ARTS: a scalable data set for Arctic Retrogressive Thaw Slumps

Abstract

Retrogressive thaw slumps (RTS) are one of the most rapid abrupt thaw events that have a positive feedback on climate warming. RTS are not yet well understood because of the lack of geospatial products describing abrupt thaw distribution and changes over time in the Arctic. Although many standalone RTS digitisation data sets have been archived, it is challenging to find, access and pool the existing data sets into a comprehensive and unified one due to the lack of common data curation standards. Therefore we collected the existing RTS digitisation data sets known to date and compiled them into a scalable and uniform data set - Arctic Retrogressive Thaw Slumps (ARTS). Besides, we developed an RTS data curation framework, which provides guidelines for RTS remote sensing data digitisation, metadata formatting, RTS indexing, storage format, contribution guidelines and more. So far the ARTS data set contains around 24,000 RTS digitisations and 3,300 non-RTS background labels. This data set will empower a wide range of Arctic studies, especially beneficial for deep learning studies that are highly data-intensive.