Ice marginal lakes were identified using three independent remote sensing methods:
1) multi-temporal backscatter classification from Sentinel-1 synthetic aperture radar imagery;
2) multi-spectral indices classification from Sentinel-2 optical imagery;
and 3) sink detection from the ArcticDEM (v3). (The ArcticDEM is an NGA-NSF public-private initiative to automatically produce a high-resolution, high quality, digital surface model (DSM) of the Arctic using optical stereo imagery, high-performance computing, and open source photogrammetry software.)

All data were compiled and filtered in a semi-automated approach, using a modified version of the MEaSUREs GIMP ice mask (https://nsidc.org/data/NSIDC-0714/versions/1) to clip the dataset to within 1 km of the ice margin. Each detected lake was then verified manually.