Analysis of the suitability of the German soil texture classification for the regional scale application of physical based hydrological model

Abstract. Regional scale hydrological simulations are mostly based on the use of standard data sets such as soil maps which are based on soil texture classification schemes. This paper analyses the suitability of the German soil texture classification for the application of a physically based soil-vegetation-atmosphere-transfer scheme. Theoretical soil columns are defined to be able to represent the entire soil texture triangle by a 1% grid of the three particle size classes: sand, clay and silt. These theoretical soil columns are characterized by a homogenous soil texture and consist of two layers of increasing bulk density and decreasing content of organic matter with depth. Soil hydraulic parameterisation is derived by applying a pedotransfer function. Continuous water balance calculations are carried out for a ten year period for all grid cells of the 1% grid. The results of the water balance calculations are compared to the simulation results of the centre of gravity of the respective soil texture class. Texture class specific mean deviations and root mean squared deviations are calculated from the differences between the 1% pixels and texture class representatives. The results reveal that the loam and silt texture classes show only small deviations from the centres of gravity. For a few sand texture classes and most of the clay texture classes deviations are considerably large. Assuming an equal distributed probability of occurrence of all realisations within a soil texture class, an uncertainty of more than 100 mm/a with respect to runoff and actual evapotranspiration is detected for four clay texture classes, two sand texture classes and one silt texture class. These results are confirmed by a sensitivity analysis investigating the model response for a grid cell compared to the neighboured grid cells. High sensitivities mainly appear for sandy and clayey soils while the sensitivity of the model for loam and silt soils is smaller. Resuming it can be stated that most of the texture classes of the German texture classification scheme are suitable for the application of a physically based model, on regional scale in particular. Clay texture classes can be expected to cause high simulation uncertainties.