Assessing fire-induced permafrost ground deformation and environmental drivers using interferometric synthetic aperture radar and machine learning

In recent years, fires in the Arctic have increased, especially in Siberia, affecting permafrost ground that stays frozen year-round. While fires have immediate impacts, they may also cause longer-term changes in the permafrost ground, as burning removes the insulating organic layer and exposes charcoal, leading to deeper seasonal thawing and long-term deformation that can persist for decades. We used satellite data from Sentinel-1 to measure permafrost ground deformation after five tundra fires in northeastern Siberia in 2019 and 2020. We found that burned areas sank three times faster than unburned ones. To better understand why, we used a machine learning model, and we found that surface reflectivity (albedo) after fires was the biggest factor driving this sinking, explaining over half of the changes. Fire-related landscape changes were more important than terrain characteristics in explaining permafrost ground deformation. We also used numerical model simulations, which showed that darker surfaces and lost organic layers speed up permafrost ground deformation. Our study highlights how tundra fires are impacting permafrost, and the satellite-based approach can help track fire impacts on permafrost across the Arctic, which is important as wildfires are becoming more frequent there. This dataset contains supporinting raw data for the study.