Effect of data length on rainfall-runoff modelling

A 64-year data set of daily rainfall and runoff, and average monthly potential evapotranspiration (PET) was split into subsets of 2, 5, 10, 20 and 30 years. Each subset was used to calibrate the AWBM daily rainfall-runoff model. Each subset calibration was then used to estimate runoff from the 64 years of rainfall and PET data. The ratios of calculated to actual total runoff were used to determine probabilities of error from the different lengths of data used for calibration. There was little difference in results from the 2 and 5 year subsets with 10% chance of error in estimating long term runoff in the range of -21% to +31%. The probabilities of overestimating long term runoff reduced with length of calibration data of 10+ years; however, the chances of underestimating were above 15% even with 30 years of calibration data. Some limited repetition of the calculations with the Curve Number rainfall-runoff model indicated that the error characteristics were inherent in the data set and not an artifact of the model used. The ramifications for applications of rainfall-runoff modelling are briefly discussed.