Calibrations of a daily rainfall-runoff model with poor quality data

Abstract

A set of 99 rainfall-runoff data sets that were previously discarded because of poor quality are used to test several methods of enhancing the calibration of a daily rainfall-runoff model. The main enhancement occurred when rainfall and PET input data were scaled to improve the consistency with runoff, measured by the coefficient of efficiency based on monthly runoff. A common feature among the data sets is that the maximum error between actual and calculated monthly runoff occurs in the month of maximum recorded runoff, due to the difficulties in measuring rainfall and runoff in big storms. When the input data were adjusted and the month of maximum error omitted from the calculation, the average coefficient of efficiency of 68 data sets increased from 0.629 to 0.829. The calibrations of the other 31 data sets could not be enhanced because of transmission loss in 21 catchments with small rainfall and runoff, significant variations in 3 data sets over the period of record, and very poor quality data in 7 data sets. The methods gave substantial enhancement of calibrations in more than two-thirds of poor quality data sets, and offer potential for improvement of the calibrations of other daily rainfall-runoff models and even better quality data sets.