Does the incorporation of process conceptualisation and tracer data improve the structure and performance of a simple rainfall-runoff model in a mesoscale catchment?

A geomorphological instantaneous unit hydrograph (GIUH) rainfall-runoff model was applied in a 31 km(2) montane catchment in Scotland. Modelling was based on flow path length distributions derived from a digital terrain model (DTM). The model was applied in two ways; a single landscape unit response based on the DTM alone, and a two-landscape unit response, which incorporated the distribution of saturated areas derived from field-validated geographic information system (GIS) analysis based on a DTM and soil maps. This was to test the hypothesis that incorporation of process-information would enhance the model performance. The model was applied with limited multiple event calibration to produce parameter sets which could be applied to a spectrum of events with contrasting characteristics and antecedent conditions. Gran alkalinity was used as a tracer to provide an additional objective measure for assessing model performance. The models captured the hydrological response dynamics of the catchment reasonably well. In general, the single landscape unit approach produced the best individual model performance statistics, though the two-landscape unit approach provided a range of models, which bracketed the storm hydrograph response more realistically. There was a tendency to over-predict the rising limb of the hydrograph, underestimate large storm event peaks and anticipate the hydrograph recession too rapidly. Most of these limitations could be explained by the simplistic assumptions embedded within the GIUH approach. The modelling also gave feasible predictions of stream water chemistry, though these could not be used as a basis for model rejection. Nevertheless, the study suggested that the approach has potential for prediction of hydrological response in ungauged montane headwater basins. Copyright (c) 2007 John Wiley & Sons, Ltd.