snow depth observations. The performance is evaluated over the European Alps, using a large dataset of in situ time series measurements for the period 2015-2022. The results show that the physics-based snow density modeling approach outperforms the empirical approaches, yielding high spatio-temporal correlation between S1 SWE retrievals and in situ measurements. Therefore, the study demonstrates the capability of the Sentinel-1 satellite mission, in combination with a physics-based snow model, to accurately represent the spatio-temporal distribution of SWE in mountainous regions, which can benefit a large range of applications, including hydropower generation, water management, flood forecasting, and numerical weather prediction.

How to cite: Boeykens, L., Lievens, H., Beernaert, E., Jans, J.-F., and Verhoest, N.: SWE retrieval in the European Alps based on Sentinel-1 snow depth observations and modeled snow density, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-7050, https://doi.org/10.5194/egusphere-egu23-7050, 2023.

Snowmelt can vary largely across time and space, especially in complex terrain. However, hydrological models often represent snowmelt using a single static degree-day factor that relates the melt runoff with air temperature. Seasonally or spatially varying degree-day factors have been shown to better capture the snowmelt heterogeneity, but still rely on simplified parameterizations. One interesting solution proposed in the literature is to use MODIS satellite imagery to capture the true snowmelt heterogeneity, and use it to inform hydrological models on the temporal and spatial evolution of the degree-day factor on a near-daily basis. However, the limited spatial resolution of MODIS makes this process difficult to apply in complex mountainous terrain. Meanwhile, Landsat or Sentinel 2 satellite imagery could be an interesting alternative as they have a much higher spatial resolution but fall short in terms of temporal resolution. In this study, we overcome both these obstacles with a synthetically generated daily snow cover time series based on Landsat resampling. We use the daily synthetic snow cover maps to derive the snow cover depletion in each coarse resolution hydrological model grid cell, which in turn defines the degree-day factor for each cell using a transfer function. To capture the inherent uncertainty of this methodology, we run an ensemble of models using different meteorological forcings and different stochastic realizations of the synthetic snow cover maps. The resulting degree-day factors are evaluated through the skill of the modeled streamflow and snow water equivalent, using different transfer functions in several snow-influenced catchments in Switzerland. 

How to cite: Wiersma, P., Zakeri, F., and Mariéthoz, G.: Using synthetic snow cover maps to determine the degree-day snowmelt factor of a distributed hydrological model, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-7343, https://doi.org/10.5194/egusphere-egu23-7343, 2023.