ICESat-2 melt depth retrievals: Revisiting surface melt on Amery Ice Shelf, East Antarctica

Abstract:

The volume of surface melt produced on the Antarctic and Greenland ice sheets has been difficult to monitor accurately because depth estimates are challenging. NASA’S ICESat-2 brings a new capability to surface melt detection: 532 nm photons penetrate through water and are reflected from both the water and the underlying ice; the difference between the two provides a depth estimate. Several ICESat-2 tracks sampled Amery Ice Shelf during the January 2019 melt season. We initiated a pilot project with several investigators who contributed depth estimates for four melt features with depths ranging to 6m, using seven different algorithms based on ICESat-2, and compared them with estimates from two algorithms based on coincident Landsat MSS and Sentinel-2 images. We compared depth estimates with a manual dataset created as a baseline, in the absence of ground truth. We show that all algorithms are able to successfully identify the presence of surface water, at the same locations. The algorithms based on ICESat-2 data produce the most accurate depths, with each algorithm capturing different degrees of structural detail, and the image-based algorithms tending to underestimate depths. This implies that previous estimates of meltwater volumes have been too low, and that ICESat-2 depths can be used to tune image-based algorithms, moving us close to accurate estimates of meltwater volumes across Antarctica and Greenland.