Search

Retrogressive thaw slumps (RTS) are typical landscape processes of thawing and degrading permafrost. To this point, their distribution and dynamics are almost completely undocumented across many regions in the permafrost domain, partially due to the lack of data and monitoring techniques in the past. We are tackling this shortcoming by creating a deep learning based semantic segmentation framework to detect RTS, using multi-spectral PlanetScope, derived topographic (ArcticDEM) and multi-temporal Landsat Trend data. We created a highly automated processing pipeline, which is designed to create reproducible results and to be flexible for multiple input features. The processing workflow is based on the pytorch deep-learning framework and includes a variety of different segmentation architectures (UNet, UNet++, DeepLabV3), backbones and includes common data transformation techniques such as augmentation or normalization. We tested (training, validation) our DL based model in six different regions of 100 to 300 km² size across Canada (Banks Island, Tuktoyaktuk, Horton, Herschel Is.), and Siberia (Kolguev, Lena). We performed a regional cross-validation (5 regions training, 1 region validation) to test the spatial robustness and transferability of the algorithm. Furthermore, we tested different architectures backbones and loss-function to identify the best performing and most robust parameter sets. For training the models we created a training database of manually digitized and validated RTS polygons. The resulting model performance varied strongly between different regions with maximum Intersection over Union (IoU) scores between 0.15 and 0.58. The strong regional variation emphasizes the need for sufficiently large training data, which is representative for the massive variety of RTS. However, the creation of good training data proved to be challenging due to the fuzzy definition and delineation of RTS, particularly on the lower part. We have recently expanded our analysis to several RTS-rich regions across the Arctic (Fig.X) for the year 2021 and annual analysis (2018-2021) for RTS hot-spots, e.g. Banks Island, Peel Plateau and others. First model inference runs are promising for detecting RTS, but are still strongly overestimating the number and area of RTS, due to an excessive number of false positives. Model performance however, varies strongly between regions. Due to the strong variability of landscapes with RTS, we expect an improvement in model performance with an increase in the number and spatial distribution of training datasets. The community driven formation of the IPA Action Group RTSIn, which aims to create standardized RTS digitization protocols and training datasets for deep/machine-learning purposes will be a great boost for our purpose. With our standardized processing pipeline (preprocessing, training, inference), which allows to add more features based on user interest and data availability,, we tested our workflow for surface water and pingos with a mixture of publically available (Jones et al) and digitized data (Grosse pingos, Nitze water). These tests produced very good results and showed that the designed workflow is transferrable beyond the segmentation of RTS only. In the near future, we are aiming to integrate the community based training data and further gradually improve our training database. Within the framework of the ML4Earth project, we will create a temporal and pan-arctic monitoring system for RTS based on our highly automated processing chain. This will enable us to better understand pan-arctic RTS dynamics, their influencing factors, and consequences. Combining these spatial-temporal datasets with volumetric change information and carbon stock information will enable us to better quantify the consequences of thaw slumping across the permafrost domain.

While temperatures are increasing on the global scale, the Arctic regions are especially vulnerable to this changing climate and landscapes underlain by permafrost experience increased thaw and degradation. The enhanced warming of organic-rich frozen ground can have severe consequences on infrastructure and ecosystems and is projected to become a highly relevant driver of greenhouse gas fluxes into the atmosphere. Degrading permafrost landscapes occur extensively in vast areas of the North American Arctic, directly affecting communities and ecosystems. To identify and quantify these widespread degradation phenomena over vast areas, we require highest-resolution Earth observation dataset that we collect during aerial imaging campaigns. We here report on observations and first results from three airborne campaigns in 2018, 2019 and 2021. We performed large-scale monitoring of permafrost-affected areas in northern Canada and Alaska, focusing on sites that experienced disturbances in the past or recently. This included sites with vulnerable settlements, coastal erosion, thaw slumping, lake expansion and drainage, ice-wedge degradation and thaw subsidence, fire scars, pingos, methane seeps, and sites affected by beaver activities. All surveys were flown with the Alfred Wegener Institute's Polar-5 and -6 scientific airplanes at 500-1500 m altitude above terrain. The onboard sensor, the Modular Aerial Camera System (MACS), a very-high-resolution multispectral camera developed by the German Aerospace Center, operated in the visible (RGB) and near-infrared (NIR) domain. From the comprehensive collection of multiple TB of gathered data, super-high-resolution (up to 7 cm/px) RGB+NIR image mosaics and stereophotogrammetric digital surface models were derived. By presenting the data and first analyses, we would like to invite the community to discuss best use for maximized benefit of the data, in order to substantially contribute to our understanding of permafrost thaw dynamics.

While temperatures are rising globally, they are rising more than twice as fast in the Arctic. Landscapes underlain by permafrost are especially vulnerable to this changing climate and experience increased thaw and degradation. The proceeding warming of organic-rich frozen ground is a highly relevant driver of carbon release into the atmosphere. Retrogressive Thaw Slumps (RTSs) are dynamic thermokarst features which develop when ice-rich permafrost thaws and thus are important indices when it comes to the assessment of potential carbon sources in permafrost landscapes. Thousands of RTSs have been inventoried in northwestern Canada. These inventories showed that thaw slumping modifies terrain morphology and alters the discharge into aquatic systems resulting amongst others in infrastructure instabilities and ecosystem changes. Furthermore, recent studies project that abrupt thermokarst processes contribute significant amounts of greenhouse gas emissions. As observed in most arctic regions, RTS activity has increased in the Russian High Arctic, however, little research has been done on RTSs in this region. The objective of this study is to better understand growth pattern and development rates of RTSs in northern Russia during the last decade. The study area consists of five different sites in the Russian High Arctic covering an area of more than 600 km². The sites are located on the Novaya Zemlya Archipelago, Kolguev Island, Bol’shoy Lyakhovsky Island and Taymyr Peninsula in ice-rich permafrost characterized by either buried glacial ice deposits or syngenetically formed Yedoma permafrost. To assess changes in number and extent, a GIS based inventory of manually mapped RTSs was created. The inventory is based on multispectral imagery of high-resolution satellite sensors, including PlanetScope, RapidEye, Pléiades and SPOT. Cloud free images were acquired between 2011 and 2020 and exist for each or every few years depending on their availability. Additional data sets such as ArcticDEM, Esri Satellite base map and Tasseled Cap Landsat Trends were used to support the mapping process. From the extracted individual RTS objects, changes in number and surface area were calculated. Furthermore, for coastal slumps thermal denudation and thermal abrasion rates were computed. The results show that RTS activity was high at the study sites during the investigation period and that the diverse sites revealed different RTS characteristics, with non-coastal RTSs showing a much larger increase in area. At the non-coastal sites, RTS-affected area increased by a factor of 2 (100 %) in West Taymyr, a factor of 4 (400 %) in Novaya Zemlya, and a factor of 33 (3300 %) in East Taymyr, with particularly large increases in more recent years. At the coastal sites, total RTS area increased by a factor of 1.2 (20%) in North Kolguev, remained the same in South Kolguev, and decreased slightly by a factor of 0.95 (5%) in Bol’shoy Lyakhovsky. Headwall and base of the coastal slumps retreated at different rates. However, at all coastal sites, erosion of the headwall and base progressed, demonstrating that RTS activity cannot be determined by area changes alone because coastal RTSs are strongly influenced by thermal abrasion and thermal denudation which diminishes areal changes. Moreover, the number of RTS did not necessarily increase with increasing RTS activity. At all study sites except East Taymyr, increased RTS activity resulted from RTS growth rather than new RTS initiation. In addition, climate analysis revealed that the mean temperature increased significantly, within the last decade at all sites, potentially favouring RTS initiation and growth. The findings of this study contribute substantially to our understanding of regional permafrost thaw in the Russian High Arctic. Nevertheless, further research is needed to quantify volumetric permafrost loss and associated carbon release comprehensively throughout the Russian High Arctic to better understand RTS dynamics and their impact on greenhouse gas release.

Artificial Intelligence for Cold Regions (AI-CORE) is a collaborative project of the DLR, the AWI, the TU Dresden, and is funded by the Helmholtz Foundation since early 2020. The project aims at developing AI methods for addressing some of the most challenging research questions in cryosphere remote sensing, rapidly changing ice sheets and thawing permafrost. We apply data analytics approaches to discover the data variable from data set simulated with an ice sheet model, observe the migration, and time Series analysis to predict and contrast this to simulated grounding line position. For the data assimilation in simulations of the Greenland ice sheet, we engage a level set method, that allows to derive a continuous function in time and space from discrete information at satellite acquisition time steps. We use an alpha-shape method to derive a seamless product of the margin at each time step to be used in the level set method driving the simulations. We develop AI algorithms and tools that allow scaling of our analyses to very large regions. Here we focus on the detection of Retrogressive Thaw Slumps (RTS), highly dynamic erosion processes caused by rapid permafrost thaw. We apply deep-learning based object detection on dense time-series of high-resolution (3m) multi-spectral PlanetScope satellite images and auxiliary datasets such as digital elevation models. RTS detection is challenging, as they are difficult to define semantically and spatially and are highly dynamic and embedded in different landscape settings. The results will help to understand, quantify and predict RTS dynamics and their landscape-scale impacts in a rapidly warming Arctic. We upgrade the base IT-infrastructure at AWI by integrating new GPU computing hardware into the on-premise IT-infrastructure to speed up the computing, data storage capabilities, and parallel processing, supporting the analytical workflows specifically.