Tracking rapid permafrost thaw through time

This research paper describes the application of convolutional neural network (CNN)-based machine learning models to detect and monitor retrogressive thaw slumps (RTS) in high latitude northern permafrost using Sentinel-2 satellite data.

• The authors developed site-specific CNN models using open-source Sentinel-2 satellite data to remotely detect retrogressive thaw slumps (RTS), which serve as indicators of rapid permafrost thaw (RPT) and the accelerated release of greenhouse gases (GHG).
• The CNN models achieved precision, recall, and F1 values greater than 0.8 for the assessed sites, enabling the production of time series data on RTS development to approximate associated GHG emissions.
• Current analysis of short time series does not show clear trends in RTS development, a limitation attributed to the 10 m resolution of Sentinel-2 data and a lack of diverse, validated training data.

APA

Stockholm Environment Institute (2022). Tracking rapid permafrost thaw through time. https://www.sei.org/publications/tracking-rapid-permafrost-thaw-through-time/

Chicago

Stockholm Environment Institute. Tracking rapid permafrost thaw through time. 2022. https://www.sei.org/publications/tracking-rapid-permafrost-thaw-through-time/.

Wikipedia

{{cite report |author=Stockholm Environment Institute |title=Tracking rapid permafrost thaw through time |date=20 October 2022 |url=https://www.sei.org/publications/tracking-rapid-permafrost-thaw-through-time/ |access-date=17 August 2026 |via=Climate Insights Directory}}

BibTeX

@techreport{stockholmenvironmentinstitute2022tracking, author = {{Stockholm Environment Institute}}, title = {{Tracking rapid permafrost thaw through time}}, institution = {Stockholm Environment Institute}, year = {2022}, month = oct, url = {https://www.sei.org/publications/tracking-rapid-permafrost-thaw-through-time/}, urldate = {2026-08-17}, note = {Indexed by Climate Insights Directory} }