ABSTRACT: A lumped-parameter rainfall-runoff model, IHACRES, has been used to predict the long-term natural variability of runoff from approximately 100 years of daily rainfall and evaporation data for eight different catchments. The model efficiently encapsulates the response dynamics of a catchment and is a good predictor of stream discharge. It is well suited to illustrate the factors influencing the quality of stream discharge predictions, in terms of rainfall-runoff model fits to daily discharge in calibration and validation mode and to the flow duration curve. The predominant factors are rain gauge density, stream gauge rating quality, catchment response dynamics (especially slowflow/baseflow volume) and the sampling interval of rainfall discharge (always daily here). These factors manifest themselves to sufficiently different degrees in the range of catchments studied to reveal a useful appreciation of their individual and combined contributions to the quality of streamflow prediction. The predictive benefits of improvements in rain gauge coverage or in stream discharge rating are indicated. A large proportion of slowflow/baseflow in a catchment is a factor which can compensate for data quality problems. The analysis and findings in this paper can be used to improve rainfall-runoff model performance and the synthesis of long-term stream discharge records in a large number and variety of catchments.
Environmental Software. 11:193-202.