Advancing airborne and spaceborne synthetic aperture radar measurements of ice and snow in the northern Great Plains

Thumbnail Image



Journal Title

Journal ISSN

Volume Title


Montana State University - Bozeman, College of Letters & Science


The cryosphere is responding to climate change in ways that have negatively impacted socio-environmental systems. Accurate and timely observations of the cryosphere are critical to adapting our infrastructure to these rapid changes. This dissertation contributes novel approaches to validating synthetic aperture radar (SAR) measurements over river ice and seasonal prairie snow. Previous C-band SAR-based river ice studies typically validate regional ice cover maps using aerial photos of frozen rivers. This qualitative approach relies on the principle that visually rougher ice should result in stronger SAR backscatter. In Chapter 2 of this dissertation I present the first systematic, quantitative investigation of the effect of river ice surface roughness on C-band Sentinel-1 backscatter. I employ Random Forest algorithms first to replicate qualitative classification results from previous studies, and then as regression models to explore relationships between Sentinel-1 backscatter and novel, quantitative surface roughness metrics derived from drone-based Structure-from-Motion datasets. Classification accuracies are similar to those reported in previous studies, but poor regression performance indicates a weak relationship between river ice roughness and Sentinel-1 backscatter. In Chapter 3, I extend these drone-based surface measurements of river ice with GPR-based subsurface measurements. Results from this smaller, richer dataset demonstrate that Sentinel-1 VV backscatter is correlated with ice thickness and VH backscatter with structural properties, but results are site-specific and more work is necessary to create generalized river ice models from Sentinel-1 measurements. Interferometric SAR techniques have been used to estimate snow water equivalent (SWE) using L-band measurements from the UAVSAR platform. These methods have been developed in mountainous areas and have not been investigated over prairie snowpacks, which typically feature exposed agricultural vegetation and greater spatial variability than found in mountain snowpacks. In Chapter 4 I develop a rigorous statistical framework to demonstrate that UAVSAR measurements over prairie snowpacks are sensitive to small changes in SWE, and are relatively unaffected by exposed agricultural vegetation. However, sub-pixel snow depth variability decreases the accuracy of SWE estimates derived from UAVSAR measurements. The upcoming NISAR satellite mission provides an opportunity to extend this work with repeated L-band measurements over a wider range of prairie snow conditions.





Copyright (c) 2002-2022, LYRASIS. All rights reserved.