Snow depth, density, and SWE estimates derived from GPS reflection data: Validation in the western U.S.

Geodetic-quality GPS systems can be used to measure average snow depth in the approximate to 1000 m(2) area around the GPS antenna, a sensing footprint size intermediate between in situ and satellite observations. SWE can be calculated from density estimates modeled on the GPS-based snow depth time series. We assess the accuracy of GPS-based snow depth, density, and SWE data at 18 GPS sites via comparison to manual observations. The manual validation survey was completed around the time of peak accumulation at each site. Daily snow depth derived from GPS reflection data is very similar to the mean snow depth measured manually in the approximate to 1000 m(2) scale area around each antenna. This comparison spans site-averaged depths from 0 to 150 cm. The GPS depth data exhibit a small negative bias (-6 cm) across this range of snow depths. Errors tend to be smaller at sites with more usable GPS ground tracks. Snow bulk density is modeled using the GPS snow depth time series and model parameters are estimated from nearby SNOTEL sites. Modeled density is within 0.02 g cm(-3) of the density measured in a single snow pit at the validation sites, for 12 of 18 comparisons. GPS-based depth and modeled density are multiplied to estimate SWE. SWE estimates are very accurate over the range observed at the validation sites, from 0 to 60 cm (R-2=0.97 and bias=-2 cm). These results show that the near real-time GPS snow products have errors small enough for monitoring water resources in snow-dominated basins.