Statistical self-similarity of spatial variations of snow cover and its application for modelling snowmelt runoff generation in basins with a sparse snow measurement network

Abstract An analysis of snow cover measurement data in a number of physiographic regions and landscapes has shown that fields of snow cover characteristics exhibit statistical self-similarity property. This property is useful when, because of a sparse measurement network, the spatial variability of snow cover can be determined only for large enough basins. Small-scale variability of snow cover can be estimated by scaling of the spatial variance determined for a large basin. A physically-based distributed model of snowmelt runoff generation developed for the Kolyma River and the Sosna River basins has been used to estimate the sensitivity of snowmelt dynamics and flood hydrographs to scaling of maximum snow water equivalent variance. It was shown that this scaling allows improvement of the description of snowmelt dynamics both within small areas and over the entire river basin. The flood hydrographs appeared to be sensitive to scaling of snow water equivalent mainly for small river basins and at certain hydrometeorological conditions.