Remote Sensing of Cryospheric Surfaces : Small Scale Surface Roughness Signatures in Satellite Altimetry Data

Abstract: The Arctic cryosphere is experiencing a higher rate of warming compared to the rest of the world due to Arctic amplification. As glacier elevation change provide reliable evidence of climate change it is routinely measured by satellite altimeters. Satellite altimetry, while a valuable tool for monitoring elevation change over time, is subject to inherent uncertainties caused by, among other factors, the small scale surface roughness of the target surfaces. Previous studies have identified surface roughness as a key source of uncertainty when measuring sea ice freeboard and studies suggest the surface roughness strongly influences the Synthetic Aperture Radar (SAR) signatures of sea ice. Similar studies over snow- and glacier surfaces, are rare. In this context, we attempt to conduct a small scale calibration and validation (cal/val) campaign over glacier surfaces, using the ideal location and infrastructure of the University Centre in Svalbard. We demonstrate the process, from planning through field data collection and data analysis. By doing so, we identify good as well as bad practices. Using high resolution in-situ LiDAR data, collected under two ICESat-2 (IS2) overpasses in Svalbard we generated Digital Elevation Models (DEM) and calculated surface roughness estimates across glacier- and snow surfaces. The surface roughness was quantified by calculating the Root Mean Square (RMS) of deviations from the overall topography of the surfaces. The DEMs were used for direct comparison with the satellite elevation retrievals and the observed elevation differences were tested for correlation with surface roughness at different length scales. We then investigated the effect of surface roughness on the photon cloud of the lower level ATL03 ICESat-2 data products, by quantifying the precision in the data. We found little to no correlation between RMS roughness and the observed elevation differences between in-situ and satellite data sets, possibly explained by errors in georeferencing the DEMs. We show moderate to strong correlation between photon cloud precision and along- and across-track absolute surface slopes, with correlation coefficients of 0.6–0.8. Correlation between photon cloud precision and RMS roughness was found, with a maximum correlation coefficient of 0.9 for a roughness length scale of 1m. The results suggest IS2 is sensitive to surface roughness at similar length scales but we identify a need for more data, covering a wider range of surfaces and potential roughness scenarios, to draw strong conclusions. We demonstrate how a small team can carry out a cal/val campaign in the high arctic and collect coincident data under satellite overpasses, data which is typically rare for the remote high Arctic regions.