Sensitivity of glacier elevation analysis and numerical modeling to CryoSat-2 SIRAL retracking techniques

• Main Authors: Trantow, Thomas, Herzfeld, Ute C., Helm, Veit, Nilsson, Johan
• Format: Article in Journal/Newspaper
• Language: unknown
• Published: 2021
• Subjects: glacier; Alaska
• Online Access: https://epic.awi.de/id/eprint/53265/; https://www.sciencedirect.com/science/article/pii/S0098300420305926?via%3Dihub; https://hdl.handle.net/10013/epic.77800f91-ebe0-4459-af95-1713876ea3f2

The CryoSat-2 radar altimetry mission, launched in 2010, provides key measurements of Earth's cryosphere. CryoSat-2's primary instrument, the Synthetic Aperture Interferometric Radar Altimeter (SIRAL), allows accurate height measurements of sloped ice-surfaces including the highly crevassed Bering-Bagley Glacier System (BBGS) in southeast Alaska. The recent surge of the BBGS in 2011–2013, which resulted in large-scale elevation changes and wide-spread crevassing, presents an interesting challenge to the processing of the SIRAL measurements. Derivation of surface height is achieved by retracking the received waveform of the altimeter signal. Several such retracking methods have been developed. In this paper, we investigate the influence of six unique SIRAL retracking methods on (1) Digital Elevation Model (DEM) generation, (2) analysis of ice-surface topography, and (3) numerical modeling results of the BBGS during surge. First, we derive a surface DEM for each retracked dataset using kriging. The swath-processed dataset provides 100–250 times more points than the other datasets, which decreases DEM uncertainty associated with data coverage by a factor of 2–4. Differences between the six resulting DEMs imply that retracking methods can have significant effects on elevation and elevation-change analysis, but we find that lower-level processing has larger effects. Next, the sensitivity of the data-model connection is evaluated using a finite element model of the BBGS surge. We set up six modeling experiments, each initiated with a unique input surface DEM derived from the various retracking methods. While retracking choices effect estimation of unknown model parameters related to crevasse simulation, we have developed a procedure to limit these effects resulting in remarkably consistent parameter optimization across modeling experiments. Each model experiment yields an optimal friction coefficient in the sliding law of 10^-5 MPa*a/m, while estimates of the optimal von Mises stress threshold for crevasse initiation ranged between 230 and 240 kPa.