Ensemble matching of repeat satellite images applied to measure fast-changing ice flow, verified with mountain climber trajectories on Khumbu icefall, Mount Everest

Abstract

Velocities within an icefall are typically the fastest within a glacier system and experience complex flow. The combination of convergent and fast flow, and steep slope generate a quickly changing and intensely fractured surface. This complicates velocity extraction from repeat satellite images, especially when common pattern matching procedures are used. In this study, we exploit the high temporal revisit of medium-resolution satellite images using a novel image matching technique, ensemble matching, making it possible to generate a high-resolution (30 m) velocity field from high-repeat image sequences despite challenging image conditions. We demonstrate this technique for the first time in the glaciology domain using repeat Sentinel-2 optical data over the famous Khumbu icefall, situated on the southern slopes of Mount Everest. Estimates of velocity go just over 1m d−1, which is slower than summer velocities from noisy single pair image matching. This icefall is frequently crossed by high-altitude mountaineers who use a route confined by fixed ropes and ladders set out every season. The mountain climbers typically record their trajectory on their personal satellite navigation device. We use such volunteered geographic information to verify our velocity estimates, confirming our underestimation with ensemble matching. Besides unprecedented remotely sensed surface velocities over the icefall, we also note that the generated velocity field can aid with the planning of a safe passage through this icefall.