Satellite radar interferometry and feature tracking for measuring Everest region glacier recession

The application of satellite radar interferometry (SRI) to high-mountain areas has recently become more viable because of improved digital elevation data and precision orbit models, which now allow 24-hour snapshots of Himalayan glacier velocity from the mid 1990s to be retrieved from ERS imagery without ground control information. Feature tracking between Synthetic Aperture Radar (SAR) images can yield complementary results, even on the small glaciers of this region, providing annually averaged velocity data from 1992 through to present day. In unison, these two techniques now provide a means of monitoring glacier dynamics and their response to climatic warming in areas where in-situ measurements are understandably rare.

In this paper, we compare the two techniques of SRI and feature tracking to measure flow rates in Everest region glaciers. We demonstrate the varying spatial coverage provided by each method, with SRI being more successful in deriving velocities high on the glaciers where coherence is at a maximum and supraglacial debris is not so abundant, while feature tracking tends to work better on the glacier snouts where the debris layer is thick and recorded velocities are low. When combined, the results reveal a distinct spatial variability in flow in the region, with northern flowing (predominantly clean ice) glaciers exhibiting surface displacements of the order of 0.05 to 0.1 m/d-1 and southern flowing (debris covered) glaciers being largely stagnant, with most surface displacement recorded in these areas being below the expected error in the data. In support of the velocity data, we also propose the use of topographic profiling as a proxy for glacier health, with active glaciers exhibiting a convex surface profile in comparison to those wasting in-situ, which display profiles of a more concave form.

We conclude that both radar techniques can generate valuable glacier dynamics information independently, but their true worth is best demonstrated when employed in tandem. This combined approach ensures that velocity data can be derived over both clean and debris-covered ice, irrespective of glacier flow direction, over daily as well as annual timescales. The specific results demonstrated for the Everest region show that a distinct north-south divide exists between active and stagnant ice in this area, and that in-situ downwasting is the predominant symptom of recession, manifested by strongly concave surface profiles. The methods and results presented here will be of interest to glaciology and remote sensing communities alike, as well as practitioners working in such high-mountain areas with interests in water resource issues in particular.