Parameterization of sub-grid effects in a large-scale hydrological model

The methods of parameterization of sub-grid effects and their effectiveness are investigated using a distributed physically-based model of the hydrological cycle for two river basins situated in the steppe-forest zone of Russia. To estimate the sub-grid effects, it is assumed that these values are stochastic fields and two-parameter statistical distributions are used to describe variations of these characteristics inside the grid domain. Different assumptions on the change of the coefficient of spatial variation depending on the size of grid domain and the magnitude of characteristic under consideration were tested. It is shown that the dependence of the spatial variance of snow cover on the size of area can be obtained on the basis of the hypothesis of statistical self-similarity and the application of this dependence can essentially improve snowmelt runoff modelling. The rainfall, runoff and basin-averaged evapotranspiration are sensitive to the procedure of accounting for the sub-grid variation of saturated hydraulic conductivity and stomatal resistance, but the scaling of these parameters has comparatively small influence on the hydrographs.