ALBERT

Description: “Artificial Intelligence for Cold Regions” (AI-CORE) is a collaborative project of the German Aerospace Center (DLR), the Alfred Wegener Institute (AWI), the Technical University Dresden (TU Dresden), and is funded by the Helmholtz Foundation since early 2020. The project aims at developing artificial intelligence methods for addressing some of the most challenging research questions in remote sensing of the cryosphere. Rapidly changing ice sheets and thawing permafrost are big societal challenges, hence quantifying these changes and understanding the mechanisms are of major importance. Given the vast extent of polar regions and the availability of exponentially increasing satellite remote sensing data, intelligent data analysis is urgently required to exploit the full information in satellite time series. This is where AI-CORE comes into play: Four geoscientific use cases have been defined, including a) change pattern identification of outlet glaciers in Greenland; b) object identification in permafrost areas; c) edge detection of calving fronts of glaciers/ice shelves in Antarctica; d) firn line detection and monitoring: The glacier mass balance indicator. For these four use cases, AI-methods are being developed to allow for an accurate, efficient, and automated extraction of the desired parameters. Once these methods have been successfully developed, they will be implemented in processing infrastructures at AWI, TU Dresden, and DLR, and subsequently made available to other research institutes. The presentation will outline the specific goals and challenges of the four use cases as well as the current state of the developments and preliminary results.

Description: Iron (Fe) plays a key role in mediating organic carbon (OC) decomposition rates in permafrost soils. Fe-bearing minerals stabilize OC through complexation, co-precipitation or aggregation processes and thus hinder degradation of OC. In addition, Fe(III) reduction can inhibit methanogenesis and decrease warming potential of greenhouse gases release. Ice-rich permafrost is subject to abrupt thaw and thermokarst formation, which unlocks OC and minerals from deep deposits and exposes OC to mineralization. These ice-rich domains include Yedoma sediments that have never thawed since deposition and Alas sediments that have undergone previous thermokarst processes during the Lateglacial and Holocene warming periods. The post-depositional history of these sediments may affect the distribution and reactivity of Fe-bearing minerals and the role Fe plays in mediating present day OC mineralization. Here we quantify Fe concentrations, Fe spatial and depth distribution, and Fe mineralogy in unthawed Yedoma and previously thawed Alas deposits from the Yedoma domain (West Siberia, Laptev Sea region, Kolyma region, New Siberian Islands and Alaska). Total Fe concentrations of ice-rich Yedoma deposits and previously thawed Alas deposits were determined using a portable X-ray fluorescence (XRF) device. This non-destructive method allowed a total iron concentration assessment of Yedoma domain deposits based on 1292 sediment samples. Portable XRF-measured concentrations trueness were calibrated from alkaline fusion and inductively coupled plasma optical emission spectrometry (ICP-OES) measurement method on a subset of 144 samples (R² = 0.81). Fe extractions of unthawed and previously thawed deposits display that, on average, 25% of the total iron is considered as reactive species, either as crystalline or amorphous oxides, or complexed with OC, with no significant difference between Yedoma and Alas deposits. We observe a constant total Fe concentration in Yedoma deposits, but a depletion or accumulation of total Fe in Alas deposits, which experienced previous thaw and/or flooding events, suggesting that redox driven processes during the Lateglacial and Holocene thermokarst formation impact the present day distribution of reactive Fe and its association with organic carbon in ice-rich permafrost.

Description: Thematic Open Access data portals foster and support an open data culture in order to reduce knowledge gaps and data uncertainty. Here we present the Arctic Permafrost Geospatial Center (APGC), which provides open access, high quality spatial data in the field of permafrost research. The distribution and easy access of a wide range of permafrost-related data products supports multi-scale and interdisciplinary analysis of combined field, remote sensing and modelling data. The APGC mission is to provide data that is of high usability, significance and impact, and to facilitate data discovery, data view and supports metadata documentation and exchange via the APGC data catalogue at https://apgc.awi.de/. The catalogue structure can host data models of varying themes, formats, and spatial and temporal extents. Data can be searched by interactively selecting locations on a base map and by many predefined metadata filters. Data can be downloaded directly through a link to the publishing data repository such as PANGAEA. The Catalogue is based on the open source CKAN catalogue architecture, which allows on-the-fly access to catalogued data in QGIS. The APGC currently features over 200 selected datasets from projects such as ERC PETA-CARB, ESA GlobPermafrost, and others. Data products provide information about surface and subsurface permafrost characteristics in the Arctic, Antarctica, or mountain permafrost areas, e.g., soil temperatures, soil carbon, ground ice, land cover, vegetation, periglacial landforms, subsidence and more. Collections of datasets allow users to easily get an overview of the spatial distributions of datasets or their availability in different formats. An additional WebGIS application allows users to explore most of the data interactively (https://maps.awi.de). Data submissions are evaluated according to the following access criteria: permafrost focus, broader significance and impact, open access, high quality, and available metadata.

Description: Lakes and drained lake basins (DLBs) are dominant landforms across Arctic lowland regions. The long-term dynamics of lake formation and drainage is evident in the abundance of lakes and DLBs covering as much as 80% of the landscape in various regions of Arctic Alaska, Russia, and Canada. Lake drainage can be triggered through different mechanisms such as lake tapping by an adjacent stream, bank overflow or ice wedge degradation. Following drainage, DLBs can become valuable grazing land for caribou and reindeer as well as usable land for infrastructure development due to low ground ice content in recent DLBs. In addition, DLBs can be sites for soil organic carbon accumulation in the form of peat which also play a role for carbon cycling. Comprehensive and accurate mapping of DLB distribution, age and drainage mechanism, will further inform our understanding of their role in permafrost landscape evolution across varying timescales. DLBs differ from the surrounding terrain in vegetation structure and composition, soil moisture, elevation, size and types of ice-wedge polygons and other parameters that make them an identifiable target based on remote sensing data. Here, we present a novel approach to map DLBs in permafrost landscapes with a specific focus on the North Slope of Alaska as well as select areas in Siberia and northwestern Canada. To map DLBs, we combined multispectral satellite imagery (Landsat-8 and Sentinel-2), Synthetic Aperture Radar (SAR) acquisitions (Sentinel-1), and DEM data (ArcticDEM). To cover the entire study area in each region, we included Landsat-8 acquisitions from all available years and Sentinel-2 for 2016 and 2018 to create cloud-free mosaics. The classification combines methodologies from pixel-based and object-based image analysis. To allow for processing of these large datasets that cover more than 200.000 km2, a classification workflow was developed in Google Earth Engine. Preliminary results show good agreement of our classification with previously published data sets for subsets of our North Slope study area. This work marks the first attempt to map DLBs at the pan-Arctic scale. Our results highlight the importance of treating areas of different surficial geology and vegetation communities separately in the classification process to ensure higher classification accuracy.

Description: Uncertainty in carbon cycling in terrestrial ecosystems contributes to overall uncertainty in Earth System Models. In particular, polar terrestrial ecosystems are understudied. Here, we focus on optical and radar remote sensing approaches to understand above-ground carbon dynamics related to vegetation as primary producers in tundra permafrost landscapes. In the ongoing Russian-German research cooperation and joint field expeditions we evaluate the applicability of remote sensing for assessing vegetation stocks and short-term fluxes in the Lena River Delta in the Siberian Arctic. New spaceborne satellite missions such as Sentinel-1, Sentinel-2 and ESA Data User Element DUE Permafrost provide useful services and data for this investigation. i) We evaluated and ground-truthed circumarctic-harmonized geospatial products of land cover and vegetation height from the ESA GlobPermafrost program for the Lena Delta region. The remote sensing products were derived from radar Sentinel-1 and optical Sentinel-2 satellite data. They are findable in the Arctic Permafrost Spatial Center (APGC) (apgc.awi.de) and are published under 10.1594/PANGAEA.897916, [Titel anhand dieser DOI in Citavi-Projekt übernehmen] and 10.1594/PANGAEA.897045 [Titel anhand dieser DOI in Citavi-Projekt übernehmen] . ii) We classified land cover using Sentinel-2 data based on in-situ vegetation data and optimized on biomass and wetness regimes. iii) We investigated the applicability of different land cover products for upscaling in-situ field-based biomass estimates to landscape-scale above-ground vegetation carbon stocks. iv) We investigated how disturbances enhance above-ground vegetation carbon cycling using in-situ data on vegetation community, biomass, and stand age and including remote sensing observations. Our research suggests that subarctic land cover needs to show biomass and moisture regimes to be applicable. Sentinel-1 and Sentinel-2 satellite missions provide adequate spatial high resolution to upscale vegetation communities and biomass in permafrost tundra landscapes. Biomass is providing the magnitude of the carbon flux, whereas stand age is irreplaceable to provide the cycle rate. High disturbance regimes such as floodplains, valleys, and other areas of thermo-erosion are linked to high and rapid carbon fluxes compared to low disturbance on Yedoma upland tundra and holocene terraces with polygonal tundra.

Description: Ebullition (bubbling) is often the dominant form of methane (CH4) emission from Arctic lakes. Understanding the dynamics of CH4 ebullition in these lakes is important to the global atmospheric CH4 budget and climate models. Lake CH4 ebullition bubbles generally originate from either ecologic or geologic sources. Ecologic CH4 is produced through anaerobic microbial decomposition of organic matter within lake sediments and the talik - a thawed zone beneath lakes in permafrost regions. Emissions from these seeps can be quantified and scaled based on existing field-based and remote-sensing methods. The other type of ebullition has not been well quantified, yet emits gas at a much higher rate than ecologic seeps. Geologic CH4 seeps originate from microbial, thermogenic, or a combination of both processes altering buried organics in ancient sedimentary basins. Bubbling rates of geologic seeps are strong enough to maintain holes in thick (〉1 m) lake ice – creating winter traveling hazards in the Arctic and sub-Arctic. While ecologic CH4 seeps produced in surficial sediments have modern to Holocene radiocarbon (14C) ages and those produced deeper in the talik have Pleistocene to early Holocene 14C ages, geologic CH4 seeps are often 14C-depleted due to the large contribution of carbon from fossil sources. Quantification and upscaling of geologic CH4 seepage is challenging because CH4 accumulations are distributed beneath complex, site-specific geologic and cryospheric settings. Previously, geologic seeps were studied through aerial surveys and ground truthing of open holes in winter lake ice along a north-south Alaskan transect. However, this is not efficient for quantifying these “superseeps” on a larger scale. Therefore, a remote sensing approach is needed. This work aims to detect superseeps using space borne Synthetic Aperture Radar (SAR). Engram et al. (2013) showed that L-band SAR backscatter correlates with roughness caused by stratigraphically-layered ecologic CH4 bubbles trapped during freeze-up – the greater the ebullition, the stronger the backscatter. Using this correlation, we developed a new method that identifies superseeps as perennial backscatter anomalies in lake ice on a landscape scale. Results from three regions in Alaska will be presented and compared to other methods of studying superseeps.

Description: Permafrost thaw has been observed at several locations across the pan-Arctic in recent decades, yet the pan-Arctic extent and potential spatial-temporal variations in thaw are poorly constrained. Thawing of ice-rich permafrost can be inferred and quantified with satellite imagery due to the subsequent differential ground subsidence and erosion that in turn affects land surface cover. Information contained within existing and rapidly growing collections of high-resolution satellite imagery (Big Imagery) is here extracted across the Arctic region through a collaboration between software engineers, computer- and earth scientists. More specifically, we are a) developing geospatial data down to sub-meter resolution, and also b) enabling discovery and knowledge-generation through visualization tools. This cyberinfrastructure platform, the Permafrost Discovery Gateway (PDG), is being designed with input from users of the PDG, e.g. primarily the Arctic earth science community but also the general public. The PDG builds upon other NSF supported data management resources (Arctic Data Center and Clowder) and the Fluid Earth Viewer. The Fluid Earth Viewer, which is the first visualization tool implemented into the PDG, was initially created for the public to explore atmospheric and oceanographic visualizations and is here modified to support permafrost geospatial products, and a number of community built analytic tools to identify permafrost artifacts within satellite imagery. The effort also includes workflow optimization of remote sensing code for pan-Arctic sub-meter scale mapping of ice-wedge polygons from optical imagery. We are additionally actively engaging with the user-community to ensure that the PDG becomes useful, both in terms of the type of data contained within the PDG and the design of the visualization tools. The PDG has the potential to fill key Arctic science gaps, such as bridging plot to pan-Arctic scale findings, while also serving as a resource informing decisions regarding the economy, security, and resilience of the Arctic region.

Description: While increasing Arctic temperatures have been identified to induce widespread thermokarst development in permafrost lowland landscapes over only several decades, disturbances, such as tundra fires can cause similar impacts within a few years. Transition from low-centered to high-centered polygons through the formation of troughs is an immediate result of melting ice wedges 3-4 years after a fire (Jones et al., 2015). Liljedahl et al (2016) have shown that widespread ice-wedge degradation can lead to hydrological connectivity and increased drainage of entire landscapes through newly developing trough networks. Quantifying such dynamics is important for projecting the hydrological outcomes of climate change impacts across vast Arctic landscapes. New VHR remote sensing approaches allow assessing ice wedge polygonal structures and their change in unprecedented detail. Data science methods provide valuable tools for understanding and modeling resulting very large datasets of changing ice wedge networks. Here we quantify thermokarst development representing the network of troughs as a graph, a concept from discrete mathematics used to model complex networks. Our analysis is based on optical VHR aerial imagery of the DLR MACS sensors and DSMs derived from LiDAR. Datasets are available for 2009, 2014 and 2019 of the northern Anaktuvuk River Fire scar in Alaska, which formed due to a large tundra fire in 2007. In particular, the post-fire permafrost degradation is observable in the northern ice-rich region of the fire scar on short timescales, offering an ideal site for the monitoring of degradation processes. We use morphological image analysis to extract a graph from the imagery and further deduce trough parameters, such as soil volume, depth, and water availability. Quantifying these factors for the study area shows that soil erosion and ice melt within individual troughs have progressed, while the overall connectivity of the network has increased, implying strong thermo-erosion since 2009. Using graphs to monitor the ongoing development offers a detailed and computationally efficient method that will allow quantification of ice-wedge degradation over very large spatial and temporal scales and may provide useful metrics for projecting landscape trajectories in thaw-vulnerable permafrost environments.

Description: Lakes in the northern permafrost region are a significant source of atmospheric methane (CH4), a potent greenhouse gas, yet large uncertainties exist in quantifying lake-source CH4. In thermokarst (thaw) lakes, the dominant pathway of CH4, ebullition (bubbling), is sporadic and spatially irregular. These lakes are also generally remote and difficult to access, resulting in challenging and costly field measurements. Scaling up field measurements from a few study lakes to regional and pan-Arctic scales relies on the assumption that the sampled lakes are a fair representation of all lakes across a landscape, which is not always the case. We present an innovative new method of quantifying lake-source CH4 using space-borne synthetic aperture radar (SAR), an instrument which can image at night, through clouds and dry snow, valuable attributes for Arctic remote sensing. Our recent work using satellite-based SAR data showed a significant correlation between polarimetric L-band SAR backscatter from lake ice and field-measured ebullition bubbles: L-band SAR backscatter intensity increases with the amount of ebullition bubbles trapped by early winter lake ice. We developed a regionally robust empirical model based on this correlation to quantify ebullition across surfaces of over 5,000 individual Alaskan lakes in satellite SAR scenes. We produced SAR-based ebullition fluxes from each lake across the landscape and created CH4 maps for five sub-regions in Alaska. Our SAR-based lake-source CH4 fluxes compare favorably with airborne CH4 measurements on the Barrow Peninsula and Atqasuk regions, and with scaled-up field measurements. We examine how our SAR remote sensing application can 1) improve selection of study lakes for field work, 2) provide regional estimates of CH4 ebullition from lakes in remote areas where field work is limited, 3) improve lake-size vs. flux relationships for upscaling field measurements and 4) shed light on the discrepancy of top-down vs. bottom-up CH4 flux estimates in the Arctic. This new approach to estimate lake-source CH4 from ebullition offers a unique opportunity to improve knowledge about CH4 fluxes for seasonally ice-covered lakes globally.

Description: Thermokarst lakes are one of the most abundant landforms in periglacial landscapes. They develop in regions underlain by permafrost as a consequence of soil subsidence triggered by the melting of excess ground ice. As a result of further permafrost degradation and shoreline erosion, thermokarst lakes increase in size, expanding vertically and laterally. This growth process has strong impacts on local to regional hydrological networks and ecological functions of the surrounding landscape. Previous research on the lateral growth of thermokarst lakes usually focused on decadal time scales which results in averaged expansion rates. These averages mask the temporal and spatial variations of lateral thermokarst expansion that occur over shorter time periods of only a few years. The short-term variability results from complex interactions between local erosion processes and meteorological and permafrost conditions. The aim of our study is to quantify these short-term changes of lake shorelines to provide a better understanding of permafrost landscape processes using multi-temporal high-resolution satellite imagery. The images are in the visible and near-infrared spectrum with a resolution of 0.3 to 0.7 m. They cover the period from 2006 to 2017 with acquisitions every 2 to 4 years. In order to map the lake shoreline changes we developed a fully-automated, open-source workflow for analyzing the changes of waterbodies larger than 1000 m². First, all necessary pre-processing steps are implemented such as pansharpening and smoothing of any speckle over waterbodies. Then, the normalized difference water index (NDWI) is applied to extract waterbodies from the imagery and derive their shoreline geometry. After filtering for potentially misclassified elements that originate from infrastructure, shoreline movement rates are calculated using a nearest point analysis. The workflow is independent of scale, image spatial resolution, and waterbody geometry. Preliminary findings demonstrate that the approach provides reliable shoreline recognition for every time step in the different study areas even under difficult light conditions. Changes can be detected on a sub-meter scale. Finally, we discuss the influence of the waterbody’s size and geometry on the shoreline change processes.