Design flood estimation using Monte Carlo Simulation and RORB Model: stochastic nature of RORB model parameters

Rainfall"based flood estimation method is often adopted when a complete design hydrograph is required and/or in the situations where the recorded streamflow data are not long enough to characterize the underlying at"site flood frequency distribution with sufficient accuracy. The Design Event Approach (DEA) is currently recommended rainfall"based flood estimation method in Australia according to Australian Rainfall and Runoff "” the national guide to flood estimation. However, DEA does not account for the probabilistic nature of the key flood producing variables except for the rainfall depth. This arbitrary treatment of key inputs and model parameters in DEA can lead to inconsistencies and significant bias in flood estimates for a given average recurrence interval (ARI). A significant improvement in design flood estimates can be achieved through a Joint Probability Approach (JPA), which is more holistic in nature that uses probability"distributed input variables/model parameters and their correlations to obtain probability"distributed flood output. More recently, there have been notable researches in Australia on Monte Carlo simulation technique (MCST) for flood estimation based on the principles of Joint Probability that can employ many of the commonly adopted flood estimation models and design data in Australia. Recently, the National Committee on Water Engineering in Australia has resolved that MCST should replace the DEA as the preferred method of flood hydrograph modeling in Australia. Based on the previous researches on MCST in Australia, the industry"based software URBS has integrated MCST within the software, which however, needs further enhancement for applications under a wide range of hydrologic and catchment conditions. Application of MCST with RORB model, the most widely used hydrologic model in Australia, has not been well investigated. At present, the RORB model has a limited capability in terms of implementation of the MCST in flood modeling. This paper investigates the applicability of the MCST with RORB model, in particular, this examines the probabilistic nature of key RORB model parameter kc and its impacts on design flood estimates. A large number of storm and runoff events were selected from Lismore catchment in New South Wales, Australia and values of kc, initial loss (IL) and continuing loss (CL) were estimated. Values of kc were categorized according to their goodness"of"fit in the FIT run and were validated using a number of independent storm events. Finally, peak flows were estimated for many combinations of kc, IL and CL values. In the RORB modeling, kc is considered to be a fixed parameter. It has been found that the value of kc may exhibit a high degree of variability, and they result in quite different flood peak estimates and hence it should be considered as a random variable in rainfall runoff modelling.