The Use of Forecast Rainfall for Dam Operations in Extreme Flood Events

The use of forecast information in the operation of dams has proved a controversial issue in Australia, following the Brisbane River flood of 2011 and the subsequent Queensland Flood Commission of Inquiry. A persistent group of dam engineers have suggested that the simplistic use of forecast rainfall would have lead to better operational outcomes. This view is rarely based on detailed modelling or operational testing but on a belief that the use of a rainfall forecast will result in an improved starting position for each operational decision. The uncertainty in forecasts and how this affects the operation decisions has never properly been investigated.

When the key operational objective is reducing peak flow, mitigation is optimized when the maximum amount of operational air space or temporary flood storage is used to attenuate the peak of the inflow hydrograph. When forecast rainfall data is used it is assumed that using this information will result in each operational decision happening earlier and closer to the optimal position that would have occurred with a perfect forecast.

This paper investigates the operational value of the use of forecast rainfall information and the asymmetrical influence of forecast error on operational outcomes. The 2011 Brisbane River Wivenhoe event is used as the basis for this investigation. To investigate this issue a large set of variations for the 2011 event are modelled. The uncertainty in the forecast is explored by first using the difference between predicted and observed and then by applying the different prediction errors based on other large events.