Characterizing Small-Scale Variability of Snow Thickness Using GPR on Taku Glacier, Alaska

Spatial variability of snow accumulation is controlled by terrain type, time, and other factors, and has proven challenging to measure. Yet snow accumulation plays an important role in hydrological and geodetic issues. For example, glacier-wide mass balance estimates rely on sparsely distributed point measurements, under the assumption that spatial variability around each measurement site is negligible. Moreover, errors in accumulation estimates are substantially greater than those in ablation estimates, primarily due to our inability to model the aforementioned variability.

To fill this information gap, we used 500 MHz common offset ground-penetrating radar (GPR) to examine the local representativeness of direct point measurements of snow depth used to estimate glacier-wide mass balance at Taku Glacier in Alaska. During summer 2013, we measured snow depths in four dense 120 m square radar grids centered on ground truth snow pits, and also along longitudinal profiles between these pits. We used the results to characterize accumulation variability over multiple length scales in this maritime climate. Processed GPR traces adjacent to our snowpits resolved depths of wet, isothermal snow within the nominal error of the instrument: approximately ± 15cm. Throughout each grid, the interquartile range (IQR) of GPR-estimated snow depths was less than 10cm. This suggests that in this setting a single snow pit adequately represents nominal snow thickness within each grid, and that elevation provides a dominant control on snow accumulation.