Glacier melt is an important fresh water source. Seasonal changes can have impacting consequences on downstream water resources management. Today’s glacier monitoring lacks an observation-based tool for sub-seasonal observation of glacier surface mass balance and a quantification of the associated meltwater release at high temporal resolution on mountain range to regional scale.
The snowline on a glacier marks the transition between the ice and snow surface, and is, at the end of the summer, a proxy for the annual glacier mass balance. Using transient snowlines for model calibration to derive annual mass balance time series for glaciers on regional scale has shown great potential to better grasp the glacier response to climate change for remote regions. Model simulations directly integrating sub-seasonal snowline time series based on optical satellite imagery are improving conventional modelling, but glacier-specific snowline observation remained spares.
We developed an approach that can automatically handle classification of multi-source and multi-resolution satellite image stacks. The combination of SAR and optical Sentinel 2 and MODIS data in a complementary way improves the temporal and spatial resolution of snow depletion monitoring on glacier scale.
We applied a change detection algorithm for generating maps of wet snow from Sentinel-1 images. A Support Vector Machine (SVM) classification algorithm was applied to the Sentinel-2 images by employing all the spectral bands with a spatial resolution equal to 10 or 20 meters as input features. From both the Sentinel-1 and Sentinel-2 snow maps, we calculated the fraction of snow cover in relation to the total glacier area. This fraction was then related to mean NDSI from MODIS for common observation dates. With a linear regression between the different products (R2 = 0.38 to 0.79), we estimated a function to reproduce close-to-daily snow cover area per glacier from the mean NDSI of MODIS images covering the period 2000 to present. Validation with manually delineated snow-covered area fractions revealed RMSEs between 7 to 15%.
This provides a unique solution for continuous snowline mapping since the beginning of the century when sensor availability and quality was still limited. With the provided close-to-daily transient snowlines, we provide the basis for a new strategy to directly integrate multi-source satellite image classification into glacier mass balance modelling. To estimate glacier mass balance and meltwater input to the total river runoff, openAMUNDSEN (Alpine Multiscale Numerical Distributed Simulation Engine), an open source, physically based process model designed to quantify the energy and mass balance of ice and snow is used. Thereby, we use a calibration strategy for openAMUNDSEN, using the sub-seasonal snowline maps for annual model calibration. This setup is tested and validated with sub-seasonal mass balance measurements at Vernagtferner, Austria. We aim for a highly resolved, observation-based glacier monitoring and the detection of sub-seasonal changes of glacier meltwater contribution into the river system for the past two decades. The developed approach is applicable for remote and inaccessible glaciers and will help to better understand the impact of climate change on regional water availability for remote and so far, unmeasured regions.

Glaciers all over the world are reported to be undergoing a "darkening" process. This darkening has multiple causes, including deposition of lgiht absorbing particles, increased debris input and glacier algae, and influences glacier melt by decreasing the albedo and initiating a a feedback cycle. The extent of this albedo decrease and its spatial variability in different mountain ranges however is poorly known. In this study, we assess albedo changes on a selection of Alpine glaciers (542 glaciers > 0.1 km2); we do so by investigating changes in 8 albedo metrics from Modis MOD10A1 satellite observations over a time span of 20 years (2000-2019). Albedo metrics include concepts borrowed from vegetation phenology, i.e. the start and end of the ablation season and its length (end-start), calculated through the selection of a threshold to separate melting ice from snow; we further calculated the number of ice melt days, integral of the length of the ablation season, and more conventional metrics such as minimum annual albedo, mean summer albedo and mean annual albedo.
We identified significant negative trends in all metrics for a large percentage of glacier pixels, highest for minimum albedo (55% of glacier pixels with a significant decrease, 67% when averaging over the entire glacier) and slightly lower for the other metrics. Spatially, changes in minimum albedo occur mostly around the glacier ELA, while for other metrics changes are more apparent at the termini. We further investigated the relationship between the metrics and climate variability from ERA5, finding temperature and elevation as the main proxies (0.41 R2). Finally, we looked at the ability of the albedo metrics to represent annual and summertime glacier mass balance of 31 selected Alpine glaciers in the WGMS database, finding very good correlations especially between minimum and summertime albedo with the mass balance of most glaciers (e.g. Argentiere: 0.84 R2, Saint Sorlin 0.88). Overall, these metrics performed better than phenological metrics, possibly because of the large noise in the MODIS albedo product. Nevertheless, our findings point to an increase in the length of the glacier ablation season and decrease in glacier minimum and summertime albedo over the Alps

In Antarctica dynamic ice loss dominates the continents’ contribution to sea level, and the magnitude of this depends in part on the ice speed of marine-terminating glaciers. Ice speed variations in Antarctica have been observed on multi-year timescales, most notably the long-term increase in the speed of glaciers in the Amundsen Sea sector, Getz basin and Antarctic Peninsula. In Greenland widespread seasonal ice velocity variability has been observed on tidewater glaciers, but there is limited evidence of seasonality on Antarctic ice streams. In this study we exploit the entire Sentinel-1 record to measure ice speed across the West Antarctic Peninsula coastline, from December 2014 to May 2021. The Antarctic Peninsula is characterized by its steep, 3,000 m high mountainous topography, inland plateaus, small tidewater glaciers, large ice shelves and ice-free areas. The weather is variable with summer air temperatures above 0°C and significant precipitation, including rain, on the Northern Peninsula. For these reasons, ice motion tracking has historically been difficult in this region. We measure ice speed in 10,434 image pairs using offset tracking by frequency domain intensity cross correlation in 6 or 12-day separated Sentinel-1 frames. On the West Antarctic Peninsula coast we extract a time series of ice speed in the Sentinel-1 epoch for 106 tidewater glaciers and post-process these measurements using a Bayesian recursive smoother.

Time series analysis of the data from 106 glaciers reveals previously unreported widespread seasonal ice speed variability throughout the Sentinel-1 period, with glaciers displaying speed maxima in the austral summer and minima the following winter being common, particularly at the Northern Peninsula. Our results show that for glaciers flowing faster than 500 m/yr, there is a mean intra-annual speed variability of 130 m/yr, 13.0%, and a mean intra-annual speed interquartile range of 57 m/yr, 5.2%. Analysis of these speed trends against potential forcing mechanisms shows good correspondence with terminus position change, modelled snowmelt and modelled upper ocean heat content, suggesting that the observed seasonal ice speed variability is due to increased heat in the ice-atmosphere-ocean system driving glacier dynamics. Studies of mass balance on the Antarctic Peninsula must take account of changes in ice speed on intra-annual timescales to avoid over or under-estimating the total sea level contribution from this region. Future work is needed to understand the historic prevalence of seasonal speed changes on the Peninsula, to improve future projections of the Antarctic response to warming and its contributions to sea level rise.

For studying the flow of glaciers and their response to climate change it is important to detect glacier surges. Glacier surges disturb the link with climate and thus climatic interpretation of changes in glacier size or mass balance. Typically, surge-type glaciers have to be excluded from climatic glacier studies. Surges can also constitute potential natural hazards, for instance by damming up rivers and fjords, or by leading into ice avalanches. Questions also arise whether climate change impacts the frequency of surge-like glacier instabilities. To detect increasing or decreasing surge activity of individual glaciers world-wide, we exploit here variations in the strength of radar backscatter over time. As a surge implies substantial increase in the flow velocity and velocity gradients of a glacier, the glacier surface typically becomes more crevassed and radar backscatter is subsequently increased during an active surge. Likewise a glacier surface will become less crevassed and therefore the backscatter will decrease at the end of an active surge phase. We compute within Google Earth Engine the normalized differences between winter maxima of Sentinel-1 C-band radar backscatter image stacks over subsequent years. That way we arrive at a global map of annual backscatter changes, which are for glaciers in most cases related to changed crevassing associated with surge-type activity. In our manual detection exercise based on these data we find many glacier surges, most of them not classified so far as surge type. Comparison with glacier surface velocities shows that we reliably find known surge activities. Also visual comparison to optical Sentinel-2 data confirms in by far most cases our backscatter-based detection. In our study we also investigate the influence of polarization and ascending/descending acquisition geometry on the performance of our approach. We find that most glacier surges or surge-like glacier activities become visible in our backscatter differences before they become apparent in optical images or velocity time series. It is clear, however, that the latter velocity time series based on repeat offset tracking, e.g. in Sentinel-1 data, provide much more detailed information about surge development and they quantify ice flow speed, which our approach is not able to do. Our method turns out very helpful to support operational monitoring of glacier surges through their initial detection and steps to include our world-wide map of glacier surges over 2017-2021 into the Randolph global glacier inventory are underway. While focusing on glacier surges in this contribution we also recognize that some other special events such as large rock and snow avalanches or glacier lake outbursts become well visible in our approach.

Consistent monitoring of the changing topography of glaciers and icecaps can be done on a global scale through satellite remote sensing. Such backbone missions, rely on data from CryoSAT and interferometric altimetry or the stereoscopic configuration of ASTER and photogrammetric processing. However, if this high level of monitoring will persist is unclear, as a follow-on mission for CryoSAT is not envisioned. Similarly, the Terra satellite, with ASTER onboard, has started to de-orbit and is in its end-of-life, while no photogrammetric mission is planned in the near future. Alternative methodologies for topographic retrieval are thus needed to fill this information gap.

Here we introduce a new method to derive glacier elevation change from multi-spectral data of Sentinel-2. It is a monoscopic retrieval, making use of the potential of shadow casting and straight forward triangulation. It makes it a candidate methodology to roll out to regional to global-scale product generation, as it is based upon a simple principle. The generated topographic information can also be merged, so it can enrich and complement available time-series, as it is able to measure elevation change on very small mountain glaciers.

Mountain glaciers are important sources of fresh water for agriculture, irrigation, hydro power and drinking water, particularly in arid and semiarid regions. During recent decades, mountain glaciers were strongly affected by global climate change. Increasingly negative glacier mass change and a significant contribution to global sea level rise have been reported by numerous studies from different regions. However, the accuracy and intercomparability of glacier change measurements in different regions can be hampered by a broad variety of applied methods or a lack of available measurements in remote regions.
Within the SATELLITE2 project, elevation and volume changes of almost all mountain glaciers, outside the Antarctic regions, are derived from space-borne interferometric synthetic aperture radar (InSAR) remote sensing data of the bistatic TanDEM-X satellites and the Shuttle Radar Topography Mission (SRTM). Therefore, we calculate changes in glacier surface elevation by differentiating vertically and horizontally co-registered TanDEM-X X-Band DEMs (2011-2020) and SRTM C-Band (2000) to estimate changes in glacier volume and mass over the last two decades. For all glacierized areas, surface elevation changes are measured over the period ~2012 to ~2018 (TanDEM-X – TanDEM-X). In addition, for mountain regions within the coverage of the SRTM C-Band DEM, 60°N to 56°S, glacier changes are measured for the period 2000 to ~2012 (SRTM – TanDEM-X). As reference surface for the interferometric SAR DEM creation, we use the SRTM DEM and the Copernicus DEM and TanDEM-X Global 90m DEM for the polar regions.
First results have been already derived for the South American Andes, South Georgia, the European Alps and the Russian Arctic. With the help of High Performance Computing all glacierized regions outside the Antarctic Ice Sheet will be calculated by the time of the poster presentation.
Based on a standardized methodology and an extensive and homogenous database, the resulting SAR glacier change dataset will enable a high comparability of regional and global glacier volume and mass change and its contribution to sea level rise.