Abstract:Reliable hydrological forecasts of snowmelt runoff are of major importance for many areas. Ground-penetrating radar (GPR) measurements are used to assess snowpack water equivalent for planning of hydropower production in northern Sweden. The travel time of the radar pulse through the snow cover is recorded and converted to snow water equivalent (SWE) using a constant snowpack mean density from the drainage basin studied. In this paper we improve the method to estimate SWE by introducing a depth-dependent snowpack density. We used 6 years measurements of peak snow depth and snowpack mean density at 11 locations in the Swedish mountains. The original method systematically overestimates the SWE at shallow depths (C25% for 0Ð5 m) and underestimates the SWE at large depths ( 35% for 2Ð0 m). A large improvement was obtained by introducing a depth-density relation based on average conditions for several years, whereas refining this by using separate relations for individual years yielded a smaller improvement. The SWE estimates were substantially improved for thick snow covers, reducing the average error from 162 š 23 mm to 53 š 10 mm for depth range 1Ð2-2Ð0 m. Consequently, the introduction of a depth-dependent snow density yields substantial improvements of the accuracy in SWE values calculated from GPR data.

No abstract