Surface meltwater runoff on the Greenland ice sheet estimated from remotely sensed supraglacial lake infilling rate

Abstract

Abstract: Surface runoff constitutes a large percentage of Greenland Ice Sheet (GrIS) mass loss at present but is difficult to measure directly. This study provides a novel method to estimate surface runoff through remote sensing of supraglacial lake volumes. Because terminal, non-draining (consistently expanding during the melt season) lakes impound runoff from their surrounding contributing catchments, such changes reflect runoff produced within the catchment. To estimate supraglacial lake volumes, multi-temporal lake maps derived from Landsat-8 images are intersected with dry lake-bed topographic depressions (showing lake bathymetry) identified for two supraglacial catchments (~10 km2) in southwestern GrIS, using high-resolution (2 m) ArcticDEMs. Intersecting remotely sensed lake shorelines with their underlying ice surface topography yields multi-temporal lake volume changes, which are then compared with cumulative runoff as simulated by four Surface Mass Balance (SMB) models (HIRHAM5, MAR3.6, RACMO2.3, and MERRA-2). Comparison of cumulative lake infilling with SMB simulations for these two lakes over the period 8–31 July 2015 indicates that SMB models overestimated surface runoff by 106 – 123%. These large offsets improved after early July, overestimating runoff by 40 – 55%. The runoff delay function incorporated into the MAR3.6 model improves simulation of early melt season runoff, signifying the importance of integrating meltwater routing schemes into SMB models for improved understanding of Greenland supraglacial hydrology and surface mass balance.