Estimating Lake Mulwala Diversions for calibration of a semi-distributed hydrologic model of the Murray River

In hydrologic model calibration, storage diversions are often neglected due to the lack of sufficient data. While it may make little difference to include diversions where the points of interest are far downstream of the diversions, the same is not true where the distances between the former and latter are small. In this paper, we develop a series of equations using least squares optimisation to estimate the flow diversions from Lake Mulwala on the Murray River to Mulwala Canal in the north and Yarrawonga Main Channel in the south. We then use the equations to estimate the diversions for times when historical observations are unavailable. This is to provide data for calibrating a hydrologic model of the section of the Murray River between Hume Dam and Lake Mulwala. The hydrologic model is developed using the Short-term Water Information and Forecasting Tools (SWIFT2) and is intended for generating forecasts of the hourly inflow to Lake Mulwala when provided rainfall forecasts. The results show the equations to be most accurate when predicting the diversions in summer when the flows are highest, and when predicting the diversion to Mulwala Canal, the larger of the two diversions. Further, results of the calibration and validation performance of the hydrologic model show including the Lake Mulwala diversions (as opposed to neglecting them as is often the case where real data are unavailable) substantially improves the model fit, thus increasing the model credibility.