No abstract

Glacier surges are periodic episodes of mass redistribution characterized by dramatic increases in ice flow velocity and, sometimes, terminus advance. We use optical satellite imagery to document five previously unexamined surge events of Sít’ Kusá (Turner Glacier) in the St. Elias Mountains of Alaska from 1983 to 2013. Surge events had an average recurrence interval of ~5 years, making it the shortest known regular recurrence interval in the world. Surge events appear to initiate in the winter, with speeds reaching up to ~25 m d−1. The surges propagate down-glacier over ~2 years, resulting in maximum thinning of ~100 m in the reservoir zone and comparable thickening at the terminus. Collectively, the rapid recurrence interval, winter initiation and down-glacier propagation suggest Sít’ Kusá's surges are driven by periodic changes in subglacial hydrology and glacier sliding. Elevation change observations from the northern tributary show a kinematic disconnect above and below an icefall located 23 km from the terminus. We suggest the kinematic disconnect inhibits drawdown from the accumulation zone above the icefall, which leads to a steady flux of ice into the reservoir zone, and contributes to the glacier's exceptionally short recurrence interval.

Theory and observation show that glacier-flow regimes characterized by high basal slip enhance the projection of topographic detail to the surface, motivating this investigation into the efficacy of using glacier surges to improve bed estimation. Here we adapt a Bayesian inversion scheme and apply it to real and synthetic data as a proof of concept. Synthetic tests show a reduction in mean RMSE between true and inferred beds by more than half, and an increase in the mean correlation coefficient of ~0.5, when data from slip- versus deformation-dominated regimes are used. Multi-epoch inversions, which partition slip- and deformation-dominated regimes, are shown to outperform inversions that average over these flow regimes thereby squandering information. Tests with real data from a surging glacier in Yukon, Canada, corroborate these results, while highlighting the challenges of limited or inconsistent data. With the growing torrent of satellite-based observations, fast-flow events such as glacier surges offer potential to improve bed estimation for some of the world's most dynamic glaciers.