Comparison of selected methods used for the calculation of the snowpack spatial distribution, BystŘice River Basin, Czechia

Abstract

The knowledge of the water volume stored in the snowpack, including its spatial distribution, is vital for many hydrological applications. Such information is useful for hydrological forecasts and it is often used for the calibration of snowmelt runoff models. Data from four field measurements of the snow water equivalent (SWE) carried out in two winter seasons were assessed by ten interpolation methods. Measurements from both snow accumulation and snowmelt periods were evaluated. The ability of methods to predict SWE at unmeasured locations was assessed by the means of cross validation. The best prediction accuracy of SWE was achieved by means of multiple a simple linear regressions, residual kriging and cokriging methods. The accuracy was enhanced by the use of elevation, aspect, slope and vegetation as variables in the calculation of the SWE. Elevation and vegetation show a significant correlation with the SWE in the study area. The multiple regression gave best results for snow accumulation period. However, the spatial variability of SWE was not successfully explained for snowmelt periods.