Numerical experiments on the sensitivity of runoff generation to the spatial variation of soil hydraulic properties

详细信息    查看全文

• 作者：Michael Herbst ; Bernd Diekkrü ; ger and Jan Vanderborght
• 关键词：Conditional stochasic simulation ; Soil mapping ; Macropore runoff ; Soil hydraulic properties ; Spatial variation ; Soil heterogeneity
• 刊名：Journal of Hydrology
• 出版年：2006
• 出版时间：15 July 2006
• 年：2006
• 卷：326
• 期：1-4
• 页码：43-58
• 全文大小：1318 K

文摘

Spatially distributed soil hydraulic properties are required for distributed hydrological modelling. These soil hydraulic properties are known to vary significantly in space, and considering the non-linearity of runoff generation, the question arises how the spatial variation of soil hydraulic parameters affects the continuous runoff modelling for a micro-scale catchment. This was analysed by applying a three-dimensional hydrological model to the 28.6 ha ‘Berrensiefen’ catchment, Germany, for a simulation period of one year. The model was based on an observed distribution of soil hydraulic properties, which were assumed to be layered in vertical and to vary continuously in horizontal direction, and validated for total runoff. Numerical experiments with five spatial distributions of soil hydraulic parameters derived from the observed spatial distribution, which was supposed to be the ‘true’ underlying spatial variation, were carried out. These five spatial concepts were: choropleth map, spatially homogeneous case, random distribution, stochastic simulation and conditional stochastic simulation. The comparative modelling revealed a significant sensitivity of runoff generation towards the spatial variation of soil hydraulic properties. The comparison of the hydrograph of surface and macropore runoff to the initial model runs exhibited the highest root mean square error with 1.3 mm h⁻¹ for the homogeneous case. Further we detected, that the frequency distribution of soil hydraulic properties played an important role for the reproduction of runoff amounts. But also the spatial topology (deterministic spatial variation) was relevant for an adequate description of runoff generation. Conditional stochastic simulation is seen as a promising approach, because it preserved both, the frequency distribution and the deterministic variation.