A First Approach for an Approximate Calculation of Surface Heights for Ice Caps from TanDEM-X InSAR Measurements for Devon Island

Abstract

Surface elevation information of glaciers and ice caps is an important Essential Climate Variable (ECV) in particular for ice sheet mass balances. Due to their remote location and immense extension this information is hard to be gained. Selective ground measurements, airborne campaigns and spaceborne mission are the alternatives. Within the TanDEM-X mission a global digital elevation model was generated from single-pass X-Band synthetic aperture radar (SAR) interferometry. It is believed that the TanDEM-X DEM will widely be used for ice mass balances as those data are globally available. SAR systems penetrate the surface of snow and ice covered areas to some degree before being reflected. The most important parameters for the penetration into the ice sheet are the structure, the surface roughness, and the water content. In this study we present a procedure how corrections for the penetration depth into ice caps can be modelled and applied to the TanDEM-X DEM. The Devon Ice Caps in the Canadian Arctic serves as example to demonstrate effects and possible improvements. As verification data set ICESat is used. TanDEM-X DEM data were acquired between December 2010 and January 2015. The DEM has a 12.5 m posting with 10 m absolute and 2-4 m relative height accuracy (depending on the terrain slope). In the Arctic most of the acquisitions took place in winter. Therefore, we generated in addition to the TanDEM-X DEM mosaic two DEMs solely containing Winter 2010/11 and Winter 2011/12 scenes. The first step in our procedure is a classification of different snow facies. Dealing with InSAR data the intensity as well as interferometric coherence can be used for the classification. The intensity is related to surface roughness and water content and the coherence provides an estimate for the volume decorrelation. The goal is to develop a method to determine the penetration depth directly from the TanDEM-X DEM product. This method is applied to the global DEM mosaic, but also separately to two winter DEM mosaics as well as to all individual input InSAR scenes. In a second step snow and ice densities are assigned to the individual classes. Then an existing approach is used to calculate the approximate penetration depth. Here, the knowledge of a specific snow facies allows the estimation of the penetration depth by inverting an interferometric volume decorrelation model and assuming a constant and real dielectric constant. For each ice class the corresponding height offset is applied, i.e. the DEM heights are corrected accordingly. The results for TanDEM-X mosaic, the winter DEM mosaics, and the individual scene-corrected DEM mosaic are compared. For further verification the penetration-corrected DEMs are compared with ICESat data. Resulting differences to ICESat are in the order of 1-2m instead of up to 6m penetration depth. This procedure shows great potential as an innovative approach for assessing surface heights from TanDEM-X DEMs or measurements to be compared with other data measuring surface heights.