A method for the regional estimation of runoff separation parameters for hydrological modelling

The generation and a subsequent separation of runoff into different components – such as surface flow or base flow – are key processes for an accurate simulation of catchment runoff with precipitation-runoff models. However, runoff separation parameters that determine the relative importance of the runoff components usually have to be calibrated with observed runoff, and therefore an application in ungauged basins is complicated. Also for monthly water balance models this is an important issue because of delayed flow due to base flow. We present a simple method for the regional estimation of runoff separation parameters of a spatially distributed, monthly water balance model. The method takes into account the base flow index, catchment characteristics (in this case soil type), and threshold processes with increasing runoff-depth. The method is used to establish the spatio-temporal water balance of whole Austria (84,000 km2), including ungauged basins. As a test, the runoff separation parameters are optimized locally in 140 catchments. The model performance for the simulation of runoff of the period 1961–1990 increases considerably from initial (lumped) to regional parameters, and only slightly from regional to optimized parameters – from 0.68, 0.74, to 0.77 for the mean Nash–Sutcliffe measure. The results are less reliable in catchments dominated by Karst formations, but an evaluation of the feasible parameter space shows that in all 140 catchments the regional parameter estimation at least points in the right direction. It is concluded that the diverse regional patterns of runoff in Austria – characterized by the relative importance of fast and slow runoff components – are described well with the presented method. Because of the generic approach, the method should also be useful for estimation of similar parameters of precipitation-runoff models operating on e.g. the daily time-scale.