The effect of spatial rainfall variability on streamflow prediction for a south-eastern Australian catchment

Abstract

Spatial rainfall variability is considered an important factor affecting the accuracy of streamflow prediction. How does spatial rainfall variability affect catchment water balance? Can information of spatial rainfall variability improve streamflow prediction? This study was conducted to answer these questions. The modelling experiment was carried out using daily climate and streamflow data from a typical high yielding catchment in south-eastern Australia using the semi-distributed VIC (Variable Infiltration Capacity) model. Different spatial rainfall variability is generated and used as input into the VIC model to assess the sensitivity of streamflow to the spatial rainfall input. The VIC model was first driven with 0.05° gridded (distributed) daily rainfall series and calibrated against the observed daily streamflow data. The calibrated VIC model was used to simulate streamflow using both the distributed rainfall with different generated spatial rainfall variability and catchment-averaged rainfall. The effects of spatial rainfall variability on the streamflow prediction were assessed using a modified coefficient of efficiency and volumetric bias by comparing the simulated streamflow using the distributed rainfall and using the catchment-averaged rainfall. The differences in the streamflow volume and the daily streamflow distribution (daily flow duration curve) were assessed. The importance of accounting for spatial rainfall variability under different antecedent catchment condition (dry versus wet) was also assessed. The modelling experiments show that for small spatial rainfall variability, modelling using the distributed rainfall and the catchment-averaged rainfall gives very similar results. For large spatial rainfall variability, not accounting for the spatial rainfall variability will lead to an underestimation of the total streamflow volume and the entire daily streamflow distribution. The modelling results also show that under wet antecedent catchment conditions and/or high rainfall events, the spatial runoff distribution is controlled mainly by the spatial rainfall falling on catchment, and under dry antecedent conditions, the spatial runoff across the catchment is controlled by the soil moisture distribution across the catchment.