Estimation of Design Floods Using Continuous Simulation

A common problem encountered by catchment managers is the estimation of design peak flow quantiles. Current practice, as discussed by Pilgrim (1987), is based either on the use of statistical approaches such as an at-site flood frequency analysis, or the use of a catchment modelling system to simulate catchment response to design rainfall. For many practical problems, the option of an at-site flood frequency analysis is not available due to the absence of recorded data at the desired location. In these situations, the second alternative (catchment simulation) must be employed to generate the information necessary for estimation of the design peak flow quantiles. While there are many alternative approaches to the generation of information for design flood flow quantile estimation, one approach that has been proposed previously (see, for example, Cameron et al., 2001 and Droop and Broughton, 2003) is the use of continuous simulation techniques for generation of a flow sequence suitable for subsequent flood frequency analysis. As part of the revision of ARR, continuous simulation techniques for design flood estimation in urban areas have been investigated. The focus of these investigations has been on techniques for prediction of flood quantiles together with the uncertainty of these predictions. Sources of the uncertainty considered were the rainfall sequences, parameter uncertainty, and the calibration metric. Presented herein will be the results of this investigation.