In order to fully utilise the wealth of satellite and close-range sensing data available, the last decade has seen reliance on new approaches for (i) accessing, (ii) processing, (iii) interpreting, and (iv) distributing results from large-scale datasets. This includes new technologies for data access including cloud-optimised datasets; cloud geoprocessing platforms such as Google Earth Engine, Microsoft Planetary Computer, and community JupyterHubs; the increasing use of large-scale data pipelines and machine/deep learning methods to understand and monitor entire ice sheets, ice shelves, or glaciated regions; and a widespread philosophy of open data and code sharing to enable rapid dissemination of new approaches.
This session seeks contributions from anyone working on remote sensing and close-range sensing of any element of the cryosphere, including glaciers (both land-based or calving), ice sheets, snow and firn, glacial and periglacial environments, and sea ice. In particular, we welcome submissions from those researching the cryosphere using cloud data and processing, large-scale data pipelines, machine and deep learning, open code/data, and other contemporary approaches. In addition, contributions may include the use of close-range sensing methods, including, but not limited to, uncrewed aerial vehicles (UAVs), radar, time-lapse photography, TLS and LiDAR.
Orals: Fri, 8 May, 14:00–17:25 | Room L2
The growth of ice slabs influences both the spatial extent and the rate at which the Greenland Ice Sheet’s wet snow zone transitions from storing meltwater in firn, to exporting it as surface runoff. These changes reorganize meltwater storage and flow pathways, shaping the ice sheet’s contribution to sea level rise. In this work, we combine Sentinel-1 (S1) C-band satellite radar mosaics with Operation IceBridge (OIB) airborne radar profiles to produce a decade long time series of ice slab expansion in Greenland. To efficiently process the large S1 data volume, we use Google Earth Engine to compile Interferometric Wide and Extra Wide Swath data acquired during boreal winters (1 November – 30 March) from 2015 to 2025, producing annual co-polarized (HH) and cross-polarized (HV) backscatter mosaics that are multilooked and linearly corrected to a common incidence angle. We then use logistic regression to optimize ice slab detection thresholds and to quantify classification uncertainty.
Our time series reveals pronounced differences in the rate of ice slab expansion between northern and southwest Greenland. Isolated OIB radargrams also suggest marked differences in ice slab geometry between these two regions. In northern Greenland, thick downflow ice slabs transition abruptly into laterally extensive, thin ice slabs that extend tens of kilometers upslope into the percolation zone. In contrast, ice slabs in southwest Greenland either remain thick at their upflow fronts or, when thin, occur deeper in the firn column and are rapidly buried by subsequent accumulation events. To capture these contrasting ice slab front geometries, we develop a refined classification scheme to map thick and thin ice slabs across the Greenland Ice Sheet using Sentinel-1 backscatter thresholds. The observed spatial and temporal patterns point to regions where atmospheric forcings and percolation zone firn conditions have restricted meltwater infiltration depth, accelerating shallow ice slab growth and altering the near-surface hydrologic regime.
How to cite: Mutter, E. and Culberg, R.: Sentinel-1 based time series of ice slab extent reveals regional divergence in ice slab evolution, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6029, https://doi.org/10.5194/egusphere-egu26-6029, 2026.
The Greenland Ice Sheet is a major contributor to global sea-level rise, with its mass loss raising global sea-levels by 13.59 ± 1.27 mm from 1992 to 2020. Changes to the ice dynamics of the Greenland Ice Sheet make up a significant part of this mass loss, meaning that monitoring and understanding changes in ice dynamics is crucial to assessing contemporary sea-level rise and making projections of its future.
Since 2014, the European Space Agency’s Sentinel-1 synthetic aperture radar constellation has provided near-continuous, weather-independent coverage of the Greenland Ice Sheet, with 6–12 day repeat intervals over almost all outlet glaciers. We process the full Sentinel-1 archive using an intensity feature offset tracking workflow implemented in GAMMA Remote Sensing to produce a dense, decade-spanning time series of ice velocity at 100 m resolution. This Sentinel-1 derived ice velocity dataset is extremely valuable, capturing long-term speed trends and short-term interannual variability due to seasonal drivers. However, analysing this 60 TB dataset efficiently and obtaining maximum scientific value from it is a substantial data engineering challenge, as the data volume greatly exceeds available memory quotas on HPC systems, even for relatively small geographic areas.
Here, we address this challenge with an analysis pipeline built on the Xarray, Dask, and Zarr python packages, and deployed on a HPC service. This pipeline allows both large scale interactive analysis in Jupyter notebooks, streaming to GIS software, and traditional batch processing. We leverage these tools we calculate ice-sheet wide speed change, seasonal speed variability and ice discharge in a fast, reproducible and scalable manner.
Our results reveal a highly heterogeneous dynamic response across Greenland, with neighbouring glaciers often exhibiting contrasting behaviour over the past decade. We find that short-term and seasonal variability dominates the velocity signal for most glaciers, often exceeding long-term speed changes. These findings highlight the importance of resolving short-term ice-dynamic processes when assessing Greenland’s contribution to future sea-level rise and the benefit of efficient big-data processing workflows.
How to cite: Wallis, B., Hogg, A., Rigby, R., and Slater, R.: Patterns and trends of ice dynamic variability on the Greenland Ice Sheet from a decade of high-resolution synthetic aperture radar data, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11163, https://doi.org/10.5194/egusphere-egu26-11163, 2026.
Buried lakes are widely distributed on the Greenland Ice Sheet (GrIS) after summer. Some of these lakes may drain over winter, thereby delivering meltwater into the ice sheet and potentially influencing ice flow dynamics. However, to date, only a limited number of buried lake drainages (BLDs) have been identified and their spatiotemporal dynamics across the GrIS remain unclear. Here we first detect pan-GrIS wintertime BLDs by integrating Sentinel-1 and -2 satellite imagery and ArcticDEM data, and then investigate potential ice velocity anomalies (IVAs) triggered by BLDs using ice velocity data. The results show that a total of 167 complete and partial BLDs are identified over seven winters from 2017 to 2023 on the GrIS, including 25 cascade drainages. Ten significant IVAs are observed in association with BLDs, and they may lead to net increases in annual ice motion. A representative cascade drainage further reveals that partial BLDs may result from the combined effects of the fracture location and depression topography. Meanwhile, the depression topography and associated water storage capacity may be markedly altered by wintertime drainage. Notably, beyond triggering IVAs of up to ~50%, some BLDs can even trigger rerouting of subglacial hydrologic pathways within a few days.
How to cite: Wei, J., Yang, K., Zhu, Y., Wang, Y., and Guo, X.: Greenland wintertime buried lake drainage and ice dynamics response, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6153, https://doi.org/10.5194/egusphere-egu26-6153, 2026.
Three decades of routine satellite altimetry have provided an ice-sheet-wide, near-continuous observational record of polar topography, offering unparalleled insights into ice sheet elevation change. The ability of CryoSat-2, Sentinel-3 and ICESat-2 to simultaneously and continually monitor Earth’s ice surfaces is critical towards understanding the ongoing and future imbalance of the icy continents in a changing climate. To capitalise fully on the vast quantity of altimetry data from numerous coincident missions, it is important for robust, consistent and traceable uncertainties to be provided alongside measurements of ice sheet elevation. Such information is crucial for the successful combination of measurements across missions and to enable their use in downstream applications, such as the constraint of numerical ice sheet models. At present, such uncertainties are largely absent from existing ice sheet elevation products, and for the subset of products where uncertainties are provided, there is neither a standardised approach to uncertainty generation nor a method to evaluate their robustness.
Here, we present a new framework for generating and evaluating the performance of uncertainties for altimetry-based ice sheet elevation measurements and provide a comprehensive assessment of uncertainty generation for ice-sheet-wide altimetry-based ice sheet elevation datasets. Overall, we find that calculating uncertainty as a parameterisation of topographic complexity (characterised by surface slope and roughness) and measurement quality (characterised by backscattered power and coherence) improves performance relative to solutions that use fewer co-variates. Ultimately, the framework presented here will enable the systematic characterisation of ice-sheet-wide elevation uncertainties associated with historical, current and future polar radar altimeter missions, which will be particularly important as the portfolio of polar radar altimeters continues to grow with the planned launch of the Copernicus Polar Ice and Snow Topography Altimeter (CRISTAL) in 2027. Such uncertainty information will aid the successful amalgamation of the vast array of multi-mission altimetry measurements using techniques such as Kalman Filtering and improve the constraint of numerical ice sheet models, which will in turn enable more refined estimations of current and future ice sheet mass balance and global sea-level rise.
How to cite: Boxall, K., McMillan, M., Muir, A., Appleby, S., Dubber, S., Gourmelen, N., Willis, C., and Phillips, J.: A framework for evaluating ice-sheet-wide altimetry uncertainty estimates, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12980, https://doi.org/10.5194/egusphere-egu26-12980, 2026.
Satellite radar altimetry provides three decades of continuous ice sheet observations, yet non-interferometric systems face a fundamental constraint: each waveform records only the return time of reflections, not their origin within the beam footprint. As a result, current approaches - which have remained largely unchanged for 30 years - rely on simplifying assumptions and fail to extract the full information content from the recorded echoes. Interferometric processing resolves this by measuring arrival angle with two antennas, enabling swath processing with multiple across-track elevation points. However, historical missions (ERS-1/2, Envisat), current operations (Sentinel-3), and planned systems (CRISTAL Ka-band) lack this capability.
Conventional processing instead eliminates ambiguity through dimensionality reduction: retracking identifies a single range from the leading edge, while slope correction locates the point of closest approach (POCA) to attribute this to. This approach discards most of the waveform information, extracting only singular elevation estimates.
Here, we present a fundamentally alternative approach to processing radar altimetry echoes over ice sheets using probabilistic deep learning to extract the distribution of plausible surface elevations encoded within each waveform. Rather than reducing ambiguity through simplifying assumptions, we treat the full waveform as exploitable information, directly modelling the distribution of surface elevations consistent with each measurement.
Specifically, we train an ensemble of 16 ResNet-RS models on 600,000 CryoSat-2 SARIn power waveforms (2012–2022) over Antarctica using the Reference Elevation Model of Antarctica (REMA) as ground truth. The framework predicts 5th, 50th, and 95th elevation quantiles across 150 points spanning the 15 km swath, providing 150 elevation estimates where conventional methods yield one. Importantly, our new probabilistic approach allows us to directly quantify both aleatoric (surface ambiguity) and epistemic (model confidence) uncertainty.
Once trained, we validated the models against unseen ice-sheet-wide REMA data (2023), demonstrating robust prediction of swath elevation distributions. Prediction interval coverage (5th–95th) averaged ~80% (10 points below the 90% nominal target) ice-sheet-wide, with consistent performance achieved with respect to ICESat-2 over Pine Island Glacier (2019–2025) - a region entirely withheld from training. The upper 95th percentile exhibited strong stability across all conditions, reflecting its physical anchoring in leading-edge returns, while lower quantiles showed systematic under-coverage increasing with topographic complexity.
Practical utility of this new approach was demonstrated through elevation change monitoring over Pine Island Glacier, where our deep learning framework reproduced established spatial thinning patterns of 2-3 m yr-1 with temporal elevation change differences of -0.03 ± 0.11 m yr-1 relative to ICESat-2. Despite using only power waveforms, performance matched CryoSat-2's interferometric POCA and EOLIS products - detecting sub-metre annual changes from probability distributions spanning tens of metres.
This represents the first swath processing from non-interferometric altimetry using power waveforms alone. By reframing waveform ambiguity as quantifiable distributional information, rather than a processing limitation, this deep learning approach demonstrates how machine learning can fundamentally rethink conventional altimetry processing, establishing new capabilities for ice sheet monitoring across past, present, and future missions.
How to cite: Phillips, J., McMillan, M., and Maddalena, J.: Extracting Swath Elevation Information from Non-Interferometric Radar Altimetry using Probabilistic Deep Learning, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15091, https://doi.org/10.5194/egusphere-egu26-15091, 2026.
Antarctic ice shelves buttress grounded ice and play a critical role in regulating glacier stability and the mass balance of the Antarctic Ice Sheet. Over recent decades, many Antarctic ice shelves have exhibited significant mass loss. Crevasses, which formed in response to internal stresses, are key indicators of ice shelf stability and their development is intricately coupled to the evolving ice dynamics. However, consistent long-term and high-resolution records of crevasse evolution over Antarctic ice shelves remain limited.
Here, we present a continuous and automated mapping of crevasses on representative Antarctic ice shelves from 1999 to 2024 using Landsat 7 and Landsat 8 imagery and a deep-learning-based segmentation framework at 30 m resolution. A manually delineated dataset based on Landsat 8 RGB imagery from multiple ice shelves, encompassing a wide range of crevasse morphologies, was constructed for model training and validation. To address the scan-line corrector (SLC) failure of Landsat 7 ETM+ since 2003, we developed a diffusion-based gap-filling approach trained on a dataset specifically constructed for this study, enabling consistent crevasse mapping across the full Landsat 7/8 archive.
Our results reveal pronounced crevasse development on Pine Island, Thwaites, and Larsen B Ice Shelves over the past two decades, while other mapped ice shelves exhibit more moderate or minimal changes. This long-term, high-resolution crevasse mapping provides new insights into ice shelf damage evolution and offers valuable constraints for damage parameterization and assessments of ice shelf stability. The developed pipeline is readily extendable to additional ice shelves and remains computationally efficient.
How to cite: Tang, L., Bamber, J. L., and Qiao, G.: Deep-learning-based mapping reveals multi-decadal crevasse evolution on Antarctic ice shelves from Landsat imagery, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-5118, https://doi.org/10.5194/egusphere-egu26-5118, 2026.
The albedo lowering of glacier and ice sheet surface is driven by a complex combination of biogeophysical processes, including the accumulation of biologically active impurities such as glacier ice algae and dispersed cryoconite. It is therefore important to understand darkening processes by ground observations and sampling. In situ sampling requires a simple yet effective approach to quantify surface ice darkness and estimate the possible range of impurity concentrations. Traditional post-processing of snow and ice samples for cell and mineral counts can also be time and resource consuming.
In this study, we present a novel toolbox for calibrating field images acquired with consumer-grade cameras (e.g., smartphones or digital cameras) to estimate surface ice darkness. Using a data-driven approach, we tested the method on snow and ice samples collected in Greenland to predict impurity ranges. The toolbox also integrates advanced AI models for automated segmentation of microscope images and classification of snow algae, glacier ice algae, and mineral particles, enabling rapid impurity quantification. The new toolbox can enable researchers across fields to cross compare fieldwork samples. It also offers potential for integration into automated weather stations for long-term monitoring programs, advancing glacier surface darkening characterization and biogeophysical research.
How to cite: Anesio, A., Feng, S., Olivas, B. G., Broadwell, E. L. M., Peters, R. S., Benning, L. G., and Tranter, M.: From field photos to microscope images: a low-cost toolbox for glacier surface darkening and biological impurity analysis, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16858, https://doi.org/10.5194/egusphere-egu26-16858, 2026.
One of the main consequences of climate change is undoubtedly rising temperatures, which cause the polar ice caps in the Arctic and Antarctic regions to melt and, as a result, icebergs to break off. Due to their high variability in shape and size, their movements and trajectories are not easy to predict, posing a threat to maritime safety and offshore activities. Particularly, along the western coast of Greenland (e. g. in the Davis Strait), in addition to the growing number of icebergs, there has also been an increase in maritime traffic, which has tripled due to the opening of new trade and tourist routes.
Historically, icebergs were detected exclusively by naval sightings or aerial patrols, which were inherently limited in terms of accuracy and unable to reach remote or inaccessible regions. In this contemporary era, advancements in satellite technology have significantly transformed the way in which icebergs are observed, ensuring high temporal and spatial resolution, global and large-scale acquisition, as well as accessible free near-real-time data. The majority of past and present studies focusing on the detection and tracking of icebergs using satellite imagery have employed fixed threshold methodologies. These tend to be prone to false alarms, offer limited sensitivity, and are not easily exportable or automatable due to their heavy dependence on observation time and location. The integration of multispectral information with diagnostic elements from a spectral perspective is a fundamental aspect of the aforementioned techniques.
In order to overcome the limitations described above, this study proposes a multi-temporal differential approach for detecting and mapping icebergs along with other objects that could potentially compromise navigation. This methodology is referred to as Robust Satellite Techniques (RST), and it has been developed to identify statistically significant variations in the signal under investigation, at the pixel level. The RST model, already widely utilised in the domain of natural hazards, is being employed for the first time in the detection of icebergs. The test was specifically conducted within a segment of the ocean in the surrounding area of Nuuk (Greenland), located in the Davis Strait. A preliminary application of the RST approach, just to the visible band of Sentinel-2/MSI, demonstrated, compared with other multi-spectral, fixed threshold approaches, higher sensitivity and reliability, together with an easy and immediate exportability in different geographic areas and observation periods. The methodology was implemented in the Google Earth Engine (GEE) environment, which allows the process to be fully automated, easily exportable and rapidly executable.
How to cite: Mancusi, I., Colonna, R., Filizzola, C., Genzano, N., and Tramutoli, V.: Robust Satellite Techniques for detecting and monitoring icebergs, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9074, https://doi.org/10.5194/egusphere-egu26-9074, 2026.
Modern constellations of Earth observation satellites offer exciting opportunities to observe and understand cryosphere change with unprecedented spatial and temporal coverage, resolution, and accuracy. This ever-growing data firehose requires new scalable processing approaches to extract local, regional, and global results and actionable insights for downstream applications. Mulit-modal data fusion (e.g., integration of commercial very-high-resolution optical stereo, laser altimetry, and high-resolution SAR/InSAR) can capture mm- to meter-scale surface change, offering new insight for a range of geophysical processes responsible for cryosphere change.
We highlight a set of scalable, open-source satellite data processing tools and approaches for high-mountain cryosphere science applications, including:
1) Sliderule Earth - an open-source service for on-demand, parallel processing of science data archives in the cloud. SlideRule allows the user to create customized, high-level, analysis-ready data products in near-real-time. SlideRule currently supports a suite of data products and algorithms for the NASA ICESat-2 and GEDI satellite laser altimetry missions, as well as many cloud-hosted raster datasets (e.g., PGC ArcticDEM/REMA strips and mosaics, Harmonized Landsat-Sentinel, ESA WorldCover 10m, USGS 3DEP airborne lidar).
2) Intra- and interannual snow depth measurements from ICESat-2 satellite laser altimetry and commercial optical stereo
3) Global annual high-resolution snow melt runoff onset maps for the past decade from the cloud-hosted archive of radiometric terrain-corrected (RTC) Sentinel-1 SAR backscatter
4) Fused InSAR and SAR feature tracking to map debris-covered ice and evolving high-mountain hazards, including the seasonal and interannual motion of glacier lake moraine dams.
These satellite datasets and approaches offer an improved understanding of the processes driving recent cryosphere change, which is essential to improve the models used for future projections, and to understand connections with the hydrologic cycle.
How to cite: Shean, D., Swinski, J.-P., Gagliano, E., Brencher, G., Besso, H., and Henderson, S.: Scalable satellite data processing methods for ICESat-2, optical stereo and SAR/InSAR to understand cryosphere change, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-16192, https://doi.org/10.5194/egusphere-egu26-16192, 2026.
Glaciers in the Alps are receding at unprecedented rates. Firn area is declining even at high elevations and many glaciers have experienced completely snow free summers in recent years. Glacier albedo is an important control for the amount of energy available for melt and, hence, mass balance and future glacier evolution. Glacier-scale studies have shown considerable variability of glacier-wide and bare-ice albedo in both space and time. At regional and Alps-wide scales, analyses of MODIS data have found decreasing albedo trends over time. However, the relatively coarse resolution of MODIS (500 m) does not resolve small-scale variability and limits applicability especially for very small glaciers, which are numerous in the Alps. Deriving glacier-wide albedo products and addressing albedo variability over bare-ice areas using Landsat and Sentinel-2 multispectral surface reflectance (10 to 30 m resolution) has the potential to improve understanding of albedo driving factors and, for example, resolve regional impacts of heatwaves and other meteorological forcings. Landsat and Sentinel-2 surface reflectance products are available via Google Earth Engine (GEE) and the computational accessibility enabled by server-side operations within GEE allows flexible analyses at scale. For example, a glacier-wise analysis of the Sentinel-2 record indicates that the median snow cover fraction for glaciers in Austria dropped to around 10 % during the record-breaking summer of 2022, compared to values between 20 and over 30 % in the previous years. Applying a broadband albedo conversion to the multispectral reflectance data, we find the median glacier area fraction with very low albedo values below 0.2 increased to over 35 % in 2022. At elevations above 3000 m, median glacier albedo ranged from 0.4 to 0.5 in years prior to 2022 and dropped to below 0.3 in 2022, with persistently low values through 2025.
In principle, these GEE workflows can be scaled to much larger regions and thousands of individual glaciers without great difficulties. However, “traditional” challenges related to cloud cover, local topographic shading, data availability, and validation approaches remain. It has become relatively simple to produce values like the preliminary results given above and, for example, greatly extend the dataset by including the entire Landsat record. However, given the large amount of readily available data across different collections and generations of satellites, care should be taken in accounting for issues such as sensor comparability and level-2 product consistency, and developing meaningful validation metrics seems particularly important. We will present our ongoing work related to ice albedo and firn loss in the European Alps and aim to foster discussions of challenges and limitations that arise when scaling analyses from individual glaciers to larger regions.
How to cite: Hartl, L., Di Mauro, B., Fugazza, D., and Naegeli, K.: Deriving glacier albedo time series from multispectral satellite data in the Alps - insights and challenges from regional applications, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17314, https://doi.org/10.5194/egusphere-egu26-17314, 2026.
Current GLOF (Glacial Lake Outburst Floods) risk assessments in High Mountain Asia (HMA) are often limited by static glacial lake inventories and unreliable Area-Volume (A-V) scaling models, which are not widely validated in the region. In this study, we developed a fully automated framework integrating multi-source remote sensing (e.g. Landsat, Sentinel-1/2, Copernicus DEM) and satellite altimetry (ICESat-2) to monitor glacial lake dynamics. We map nearly 32,000 glacial lakes across HMA for 2022 with a detection accuracy of 96% and relative boundary accuracy of 98% for lakes >20,000 m², offering a significant improvement over manual inventories. Our analysis shows three important results. First, regional heterogeneity is pronounced across the HMA sub-regions. For instance, East Kun Lun exhibited the highest expansion rate (8% area increase, 2016–2022), highlighting the need for targeted hazard assessment, while West Himalaya showed minor change (0.04%). Second, by processing ICESat-2 data for >16,000 lakes, we validated widely used A-V scaling models. Our results demonstrate that large supraglacial and extra-glacial lakes exhibit volume estimation errors exceeding 1500% in standard models, highlighting a significant bias in current flood volume potential estimates. Third, our automated temporal monitoring identified four GLOF events during the study period that were not previously documented. Analysis of these events reveals distinct area peaks before GLOFs, providing quantifiable indicators for early warning. This framework utilizes open-source remote sensing data on the Google Earth Engine cloud platform for regular monitoring of glacial lakes with higher accuracy while providing a reproducible, scalable method to correct volume estimates and detect hidden GLOF events in high-mountain valleys.
How to cite: Kumar, R. and Vijay, S.: Automated Multi-Sensor Framework for Glacial Lake Dynamics and Unreported GLOF Detection across High Mountain Asia, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-702, https://doi.org/10.5194/egusphere-egu26-702, 2026.
Accurate and up-to-date glacier outlines are essential for quantifying glacier retreat and enabling downstream applications, such as mass balance assessments and glacier modeling. While manual delineation of glaciers remains common, automated approaches offer faster and more consistent mapping. However, even sophisticated deep learning methods show limited performance on debris-covered glaciers, which are difficult to differentiate from periglacial terrain. Optimizing input features is a key strategy to overcome this barrier. Synthetic Aperture Radar (SAR) interferometric coherence complements optical data effectively, as moving ice surfaces typically exhibit lower coherence than stable terrain.
We present a new dataset covering glaciated areas of the European Alps from 2015 to 2025, consisting of annual composites of Sentinel-1 coherence and backscatter. We have developed an automated pipeline to process multiple TB of Sentinel-1 Single Look Complex (SLC) data. For each year, bursts acquired during minimal snow cover were selected and combined to ensure optimal visibility of glacier ice. Backscatter was radiometrically terrain-flattened, and both coherence and backscatter were geocoded to the UTM coordinate system. To mitigate the high spatial resolution variability inherent to SAR imaging in mountainous terrain, we applied the local resolution-weighting approach (Small, 2012). This method combines ascending and descending acquisitions, weighting contributions by the inverse of the local contributing area to preserve the highest resolution available.
The resulting dataset provides analysis-ready annual composites at 10 m grid spacing and will be made publicly available. This enables the development and benchmarking of more robust glacier mapping methods across the European Alps, particularly in challenging debris-covered areas.
How to cite: Gentner, L. Q., Small, D., Piermattei, L., and Wulf, H.: Optimizing SAR for Glacier Mapping: A Resolution-Weighted Coherence and Backscatter Dataset of the European Alps, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-17377, https://doi.org/10.5194/egusphere-egu26-17377, 2026.
snowMapper is a physics-informed model developed to generate daily reconstructions of snow cover across complex mountain terrain. The system integrates in situ measurements with gridded meteorological inputs and incorporates binary snow presence/absence observations derived from high-resolution satellite imagery. Its modular design enables users to tailor configurations to specific study sites, producing daily snow-cover maps at spatial resolutions typically ranging from 20 to 100 meters. The workflow includes a preprocessing pipeline compatible with imagery from Landsat 4–9 and Sentinel-2; multiple terrain- and land-cover–based masking options (i.e., forest, glaciers, surface water, elevation, urban areas); five configurable schemes for converting satellite reflectance to binary snow cover; and a quasi–physically based downscaling of climate variables. Snow-cover reconstruction is accomplished through two sequential, configurable gap-filling procedures: an initial decision-tree step followed by a machine-learning classifier. The classifier can be trained either with local field observations or with in situ data originating from other regions, allowing the model to operate in a fully physics-informed mode in the absence of a local monitoring network. A built-in evaluation module compares model outputs with satellite-derived snow cover, providing accuracy metrics directly within the final product. Optional aggregation routines allow fractional snow-cover metrics to be generated across temporal and spatial scales. The system operates entirely on Google Earth Engine via its Python API, reducing dependence on local data storage and eliminating local computational demands. We applied snowMapper to generate a 41-year snow-cover climatology for Greek mountains exceeding 2,000 m a.s.l. The resulting daily 100 m climatology consisted of over 90 % modeled values, and achieved a mean overall accuracy of 93 % when assessed against 1.1 billion clear-sky, pixel-level satellite observations. The model code and example data are available as an open-source project on GitHub (https://github.com/snowMapper/snowMapper, last access: 25 November 2025).
How to cite: Alexopoulos, K., Willis, I. C., Pritchard, H. D., Kyros, G., Kotroni, V., and Lagouvardos, K.: snowMapper v1.0: a model for daily mountain snow cover reconstruction in high resolution, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-347, https://doi.org/10.5194/egusphere-egu26-347, 2026.
Many snow processes operate at the slope scale and sub-daily timescales, yet observations at these scales remain scarce. We demonstrate a mobile, automated monitoring approach that captures snowpack evolution in mass, thermal state, and liquid water content at the slope scale and sub-daily resolution.
The system combines lidar and photogrammetry with a new RFID-based microwave method, which uses a cable car to repeatedly observe an entire slope. The RFID approach relies on passive tags installed on the ground and covered by snow. When the cable car passes above them, a mobile reader interrogates them at 865–868 MHz, and they respond using backscattering communication. The variation in the signal's phase and signal strength over time is used to monitor snow water equivalent and liquid water content. In parallel, lidar and photogrammetry resolve snow depth and microtopography, while temperature sensors embedded in the tags measure basal snow temperature.
The approach was first implemented on the Pischa slope near Davos, Switzerland. During the 2024–2025 winter, more than 1,000 runs were completed along a 2 km transect spanning 700 m of elevation, and a second campaign is ongoing in the 2025–2026 winter.
The resulting dataset, unique in its spatiotemporal resolution, opens new possibilities to observe rapid snow processes at the slope scale.
Related references:
- Le Breton, Mathieu et van Herwijnen, Alec. 2025. « High-Resolution Aerial Snowpack Monitoring via Passive RFID (GONDO-RFID) ». P. 1‑6 in 2025 IEEE International Conference on RFID Technology and Applications (RFID-TA).
- Charléty, Arthur, Mathieu Le Breton, Morgane Magnier, Éric Larose, Laurent Baillet, Ludovic Moreau, et Alec van Herwijnen. 2025. « Locating RFID Tags Under Snow and Vegetation ». P. 1‑5 in 2025 IEEE International Conference on RFID Technology and Applications (RFID-TA).
How to cite: Le Breton, M. and van Herwijnen, A.: High-Resolution Snowpack Monitoring from a Multi-Sensing Gondola , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8399, https://doi.org/10.5194/egusphere-egu26-8399, 2026.
The ESA CCI Soil Moisture product provides a global long-term consistent data record of soil moisture from 1978 up to the present day. It is produced by fusing measurements from a total of 19 both passive and active satellite microwave observations. Determining the physical state of water in the soil is crucial for this product, as retrievals of soil moisture are unreliable in frozen or snow-covered conditions. Beyond its role as a quality control variable, the freeze/thaw state is itself a key environmental indicator, as it impacts the exchange of energy and water between land and atmosphere, shapes seasonal hydrological cycles, and influences agriculture, ecosystems, and climate feedbacks.
As such, as of version 9.2, the ESA CCI Soil Moisture project is providing an additional global Freeze/Thaw dataset, covering the period from November 1978 to December 2024 with daily temporal and ~25km spatial resolution. This dataset consists of a binary classification (frozen or thawed), the total number of available sensors, the number of sensors detecting frozen soils, and an agreement index for each datapoint. The product combines frozen flags from both active and passive sensors in a conservative manner, i.e. a datapoint is classified as frozen if the frozen soil classification of at least one sensor yields a positive result.
For passive sensors, the classification of frozen soils follows a decision-tree based algorithm, which incorporates vertically polarized brightness temperature measurements at three different frequencies (Ka/K/Ku-band). These frequency bands are consistently available from 1978 onwards for all sensors included in the CCI SM product except for the L-band sensors SMOS and SMAP, which are thus excluded from the processing of the Freeze/Thaw product. For active sensors, the surface state flag indicator provided by EUMETSAT H SAF is incorporated directly and a datapoint is marked as ‘frozen’ if its value is other than 1, which includes temporarily frozen soils, permanent ice, and melting water on the surface.
The current product achieves an estimated accuracy of 75% against in situ surface temperature observations from ISMN and 92% compared to ERA5 reanalysis temperature fields data. The unanimity rule leads to some over-flagging and will thus be refined in future versions. Furthermore, the classification algorithm will be optimised for each sensor and its uncertainty quantified in order to merge individual classifications more robustly.
How to cite: Formanek, M., Preimesberger, W., Lems, J., and Dorigo, W.: The ESA CCI SM Freeze/Thaw Product: Global detection of frozen grounds from 1987 to the present , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13682, https://doi.org/10.5194/egusphere-egu26-13682, 2026.
Persistent cloud contamination in MODIS normalized difference snow index (NDSI) products severely limits reliable snow cover monitoring across the Northern Hemisphere, making effective gap-filling crucial. However, existing approaches often oversimplify snow temporal dynamics and fail to capture the cumulative nature of snow–climate interactions. To address these limitations, we propose PredFormer, a novel sequence-based framework that extends the self-attention mechanism to the temporal dimension, thereby explicitly modeling the nonlinear temporal evolution of NDSI coupled with cumulative meteorological effects. The framework further incorporates vegetation growth, topographic conditions, and cloud mask information to account for complex environmental dependencies. Validation results demonstrate that PredFormer achieves superior reconstruction performance across the Northern Hemisphere, with MAE and RMSE values of 1.110 and 2.946, respectively. Notably, the distinct mechanism of incorporating nonlinear cumulative meteorological effects reduces RMSE by 11.57%, and comparative analyses against baseline models (e.g., LSTM and standard Transformers) reveal performance gains exceeding 19%. The framework also demonstrates substantial improvements in high-elevation and forested regions—including the Hindukush–Himalaya and the West Coast. Leveraging this framework, we generate the first hemispheric-scale, daily cloud-free NDSI dataset (5-km resolution, 2003–2023). This work not only advances the methodological handling of nonlinear snow dynamics but also delivers a foundational dataset for hydrological modeling and climate change assessment.
How to cite: Xu, J., Huang, Y., Lhermitte, S., Hua, R., and Yu, B.: Daily Cloud-Free NDSI Reconstruction Across the Northern Hemisphere Driven by Nonlinear Meteorological Forcing, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-308, https://doi.org/10.5194/egusphere-egu26-308, 2026.
Posters on site: Thu, 7 May, 14:00–15:45 | Hall X5
Monitoring glacier surface elevation - and thus thickness - changes at high spatiotemporal resolution is essential for improving projections of future mass loss, constraining ice-flow models, and understanding the processes driving observed variability. Here, we present U-DEM, a new ice surface elevation dataset for the Canadian Arctic at 100 m spatial and 3-monthly temporal resolution, spanning the period from the launch of Sentinel-1 in 2014 through 2024. The dataset is generated using a deep-learning framework that combines CryoSat-2 swath altimetry, Sentinel-1 SAR imagery, and ArcticDEM stereo-derived elevation strips to produce a continuous, high-resolution record of surface elevation change.
First results demonstrate that the approach successfully captures spatial and temporal variability in glacier surface elevations that is not resolved by CryoSat-2 alone. The resulting DEMs show a substantially lower root-mean-squared-error than CryoSat-2, while reproducing small-scale topographic features such as narrow outlet glaciers and complex marginal zones. These improvements are particularly important in regions characterized by steep gradients and dynamic glacier behaviour.
By providing a consistent, high-resolution elevation time series across the Canadian Arctic, U-DEM opens new possibilities for a wide range of glaciological applications. These include transient modelling of glacier evolution, inverse modelling to constrain bedrock topography, particularly in regions affected by surging glaciers, and investigations of seasonal variability and the processes driving surface elevation changes. Beyond the Canadian Arctic, the U-DEM framework is designed to be transferable and can, in principle, be applied to any region of interest where Sentinel-1 and CryoSat-2 data are available.
How to cite: Zinck, A.-S. P., Melcher, J. O., and Dahl-Jensen, D.: U-DEM: High spatiotemporal resolution surface elevations of the Canadian Arctic, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6817, https://doi.org/10.5194/egusphere-egu26-6817, 2026.
It is known for a long time that the solar light reflectance around 1030nm ice absorption band is sensitive to the liquid water content (LWC) in snow [1]. In particular, this feature has been used to derive the abundance of water in melting snow using an imaging spectrometer [2]. In this work we propose a new technique to derive the snow LWC measuring snow reflectance at three wavelengths (1000, 1020, and 1060nm) together with independent measurements of snow density. The spectral measurements are used to derive the relative snow LWC defined as the ratio of volumetric concentration of water to that of ice in melting snow. The independently measured snow density is needed for the calculation of the snow liquid water content. The developed retrieval technique is fast, simple and can be easily implemented for the express analysis of the snow LWC distribution in the field. The theoretical model is based on the assumption that wet snow can be presented as a collection of air pockets in ice-water matrix [3]. The technique is validated using independent LWC measurements over melted snow performed in combination with snow density and snow hyperspectral reflectance measurements in Italian Alps [4]. The close correspondence of the measured and retrieved LWC in the range 8 - 18% is found.
References
[1] R. O. Green, J. Dozier, D. Roberts, T. Painter, ''Spectral snow - reflectance models for grain size and liquid water fraction in melting snow for the solar-reflected spectrum'', Ann. Glaciol., vol. 34, pp. 71–73, 2002, https://doi.org/10.3189/172756402781817987.
[2] R. O. Green, T. H. Painter, D. A. Roberts, J. Dozier, ''Measuring the expressed abundance of the three phases of water with an imaging spectrometer over melting snow'', Water Resour. Res., vol. 42, W10402, 2006, doi:10.1029/2005WR004509.
[3] A. A. Kokhanovsky, K. Segl, J. Bendix, ''Reflectance of solar light from wet snowpack: direct and inverse problems'', IEEE Trans. Geosci. Remote Sens.., 2026, doi: https://ieeexplore.ieee.org/document/11322691.
[4] C. Ravasio, R. Garzonio, B. Di Mauro, E. Matta, C. Giardino, M. Pepe, et al. , ''Retrieval of snow liquid water content from radiative transfer model, field data and PRISMA satellite data'', Remote Sens. Env., vol. 311, 2024, 10.1016/j.rse.2024.114268.
How to cite: Kokhanovsky, A., Segl, K., DI Mauro, B., Ravasio, C., Colombo, R., and Bendix, J.: The retrieval of the liquid water content in snow from solar light spectral reflectance measurements, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-6902, https://doi.org/10.5194/egusphere-egu26-6902, 2026.
The ice-free areas of the Antarctic Peninsula are undergoing significant ecological changes due to regional warming, leading to the expansion of opportunistic vegetation, primarily mosses and lichens. Mapping these communities is essential for long-term ecological monitoring, yet it remains a challenge due to the high spatial fragmentation and spectral similarity of the land cover types.
This study evaluates the effectiveness of different satellite sensors and machine learning algorithms for automated vegetation classification in the ice-free areas of Barton Peninsula (King George Island). Using high-resolution WorldView-2 (2020) and medium-resolution Sentinel-2 and Landsat 8 (2023) imagery, we compared the performance of Support Vector Machines (SVM), Random Trees (RT), and k-Nearest Neighbours (kNN) classifiers through both pixel-based and object-based approaches.
Results indicate that the kNN classifier achieved the highest overall accuracy (OA = 0.91; Kappa = 0.87) when applied to WorldView-2 data, outperforming the traditionally favoured SVM in this specific environment. The study also highlights the limitations of coarser resolution sensors (Sentinel-2 and Landsat 8) in capturing small, fragmented patches of vegetation, where the "mixed pixel" effect remains a significant hurdle, assessing how the results can still be considered meaningful.
The developed methodology demonstrates that multi-sensor remote sensing is a robust tool for creating baseline vegetation maps. These maps are crucial for quantifying the "greening" of Antarctica and provide a scalable framework for environmental conservation efforts under the Antarctic Treaty System.
How to cite: Correia, M. and Pina, P.: Multi-sensor satellite-based vegetation mapping in the Antarctic Peninsula through machine learning, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-7471, https://doi.org/10.5194/egusphere-egu26-7471, 2026.
Beyond lidar’s established role in snow hydrology for mapping snow depth, lidar intensity measurements offer a promising means of informing surface energy balance through retrieval of snow grain size, a primary control on snow albedo and net shortwave radiation. In particular, near-infrared aerial lidar intensity has demonstrated strong potential for retrieving snow reflectance and grain size over mountainous watersheds when corrections for range and incidence angle are applied. However, the accuracy of lidar intensity correction (and subsequent grain size retrieval) is highly sensitive to the quality of calibration data, which has typically relied on coincident imaging spectroscopy reflectance measurements. To improve the robustness and transferability of lidar intensity calibration approaches, a clearer understanding of how the laser pulse interacts with snow surface properties is needed. Here, we address this gap using a time series of UAV lidar flights conducted in the Wasatch Mountains near Sundance, Utah using a DJI M300 equipped with a Zenmuse L1 sensor. Each flight is accompanied by detailed snow pit observations and comprehensive in situ measurements of snow physical and optical properties. We apply machine learning techniques to model UAV lidar intensity returns and to quantify the relative influence of measured snow properties on the laser signal. These results provide new insight into the dominant controls on UAV lidar intensity over snow and identify key snow surface properties governing the laser-snow interaction. Together, these findings suggest a pathway toward simplified and more transferable calibration strategies that do not require coincident imaging spectroscopy. As a result, UAV lidar intensity can directly complement existing methods by enabling fine-scale snow grain size estimation independent of solar illumination.
How to cite: Ackroyd, C. and Olson, M.: Identifying Snow Surface Controls on UAV Lidar Intensity for Grain Size Retrieval, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-8407, https://doi.org/10.5194/egusphere-egu26-8407, 2026.
We show the development of an internet-connected GNSS to study the short-term changes in surface velocity of two adjacent Icelandic glaciers, using an RTK methodology. The system is low power, cost effective, with centimetre-level accuracy (using Ublox ZED-F9P) and transmits data to the web server daily from multiple trackers. We show that we were able to deploy trackers using a UAV (Matrice 300), which allowed inaccessible areas of the glacier to be studied. Our data indicate clear similarities in temporal velocity variations between the trackers at the individual sites, both at Breiðamerkurjökull where the units were only ~200m apart as well as Fjallsjökull where they were ~1 km apart. This demonstrates how velocity patterns were similar across the glacier.
How to cite: Martinez, K. and Hart, J.: An RTK GNSS system for measuring glacier motion, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9735, https://doi.org/10.5194/egusphere-egu26-9735, 2026.
Accurate monitoring of glacier extent at high spatial resolution is essential for understanding cryospheric responses to climate change, particularly in sensitive and data-sparse regions such as the Indian Himalaya. In this study, we present an updated glacier inventory for the Bhagirathi basin in the NW Himalaya using an open-source, reproducible workflow that integrates optical and Synthetic Aperture Radar (SAR) observations. The Randolph Glacier Inventory (RGI v7, ~2000) was used as the baseline dataset. Although RGI has enabled global-scale glacier assessments, its application in the Bhagirathi basin reveals spatial inaccuracies, particularly along ice divides and within debris-covered glacier zones, largely due to resolution limitations. These shortcomings were systematically addressed using high-resolution LISS-IV imagery (5.8 m) from Resourcesat-2/2A, allowing precise manual refinement of clean-ice margins and narrow glacier tongues. To improve delineation of debris-covered glaciers, we incorporated Sentinel-1A/B Single Look Complex (SLC) data from ascending and descending passes to generate interferometric coherence maps using open-source InSAR processing routines. SAR coherence, sensitive to surface motion and temporal change, proved effective in identifying actively flowing glacier ice beneath debris and in shaded terrain. Integration of coherence information with optical data significantly reduced misclassification associated with seasonal snow, surface debris, and shadow effects. The RGI baseline contained 124 glacier polygons within the Bhagirathi basin. Following detailed boundary correction and flow-divide refinement, the updated inventory identifies 103 distinct glaciers, reflecting the merging of previously mis-segmented units. The revised total glacierized area is 397.62 km², exceeding the RGI estimate due to improved detection of debris-covered ice. The updated inventory provides a robust baseline for multi-temporal glacier change analysis and demonstrates the potential of SAR-supported, high-resolution approaches for next-generation glacier inventories in the Himalaya and beyond.
How to cite: Malviya, A., Garg, V., and Panigrahi, R. K.: High-Resolution Glacier Inventory Update for the Bhagirathi Basin (NW Himalaya) Using Integrated Optical and SAR Coherence Analysis, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-10129, https://doi.org/10.5194/egusphere-egu26-10129, 2026.
Climate warming has profoundly impacted the cryosphere of the eastern Himalaya, leading to rapid glacier retreat, the formation and expansion of proglacial lakes, and the degradation of alpine permafrost. This study examines how proglacial lakes influence the dynamics of both glaciers and rock glaciers (ice-rich debris) in the Sikkim Himalaya, with special emphasis on South Lhonak Lake. We integrate multi-temporal remote sensing datasets – including DEM differencing (TanDEM-X), optical imagery (PlanetScope), and SAR interferometry (Sentinel-1 SBAS InSAR) – to quantify landscape changes from 2016 to 2025. Results show that South Lhonak Lake expanded by ~45% in area (from ~1.12 to 1.63 km²) between 2016 and late 2023 (Qu et al., 2025), before a sudden glacial lake outburst flood (GLOF) in October 2023 drained nearly half of its volume. Pre- and post-GLOF analyses reveal accelerated glacier retreat and pronounced surface lowering at the glacier terminus adjacent to the lake. Rock glaciers in contact with the lake experienced greater surface elevation loss and deformation compared to those in non-lake settings, suggesting that thermal and mechanical erosion by the lake has exacerbated permafrost degradation. Time-series deformation mapping using SBAS-InSAR captured ongoing ground motion on periglacial slopes and rock glaciers, with line-of-sight displacement rates on the order of several cm/year in active zones. Notably, slopes fringing South Lhonak Lake showed progressive subsidence and destabilization signals before the 2023 GLOF, indicative of creeping movement in ice-rich moraine and permafrost materials. These findings highlight a coupling between proglacial lake evolution and the stability of surrounding cryospheric landforms. The study demonstrates the value of synergistic remote sensing for hazard monitoring in inaccessible high-mountain environments. It provides a first regional assessment of how an expanding (and abruptly draining) lake can influence glacier mass loss, rock glacier kinematics, and permafrost stability. Our multi-sensor approach offers a template for early detection of glacial lake outburst precursors and periglacial slope failures, information that is critical for climate change adaptation and disaster risk reduction in the Himalaya.
How to cite: Kumar, A., Malviya, A., and Singh, G.: SAR-Driven Detection of Critical Glacial Lakes for GLOF Forecasting, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-11298, https://doi.org/10.5194/egusphere-egu26-11298, 2026.
Antarctic Ice Shelves continuously modifies land ice and ocean boundaries and influences global climate change and sea-level fluctuations. The Getz Ice Shelf (GIS), which fringes nearly half of the West Antarctic coastline in the Amundsen Sea sector, experiences substantial basal melting. This study presents a long-term, satellite-based assessment of morphological changes in the GIS using multi-temporal austral summer (January–March) observations spanning 22-years (2003–2022). To capture spatial variability, the ice shelf was divided into three sectors (I–III) based on physiographic setting and dominant oceanographic processes, and changes in ice shelf extent were quantified along uniformly spaced transects at 5 km intervals The average of rates, end point rate, and linear regression method were employed to estimate the rate of change, with the linear regression showing the strongest correlation. Past ice shelf extents were reconstructed, and predicted ice shelf positions for the next 5 and 10 years and cross-validated with correlation coefficient and root-mean-square error. The result reveals a mean recession rate of -41.6 m/year during the austral summer of 2003–2022, and about 70% of transects show recession, while only 30% of transects are associated with progradation. LR-based projections indicate progradation in Sector I and continued recession in Sectors II and III, with pronounced retreat in Sector III between Wright Island and Martin Peninsula (~2.1 km by 2027 and ~2.5 km by 2032). Approximately 45% of transects in Sectors II and III have RMSE values of ±200 m, indicating good agreement between the estimated and satellite-based ice-shelf positions. Observed changes are linked to large-scale climate forcing, including variations in wind speed, sea surface temperature, and the Southern Annular Mode (SAM). Positive SAM and increased zonal winds may cause warm water upwelling near the Antarctic coast, impacting the extent. Changes in sea ice mass and accelerated basal melting are largely caused by ocean warming and geomorphological features, viz., bays, inlets, and islands. The study emphasizes the significant impact of ocean-atmospheric factors on Antarctic ice shelf dynamics and highlights the necessity for ongoing satellite observations and enhanced comprehension of these processes to accurately predict future changes in Antarctic ice shelves.
Keywords:
Getz Ice Shelf; West Antarctica; glacier monitoring; remote sensing; MODIS; ice shelf retreat; transect analysis; atmosphere-ocean forcings; future predictions.
How to cite: Nigam, A. and Gupta, D. C.: Satellite-based assessment of Getz Ice Shelf extent changes in the Amundsen Sea sector, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-12079, https://doi.org/10.5194/egusphere-egu26-12079, 2026.
The rapid retreat of glaciers in the last decades highly impacts the regional hydrological cycle and water supply and increases the potential for cryosphere-related hazards. Expansion of debris-cover and changes in glacier dynamics are indicators of mass loss. While debris-cover has an insulating effect, ice-cliffs exhibit fast backwasting and therefore represent melt hotspots in the debris-covered areas. Their relative contribution has only been quantified for few glaciers and for global and regional glacier models they have therefore not yet been parameterized. The number and distribution of moulins impacts the (seasonal) glacier dynamics. Monitoring alpine glaciers is essential to quantify ongoing mass loss and project their future evolution and water supply. However, the small-scale characteristics and dynamic nature of these features cannot be captured by traditional approaches such as satellite remote sensing. Mass loss can be quantified reliably with ablation stakes at selected locations, but to investigate spatio-temporal variations across the glacier, a distributed approach is needed.
With repeated UAV photogrammetry, the annual variability of elevation change, velocity and mass balance can be studied. On the debris-covered area, thermal imaging is employed for spatial surface temperature and debris thickness mapping. In this study, we use a structure-from-motion and multi-view stereo approach to create high-resolution orthophotos and DEMs from visual imagery of the partially debris-covered tongue of Kanderfirn in the Swiss Alps. UAV surveys have been conducted on a seasonal to annual basis since 2017 and cover the glacier tongue, which features debris-free and debris-covered surfaces. After coregistration of image pairs, we derive annual surface velocity and elevation change maps. Based on additional ice thickness information, the mass continuity method and an inversion technique is applied to investigate the glacier's dynamical evolution. The result is a SMB time-series that portrays surface processes with and without debris-cover. Additionally, the surface roughness and brightness are examined. The debris thickness maps provide a basis for estimating subdebris melt rates. Our UAV-based monitoring approach provides detailed insights into characteristic mass balance patterns and dynamic processes of (partially) debris-covered glaciers that impact their evolution and response to climatic changes. It will help with parameterizing glacier models for future mass loss projections on alpine and debris-covered glaciers.
How to cite: Zöller, A., Fürst, J. J., and Groos, A. R.: Monitoring the surface mass balance and dynamic evolution of a partially debris-covered glacier using multi-temporal visual and thermal UAV imagery, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-13044, https://doi.org/10.5194/egusphere-egu26-13044, 2026.
Snow strongly influences the climate system through its albedo and insulating properties, while also representing a critical freshwater resource. Yet, its spatial and temporal variability remain poorly constrained due to limitations of in-situ and satellite observations. Unoccupied Aerial Vehicle (UAV)-mounted surface-penetrating radars provide a solution for high-resolution snow surveys, but their data are often difficult to interpret because of variable flight conditions which, combined with the complicated EM interactions with snow, result in complex noise and signal patterns.
We present Pathfinder, an open-source algorithm for automatic detection of snow interfaces in radar echograms. The method formulates interface tracking as a path-finding problem, combining cost maps derived from reflection strength, ridge detection, and horizontal continuity, and solves it using an efficient dynamic-programming scheme. Pathfinder retrieves the air–snow and snow–ground (or snow–ice) interfaces, and can additionally identify internal layers when present. Validation against coincident in-situ snow depth measurements shows performance scores of R2 = 0.96 and RMSE = 8 cm. The algorithm is computationally efficient, enabling real-time application during surveys. Pathfinder was developed for the Ultra Wide-band Snow Sensor (UWiBaSS) from NORCE Research but we show it to be transferable across different UAV-mounted and ground-based radar systems. Case studies from Svalbard, the Alps, and Arctic- and Antarctic sea ice illustrate its robustness across diverse snow conditions. By providing a reliable, efficient, and operational interface detection method, Pathfinder advances (UAV-mounted) radar as a practical tool for snow depth and stratigraphy mapping, supporting both scientific research and in-the-field decision-making.
How to cite: Kagel, T., Stuurman, A., Siebenbrunner, A., Ricker, R., and Jenssen, R.-O.: Advances in interface detection for high-resolution snowpack analysis with (UAV-mounted) UWB radar, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14100, https://doi.org/10.5194/egusphere-egu26-14100, 2026.
InSAR is a key technology for monitoring ground movements such as subsidence, uplift, unstable slopes, and natural hazards, and it plays a critical role in reducing risks to infrastructure and communities. Since the launch of InSAR Norway in 2018 and European Ground Motion Service (EGMS) in 2022, these datasets have been freely available. However, Svalbard has not been covered by any of these services. With temperatures rising at a rate seven times faster than the global average, the need for consistent and long-term InSAR data in Svalbard and the Arctic is evident.
With funding from the Norwegian Space Agency and in collaboration with NORCE, the Geological Survey of Norway (NGU) launched the InSAR Svalbard Ground Motion Service in February 2026. The pilot products provide the first open-access InSAR dataset for Svalbard using Sentinel-1 data. Tailored to Arctic conditions, the service offers (1) seasonal data during snow-free periods aimed to detect fast-moving areas and to monitor seasonal patterns, and (2) interannual data between snow-free periods to monitor slower ground motion. It delivers essential data for monitoring permafrost degradation, slope creep processes, and freeze-thaw processes, providing critical information for research and risk management.
The pilot products are currently available for selected areas around settlements and research stations and are accessible through a web-GIS platform that provides easy visualization and analysis tools. Future development will expand coverage across the entire archipelago, integrate new satellite data, and progress towards a comprehensive ground and ice motion service.
How to cite: Hauglin, E., Aslan, G., Bredal, M., Rouyet, L., Dehls, J., Lauknes, T. R., Wendt, L., Stødle, D., Hindberg, H., von Oostveen, J., and Larsen, Y.: InSAR Svalbard Ground Motion Service , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-20200, https://doi.org/10.5194/egusphere-egu26-20200, 2026.
The Antarctic Ice Sheet is the largest potential contributor to future global sea level, with a maximum contribution of approximately 58 m. The flow of grounded ice towards the ocean is largely restricted by floating ice shelves, which fringe ~75% of the Antarctic coastline. This buttressing effect can be diminished or lost entirely following partial or complete ice-shelf disintegration events, leading to an acceleration of ice discharge. The vulnerability of ice shelves to fracture and disintegration is influenced by a range of factors, including the ponding of surface and subsurface meltwater, which can induce flexural stresses and promote fracture through the ice-shelf column.
While the widespread extent of surface meltwater systems across numerous Antarctic ice shelves during the austral summer is now well documented, meltwater storage within the ice-shelf subsurface remains poorly understood. Liquid water can persist perennially beneath the ice-shelf surface if sufficiently insulated by surrounding and overlying layers of firn, snow, and/or ice. Due to their year-round persistence, buried meltwater lakes introduce a potential mechanism for hydrofracture outside of the melt season, with important implications for ice-shelf stability.
In situ surveys in Antarctica are logistically challenging and limited in spatial extent, rendering spaceborne remote sensing an indispensable tool for monitoring ice-shelf processes at a continental scale. However, the growing volume of satellite observation poses data-analysis challenges typical of the ‘Big Data’ era; remote-sensing datasets are often high-dimensional, unstructured, and large in sample size, with rapidly increasing spatiotemporal coverage and low-cost availability. These characteristics limit the scalability of manual interpretation or traditional thresholding approaches for pan-Antarctic applications. In contrast, machine-learning methods are ideal for extracting patterns and features from large databases of satellite imagery. In light of this, machine learning offers great potential for the detection of subsurface meltwater across Antarctic ice shelves at a continental scale, if challenges relating to the sampling and labelling of training data, as well as class imbalance, are addressed carefully. An active learning strategy can help to reduce data redundancy and labelling requirements in deep learning, while also improving model performance in the presence of class imbalance.
Here, we present a systematic training-data sampling strategy applied to all major Antarctic ice shelves. We employ a stratified random sampling approach to mitigate strong regional imbalances in data availability and create a curated training data subset that combines an equal number of random and expert-selected samples. This dataset is used to initialise an active learning framework for training a deep-learning model to detect subsurface lakes in Antarctica. We evaluate model performance across multiple configurations and present fine-tuning of model hyperparameters.
How to cite: Suchantke, P., Dell, R., and Arnold, N.: A Deep Active Learning Framework for the Detection of Subsurface Meltwater Lakes on Antarctic Ice Shelves, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-18761, https://doi.org/10.5194/egusphere-egu26-18761, 2026.
At present, 10% of the Greenland Ice Sheet (GrIS) margin terminates in a lake, but this is forecast to increase significantly over coming decades with ongoing retreat and thinning of the ice margin. This is important because the existence of more and larger lakes implies greater mass loss at the margin as well as an increase in dynamic thinning up-ice. At the same time, non-glacial lake surface waters worldwide have been shown to be warming yet the record of temperature change in ice-marginal lakes is extremely sparse. These ice-marginal lakes are likely to become increasingly important drivers of mass loss, and so there is an urgent need to investigate more closely the connection between their evolution and changes in the ice sheet. Therefore, a key objective is to explore the thermal properties of these lakes.
The majority of Greenland’s ice marginal lakes are small (<0.5km2) thus limiting the role of remote sensing approaches. To fully characterise and understand the detailed thermal properties of these lakes over time and space we thus developed an in-situ approach for high resolution monitoring of the temperature evolution of two Greenlandic ice marginal lakes. Our monitoring approach involved the installation of a series of thermistor strings which recorded water temperature at several locations and at a range of depths every 30 minutes for more than one year. Over the same period we deployed a series of 15 time-lapse cameras capturing hourly imagery so as to build 3D models of ice front change in response to changing lake properties. Finally, we carried out several drone-based surveys using a thermal camera to record spatially distributed lake temperatures. As well as discussing our monitoring network, we also present data that reveals Greenlandic lake evolution at a resolution not previously seen, showcasing an observational framework that could be replicated at other lake-terminating sites elsewhere in the world.
How to cite: Rippin, D., Tuckett, P., Quincey, D., Harpur, C., Scoffield, A., Abraham, J., Rawlins, L., Mallalieu, J., Barnett, H., Sutherland, J., Woolway, I., Merchant, C., McCarroll, N., Carrea, L., and Wang, W.: High resolution monitoring of Greenlandic ice marginal lakes, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-1572, https://doi.org/10.5194/egusphere-egu26-1572, 2026.
Seasonal variability in the ice velocities of slow-moving Himalayan glaciers (≤100 m yr⁻¹; >4000 m a.s.l.) is largely unknown, primarily due to the scarcity of high-resolution observations and the substantial uncertainties associated with satellite-based velocity products. In this study, we present high-frequency terrestrial time-lapse camera (TLC) observations of glacier ice motion from Drang Drung Glacier (33.76° N, 76.30° E), spanning October 2023 to April 2025. The glacier terminus is predominantly lake-terminating, with a smaller land-terminating component, enabling a comparative assessment of spatial variations in ice dynamics.
Our results show spatial heterogeneity in surface velocity. Annual mean velocities at the lake-terminating section (35.4 m yr⁻¹) are nearly twice those observed at the adjacent land-terminating segment (19.5 m yr⁻¹). A primary seasonal cycle is evident in both regions, characterised by summer speedup (June–September) followed by autumn slowdown (September–November). These variations correspond closely with increases in air temperature and solar radiation, and are consistent with the meltwater-driven evolution of the subglacial drainage system from inefficient to efficient, channelised configurations. TLC imagery further captures signatures of active subglacial hydrology and its temporal transitions.
A secondary, modest winter speedup (November–January), followed by persistent deceleration until February, suggests that viscous deformation and associated closure of subglacial channels lead to elevated basal water pressures from trapped meltwater. Vertical ice displacement exhibits substantial seasonal variability (−0.9 to −2.0 m month⁻¹) from June to October, with minimal changes outside this period. Sub-weekly analyses reveal coherent patterns of glacier acceleration, contemporaneous increases in lake turbidity, and uplift of the ice front, indicating rapid responses to fluctuations in basal water pressure. TLC-derived velocities show strong agreement with in-situ GNSS measurements but highlight a marked underestimation of glacier motion in ITS_LIVE satellite products for this site.
How to cite: Singh, P. and Vijay, S.: Seasonal ice velocity of Drang Drung Glacier, in the western Himalaya, using terrestrial time-lapse camera imaging , EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-961, https://doi.org/10.5194/egusphere-egu26-961, 2026.
Sea ice floe size, ice concentration, snow cover, and thickness collectively drive the evolution of the marginal ice zone (MIZ) by influencing albedo, melt dynamics, wave-ice interactions, ice strength, and navigation conditions for icebreakers. Yet, reliable measurements of these parameters remain scarce, significantly limiting process understanding and model validation in the Antarctic. Satellite-based products are constrained by spatial resolution, and the scarcity of ground truth data prevents thorough validation and refinement of remote sensing retrieval algorithms. We demonstrate the use of the Segment Anything Model (SAM), a foundation vision model, to extract multiple physically meaningful sea ice properties from close-range, shipborne imagery. Using extensive datasets of high-resolution images collected during multiple Antarctic icebreaker voyages, SAM identifies and delineates individual ice floes, facilitating accurate estimation of floe sizes and sea ice concentration. Validation against manually segmented benchmarks shows robust agreement across diverse ice conditions. For snow cover and ice thickness estimations, SAM is specifically fine-tuned on manually annotated datasets to detect overturning ice events, enabling thickness measurement from exposed vertical profiles, and to classify snow-covered versus bare ice, quantifying snow cover fraction on individual floes. Overall, SAM enables systematic, scalable observations previously challenging to obtain, bridging the critical sea ice data gap by transforming images into quantitative datasets that support Antarctic sea ice process studies and improve observational detail beyond satellite capabilities.
How to cite: Passerotti, G., Nelli, F., Tersigni, I., Alberello, A., Vichi, M., Bennetts, L., Bailey, J., Heil, P., and Toffoli, A.: Foundation vision models for Antarctic sea ice floe segmentation and quantification from shipborne imagery, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-9695, https://doi.org/10.5194/egusphere-egu26-9695, 2026.
The Himalayan glaciers continue to lose mass due to the undergoing warming climate. They play a critical role in feeding major rivers, such as the Ganga, Indus, and Brahmaputra, supporting the livelihoods of millions. While regional mass changes have been reported by several studies, the current retreat rates are rarely documented. This is primarily due to a lack of satellite data and methods in mapping debris-covered glacier fronts. They are also limited in distinguishing clean ice and perennial snow cover patches. While the global glacier database, such as the RGI (Randolph Glacier Inventory), provides a critical database, it is based on satellite images from 2000-2003.
In this study, we address challenges in mapping debris-covered glaciers by combining Deep Learning (DL) and a geometric algorithm. We apply several DL models (e.g., including UNet++, GlacierNet-2, GlaViTU, M-LandsNet, and SAU-Net) on multiple remote sensing satellite datasets, which include spectral, radar, topography, geomorphology, and glaciological dynamics. The study sites include four basins of the Himalaya (Chandra Bhaga, Pangong, Chombu Chu, and Alaknanda-Bhagirathi). UNet++ shows the most accurate results with reference outlines, with a mean Intersection over Union (IoU) of ~ 90%. DL-based retreat measurements were closely aligned with those outlined in the reference manual, with a coefficient of determination (R²) of ~ 75%. Our applied Python-based geometric algorithm calculates the average euclidean distance between frontal positions in 2010 and 2019. We find that the retreat rates of debris-covered glaciers in these basins are ~3 m/year during this period. Lake-terminating glaciers show three times higher retreat rates in the period. This algorithm has the capability to detect glaciers of all sizes, ranging from small to large glaciers, as well as highly debris-covered to clean-ice glaciers, and can identify cirques to hanging glaciers in all the basins.
This DL-based algorithm provides an automated approach with post-processing steps to monitor glacier change with high precision, accounting for the uncertainty of glacier retreat across the Himalaya. This study is important for understanding the relationship between glacier lake expansion and glacier ice mass loss, which can be further used for glacier hazards, such as lake outbursts and dry calving detachments.
How to cite: Verma, S. K., Vijay, S., and Banerjee, A.: Automated Mapping of Glacier Frontal Retreat in the Indian Himalaya using Satellite Remote Sensing and Deep Learning models, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-744, https://doi.org/10.5194/egusphere-egu26-744, 2026.
The mass balance estimates of the GRACE and GRACE-FO satellite missions have revolutionized our understanding of the cryosphere, yet the low spatial (~200 km) and temporal resolution (monthly) limit the detection of short-term events such as melt pulses and snowfall surges, or the onset and end of the ablation season. Furthermore, the relatively high noise at shorter time scales makes estimation of yearly mass balances challenging for the smaller glacier systems outside of the Greenland and Antarctic Ice Sheets. The MAss Change and Geosciences International Constellation (MAGIC) mission, consisting of the GRACE-C (NASA and DLR; scheduled for launch in 2028) and ESA’s Next-Generation Gravity Mission (NGGM; 2032) will drastically increase the temporal and spatial resolution of the gravimetric mass balance estimates.
Here, we combine daily surface mass balance (SMB) output from regional climate models with noise from end-to-end gravity simulations to assess NGGM and MAGIC’s capability to resolve sub-monthly mass changes over the Greenland and Antarctic ice sheets at basin scales, as well as seasonal mass balance and long-term trends of smaller glacier systems. We find that 5-daily MAGIC measurements enable the observation of extreme melt events comparable to the 2012 Greenland melt episode and capture storm-driven accumulation in Antarctica at sub-monthly time scales. For other glacier systems - such as Iceland, Patagonia, and High Mountain Asia - the mission allows the onset and termination of the ablation season to be identified within 2–3 days and constrains winter gain loss and summer mass loss to within a few gigatons, representing an order-of-magnitude improvement over GRACE(-FO). Uncertainties in long-term trends are reduced by a similar factor, such that the precision achieved by NGGM and MAGIC within a few years is comparable to that obtained by GRACE(-FO) only after 10–15 years of observations.
Overall, NGGM and MAGIC are expected to represent a step change in gravimetric cryosphere monitoring, enabling routine observation of short-term variability and supporting improved detection, attribution, and modelling of long-term mass-balance trends.
How to cite: Wouters, B. and Sasgen, I.: Advancing Observation of Cryospheric Mass Changes from sub-monthly to decadal time scales with the NGM and MAGIC Gravity Missions, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-14713, https://doi.org/10.5194/egusphere-egu26-14713, 2026.
Snowpack evolution in cold-region environments is governed by complex interactions between surface energy exchanges and internal stratigraphic processes that vary throughout the winter season. Capturing these dynamics at high temporal resolution remains challenging with conventional point-based or destructive methods. This study presents a season-long, predominantly non-destructive monitoring approach combining fixed-station terrestrial LiDAR and high-frequency ground-penetrating radar (GPR) to observe snowpack surface and internal dynamics across multiple winter processes.
Field measurements were conducted throughout the winter at two contrasting eastern Canadian watersheds. Sainte-Marthe (45°N) is a lowland agricultural catchment (110 m a.s.l.) with shallow, ephemeral snow cover, while the Montmorency Forest (47.3°N) is a boreal watershed (670 m a.s.l.) characterized by a deep, persistent snowpack. GPR systems operating at 1500 MHz and 500 MHz were deployed at Sainte-Marthe and Montmorency, respectively, to account for contrasting snow depths. A fixed 905 nm fixed LiDAR system operated concurrently at both sites, providing hourly measurements of snow surface elevation and properties. Manual snowpit surveys, including density and A2 permittivity measurements, were performed throughout the winter to constrain electromagnetic wave velocity and support GPR inversion.
LiDAR observations capture continuous changes in snow surface state, including accumulation, compaction, melt, deformation, and surface roughness, revealing periods of enhanced energy input at the snow–atmosphere interface. Corresponding variations in GPR signal amplitude, two-way travel time (TWT), and frequency content indicate internal stratigraphic adjustments within the upper snowpack layers.
Overall, integrating LiDAR and GPR with traditional monitoring stations demonstrates strong complementarity for continuous, non-destructive monitoring of snowpack processes over an entire winter season, providing new observational constraints relevant to cold-region hydrological and cryospheric modelling
How to cite: Trottier, K., Michel, B., and Daniel, N.: Season-long non-destructive monitoring of snowpack surface and internal dynamics using terrestrial LiDAR and GPR, EGU General Assembly 2026, Vienna, Austria, 3–8 May 2026, EGU26-15122, https://doi.org/10.5194/egusphere-egu26-15122, 2026.
