Thomas Alfred Schellenberger

• Strozzi, Tazio; Mannerfelt, Erik Schytt; Cartus, Oliver; Santoro, Maurizio; Schellenberger, Thomas & Kääb, Andreas (2026). Glacier surge activity over Svalbard from 1992 to 2025 interpreted using heritage satellite radar missions and Sentinel-1. The Cryosphere. ISSN 1994-0416. 20(3). doi: 10.5194/tc-20-1679-2026. Fulltekst i vitenarkiv Vis sammendrag

  Based on massive processing of heritage radar data from the satellite missions ERS-1/2, JERS-1, ENVISAT ASAR, ALOS PALSAR and Radarsat-2, and in combination with data from the current Sentinel-1 and ALOS-2 PALSAR-2 missions, we compiled a ∼ 30-year time series of radar backscatter over Svalbard. We exploited this data to detect glacier surges by using changes in backscatter as an indicator of increased or decreased surge-related crevassing. In this way, we reconstructed an as consistent as possible time series of surge activity on Svalbard for 1992 to 2025. We recorded 24 surge-type events during the pre Sentinel-1 period 1992–2014 (23 years) and 34 surge-type events during the post Sentinel-1 period 2015–2025 (11 years). This time series shows an approximately threefold increase in surges since 2015, from an average of about one surge per year to more than three surges per year. We show that this increase is unlikely to be explained alone by the better resolution, coverage and quality of the Sentinel-1 data compared to the data from the earlier SAR heritage missions. Simulation results indicate that the observed increase is extremely unlikely to be attributed to random perturbations in surge cyclicity, and instead suggest the influence of an external forcing mechanism. The number of surges during the recent decade seems high, but due to uncertainties in historical records, it remains unclear whether this frequency is exceptional or if earlier decades were unusually quiet. The cause of the observed threefold increase in surge activity also remains uncertain, given our incomplete understanding of surge initiation in relation to climate variability and non-climatic surge controls.

• Åkesson, Henning; Sjursen, Kamilla Hauknes; Schuler, Thomas; Dunse, Thorben; Andreassen, Liss Marie & Gillespie, Mette Kusk [Vis alle 9 forfattere av denne artikkelen] (2025). Recent history and future demise of Jostedalsbreen, the largest ice cap in mainland Europe. The Cryosphere. ISSN 1994-0416. 19(11), s. 5871–5902. doi: 10.5194/tc-19-5871-2025. Fulltekst i vitenarkiv Vis sammendrag

  Glaciers and ice caps worldwide are in strong decline, and models project this trend to continue with future warming, with strong environmental and socio-economic implications. The Jostedalsbreen ice cap is the largest ice cap on the European mainland (458 km2 in 2019) and occupies 20 % of the total glacier area of mainland Norway. Here we simulate the evolution of Jostedalsbreen since 1960, and its fate in a changing climate in the 21st-century and beyond (2300). This ice cap consists of glacier units with a great diversity in shape, steepness, hypsometry, and flow speed. We employ a coupled model system with higher-order three-dimensional ice dynamics forced by simulated surface mass balance that fully accounts for the mass-balance elevation feedback. We find that Jostedalsbreen may lose 12 %–74 % of its present-day volume until 2100, depending on future greenhouse gas emissions. With mid-range results obtained using the climate model ECEARTH/CCLM, Jostedalsbreen is projected to lose 49 % (RCP4.5) and 63 % (RCP8.5) of its contemporary ice volume by 2100. Regardless of emission scenario, the ice cap is likely to split into three parts during the second half of the 21st century. Our results suggest that Jostedalsbreen will likely be more resilient than many smaller glaciers and ice caps in Scandinavia. However, we show that by the year 2100, the ice cap may be committed to a complete disappearance during the 22nd century, under high emissions (RCP8.5). Under medium 21st-century emissions (RCP4.5), the ice cap is bound to shrink by 90 % until 2300. Further simulations indicate that substantial mass losses undergone until 2100 are irreversible; the ice cap would not recover to its contemporary volume if the future surface mass balance was reversed to that of the present-day. Our study demonstrates a model approach for complex ice masses with numerous outlet glaciers such as ice caps, and how tightly linked future mass loss is to future greenhouse-gas emissions. Finally, uncertainties in future climate conditions, particularly precipitation, appear to be the largest source of uncertainty in future projections of maritime ice masses like Jostedalsbreen.

• Maslov, Konstantin; Persello, Claudio; Schellenberger, Thomas & Stein, Alfred (2025). Globally scalable glacier mapping by deep learning matches expert delineation accuracy. Nature Communications. 16. doi: 10.1038/s41467-024-54956-x. Fulltekst i vitenarkiv Vis sammendrag

  Accurate global glacier mapping is critical for understanding climate change impacts. Despite its importance, automated glacier mapping at a global scale remains largely unexplored. Here we address this gap and propose Glacier-VisionTransformer-U-Net (GlaViTU), a convolutional-transformer deep learning model, and five strategies for multitemporal global-scale glacier mapping using open satellite imagery. Assessing the spatial, temporal and cross-sensor generalisation shows that our best strategy achieves intersection over union >0.85 on previously unobserved images in most cases, which drops to >0.75 for debris-rich areas such as High-Mountain Asia and increases to >0.90 for regions dominated by clean ice. A comparative validation against human expert uncertainties in terms of area and distance deviations underscores GlaViTU performance, approaching or matching expert-level delineation. Adding synthetic aperture radar data, namely, backscatter and interferometric coherence, increases the accuracy in all regions where available. The calibrated confidence for glacier extents is reported making the predictions more reliable and interpretable. We also release a benchmark dataset that covers 9% of glaciers worldwide. Our results support efforts towards automated multitemporal and global glacier mapping.

• Mannerfelt, Erik Schytt; Schellenberger, Thomas & Kääb, Andreas (2025). Tracking glacier surge evolution using interferometric SAR coherence-examples from Svalbard. Journal of Glaciology. ISSN 0022-1430. 71. doi: 10.1017/jog.2025.27. Fulltekst i vitenarkiv Vis sammendrag

  We present a practically simple methodology for tracking glacier surge onset and evolution using interferometric synthetic aperture radar (InSAR) coherence. Detecting surges early and monitoring their build-up is interesting for a multitude of scientific and safety-related aspects. We show that InSAR coherence maps allow the detection of surge-related instability on Svalbard many years before being detectable by, for instance, feature tracking or crevasse detection. Furthermore, we present derived data for two types of surges; down- and up-glacier propagating, with interestingly consistent surge propagation and post-surge relaxation rates. The method works well on Svalbard glaciers, and the data and core principle suggest a global applicability.

• Maslov, Konstantin; Persello, Claudio; Schellenberger, Thomas & Stein, Alfred (2024). Towards Global Glacier Mapping with Deep Learning and Open Earth Observation Data. arXiv.org. ISSN 2331-8422. doi: 10.48550/arXiv.2401.15113. Fulltekst i vitenarkiv Vis sammendrag

  Accurate global glacier mapping is critical for understanding climate change impacts. It is challenged by glacier diversity, difficult-to-classify debris and big data processing. Here we propose Glacier-VisionTransformer-U-Net (GlaViTU), a convolutional-transformer deep learning model, and five strategies for multitemporal global-scale glacier mapping using open satellite imagery. Assessing the spatial, temporal and cross-sensor generalisation shows that our best strategy achieves intersection over union >0.85 on previously unobserved images in most cases, which drops to >0.75 for debris-rich areas such as High-Mountain Asia and increases to >0.90 for regions dominated by clean ice. Additionally, adding synthetic aperture radar data, namely, backscatter and interferometric coherence, increases the accuracy in all regions where available. The calibrated confidence for glacier extents is reported making the predictions more reliable and interpretable. We also release a benchmark dataset that covers 9% of glaciers worldwide. Our results support efforts towards automated multitemporal and global glacier mapping.

• Maslov, Konstantin; Persello, Claudio; Schellenberger, Thomas & Stein, Alfred (2023). GLAVITU: A Hybrid CNN-Transformer for Multi-Regional Glacier Mapping from Multi-Source Data, IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium. IEEE (Institute of Electrical and Electronics Engineers). ISSN 9798350320107. s. 1233–1236. doi: 10.1109/IGARSS52108.2023.10281828. Fulltekst i vitenarkiv Vis sammendrag

  Glacier mapping is essential for studying and monitoring the impacts of climate change. However, several challenges such as debris-covered ice and highly variable landscapes across glacierized regions worldwide complicate large-scale glacier mapping in a fully-automated manner. This work presents a novel hybrid CNN-transformer model (GlaViTU) for multi-regional glacier mapping. Our model outperforms three baseline models—SETR-B/16, ResU-Net and TransU-Net—achieving a higher mean IoU of 0.875 and demonstrates better generalization ability. The proposed model is also parameter-efficient, with approximately 10 and 3 times fewer parameters than SETR-B/16 and ResU-Net, respectively. Our results provide a solid foundation for future studies on the application of deep learning methods for global glacier mapping. To facilitate reproducibility, we have shared our data set, codebase and pretrained models on GitHub at https://github.com/konstantin-a-maslov/GlaViTU-IGARSS2023.

• Nuth, Christopher; Gilbert, Adrien; Köhler, Andreas; McNabb, Robert Whitfield; Schellenberger, Thomas & Sevestre, Heidi [Vis alle 10 forfattere av denne artikkelen] (2019). Dynamic vulnerability revealed in the collapse of an Arctic tidewater glacier. Scientific Reports. 9. doi: 10.1038/s41598-019-41117-0. Fulltekst i vitenarkiv Vis sammendrag

  Glacier flow instabilities can rapidly increase sea level through enhanced ice discharge. Surge-type glacier accelerations often occur with a decadal to centennial cyclicity suggesting internal mechanisms responsible. Recently, many surging tidewater glaciers around the Arctic Barents Sea region question whether external forces such as climate can trigger dynamic instabilities. Here, we identify a mechanism in which climate change can instigate surges of Arctic tidewater glaciers. Using satellite and seismic remote sensing observations combined with three-dimensional thermo-mechanical modeling of the January 2009 collapse of the Nathorst Glacier System (NGS) in Svalbard, we show that an underlying condition for instability was basal freezing and associated friction increase under the glacier tongue. In contrast, continued basal sliding further upstream increased driving stresses until eventual and sudden till failure under the tongue. The instability propagated rapidly up-glacier, mobilizing the entire 450 km2 glacier basin over a few days as the till entered an unstable friction regime. Enhanced mass loss during and after the collapse (5–7 fold compared to pre-collapse mass losses) combined with regionally rising equilibrium line altitudes strongly limit mass replenishment of the glacier, suggesting irreversible consequences. Climate plays a paradoxical role as cold glacier thinning and retreat promote basal freezing which increases friction at the tongue by stabilizing an efficient basal drainage system. However, with some of the most intense atmospheric warming on Earth occurring in the Arctic, increased melt water can reduce till strength under tidewater glacier tongues to orchestrate a temporal clustering of surges at decadal timescales, such as those observed in Svalbard at the end of the Little Ice Age. Consequently, basal terminus freezing promotes a dynamic vulnerability to climate change that may be present in many Arctic tidewater glaciers.

• Winsvold, Solveig Havstad; Kääb, Andreas; Nuth, Christopher; Andreassen, Liss Marie; Pelt, Ward J.J. Van & Schellenberger, Thomas (2018). Using SAR satellite data time series for regional glacier mapping. The Cryosphere. ISSN 1994-0416. 12(3), s. 867–890. doi: 10.5194/tc-12-867-2018. Fulltekst i vitenarkiv Vis sammendrag

  With dense SAR satellite data time series it is possible to map surface and subsurface glacier properties that vary in time. On Sentinel-1A and RADARSAT-2 backscatter time series images over mainland Norway and Svalbard, we outline how to map glaciers using descriptive methods. We present five application scenarios. The first shows potential for tracking transient snow lines with SAR backscatter time series and correlates with both optical satellite images (Sentinel-2A and Landsat 8) and equilibrium line altitudes derived from in situ surface mass balance data. In the second application scenario, time series representation of glacier facies corresponding to SAR glacier zones shows potential for a more accurate delineation of the zones and how they change in time. The third application scenario investigates the firn evolution using dense SAR backscatter time series together with a coupled energy balance and multilayer firn model. We find strong correlation between backscatter signals with both the modeled firn air content and modeled wetness in the firn. In the fourth application scenario, we highlight how winter rain events can be detected in SAR time series, revealing important information about the area extent of internal accumulation. In the last application scenario, averaged summer SAR images were found to have potential in assisting the process of mapping glaciers outlines, especially in the presence of seasonal snow. Altogether we present examples of how to map glaciers and to further understand glaciological processes using the existing and future massive amount of multi-sensor time series data.

• Gong, Yongmei; Zwinger, Thomas; Åström, J.; Altena, Bas; Schellenberger, Thomas & Gladstone, Rupert [Vis alle 7 forfattere av denne artikkelen] (2018). Simulating the roles of crevasse routing of surface water and basal friction on the surge evolution of Basin 3, Austfonna ice cap. The Cryosphere. ISSN 1994-0416. 12(5), s. 1563–1577. doi: 10.5194/tc-12-1563-2018. Fulltekst i vitenarkiv Vis sammendrag

  The marine-terminating outlet in Basin 3, Austfonna ice cap, has been accelerating since the mid-1990s. Stepwise multi-annual acceleration associated with seasonal summer speed-up events was observed before the outlet entered the basin-wide surge in autumn 2012. We used multiple numerical models to explore hydrologic activation mechanisms for the surge behaviour. A continuum ice dynamic model was used to invert basal friction coefficient distributions using the control method and observed surface velocity data between April 2012 and July 2014. This has provided input to a discrete element model capable of simulating individual crevasses, with the aim of finding locations where meltwater entered the glacier during the summer and reached the bed. The possible flow paths of surface meltwater reaching the glacier bed as well as those of meltwater produced at the bed were calculated according to the gradient of the hydraulic potential. The inverted friction coefficients show the “unplugging” of the stagnant ice front and expansion of low-friction regions before the surge reached its peak velocity in January 2013. Crevasse distribution reflects the basal friction pattern to a high degree. The meltwater reaches the bed through the crevasses located above the margins of the subglacial valley and the basal melt that is generated mainly by frictional heating flows either to the fast-flowing units or potentially accumulates in an overdeepened region. Based on these results, the mechanisms facilitated by basal meltwater production, crevasse opening and the routing of meltwater to the bed are discussed for the surge in Basin 3.

• Solgaard, Anne; Messerli, Alexandra; Schellenberger, Thomas; Hvidberg, Christine; Grinsted, Aslak & Jackson, Miriam [Vis alle 9 forfattere av denne artikkelen] (2018). Basal conditions at Engabreen, Norway, inferred from surface measurements and inverse modelling. Journal of Glaciology. ISSN 0022-1430. 64(246), s. 555–567. doi: 10.1017/jog.2018.45. Fulltekst i vitenarkiv Vis sammendrag

  Engabreen is an outlet glacier of the Svartisen Ice Cap located in Northern Norway. It is a unique glacier due to the Svartisen Subglacial Laboratory which allows direct access to the glacier bed. In this study, we combine both sub- and supraglacial observations with ice-flow modelling in order to investigate conditions at the bed of Engabreen both spatially and temporally. We use the full-Stokes model Elmer/Ice and satellite-based surface-velocity maps from 2010 and 2014 to infer patterns of basal friction. Direct measurements of basal sliding and deformation of lower layers of the ice are used to adjust the ice viscosity and provide essential input to the setup of our model and influence the interpretation of the results. We find a clear seasonal cycle in the subglacial conditions at the higher elevation region of the study area and discuss this in relation to the subglacial hydrological system. Our results also reveal an area with an overdeepening where basal friction is significantly lower than elsewhere on the glacier all year round. We attribute this to either water pooling at the base, or saturated sediments and increased strain heating at this location which softens the ice further.

• Gong, Yongmei; Zwinger, Thomas; Åström, Jan; Altena, Bas; Schellenberger, Thomas & Gladstone, Rupert [Vis alle 7 forfattere av denne artikkelen] (2018). Simulating the roles of crevasse routing of surface water and basal friction on the surge evolution of Basin 3, Austfonna ice-cap. The Cryosphere. ISSN 1994-0416. 12, s. 1563–1577. doi: 10.5194/tc-12-1563-2018. Fulltekst i vitenarkiv
• Strozzi, Tazio; Paul, Frank; Wiesmann, Andreas; Schellenberger, Thomas & Kääb, Andreas (2017). Circum-arctic changes in the flow of glaciers and ice caps from satellite SAR data between the 1990s and 2017. Remote Sensing. 9(9). doi: 10.3390/rs9090947. Fulltekst i vitenarkiv Vis sammendrag

  We computed circum-Arctic surface velocity maps of glaciers and ice caps over the Canadian Arctic, Svalbard and the Russian Arctic for at least two times between the 1990s and 2017 using satellite SAR data. Our analyses are mainly performed with offset-tracking of ALOS-1 PALSAR-1 (2007–2011) and Sentinel-1 (2015–2017) data. In certain cases JERS-1 SAR (1994–1998), TerraSAR-X (2008–2012), Radarsat-2 (2009–2016) and ALOS-2 PALSAR-2 (2015–2016) data were used to fill-in spatial or temporal gaps. Validation of the latest Sentinel-1 results was accomplished by means of SAR data at higher spatial resolution (Radarsat-2 Wide Ultra Fine) and ground-based measurements. In general, we observe a deceleration of flow velocities for the major tidewater glaciers in the Canadian Arctic and an increase in frontal velocity along with a retreat of frontal positions over Svalbard and the Russian Arctic. However, all regions have strong accelerations for selected glaciers. The latter developments can be well traced based on the very high temporal sampling of Sentinel-1 acquisitions since 2015, revealing new insights in glacier dynamics. For example, surges on Spitsbergen (e.g., Negribreen, Nathorsbreen, Penckbreen and Strongbreen) have a different characteristic and timing than those over Eastern Austfonna and Edgeoya (e.g., Basin 3, Basin 2 and Stonebreen). Events similar to those ongoing on Eastern Austofonna were also observed over the Vavilov Ice Cap on Severnaya Zemlya and possibly Simony Glacier on Franz-Josef Land. Collectively, there seems to be a recently increasing number of glaciers with frontal destabilization over Eastern Svalbard and the Russian Arctic compared to the 1990s.

• Strozzi, Tazio; Kääb, Andreas & Schellenberger, Thomas (2017). Frontal destabilization of Stonebreen, Edgeøya, Svalbard. The Cryosphere. ISSN 1994-0416. 11, s. 553–566. doi: 10.5194/tc-11-553-2017. Fulltekst i vitenarkiv Vis sammendrag

  In consideration of the strong atmospheric warming that has been observed since the 1990s in polar regions there is a need to quantify mass loss of Arctic ice caps and glaciers and their contribution to sea level rise. In polar regions a large part of glacier ablation is through calving of tidewater glaciers driven by ice velocities and their variations. The Svalbard region is characterized by glaciers with rapid dynamic fluctuations of different types, including irreversible adjustments of calving fronts to a changing mass balance and reversible, surge-type activities. For large areas, however, we do not have much past and current information on glacier dynamic fluctuations. Recently, through frequent monitoring based on repeat optical and synthetic aperture radar (SAR) satellite data, a number of zones of velocity increases have been observed at formerly slow-flowing calving fronts on Svalbard. Here we present the dynamic evolution of the southern lobe of Stonebreen on Edgeøya. We observe a slowly steady retreat of the glacier front from 1971 until 2011, followed by a strong increase in ice surface velocity along with a decrease of volume and frontal extension since 2012. The considerable losses in ice thickness could have made the tide-water calving glacier, which is grounded below sea level some 6 km inland from the 2014 front, more sensitive to surface meltwater reaching its bed and/or warm ocean water increasing frontal ablation with subsequent strong multi-annual ice-flow acceleration.

• Paul, Frank; Strozzi, Tazio; Schellenberger, Thomas & Kääb, Andreas (2017). The 2015 surge of hispar glacier in the karakoram. Remote Sensing. 9(9), s. 1–18. doi: 10.3390/rs9090888. Fulltekst i vitenarkiv Vis sammendrag

  The Karakoram mountain range is well known for its numerous surge-type glaciers of which several have recently surged or are still doing so. Analysis of multi-temporal satellite images and digital elevation models have revealed impressive details about the related changes (e.g., in glacier length, surface elevation and flow velocities) and considerably expanded the database of known surge-type glaciers. One glacier that has so far only been reported as impacted by surging tributaries, rather than surging itself, is the 50 km long main trunk of Hispar Glacier in the Hunza catchment. We here present the evolution of flow velocities and surface features from its 2015/16 surge as revealed from a dense time series of SAR and optical images along with an analysis of historic satellite images. We observed maximum flow velocities of up to 14 m d−1 (5 km a−1) in spring 2015, sudden drops in summer velocities, a second increase in winter 2015/16 and a total advance of the surge front of about 6 km. During a few months the surge front velocity was much higher (about 90 m d−1) than the maximum flow velocity. We assume that one of its northern tributary glaciers, Yutmaru, initiated the surge at the end of summer 2014 and that the variability in flow velocities was driven by changes in the basal hydrologic regime (Alaska-type surge). We further provide evidence that Hispar Glacier has surged before (around 1960) over a distance of about 10 km so that it can also be regarded as a surge-type glacier.

• Schellenberger, Thomas; Wychen, Wesley Van; Copland, Luke; Kääb, Andreas & Gray, Laurence (2016). An Inter-Comparison of Techniques for Determining Velocities of Maritime Arctic Glaciers, Svalbard, Using Radarsat-2 Wide Fine Mode Data. Remote Sensing. 8(9)(785). doi: 10.3390/rs8090785. Fulltekst i vitenarkiv Vis sammendrag

  Glacier dynamics play an important role in the mass balance of many glaciers, ice caps and ice sheets. In this study we exploit Radarsat-2 (RS-2) Wide Fine (WF) data to determine the surface speed of Svalbard glaciers in the winters of 2012/2013 and 2013/2014 using Synthetic Aperture RADAR (SAR) offset and speckle tracking. The RS-2 WF mode combines the advantages of the large spatial coverage of the Wide mode (150 × 150 km) and the high pixel resolution (9 m) of the Fine mode and thus has a major potential for glacier velocity monitoring from space through offset and speckle tracking. Faster flowing glaciers (1.95 m · d − 1 –2.55 m · d − 1 ) that are studied in detail are Nathorstbreen, Kronebreen, Kongsbreen and Monacobreen. Using our Radarsat-2 WF dataset, we compare the performance of two SAR tracking algorithms, namely the GAMMA Remote Sensing Software and a custom written MATLAB script (GRAY method) that has primarily been used in the Canadian Arctic. Both algorithms provide comparable results, especially for the faster flowing glaciers and the termini of slower tidewater glaciers. A comparison of the WF data to RS-2 Ultrafine and Wide mode data reveals the superiority of RS-2 WF data over the Wide mode data.

• Strozzi, Tazio; Kääb, Andreas & Schellenberger, Thomas (2016). Frontal destabilisation of Stonebreen, Edgeøya, Svalbard. The Cryosphere Discussions. doi: 10.5194/tc-2016-213. Fulltekst i vitenarkiv Vis sammendrag

  In consideration of the strong atmospheric warming that has been observed since the 1990s in polar regions there is a need to quantify ice mass loss of Arctic ice caps and glaciers and their contribution to sea level rise. In polar regions a large part of glacier ablation is through calving of tidewater glaciers driven by ice velocities and their variations. The Svalbard region is characterized by glaciers with rapid dynamic fluctuations of different types, including irreversible adjustments of calving fronts to a changing mass balance and reversible, surge-type activities. For large areas, however, we don't have much past and current information on glacier's dynamic fluctuations. Recently, through frequent monitoring based on repeat optical and SAR satellite data, a number of zones of speed increases have been observed at formerly slow-flowing calving fronts on Svalbard. Here we present the dynamic evolution of the southern lobe of Stonebreen on Edgeøya. We observe a slowly steady retreat of the glacier front from 1971 until 2011 followed since 2012 by a strong increase in ice surface velocity along with a decrease of volume and an advance in frontal extension. The considerable losses in ice thickness could have made the tide-water calving glacier, which is grounded below sea level some 6 km inland from the 2014 front, more sensitive to surface melt-water reaching its bed and/or warm ocean water increasing frontal ablation with subsequent strong multi-annual ice-flow acceleration

• Schellenberger, Thomas; Dunse, Thorben; Kääb, Andreas; Kohler, Jack & Reijmer, Carleen (2015). Surface speed and frontal ablation of Kronebreen and Kongsbreen, NW Svalbard, from SAR offset tracking. The Cryosphere. ISSN 1994-0416. 9(6), s. 2339–2355. doi: 10.5194/tc-9-2339-2015. Fulltekst i vitenarkiv
• Dunse, Thorben; Schellenberger, Thomas; Hagen, Jon Ove Methlie; Kääb, Andreas; Schuler, Thomas Vikhamar & Reijmer, Carleen (2015). Glacier-surge mechanisms promoted by a hydro-thermodynamic feedback to summer melt. The Cryosphere. ISSN 1994-0416. 9, s. 197–215. doi: 10.5194/tc-9-197-2015. Fulltekst i vitenarkiv Vis sammendrag

  Mass loss from glaciers and ice sheets currently accounts for two-thirds of the observed global sea-level rise and has accelerated since the 1990s, coincident with strong atmospheric warming in the polar regions. Here we present continuous GPS measurements and satellite synthetic-aperture-radar-based velocity maps from Basin-3, the largest drainage basin of the Austfonna ice cap, Svalbard. Our observations demonstrate strong links between surface-melt and multiannual ice-flow acceleration. We identify a hydro-thermodynamic feedback that successively mobilizes stagnant ice regions, initially frozen to their bed, thereby facilitating fast basal motion over an expanding area. By autumn 2012, successive destabilization of the marine terminus escalated in a surge of Basin-3. The resulting iceberg discharge of 4.2±1.6 Gt a−1 over the period April 2012 to May 2013 triples the calving loss from the entire ice cap. With the seawater displacement by the terminus advance accounted for, the related sea-level rise contribution amounts to 7.2±2.6 Gt a−1. This rate matches the annual ice-mass loss from the entire Svalbard archipelago over the period 2003–2008, highlighting the importance of dynamic mass loss for glacier mass balance and sea-level rise. The active role of surface melt, i.e. external forcing, contrasts with previous views of glacier surges as purely internal dynamic instabilities. Given sustained climatic warming and rising significance of surface melt, we propose a potential impact of the hydro-thermodynamic feedback on the future stability of ice-sheet regions, namely at the presence of a cold-based marginal ice plug that restricts fast drainage of inland ice. The possibility of large-scale dynamic instabilities such as the partial disintegration of ice sheets is acknowledged but not quantified in global projections of sea-level rise.

• Dunse, Thorben; Schellenberger, Thomas; Hagen, Jon Ove Methlie; Kääb, Andreas; Schuler, Thomas & Reijmer, CH (2014). Destabilisation of an Arctic ice cap triggered by a hydro-thermodynamic feedback to summer-melt. The Cryosphere Discussions. 8, s. 2685–2719. doi: 10.5194/tcd-8-2685-2014. Fulltekst i vitenarkiv Vis sammendrag

  Mass loss from glaciers and ice sheets currently accounts for two-thirds of the observed global sea-level rise and has accelerated since the 1990s, coincident with strong atmospheric warming in the Polar Regions. Here we present continuous GPS measurements and satellite synthetic aperture radar based velocity maps from the Austfonna ice cap, Svalbard, that demonstrate strong links between surface-melt and multiannual ice-flow acceleration. We identify a hydro-thermodynamic feedback that successively mobilizes stagnant ice regions, initially frozen to their bed, thereby facilitating fast basal motion over an expanding area. By autumn 2012, successive destabilization of the marine terminus escalated in a surge of the ice cap's largest drainage basin, Basin-3. The resulting iceberg discharge of 4.2 ± 1.6 Gt a−1 over the period April 2012 to May 2013 triples the calving loss from the entire ice cap. After accounting for the terminus advance, the related sea-level rise contribution of 7.2 ± 2.6 Gt a−1 matches the recent annual ice-mass loss from the entire Svalbard archipelago. Our study highlights the importance of dynamic glacier wastage and illuminates mechanisms that may trigger a sustained increase in dynamic glacier wastage or the disintegration of ice-sheets in response to climate warming, which is acknowledged but not quantified in global projections of sea-level rise.

• Schellenberger, Thomas; Dunse, Thorben; Kääb, Andreas; Kohler, Jack & Reijmer, Carleen (2014). Surface speed and frontal ablation of Kronebreen and Kongsbreen, NW-Svalbard, from SAR offset tracking. The Cryosphere Discussions. 8, s. 6193–6233. doi: 10.5194/tcd-8-6193-2014. Fulltekst i vitenarkiv Vis sammendrag

  Kronebreen and Kongsbreen are among the fastest-flowing glaciers on Svalbard and, therefore, important contributors to the total dynamic mass loss from the archipelago. Here, we present a time series of area-wide surface velocity fields from April 2012 to December 2013 based on offset tracking on repeat high-resolution Radarsat-2 Ultrafine data. Surface speeds reached up to 3.2md-1 near the calving front of Kronebreen in summer 2013 and 2.7md-1 at Kongsbreen in late autumn 2012. Additional velocity fields from Radarsat-1, Radarsat-2 and TerraSAR-X data since December 2007 together with continuous GPS measurements on Kronebreen since September 2008 revealed complex patterns in seasonal and interannual speed evolution. Part of the ice-flow variations seem closely linked to the amount and timing of surface meltwater production and rainfall, both of which are known to have a strong influence on the basal water pressure and hence basal lubrication. In addition, terminus retreat and the associated reduction in back stress appear to have influenced the speed close to the calving front, especially at Kongsbreen in 2012 and 2013. Since 2007, Kongsbreen retreated up to 1800 m, corresponding to a total area loss of 2.5 km2. In 2011 the retreat of Kronebreen of up to 850 m, responsible for a total area loss of 2.8 km2, was triggered after a phase of stable terminus position since 1990. Retreat is an important component of the mass balance of both glaciers, in which frontal ablation is the largest component. Total frontal ablation between April 2012 and December 2013 was estimated to 0.21–0.25 Gt a-1 for Kronebreen and 0.14–0.16 Gt a-1 for Kongsbreen.

• Notarnicola, Claudia; Duguay, Martial; Moelg, Nico; Schellenberger, Thomas; Tetzlaff, Anke & Monsorno, Roberto [Vis alle 9 forfattere av denne artikkelen] (2013). Snow Cover Maps from MODIS Images at 250 m Resolution, Part 2: Validation. Remote Sensing. 5(4), s. 1568–1587. doi: 10.3390/rs5041568. Fulltekst i vitenarkiv
• Notarnicola, Claudia; Duguay, Martial; Moelg, Nico; Schellenberger, Thomas; Tetzlaff, Anke & Monsorno, Roberto [Vis alle 9 forfattere av denne artikkelen] (2013). Snow Cover Maps from MODIS Images at 250 m Resolution, Part 1: Algorithm Description. Remote Sensing. 5(1), s. 110–126. doi: 10.3390/rs5010110. Fulltekst i vitenarkiv

Se alle arbeider i NVA

• Maslov, Konstantin; Schellenberger, Thomas; Paul, Frank; Goerlich, Franz; Rastner, Philipp & Kohler, Jack [Vis alle 8 forfattere av denne artikkelen] (2026). Benchmark dataset “ICEmapper Svalbard” - Dataset - Sigma2 Data Catalogue. doi: https:/archive.sigma2.no/dataset/benchmark-dataset-icemapper-svalbard. Fulltekst i vitenarkiv Vis sammendrag

  Sigma2 Data Catalogue

• Maslov, Konstantin; Schellenberger, Thomas; PANDIT, PRASHANT H.; Persello, Claudio & Stein, Alfred (2026). Pan-European dataset for glacier facies classification. doi: 10.5281/ZENODO.18469893. Fulltekst i vitenarkiv Vis sammendrag

  The dataset includes 137592 manually annotated point labels and the corresponding image stacks collected on 92 Landsat and Sentinel-2 scenes covering 31 glaciers across the Alps, Pyrenees, Scandinavia, Iceland and Svalbard. Eight surface classes are provided: ice, snow, firn, debris, refrozen-like ice, shadow, water and cloud. Labels were produced by three expert glaciologists and refined through a cross-check procedure involving harmonisation of ambiguous transitions and addition of points in sparsely sampled areas. Pre- and post-cross-check label sets are included, along with acquisition metadata to facilitate reproducibility. The dataset is intended as a benchmark for automated glacier facies mapping and for developing machine learning or remote sensing algorithms under diverse glaciological and radiometric conditions. Please cite the accompanying paper when using these data. To cite this dataset, please follow the instructions at https://github.com/konstantin-a-maslov/glacier_facies_classification

• Schellenberger, Thomas; Maslov, Konstantin & Persello, Claudio (2025). Space-time dimension swap: a novel deep learning based approach to extract calving fronts of marine-terminating glaciers from time series of Sentinel-1 imagery. Fulltekst i vitenarkiv Vis sammendrag

  Marine-terminating glaciers are important contributors to the total the mass balance of all glaciers in the Northern Hemisphere, despite their small count of only ~1500. Several deep learning-based approaches have successfully been applied to specifically map calving fronts from optical and SAR satellite imagery. Yet, this task poses a specific challenge: accuracy decreases significantly when “fjord ice”, e.g. icebergs, seasonal sea ice or ice melange, with similar spectral and textural signature is present. As published methods are all based on single images, they are prone to artifacts of fjord-ice in the outlines. Reported general accuracies are greater 3-10 pixels under good conditions. The standard approach is to map the termini in each satellite image of a given date separately (2D in space & 0D in time). Stacking all retrieved outlines from a time series of satellite images temporally then results in the 3D calving front time series. Here we propose a novel approach, in which we will swap the time and space dimension in the image segmentation. We first extract backscatter values along a flowline of a given glacier (which crosses the terminus) from all Sentinel-1 images (1D in space & 1D in time). Each column in the resulting “time-series” image represents data along the centreline at a given date. We then manually label pixel coordinates where the flowline crosses the calving front. The front is clearly identifiable either from the strong backscatter as the front acts as a large corner reflector, or in very low backscatter due to the front’s radar shadow. The resulting binary masks of several glaciers are then used to train an adapted version of our vision-transformer convolutional neural network hybrid model GlaViTU. The model is applied to all “time series images” extracted along a dense pixel-by-pixel grid of flowlines parallel to the original flowline, covering the whole calving front of a glacier. The resulting “label-like” images are then stacked spatially to retrieve the full 3D time series of calving front of the glacier, back-swapping the dimensions. The novel approach has four advantages over current methods: Firstly, it allows to retrieve data from a single or multiple complete sensor collection(s) for a glacier or region with high computational efficiency. Secondly, labelling single points drastically increases the quality of the training data and reduces expert workload. Thirdly, “fjord ice” ends up in the “fjord” class, which is highly inhomogeneous in time and hence easier to distinguish from glacier ice. Fourthly, with expert knowledge, the maximum distance between neighbouring points in time can be constrained using a expert-informed loss functions in the deep learning model, reducing the average accuracy to an estimated 1-3 pixels, e.g. 10-30 m for Sentinel-1. We will present a time series of calving fronts of all marine-terminating glaciers in Svalbard and Alaska from 2016-2024 retrieved from Sentinel-1 data.

• Schellenberger, Thomas; Maslov, Konstantin & Persello, Claudio (2025). Space-time dimension swap: a novel deep learning based approach to extract calving fronts of marine-terminating glaciers from time series of Sentinel-1 imagery. doi: https:/2025.ieeeigarss.org/view_paper.php?PaperNum=5567&SessionID=1624. Fulltekst i vitenarkiv Vis sammendrag

  Marine-terminating glaciers are important contributors to the total the mass balance of all glaciers in the Northern Hemisphere, despite their small count of only ~1500. Several deep learning-based approaches have successfully been applied to specifically map calving fronts from optical and SAR satellite imagery. Yet, this task poses a specific challenge: accuracy decreases significantly when “fjord ice”, e.g. icebergs, seasonal sea ice or ice mélange, with similar spectral and textural signature is present. As published methods are all based on single images, they are prone to artifacts of fjord-ice in the outlines. Reported general accuracies are greater 3-10 pixels under good conditions. Here we propose a novel approach, in which we will swap the time and space dimension in the image segmentation. The standard approach is to map the termini in each satellite image of a given date separately (2D in space & 0D in time). Stacking all retrieved outlines from a time series of satellite images temporally then results in the 3D calving front time series. Instead, we first extract backscatter values along a given flowline which crosses the termini from all Sentinel-1 images of a given glacier (1D in space & 1D in time). Each column in the image represents data along the centreline at a given date. We then manually label pixel coordinates where the flowline crosses the calving front. The front is clearly identifiable either from the strong backscatter as the front acts as a large corner reflector, or in very low backscatter due to the front’s radar shadow. The resulting binary masks of several glaciers are then used to train an neural network. The model is applied to all space-time images extracted along a dense pixel-by-pixel grid of flowlines parallel to the red line, covering the whole calving front of a given glacier. The resulting “label-like” images are then stacked spatially to retrieve the full 3D time series of calving front of a given glacier, back-swapping the dimensions.

• Maslov, Konstantin; Schellenberger, Thomas; Persello, Claudio & Stein, Alfred (2025). Deep learning applied to Sentinel-1 data shows doubling of glacier area loss in Svalbard compared to 1980–2010. doi: 10.5194/egusphere-egu25-9898. Fulltekst i vitenarkiv Vis sammendrag

  Glaciers in Svalbard are undergoing rapid changes due to Arctic Amplification that demand frequent and accurate monitoring. We present a deep learning model, Intensity-Coherence-Evolution-mapper (ICEmapper) designed to extract annual glacier outlines from Sentinel-1 time series data regardless of cloud cover. The model combines SAR backscatter intensity and interferometric coherence. In extensive validation tests against manually digitised optical imagery, ICEmapper demonstrates human-expert accuracy, with intersection of union score higher than 0.95, total area discrepancies below 0.5%, median distance deviations under 15 m, and 95th percentile deviations within 250 m. Additionally, we report calibrated uncertainties of our classification results at the pixel level, allowing for detailed analysis of significant changes as well as total area uncertainty estimation. This performance allowed us to construct a continuous inventory of glacier outlines in Svalbard from 2016 to 2023. The results of area change analysis suggest a substantial escalation in the rate of glacier area loss to about 180 km2 a−1 in the last decade, nearly doubling from the previously reported ~80 km2 a−1 (1980–2010; Nuth et al., 2013). This increase is predominantly driven by enhanced calving at tidewater glaciers, although climatic signal also shows a significant correlation with the area loss of land-terminating glaciers. Surging glaciers, particularly in the Nathorstbreen system and Austfonna, exhibited unique behaviours that can temporarily increase the total glacier area. In 2016, the Nathorstbreen system gained 107.76 km2, while Austfonna, Basin-3 expanded by 86.54 km2 as compared to the Randolph Glacier Inventory, jointly offsetting net losses by approximately two years. In the last decade, however, surging glaciers lost area more rapidly (−0.57 km2 a−1 on average) than the non-surging ones (−0.09 km2 a−1). Additionally, our analysis uncovered a surge in Austfonna, Basin-7 starting in 2019 and not reported previously, emphasising the capability of annually updated inventories to complement other methods of surge detection. Our methods have the potential to be transferred to other glacierised regions and enhance monitoring and understanding of glacier area changes on larger scales using Sentinel-1 data.

• Pandit, Prashant; Schellenberger, Thomas; Callegari, Mattia & Bruzzone, Lorenzo (2025). Glacier Snowline Mapping from Sentinel-2 images by Machine Learning. doi: https:/lps25.esa.int/lps25-presentations/presentations/382/_382.pdf. Fulltekst i vitenarkiv Vis sammendrag

  Snow accumulation and ablation play a pivotal role in the mass balance of mountain glaciers. Accurate mapping of snow extends on glaciers enhance our understanding of climate change impacts (Larocca et al.) and serve as valuable input for glacier surface mass balance models (Rabatel et al.). For example, the snowline altitude at the end of the summer season serves as a proxy for the Equilibrium Line Altitude and strongly correlated with the annual mass balance (Rabatel et al.). The largest study mapped 3489 snowlines of 269 glaciers of the ~275.000 glaciers from Landsat data between year 1984 and 2022 and found an increase in snow line elevation of approximately 150 meters. This significant rise indicates a reduction in glacier accumulation zones, suggesting negative mass balance trends driven by rising temperatures and shifting precipitation patterns. Regional variability in snowline changes underscores the complex interplay of climatic and local factors, emphasizing the importance of snowline monitoring for glacier health and climate change studies (Larocca et al.). However, manual delineation is time consuming and band- and index threshold-based approaches often fails in areas of steep and complex mountainous terrain and in homogenous snow conditions. The spectral similarity to firn and ice and varying illumination conditions in steep terrain are posing significant challenges for accurate large-scale and long-term snow monitoring Advanced machine learning techniques offer promising solutions to address these limitations by automating snowline detection and improving accuracy under diverse conditions (Prieur et al.). Designed to overcome these limitations by improving accuracy and scalability, this study presents first steps towards region-scale snow cover extend mapping using machine learning. We manually digitized 312 snowlines on 41 glaciers in Scandinavia (13), Svalbard (9), and the European Alps (19) from Sentinel 2 data in the period 2015 to 2023, encompassing a wide range of seasonal snow conditions., Using this benchmark, we trained several machine learning models, including pixel-based classifiers such as Support Vector Machine, Random Forest, and XGBoost (Chen et al.), and U-Net (Ronneberger et al.), a Fully convolutional neural network , and compared them against threshold-based approaches as baselines. The results from Scandinavia demonstrate the superiority of machine learning methods. While threshold-based approaches, such as the Normalized Difference Snow Index (NDSI > 0.4) and Near-Infrared (NIR > 0.11), achieved an Intersection over Union (IoU) score of 0.7147, U-Net significantly outperformed with an IoU of 0.9456. Random Forest was the next best-performing method with an IoU of 0.8957, followed by XGBoost (0.8899) and SVM (0.8887). Adding elevation models and slope data to the classifiers resulted in only marginal performance improvements. This significant improvement highlights the potential of U-Net to accurately capture fine-scale snowline features, especially in heterogeneous and complex mountainous environments with spectrally similar classes such as firn and ice and paves the way for accurate, low-cost, automated and large-scale snow mapping on glaciers.

• Schellenberger, Thomas & Åkesson, Henning (2025). Ice-surface velocity in 1996 of Jostedalsbreen ice cap, Norway from European Remote Sensing satellite (ERS-1/-2) data. doi: 10.5281/zenodo.17541329. Fulltekst i vitenarkiv Vis sammendrag

  This dataset contains x- and y- components of the ice-surface velocity of Jostedalsbreen ice cap in winter 1996. The data covers the southern, northern and some of the central parts of the ice cap.

• Åkesson, Henning; Sjursen, Kamilla Hauknes; Schuler, Thomas; Dunse, Thorben; Andreassen, Liss Marie & Gillespie, Mette Kusk [Vis alle 9 forfattere av denne artikkelen] (2025). Recent history and future demise of Jostedalsbreen, the largest ice cap in mainland Europe. doi: 10.5194/egusphere-egu25-10323. Fulltekst i vitenarkiv Vis sammendrag

  Glaciers and ice caps worldwide are in strong decline, and models project this trend to continue with future warming, with strong natural and socioeconomic implications. The Jostedalsbreen ice cap is the largest ice cap on the European continent (500 km2 in 1966, 458 km2 in 2019) and occupies 20% of the total glacier area of mainland Norway. Here we simulate the evolution of Jostedalsbreen since 1960, and its fate in a changing climate in the 21st-century and beyond (2300). This ice cap contains glacier units with a great diversity in shape, steepness, hypsometry, and flow speed. We employ a coupled model system annually accounting for the mass-balance elevation feedback, with 3-d ice dynamics and simulated surface mass balance constrained by Bayesian inference. We find that Jostedalsbreen may lose 12-74% of its present-day volume, depending on future emissions. Regardless of scenario, the ice cap is likely to split into three parts during the second half of the 21st century. Our results suggest that Jostedalsbreen will likely be more resilient than many smaller glaciers and ice caps in Scandinavia. However, we show in simulations to the year 2300 that by the year 2100, the ice cap may be committed to a complete disappearance during the 22nd century, under high emissions. Under medium 21st-century emissions, the ice cap is bound to shrink by 90% until 2300. Further simulations indicate that substantial mass losses undergone until 2100 are irreversible. Our study illustrates a model approach for complex ice masses with numerous outlet glaciers such as ice caps, and how tightly linked future mass loss is to future greenhouse-gas emissions. Finally, uncertainties in future climate conditions appear to be the largest source of uncertainty in future evolution of ice caps like Jostedalsbreen.

• Schellenberger, Thomas; Maslov, Konstantin; Pandit, Prashant H; Persello, Claudio; Mattia, Callegari & Stein, Alfred (2025). Large scale satellite-based datasets for glacier modelling. Fulltekst i vitenarkiv
• Åkesson, Henning; Sjursen, Kamilla Hauknes; Schuler, Thomas Vikhamar; Dunse, Thorben; Andreassen, Liss Marie & Gillespie, Mette Kusk [Vis alle 9 forfattere av denne artikkelen] (2025). The future fate of the largest ice cap in mainland Europe. Fulltekst i vitenarkiv
• Maslov, Konstantin & Schellenberger, Thomas (2024). Globally Scalable Glacier Mapping by Deep Learning Matches Expert Delineation Accuracy. doi: 10.11582/2024.00168. Fulltekst i vitenarkiv Vis sammendrag

  Benchmark dataset for global glacier mapping using deep learning and satellite imagery. Details about the dataset and its use can be found here: Maslov, K.A., Persello, C., Schellenberger, T. And A. Stein: Globally scalable glacier mapping by deep learning matches expert delineation accuracy. Nat Commun 16, 43 (2025). https://doi.org/10.1038/s41467-024-54956-x The related code here: https://github.com/konstantin-a-maslov/scalable_glacier_mapping

• Åkesson, Henning; Sjursen, Kamilla Hauknes; Andreassen, Liss Marie; Schuler, Thomas Vikhamar; Dunse, Thorben & Gillespie, Mette Kusk [Vis alle 8 forfattere av denne artikkelen] (2024). History and future demise of Jostedalsbreen, the largest ice cap in mainland Europe. Fulltekst i vitenarkiv
• Maslov, Konstantin; Schellenberger, Thomas; Persello, Claudio & Stein, Alfred (2024). Glacier Mapping from Sentinel-1 SAR Time Series with Deep Learning in Svalbard. Fulltekst i vitenarkiv Vis sammendrag

  Glaciers are one of the essential climate variables. Tracking their areal changes over time is of high importance for monitoring the impacts of climate change and designing adaptation strategies. Mapping glaciers from optical remote sensing data might result in a very limited temporal resolution due to the absence of cloud-free imagery at the end of the ablation season. Synthetic aperture radar (SAR) solves this problem as it can operate in almost all weather conditions. Here, we present a deep learning strategy for glacier mapping based solely on Sentinel-1 SAR data in Svalbard. We test two options for integrating SAR image time series into deep learning models, namely, 3D convolutions and long short-term memory (LSTM) cells. Both proposed models achieve an intersection over union (IoU) of 0.964 on the test subset. Our results highlight the applicability of SAR data in glacier mapping with the potential to obtain glacier inventories with higher temporal resolution. We shared our dataset, code-base and pretrained models at https://github.com/konstantin-a-maslov/icemapper. See also: https://ieeexplore.ieee.org/document/10640676

• Maslov, Konstantin & Schellenberger, Thomas (2024). Machine Learning in Remote Sensing of Glaciers. Fulltekst i vitenarkiv
• Schellenberger, Thomas; Maslov, Konstantin & Persello, Claudio (2024). Towards global glacier mapping from satellite data by deep learning. Fulltekst i vitenarkiv Vis sammendrag

  Glaciers are under pressure in the current climate warming trend. The aerial extent and the mass balance, a measure of mass loss or mass gain of a glacier are representative of the “health state” of a glacier. They are recognized as Essential Climate Variables by the Global Climate Observing System (GCOS) with the glacier area being an important ECV product. Due to the sheer number of glaciers of roughly 250.000 worldwide and their inaccessibility, satellite data has proven as a valuable resource to map glaciers around the globe. Nevertheless, mapping large numbers of glaciers remains a highly time-consuming task and is prone to systematic biases and errors and only a limited number of ≈870,000 outlines exist in total today. This presentation will show results from the projects MASSIVE and MATS in which we built a benchmark dataset for global glacier mapping. It is based on open optical and synthetic aperture radar satellite imagery and digital elevation models covering 9% or 19,000 glaciers worldwide. We trained a newly developed hybrid convolutional neural network-vision transformer model, GlaViTU, using glacier outlines from the Global Land Ice Measurements from Space (GLIMS) database. As the dataset contains a large variety of glacier types, we tested several training strategies for global generalisation. The best performing model achieves a high accuracy in most regions, with excellent results for ‘clean ice’ glaciers. As a novelty, we also report calibrated confidence of our estimates, increasing their reliability and interpretability. Challenges remain for calving fronts and debris-covered glaciers. To address this, we are currently investigating the use of additional input datasets such as glacier velocity maps, elevation change maps and Synthetic Aperture Radar satellite image time series as well as other machine learning approaches such as LSTMs and 3D convolutional neural networks with promising results.

• Schellenberger, Thomas; Maslov, Konstantin & Persello, Claudio (2024). Towards global glacier mapping by deep learning and open earth observation data. Fulltekst i vitenarkiv Vis sammendrag

  Glaciers are under pressure in the current climate warming trend. The aerial extent and the mass balance, a measure of mass loss or mass gain of a glacier are representative of the “health state” of a glacier. They are recognized as Essential Climate Variables by the Global Climate Observing System (GCOS) with the glacier area being an important ECV product. Due to the sheer number of glaciers of roughly 250.000 worldwide and their inaccessibility, satellite data has proven as a valuable resource to map glaciers around the globe. Nevertheless, mapping large numbers of glaciers remains a highly time-consuming task and is prone to systematic biases and errors and only a limited number of ≈870,000 outlines exist in total today. This presentation will show results from the projects MASSIVE and MATS in which we built a benchmark dataset for global glacier mapping. It is based on open optical and synthetic aperture radar satellite imagery and digital elevation models covering 9% or 19,000 glaciers worldwide. We trained a newly developed hybrid convolutional neural network-vision transformer model, GlaViTU, using glacier outlines from the Global Land Ice Measurements from Space (GLIMS) database. As the dataset contains a large variety of glacier types, we tested several training strategies for global generalisation. The best performing model achieves a high accuracy in most regions, with excellent results for ‘clean ice’ glaciers. As a novelty, we also report calibrated confidence of our estimates, increasing their reliability and interpretability. Challenges remain for calving fronts and debris-covered glaciers. To address this, we are currently investigating the use of additional input datasets such as glacier velocity maps, elevation change maps and Synthetic Aperture Radar satellite image time series as well as other machine learning approaches such as LSTMs and 3D convolutional neural networks with promising results.

• Maslov, Konstantin & Schellenberger, Thomas (2024). Introduction to Machine Learning. Fulltekst i vitenarkiv
• Maslov, Konstantin; Persello, Claudio; Schellenberger, Thomas & Stein, Alfred (2023). GLAVITU: A Hybrid CNN-Transformer for Multi-Regional Glacier Mapping from Multi-Source Data. Fulltekst i vitenarkiv Vis sammendrag

  Glacier mapping is essential for studying and monitoring the impacts of climate change. However, several challenges such as debris-covered ice and highly variable landscapes across glacierized regions worldwide complicate large-scale glacier mapping in a fully-automated manner. This work presents a novel hybrid CNN-transformer model (GlaViTU) for multi-regional glacier mapping. Our model outperforms three baseline models—SETR-B/16, ResU-Net and TransU-Net—achieving a higher mean IoU of 0.875 and demonstrates better generalization ability. The proposed model is also parameter-efficient, with approximately 10 and 3 times fewer parameters than SETR-B/16 and ResU-Net, respectively. Our results provide a solid foundation for future studies on the application of deep learning methods for global glacier mapping. To facilitate reproducibility, we have shared our data set, codebase and pretrained models on GitHub at https://github.com/konstantin-a-maslov/GlaViTU-IGARSS2023.

• Maslov, Konstantin; Persello, Claudio; Schellenberger, Thomas & Stein, Alfred (2023). Towards the First Global Multitemporal Data Set of Glacier Outlines. Fulltekst i vitenarkiv
• Mattia, Callegari; Persello, Claudio & Schellenberger, Thomas (2023). Multi-source snow cover as proxy for glacier mass balance estimation. Fulltekst i vitenarkiv
• Racoviteanu, Adina E.; Miles, Evan; Davies, Bethan; Bolch, Tobias; Watson, C. Scott & Buri, Pascal [Vis alle 14 forfattere av denne artikkelen] (2023). Intercomparison of remotely sensed methods for delineation of high mountain debris-covered glaciers using machine learning. Fulltekst i vitenarkiv Vis sammendrag

  Glaciers are important contributors to water resources in many parts of the world, as well as constituting potential hazards. Monitoring of glaciers with remote sensing data is complicated by the presence of supraglacial debris which limits the effectiveness of semi-automated glacier delineation techniques. Furthermore, debris-covered glaciers are notoriously hard to survey in the field due to their complex, chaotic topography, and the presence of ephemeral ice cliffs and supraglacial lakes. As such, debris-covered glaciers are poorly represented in global inventories such as GLIMS and RGI, which consist of an assemblage of outlines from various dates. These inventories are subject to uncertainties due to mapping by multi-analysts using different methods. Despite recent efforts to map supraglacial debris cover at regional level, there remains an urgent need to develop a global, robust automated mapping approach based on open access data. Novel remote sensing data including high-resolution optical and radar data combined with emerging machine learning offer unique opportunities to advance current mapping methods. Here we evaluate current indices used in mapping supraglacial debris-cover and derive a best-practice workflow recommendation from particular combinations of these indices to derive glacier outlines for six representative subregions chosen globally: Khumbu and Manaslu regions (Nepal), Cordillera Blanca (Peru), Northern Patagonian Ice fields, Alaska Wrangell range, Hunza valley (Karakoram) and the Tien Shan. We present preliminary results of machine learning algorithms that combine the various remote sensing indices. The aim is to develop a robust, systematic method to map debris cover in a consistent manner at multi-temporal scales.

• Maslov, Konstantin; Schellenberger, Thomas; Persello, Claudio & Stein, Alfred (2022). Towards Fully-Automated Glacier Mapping in Arctic Regions using Deep Learning. Fulltekst i vitenarkiv
• Maslov, Konstantin & Schellenberger, Thomas (2022). Deep Learning for Glacier Mapping. Fulltekst i vitenarkiv
• Schellenberger, Thomas & Maslov, Konstantin (2022). Remote Sensing in Glaciology – the traditional basics. Fulltekst i vitenarkiv
• Pandit, Prashant H; Persello, Claudio; Aghababaei, Hossein; Schellenberger, Thomas & Maslov, Konstantin (2022). How precisely can we map the glaciers? Fulltekst i vitenarkiv
• Altena, Bas; Schellenberger, Thomas & Kääb, Andreas (2019). Bringing together remote sensing data to produce region-wide glacier velocities. Fulltekst i vitenarkiv Vis sammendrag

  The recent increase of medium resolution satellite programs that consistently senses the Earth surface has brought great potential for glaciology. Satellites like the Sentinel-1&2 tandems, Landsat 7&8 duo and Rapideye constellation sense mountain ranges almost without missing a day. Consequently, high resolution time-series can be constructed, for snow-cover or crack propagation, or other single scene information. Properties derived from combinations of acquisitions, such as, glacier velocity is more challenging. Velocity estimation is based upon image matching, which can be hindered by coherence loss or cloud cover. Consequently, generated velocity data have gaps and have scattered coverage at different time intervals. This steady data stream has resulted in several initiatives to use image matching techniques in batch processing mode. Resulting in glacier velocity services being set-up (e.g.: cryoportal, GoLIVE, Geodetic data portal), but these initiatives are satellite specific. If such products are merged it is possible to increase temporal and spatial coverage and increase the reliability of the velocity estimates. We developed such a workflow that merges glacier velocity products of different time spans and type. This combined dataset covers several mountain ranges and has a regular time step at a monthly pace. This consistent time-series make it possible to observe glacier dynamics that are of short nature. As the satellite recordings already span several years, the dynamics of tidewater glaciers or (partial) surges can be observed and analyzed. This study therefor shows the great potential of bringing together different satellite systems.

• Nuth, Christopher; Kohler, Jack; Köhler, Andreas; Lefeuvre, Pierre-Marie; Schuler, Thomas & Pelt, Ward Van [Vis alle 13 forfattere av denne artikkelen] (2018). The different dynamic personalities of two neighboring tidewater glaciers. Fulltekst i vitenarkiv
• Schellenberger, Thomas; Wychen, Wesley Van; Copland, Luke; Kääb, Andreas & Gray, Laurence (2017). Glacier surface velocities of the Svalbard Archipelago. Fulltekst i vitenarkiv
• Schellenberger, Thomas; Dunse, Thorben; Kääb, Andreas; Schuler, Thomas; Hagen, Jon Ove Methlie & Reijmer, Carleen (2017). Multi-year surface velocities and sea-level rise contribution of the Basin-3 and Basin-2 surges, Austfonna, Svalbard. Fulltekst i vitenarkiv
• Gong, Yongmei; Zwinger, Thomas; Åström, Jan A.; Altena, Bas; Schellenberger, Thomas & Gladstone, Rupert M [Vis alle 7 forfattere av denne artikkelen] (2017). Basal friction evolution and crevasse distribution during the surge in Basin-3, Austfonna ice-cap - offline coupling between a continuum ice dynamic model and a discrete element model. Fulltekst i vitenarkiv
• Kääb, Andreas; Treichler, Désirée Silvana; Girod, Luc Maurice Ramuntcho; Lefeuvre, Pierre-Marie; Schellenberger, Thomas & Altena, Bas [Vis alle 10 forfattere av denne artikkelen] (2016). On the Karakoram anomaly. Fulltekst i vitenarkiv
• Solgaard, Anne; Messerli, Alexandra; Grinsted, Aslak; Schellenberger, Thomas; Hvidberg, Christine & Karlson, Nanna [Vis alle 9 forfattere av denne artikkelen] (2016). Engabreen Basal Conditions from Surface Measurements. Fulltekst i vitenarkiv
• Schellenberger, Thomas; Dunse, Thorben; Kääb, Andreas; Schuler, Thomas; Hagen, Jon Ove Methlie & Reijmer, Carleen (2016). The surge must go on: extended surface velocity time series and frontal ablation estimates of Basin-3 (and Basin-2), NE-Svalbard. Fulltekst i vitenarkiv
• Altena, Bas; Dunse, Thorben & Schellenberger, Thomas (2016). Crevasse evolution of Basin-3 during its speed-up. Fulltekst i vitenarkiv
• Winsvold, Solveig Havstad; Kääb, Andreas; Nuth, Christopher; Pelt, Ward J.J. Van; Andreassen, Liss Marie & Schellenberger, Thomas (2016). Glacier mapping using combined optical and SAR time-series. Fulltekst i vitenarkiv
• Winsvold, Solveig Havstad; Kääb, Andreas; Nuth, Christopher; Andreassen, Liss Marie; Pelt, Ward Van & Schellenberger, Thomas (2016). Regional glacier mapping by combination of dense optical and SAR satellite image time-series. Fulltekst i vitenarkiv
• Dunse, Thorben; Schellenberger, Thomas; Hagen, Jon Ove Methlie; Kääb, Andreas; Schuler, Thomas & Reijmer, Carleen (2015). A surge of the Austfonna ice cap, Svalbard, promoted by a hydro-thermodynamic feedback to summer melt. Fulltekst i vitenarkiv
• Gong, Yongmei; Schellenberger, Thomas; Zwinger, Thomas & Moore, John C. (2015). Observation and model study of terminus initiated surging in southern Basin 3, Austfonna ice cap, Svalbard. Fulltekst i vitenarkiv
• Schellenberger, Thomas; Dunse, Thorben; Kääb, Andreas & Kohler, Jack (2014). Velocity variations and calving flux of Kronebreen, NW-Svalbard from SAR offset tracking. Fulltekst i vitenarkiv
• Girod, Luc Maurice Ramuntcho; Nuth, Christopher & Schellenberger, Thomas (2014). Structure from motion, a low cost, very high resolution method for surveying glaciers using GoPros and opportunistic helicopter flights. Fulltekst i vitenarkiv
• Wychen, Wesley Van; Schellenberger, Thomas; Copland, Luke; Kääb, Andreas & Gray, Laurence (2013). An inter-comparison of techniques for determining glacier velocities over Svalbard. Fulltekst i vitenarkiv
• Dunse, Thorben; Schellenberger, Thomas; Hagen, Jon Ove Methlie; Kääb, Andreas; Schuler, Thomas & Reijmer, Carleen (2013). On a major surge of Austfonna. Fulltekst i vitenarkiv
• Kääb, Andreas; Schellenberger, Thomas & Dunse, Thorben (2013). A large glacier surge on Austfonna, Svalbard, from TerraSAR-X offset tracking. Fulltekst i vitenarkiv
• Nuth, Christopher; Kohler, Jack; Schellenberger, Thomas; Schuler, Thomas; Kääb, Andreas & Hagen, Jon Ove Methlie (2013). Combining remote sensing and surface mass balance observations/modeling to partition mass changes of Kronebreen, northwest Svalbard. Fulltekst i vitenarkiv
• Schellenberger, Thomas; Heid, Torborg; Dunse, Thorben; Nuth, Christopher; Reijmer, Carleen & Kohler, Jack [Vis alle 8 forfattere av denne artikkelen] (2012). Calving flux of Svalbards tidewater glaciers estimated from remote sensing-based surface velocities. Fulltekst i vitenarkiv
• Schellenberger, Thomas; Heid, Torborg; Dunse, Thorben; Nuth, Christopher; Kohler, Jack & Hagen, Jon Ove Methlie [Vis alle 7 forfattere av denne artikkelen] (2012). Calving flux of Svalbards tidewater glaciers estimated from remote sensing-based surface velocities. Fulltekst i vitenarkiv

Se alle arbeider i NVA