Machine Learning-Based Monitoring of Surface Melt on Devon Ice Cap, Nunavut, Canada Using Sentinel-1 (2015-2024)

This dataset provides record of glacier surface melt conditions on Devon Ice Cap, Nunavut, Canada, derived from sub-weekly Sentinel-1 synthetic aperture radar (SAR) Extra Wide (EW) mode imagery and digital elevation model (DEM) derivatives for the period 2015–2024. A convolutional neural network (CNN) was trained using 8,297 manually labelled samples to classify four surface classes: glacier ice, slush, wet snow/firn, and dry snow. The repository includes ~300 classified raster products for each acquisition period. All data are georeferenced, time-stamped, and formatted for direct integration into GIS and remote sensing workflows.