Antarctic Ice Shelf Coastline delineation by combining Sentinel-1 and CryoSat-2 data using Deep Learning

Antarctic Ice Shelves exhibit dynamic daily activity making the delineation of their coastlines important to examine the health of the ice shelf and monitor calving. Currently, no solution combines the high temporal and spatial resolution of Sentinel-1 with the precision of CS2-based delineation. Our project addresses this gap by developing a deep learning-based method that fuses radar (Sentinel-1) and altimetry (CS2) data to delineate ice shelf coastlines more accurately. We show how our model siginificantly improves on the Icelines even though it used model created labels. The study focuses on delineating ISCL between 2018- 2022 for Larsen-C, Ronne, and Filchner.