Estimation of flood volumes using stochastic rainfall losses : a case study for the NSW east coast

Flooding of volume-sensitive structures has potentially devastating consequences, therefore, it is critical to accurately model flood volumes and their uncertainties. In this paper, the performance of three loss models is assessed with regards to the flood volume. Simulations were undertaken using an industry-based hydrologic model (namely RORB) coupled with three separate loss models; the initial loss-continuing loss (IL-CL) model, initial loss-proportional loss (IL-PL) model and soil water balance model (SWMOD). Uncertainties in the design losses were then investigated using a Monte Carlo framework. The results show that both the IL-CL and IL-PL models have a tendency to underestimate flood volumes, and that both models also exhibit significant variability when treating losses stochastically. Overall, SWMOD performed quite well, with more accurate volume estimates in three of the four study catchments and substantially less variability than the other two loss models.