Modeling and simulation of GPR wave propagation through wet snowpacks: Testing the sensitivity of a method for snow water equivalent estimation

Abstract

Abstract Snow water equivalent (SWE) of a snowpack is an important input to distributed snow hydrological models used for runoff predictions in areas with annual snowpacks. Since the conventional method of manually measuring SWE is very time-consuming, more automated methods are being adopted, such as using ground penetrating radar operated from a snowmobile with SWE estimated from radar wave two-way travel time. However, this method suffers from significant errors when liquid water is present in the snow. In our previous work, a new method for estimating SWE of wet snowpacks from radar wave travel times and amplitudes was proposed, with both these parameters obtained from a common mid-point survey. Here we present a custom ray-based model of radar wave propagation through wet snowpacks and results of MATLAB simulations conducted to investigate the method's sensitivity to measurement errors and snowpack properties. In particular, for a single-layer snowpack up to 2.1 m deep and with liquid water content up to 4.5% (by volume), the simulations indicate that SWE can be estimated with an error of ± 5% or less if (a) the noise (measurement errors) in resulting amplitude has a standard deviation less than 15% and(b) the noise in two-way travel time has a standard deviation less than 0.075 ns (22.5% and 0.15 ns for a snowpack less than 1.3 m deep).