Search for: All records

Abstract One of the most conspicuous signals of climate change in high‐latitude tundra is the expansion of ice wedge thermokarst pools. These small but abundant water features form rapidly in depressions caused by the melting of ice wedges (i.e., meter‐scale bodies of ice embedded within the top of the permafrost). Pool expansion impacts subsequent thaw rates through a series of complex positive and negative feedbacks which play out over timescales of decades and may accelerate carbon release from the underlying sediments. Although many local observations of ice wedge thermokarst pool expansion have been documented, analyses at continental to pan‐Arctic scales have been rare, hindering efforts to project how strongly this process may impact the global carbon cycle. Here we present one of the most geographically extensive and temporally dense records yet compiled of recent pool expansion, in which changes to pool area from 2008 to 2020 were quantified through satellite‐image analysis at 27 survey areas (measuring 10–35 km²each, or 400 km²in total) dispersed throughout the circumpolar tundra. The results revealed instances of rapid expansion at 44% (15%) of survey areas. Considered alone, the extent of departures from historical mean air temperatures did not account for between site variation in rates of change to pool area. Pool growth was most clearly associated with upland (i.e., hilly) terrain and elevated silt content at soil depths greater than one meter. These findings suggest that, at short time scales, pedologic and geomorphologic conditions may exert greater control on pool dynamics in the warming Arctic than spatial variability in the rate of air temperature increases. 

Infrastructure across the circumpolar Arctic is exposed to permafrost thaw hazards caused by global warming and human activity, creating the risk of damage and economic losses. However, losses are underestimated in existing literature due to incomprehensive infrastructure maps. Here, we mapped infrastructure from 0.5 m resolution satellite imagery of 285 Alaskan communities with a deep learning detection model. Combined with OpenStreetMap, we mapped a statewide Alaskan building footprint of 53 M m2 and a road network of 50,477 km. With deep learning, we expanded the OpenStreetMap building footprint by 47% statewide and 86% on discontinuous and continuous permafrost. Doubling the amount found in existing literature by using our improved map, we estimated that building and road losses due to permafrost thaw could cost Alaska $37B to $51B under the SSP245 and SSP585 scenarios, respectively. Finally, we highlight shortcomings in U.S. national risk assessments, which do not account for Alaskan permafrost hazards. 

Abstract The presence of ground ice in Arctic soils exerts a major effect on permafrost hydrology and ecology, and factors prominently into geomorphic landform development. As most ground ice has accumulated in near-surface permafrost, it is sensitive to variations in atmospheric conditions. Typical and regionally widespread permafrost landforms such as pingos, ice-wedge polygons, and rock glaciers are closely tied to ground ice. However, under ongoing climate change, suitable environmental spaces for preserving landforms associated with ice-rich permafrost may be rapidly disappearing. We deploy a statistical ensemble approach to model, for the first time, the current and potential future environmental conditions of three typical permafrost landforms, pingos, ice-wedge polygons and rock glaciers across the Northern Hemisphere. We show that by midcentury, the landforms are projected to lose more than one-fifth of their suitable environments under a moderate climate scenario (RCP4.5) and on average around one-third under a very high baseline emission scenario (RCP8.5), even when projected new suitable areas for occurrence are considered. By 2061–2080, on average more than 50% of the recent suitable conditions can be lost (RCP8.5). In the case of pingos and ice-wedge polygons, geographical changes are mainly attributed to alterations in thawing-season precipitation and air temperatures. Rock glaciers show air temperature-induced regional changes in suitable conditions strongly constrained by topography and soil properties. The predicted losses could have important implications for Arctic hydrology, geo- and biodiversity, and to the global climate system through changes in biogeochemical cycles governed by the geomorphology of permafrost landscapes. Moreover, our projections provide insights into the circumpolar distribution of various ground ice types and help inventory permafrost landforms in unmapped regions. 

This dataset, called DARTS, comprises footprints of retrogressive thaw slump (RTS) identified and quantified using an automated deep learning approach in RTS hotspots across the Arctic and Subarctic permafrost regions. We utilized multispectral PlanetScope imagery with a spatial resolution of ~3 meters (m), complemented by ArcticDEM (Digital Elevation Model) and derived datasets, including slope, relative elevation, and Landsat-derived change trends. The dataset covers an area of 1.6 million square-kilometers (km²), with at least one coverage between 2021 and 2023, and provides annual coverage for approximately 900,000 km². In several highly active key sites, such as Banks Island, Peel Plateau, and Novaya Zemlya, we extended the data frequency and temporal coverage to 2018-2023. We mapped a total of more than 43,000 individual RTS and ALD, many of them multiple times. We offer two levels of datasets; Level 1: RTS footprints per image with timestamps; and Level 2: annually aggregated RTS footprints. Essential metadata includes image footprints, dataset coverage, timestamps, and model-specific information. To enhance reproducibility and further use, the training labels, processing code, and model checkpoints are publicly available. This version, v1.1, is the revised first openly accessible release. The dataset will be maintained and continuously updated in both spatial and temporal extent. It can be used for mapping and quantifying RTS, analyzing spatio-temporal patterns of RTS dynamics, or serving as input for landscape dynamics models. 

Arctic permafrost is facing significant changes due to global climate change. As these regions are largely inaccessible, remote sensing plays a crucial rule in better understanding the underlying processes across the Arctic. In this study, we focus on the remote detection of retrogressive thaw slumps (RTSs), a permafrost disturbance comparable to slow landslides. For such remote sensing tasks, deep learning has become an indispensable tool, but limited labeled training data remains a challenge for training accurate models. We present PixelDINO, a semi-supervised learning approach, to improve model generalization across the Arctic with a limited number of labels. PixelDINO leverages unlabeled data by training the model to define its own segmentation categories (pseudoclasses), promoting consistent structural learning across strong data augmentations. This allows the model to extract structural information from unlabeled data, supplementing the learning from labeled data. PixelDINO surpasses both supervised baselines and existing semi-supervised methods, achieving average intersection-over-union (IoU) of 30.2 and 39.5 on the two evaluation sets, representing significant improvements of 13% and 21%, respectively, over the strongest existing models. This highlights the potential for training robust models that generalize well to regions that were not included in the training data. 

Much of the Arctic tundra is underlain by a network of ice wedges that formed during millennia of repeated frost cracking on cold winter days and later infilling of snowmelt water. Growing ice wedges push the soil upwards, forming connected ridges on the ground surface and the ubiquitous ice-wedge polygon tundra. Melting of the top of the ice wedge causes the ground surface to collapse with the rims transforming into snow- and water-collecting troughs — a phenomenon observed at multiple sites across the Arctic tundra in a decade or less. Continued melt establishes a new drainage network only a metre or two wide and less than a half-metre deep, where a doubling of runoff and reduced surface water storage is possible without changes in precipitation. Across the Arctic, lakes are disappearing, while precipitation and river runoff are increasing. So far, the sub-metre microtopographical changes have not entered the scientific analyses encompassing regional and pan-Arctic hydrology. The data and technology are now here to quantify the network of ice wedges across large regions and, though individually small, the ice wedges add up to large numbers. What at first may appear as contradicting hydrological change (for example, shrinking lakes despite increasing precipitation) could be explained by a sudden evolution of the stream network where the new channels are narrow but bountiful: the capillaries of the Arctic tundra hydrological system. 

Topographical changes are of fundamental interest to a wide range of Arctic science disciplines faced with the need to anticipate, monitor, and respond to the effects of climate change, including geohazard management, glaciology, hydrology, permafrost, and ecology. This study demonstrates several geomorphological, cryo- spheric, and biophysical applications of ArcticDEM – a large collection of publicly available, time-dependent digital elevation models (DEMs) of the Arctic. Our study illustrates ArcticDEM’s applicability across different disciplines and five orders of magnitude of elevation derivatives, including measuring volcanic lava flows, ice cauldrons, post-failure landslides, retrogressive thaw slumps, snowdrifts, and tundra vegetation heights. We quantified surface elevation changes in different geological settings and conditions using the time series of ArcticDEM. Following the 2014–2015 B´arðarbunga eruption in Iceland, ArcticDEM analysis mapped the lava flow field, and revealed the post-eruptive ice flows and ice cauldron dynamics. The total dense-rock equivalent (DRE) volume of lava flows is estimated to be (1431 ± 2) million m3. Then, we present the aftermath of a landslide in Kinnikinnick, Alaska, yielding a total landslide volume of (400 ± 8) × 103 m3 and a total area of 0.025 km2. ArcticDEM is further proven useful for studying retrogressive thaw slumps (RTS). The ArcticDEM-mapped RTS profile is validated by ICESat-2 and drone photogrammetry resulting in a standard deviation of 0.5 m. Volume estimates for lake-side and hillslope RTSs range between 40,000 ± 9000 m3 and 1,160,000 ± 85,000 m3, highlighting applicability across a range of RTS magnitudes. A case study for mapping tundra snow demonstrates ArcticDEM’s potential for identifying high-accumulation, late-lying snow areas. The approach proves effective in quantifying relative snow accumulation rather than absolute values (standard deviation of 0.25 m, bias of 0.41 m, and a correlation coefficient of 0.69 with snow depth estimated by unmanned aerial systems photogrammetry). Furthermore, ArcticDEM data show its feasibility for estimating tundra vegetation heights with a standard deviation of 0.3 m (no bias) and a correlation up to 0.8 compared to the light detection and ranging (LiDAR). The demonstrated capabilities of ArcticDEM will pave the way for the broad and pan-Arctic use of this new data source for many disciplines, especially when combined with other imagery products. The wide range of signals embedded in ArcticDEM underscores the potential challenges in deciphering signals in regions affected by various geological processes and environmental influences. 

This dataset, called DARTS, comprises footprints of retrogressive thaw slump (RTS) and active layer detachments slides (ALD) identified and quantified using an automated deep learning approach in RTS hotspots across the Arctic and Subarctic permafrost regions. We utilized multispectral PlanetScope imagery with a spatial resolution of 3 meters (m), complemented by ArcticDEM (Digital Elevation Models) and derived datasets, including slope, relative elevation, and Landsat-derived change trends. The dataset covers an area of 1.6 million square-kilometers (km²), with at least one coverage between 2021 and 2023, and provides annual coverage for approximately 900,000 km². In several highly active key sites, such as Banks Island, Peel Plateau, and Novaya Zemlya, we extended the data frequency and temporal coverage to 2018-2023. We mapped a total of more than 43,000 individual RTS and ALD, many of them multiple times. We offer two levels of datasets; Level 1: RTS footprints per image with timestamps; and Level 2: annually aggregated RTS footprints. Essential metadata includes image footprints, dataset coverage, timestamps, and model-specific information. To enhance reproducibility and further use, the training labels, processing code, and model checkpoints are publicly available. This version, v1, is the first openly accessible release. The dataset will be maintained and continuously updated in both spatial and temporal extent. It can be used for mapping and quantifying RTS, analyzing spatio-temporal patterns of RTS dynamics, or serving as input for landscape dynamics models. 

### Overview The SACHI (Sentinel-1/2 derived Arctic Coastal Human Impact) dataset has been developed as part of the HORIZON2020 project Nunataryuk by b.geos (www.bgeos.com). V1 covered a 100km buffer from the Arctic Coast (land area), for areas with permafrost near the coast. V2 covers additional selected areas extending the coverage to the south. It is based on Sentinel-1 and Sentinel-2 data from 2016-2020 using the algorithms described in Bartsch et al. (2020). It is a supplement to Bartsch et al. (2023). This dataset contains detected coastal infrastructure separated into seven different categories: linear transport infrastructure (asphalt), linear transport infrastructure (gravel), linear transport infrastructure (undefined), buildings (and other constructions such as bridges), other impacted area (includes gravel pads, mining sites), airstrip, and reservoir or other water body impacted by human activities. This SACHI version 2 dataset was post-processed by the Permafrost Discovery Gateway visualization pipeline. This workflow cleaned, standardized, and visualized the data as two Tile Matrix Sets per year. One Tile Matrix Set is the data in the form of GeoPackages, or staged tiles, and the other Tile Matrix Set is the staged tiles in the form of GeoTIFF tiles. The highest resolution tiles were resampled to produce GeoTIFFs for lower resolutions. This data is visualized on the Permafrost Discovery Gateway portal: https://arcticdata.io/catalog/portals/permafrost/Imagery-Viewer ### References Bartsch, A., Widhalm, B., von Baeckmann, C., Efimova, A., Tanguy, R., and Pointner, G. (2023). Sentinel-1/2 derived Arctic Coastal Human Impact dataset (SACHI) (v2.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.10160636 Bartsch, A., G. Pointner, I. Nitze, A. Efimova, D. Jakober, S. Ley, E. Högström, G. Grosse, P. Schweitzer (2021): Expanding infrastructure and growing anthropogenic impacts along Arctic coasts. Environmental Research Letters. https://doi.org/10.1088/1748-9326/ac317 Bartsch, A., Pointner, G., Ingeman-Nielsen, T. and Lu, W. (2020), ‘Towards circumpolar mapping of Arctic settlements and infrastructure based on Sentinel-1 and Sentinel-2’, Remote Sensing 12(15), 2368. ### Access Data files output from the visualization workflow are available for download at: [http://arcticdata.io/data/10.18739/A21J97929](http://arcticdata.io/data/10.18739/A21J97929) To download all files in the command line, run the following command in a terminal: `wget -r -np -nH --cut-dirs=3 -R '\?C=' -R robots.txt https://arcticdata.io/data/10.18739/A21J97929/` To download a subdirectory of the archived files, add the subdirectories to the end of the URL above. 

Climate change pressure on the Arctic permafrost is rising alarmingly, creating a decisive need to produce Pan-Arctic scale permafrost landform and thaw disturbance information from remote sensing (RS) data. Very high spatial resolution (VHSR) satellite images can be utilized to detect ice-wedge polygons (IWPs) – the most important and widespread landform in the Arctic tundra region - across the Arctic without compromising spatial details. Automated analysis of peta-byte scale VHSR imagery covering millions of square kilometers is a computationally challenging task. Traditional semantic segmentation requires the use of task specific feature extraction with conventional classification techniques. Semantic complexity of VHSR images coupled with landscape heterogeneity makes it difficult to use conventional classification approaches to produce Pan-Arctic scale geospatial products. This leads to adapting deep convolutional neural network (DLCNN) approaches that have excelled in computer vision (CV) applications. Transitioning domains from everyday image understanding to remote sensing image analysis is challenging. This study aims to systematically investigate two main obstacles confronted when adapting DLCNNs in large-scale RS image analysis tasks; 1) the limited availability labeled data sets and 2) the prohibitive nature of hyperparameter tunning when designing DLCNNs that can capture the rich characteristics embedded in remotely-sensed images. With a case study on the production of the first pan-Arctic ice-wedge polygon map using thousands of VHSR images, we demonstrate the use of transfer learning and the impact of hyperparameter tuning with a 16% improvement of the Mean Average Precision (mAP50).