Land Cover Patterns for selected Drained Lake Basins in Yamal, Siberia.

Circumpolar permafrost landscapes are in rapid transition and strongly affected by climate warming. Common features of permafrost landscapes are thermokarst lakes and drained lake basins (DLBs), which play an important role for the geomorphological, hydrological and the ecological development of Arctic landscapes. Landcover changes related to lake drainage events affect the carbon cycle and induce habitat changes that disturb the local biodiversity. Improving our knowledge on processes involved in drainage events in the Arctic environment associated with permafrost conditions and surrounding terrain is crucial for numerous applications (e.g., landscape models, carbon cycle processes and local infrastructure stability).

In this study, we focus on DLBs on the Yamal peninsula in northern Russia, Siberia. DLBs and associated landscape dynamics such as changes in surface water area and vegetation cover can be monitored consistently only from space due to the remoteness and expanse of the study area. Yamal is underlain by both discontinuous and continuous permafrost and is covered by different tundra vegetation communities, thaw lakes, wetlands, and river floodplains. We semi-manually selected several drained lake basins which experienced a recent drained lake event. Existing Landsat trend products going back to 1999 have been combined with a recent lake classification based on Sentinel-1 and Sentinel-2 (2016-2020, machine learning classifier) in a first step. Lakes which remained drained have been identified and manually inspected before final selection.

The selected DLBs are evaluated for annual landcover patterns from 2016 up to present. Landcover information was generated from Sentinel-1 and Sentinel-2 satellite data. The landcover classes were derived from a combination of unsupervised and supervised classification following landcover classifications developed within the ESA DUE Globpermafrost and Permafrost_cci projects. A comparison between derived landcover classes and existing in-situ data was carried out for certain drained lake basins.

The results will advance the understanding of DLBs and the corresponding change of biodiversity as well as carbon cycling as is the focus of the Horizon 2020 project CHARTER and ERC Q-Arctic respectively.