The feasibility of deriving bathymetry from data acquired with ATLAS, the photon-counting lidar on NASA’s upcoming ICESat-2 satellite, is assessed empirically by examining data from NASA’s airborne ICESat-2 simulator, MABEL. The primary objectives of ICESat-2 will be to measure ice-sheet elevations, sea-ice thickness, and global biomass. However, the 6-beam, green-wavelength photon-counting lidar, combined with the 91-day repeat period and near-polar orbit, may provide unique opportunities to measure coastal bathymetry in remote, poorly-mapped areas of the globe. The study focuses on high-probability bottom returns in Keweenaw Bay, Lake Superior, acquired during the “Transit to KPMD” MABEL mission in August, 2012 at an AGL altitude of 20,000 m. Water-surface and bottom returns were identified and manually classified using MABEL Viewer, an in-house prototype data-explorer web application. Water-surface returns were observed in 12 green channels, and bottom returns were observed in 10 channels. Comparing each channel’s mean water-surface elevation to a regional NOAA Nowcast water-level estimate revealed channel-specific elevation biases that were corrected for in our bathymetry estimation procedure. Additionally, a first-order refraction correction was applied to each bottom return. Agreement between the refraction-corrected depth profile and NOAA data acquired two years earlier by Fugro LADS with the LADS Mk II airborne system indicates that MABEL reliably detected bathymetry in depths up to 8 m, with an RMS difference of 0.7 m. In addition to feeding coastal bathymetry models, MABEL (and potentially ICESat-2 ATLAS) has the potential to seed algorithms for bathymetry retrieval from passive, multispectral satellite imagery by providing reference depths.