Gradual build up and episodic character of glacier velocity preceding and during the surge of a Karakorum glacier enabled by open-access image-processing

Glacier surging provides a unique opportunity to examine rapid changes in glacier sliding that occur when some glaciers alternate between slower-than-normal (quiescence) and faster-than-normal (surge) velocities. On surging glaciers, mechanical instabilities within the glacier set off a regime of fast glacier flow, which causes these glaciers to accelerate and advance. The precise processes that cause a surging remain uncertain and likely vary between glaciers. However, the uptake of studies on glacier surging over the past decade continues to yield invaluable insights in glacier dynamics. In this study, we combine optical remote sensing and numerical modeling to examine the recent surge of Shishper glacier, in the Pakistani Karakorum. This glacier started surging in 2018, showed a dramatic terminus advance that reached rates of several meters per day. In the process, it dammed the adjacent valley, forming a lake which drained in June 2019 flooding the downstream valley, damaging the Karakorum Highway and threatening nearby communities. We leverage a high spatio-temporal resolution dataset of glacier velocities, using roughly 100 open-access images, across the Landsat-8 and Sentinel-2 record, thus encompassing the quiescence (2013-2018) and surge (2018-2019) phases. We created the dataset in an updated and nearly automated workflow by using the COSI-Corr software package to calculate displacements between images combined with a unique algorithm to filter data and remove artifacts. The result consists in high-resolution velocity maps with resolution with time intervals as short as five days. Such dataset provide a complete time-series of the spatio-temporal evolution of ice-surface velocities during a surge. One of the most notable finding is that the surge onset occurs progressively. In the two years leading up to the surge, spring speed-ups became increasingly larger in than the long-term median. We further identify three periods with  surge velocities far higher than the long-term median that likely coincide with hydrological events. Two periods occur in the spring (2018 and 2019) and the third corresponds with the lake formation in the winter of 2018-2019. Finally, we establish that the surge termination coincided with the lake drainage at the end of June 2019. The current availability of open-access imagery and  glacier topography allow us to  make an increased quantity of observations and thus better quantify glacier dynamics.