Mapping snow depth within a tundra ecosystem using multiscale observations and Bayesian methods

This paper compares and integrates different strategies to characterize the variability of end-of-winter snow depth and its relationship to topography in ice-wedge polygon tundra of Arctic Alaska. Snow depth was measured using in situ snow depth probes and estimated using groundpenetrating radar (GPR) surveys and the photogrammetric detection and ranging (phodar) technique with an unmanned aerial system (UAS). We found that GPR data provided highprecision estimates of snow depth (RMSE = 2.9 cm), with a spatial sampling of 10 cm along transects. Phodar-based approaches provided snow depth estimates in a less laborious manner compared to GPR and probing, while yielding a high precision (RMSE= 6.0 cm) and a fine spatial sampling (4cm× 4cm). We then investigated the spatial variability of snow depth and its correlation to microand macrotopography using the snow-free lidar digital elevation map (DEM) and the wavelet approach. We found that the end-of-winter snow depth was highly variable over short (several meter) distances, and the variability was correlated with microtopography. Microtopographic lows (i.e., troughs and centers of low-centered polygons) were filled in with snow, which resulted in a smooth and even snow surface following macrotopography. We developed and implemented a Bayesian approach to integrate the snow-free lidar DEM and multiscale measurements (probe and GPR) as well as the topographic correlation for estimating snow depth over the landscape. Our approach led to high-precision estimates of snow depth (RMSE= 6.0 cm), at 0.5 m resolution and over the lidar domain (750m× 700 m).