Constraining Ice-Dynamic Model Parameters using Crevasse Observations

Traditionally, observed velocity data have been used to constrain important unknown parameters in numerical models of glacier dynamics such as the basal friction coefficient. However, during acceleration events such as a glacier surge, the rapid movement creates wide-spread crevassing and large-scale elevation changes, which tend to complicate common methods for deriving surface velocity estimates from satellite data. Therefore, estimation of parameters in a surge-glacier model requires additional observational data to fill the void of missing velocity information. To this end, we have developed a model-data comparison method that relies on observations of crevasse characteristics to constrain a surge model of the Bering Bagley Glacier System (BBGS), Alaska. We have demonstrated the usefulness of this approach when investigating the 2011-2013 surge of the BBGS. By employing a simple cost function, we optimally estimate two important parameters in our BBGS surge model: (1) the friction coefficient in the basal sliding law, and (2) the von Mises stress threshold which controls crevasse formation. We then extend the comparison method by utilizing the sparse velocity information attained during the surge, which allows estimation of a third unknown parameter: ice temperature. So far, crevasse characteristic and velocity information have been derived using only Landsat panchromatic imagery at 15 meter resolution, which is often limited by cloud cover. The combined velocity/crevasse model-data comparison method will only improve with the advent of new satellite technology, such as the short-baseline SAR imagery from Sentinel-1 and the high-resolution altimetry data from ICESat-2.