Retrieval of Supraglacial Lake Water Surface Elevation and Depth Over the Greenland Ice Sheet Using SWOT Observations

Abstract

Accurate quantification of supraglacial lake water surface elevation (WSE) and depth is essential for understanding surface mass balance processes of the Greenland Ice Sheet. However, optical remote sensing in polar regions is severely limited by cloud cover and floating ice, hindering all-weather, high-accuracy lake depth retrievals. In contrast, laser altimetry missions such as ICESat-2 provide centimeter-level vertical accuracy but are limited by one-dimensional (1-D) along-track sampling, preventing characterization of 2-D lake depth variability. To overcome these limitations, this study exploits the Ka-band interferometric capability of the Surface Water and Ocean Topography (SWOT) mission and proposes a method to retrieve supraglacial lake WSE and depth using the Pixel Cloud (PIXC) data product. Results from the five test lakes indicate the feasibility of using SWOT PIXC data under complex polar observation conditions. Specifically, SWOT-derived WSEs show strong agreement with ICESat-2 measurements, with an overall root-mean-square error (RMSE) of 0.091 m. In lake depth retrieval, the proposed method achieves an overall RMSE of 0.421 m, representing an improvement of approximately 43% compared to an eXtreme Gradient Boosting machine learning model based on Sentinel-2 imagery. Furthermore, the SWOT-based reconstruction provided spatially continuous 2-D depth estimates within the delineated lake boundaries, including areas where the optical comparison contained ice-related data gaps. Overall, SWOT not only provides an independent and traceable reference for supraglacial lake WSE and depths, but also complements optical retrievals under conditions of cloud cover and floating ice, thereby supporting a complementary SGL monitoring framework.