Investigation into probabilistic losses for design flood estimation : a case study for the Orara River catchment, NSW

Australian Rainfall & Runoff (Pilgrim, 1987) recommends the design event approach (DEA) as the preferred method for estimating design flood hydrographs, in which a single design event is adopted. More recently, Monte Carlo simulation has been used to allow for the probabilistic nature of input variables in flood modelling. This paper adopts a Monte Carlo framework to evaluate the impact of probabilistic losses on design flood estimates for the Orara River catchment in northeastern NSW. A RORB runoff routing model was used to derive loss values for both the initial loss-continuing loss (IL-CL) and initial loss-proportional loss (IL-PL) models. It has been found that the initial, continuing and proportional losses can be approximated by the Gamma, Weibull and Beta distributions, respectively. When these distributions were compared with non-parametric distributions, differences in the flood estimates were found to be minimal. Another finding was that peak floods estimated using the DEA were more biased for the IL-CL model, than for the IL-PL model. In comparison to the at-site flood frequency curve the IL-CL model produced an overall better fit of the shape of the curve, however, the IL-PL model provided a better fit to the observed flood peaks for mid-range events.