Quantifying crater morphology changes during the 2019-2020 Shishaldin Volcano eruption using SAR amplitude images

Compared to visual remote-sensing products, Synthetic Aperture Radar (SAR) data have the benefit of observing changes at volcanoes even during inclement weather conditions; InSAR utilizes the phase observations of the radar signal and is commonly applied to quantify volcano deformation and characterize eruptive activity. However, it is not always applicable if the deformation is large or if phase coherence is otherwise lost. Repeat amplitude images provide a way to observe change qualitatively but quantifying these morphological changes from a single look direction would increase our understanding of large-amplitude surface processes such as lava dome growth or collapses. We describe a new method to quantify morphological changes at volcanoes using the amplitude from SAR images with the assumption of an isotropic radar cross section. This method retrieves the gradient of elevation with respect to the radar range coordinates using a linear mapping function between amplitude values and gradient values from a digital elevation model. Integration of this gradient yields elevation values. We employ Bayesian inversions to determine the required integration constants and their confidence intervals required to compute the elevation uncertainties. We apply this method to dozens of coregistered TerraSAR-X spotlight images from both ascending and descending paths that observed the 2019-2020 eruption of Shishaldin Volcano, Unimak Island, Alaska. This eruption was characterized by months of summit scoria cone growth, lava effusion, and cone collapses, producing significant topographic change within the summit crater and ash fall on nearby communities. Our results suggest that the vertical elevation uncertainty of the method is on the order of several meters. Consequently, we were able to quantify cone growth in November 2019, collapses associated with explosions in December-January, and further changes in crater elevations into spring 2020. We conclude that this method detects and quantifies morphological changes on the order of tens of meters in a systematic way that might not be obvious during visual inspection of the imagery alone. The method can be automated to monitor morphological changes due to volcanic unrest in near-real-time as high-quality SAR images become available during an eruption.