An assessment of effective land surface parameterisation in regional-scale water balance studies

Abstract

Numerical experiments have been carried out with a soil–vegetation–atmosphere transfer (SVAT) model to study the impact of spatial variability in soil and land surface parameters on regional-scale water balance components. A statistical-dynamical approach has been used to account for the spatial variability of the selected parameters and to determine the seasonal evolution of the impact on the water budget. Point data are used to derive probability-density functions for the thickness of the permeable top soil layer, coefficients for the water retention and hydraulic conductivity curves, the leaf area index and the minimum stomatal resistance. These distributions are incorporated in the model’s mathematical framework in order to generate univariate distributions (sensitivity patterns) for evaporation, transpiration, total evaporation and runoff. The means of these univariate distributions of outputs yield catchment-scale averages. Next, the study obtains catchment-scale evaporation estimates by running simulations with aggregated parameters obtained as statistical descriptors of parameter distributions. The difference between the catchment-scale averages and values obtained with aggregated parameters describes the non-linear response of the model to spatial variability of the particular parameter. Finally, the study also investigates several effective parameters based on recently described hydrometeorological aggregation rules. Results show significant differences in sensitivity patterns between individual parameters and between seasons.