Resumen
Soil water simulations on hydrological meso- or macroscale require parameters that describe the physical characteristics of the soil. At these scales, information regarding soil properties is mostly only available on very coarse spatial resolutions with texture based soil characterisations, where it is difficult to select representative soil hydraulic parameters. We improved the parameter estimation by introducing a new soil classification system, which is based on soil hydraulic behaviour in order to realistically reproduce the soil water interaction within meso-scaled hydrological models. The time series of soil water flux were simulated based on one million different parameterisations, which were then utilised for similarity analyses while applying the k-means clustering. The resulting classes show a different pattern when compared to the United States Department of Agriculture (USDA) texture based classes. Representative time series of water flux representative of the new classes were compared to time series of the USDA texture classification. The new classes show remarkably lower uncertainties. The bandwidth of the time series within a class is orders of magnitudes higher for the USDA system when compared to the new system. The evaluation of similarity of the simulated water flux time series within one and the same class were also clearly better for the new system.