Lawrence Berkeley National Laboratory

Snowpack is a vital component of Earth's hydrological cycle and one of the most sensitive to global warming. In order to develop a predictive understanding of the hydrological and biogeochemical dynamics in snow-dominated watersheds, scientists and water resources managers need spatially and tempo- rally dense data sets of snowpack parameters, which could be obtained by networks of low-cost, low-power, distributed sensors. We investigate how 60 GHz frequency modulated continuous wave (FMCW) radar systems-on-chips (SoCs) can be used for measuring a variety of snowpack parameters. For snow depth measurements, we present a dedicated radar detection algorithm that detects the top of the observed snowpack, and we compare its performance to the established cell averaging constant false alarm rate (CA-CFAR) technique. We also evaluate the impact of non-coherent integration of radar frames over receive channels and over time. For measuring snow density and snow water equivalent (SWE), we rely on an adjustment for the signal's propagation speed in the snowpack based on prior knowledge about the radar's true distance to the ground. Our lab and field experiments show that snow layers can be detected, bulk snow density can be calculated, and the 90th percentile of snow depth measurement errors is 25 mm.