Selected Effects of Field Variability on Predictions from a Soil Water Finite Difference Model

Known field variability of soil hydrologic parameters is incorporated into a general finite difference scheme for the prediction of near surface soil water conditions, including runoff. It is shown for the situations examined that the resolution of predictions obtained is relatively robust against ‘within soil group’ changes in the soil water retention characteristics, but that a significant deterioration in resolution occurs when all input parameters are subjected to stochastic sampling from specified distributions. Nevertheless, even under these relaxed conditions spatial probability fields of relative saturation exceedence retain good resolution. Runoff predictions are shown to exhibit low variation generally, and to accord well with the SCS curve prediction method, suggesting the possibility of using finite difference schemes incorporating field variability for ‘ungauged’ catchment modelling.