Dynamics and spatial variability of Arctic and Subarctic thermokarst lakes and their biogeochemical significance

The rapidly thawing permafrost in the Arctic and Subarctic regions poses a complex scientific problem, with its mechanisms defying traditional climate modeling. Unlike gradual thaw, abrupt permafrost thaw processes, such as thermokarst lake and pond formation, depend on intricate local geological and environmental interactions, rendering them extremely challenging to integrate into Earth System Models (ESM). This is due to their abundance, small size (< 10,000 m2 ), and high optical and biogeochemical variability. In this study, spanning the different permafrost zones of Canada, a deep learning Mask-Regional-based Convolutional Neural Network (Mask R-CNN) model was trained over PlanetScope-Dove (PS-D) imagery, producing the High Latitude Water (HLWATER) model. The model was then deployed to automatically delineate 335,281 water bodies, 90% smaller than 10,000 m2 . Unmanned Aerial System (UAS) data were used to validate and assess the limitations of the HLWATER-derived products. The resultant HLWATER database was used as a spatial reference for Sentinel-2 (S2) reflectance retrievals, allowing optical assessments of water bodies (HLWATER-Optical). In the continuous permafrost zone of Paulatuk, the number of ponds increased from 164 to 454 between 1975 and 2020. In the regional sector of western Nunavik, spanning the discontinuous and sporadic permafrost zones, some landscapes, although representing only 2-7% of the total area, corresponded to over one-third of the total number of small water bodies (< 10,000 m2 ). The HLWATER-Optical allowed the identification of different Arctic and Sub-Arctic lake dominated landscapes, including thermokarst areas. Water body optical properties, ranging from oligotrophic black to brown and light-brown colors, showcased complex degradation processes of palsas and lithalsas, likely reflecting varying organic and inorganic (mineral) concentrations. The workflow presented in this thesis provides a scalable framework for mapping and characterizing thermokarst landscapes in high-latitude environments. It underlines the importance of contemplating the role of small water bodies in climate change assessments. Further, it advances the understanding of permafrost dynamics, establishing a framework for similar investigations in other vulnerable ecological systems worldwide.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.