Application of Monte Carlo simulation technique to design flood estimation : a case study for the Orara River catchment in New South Wales

![CDATA[Design of water infrastructure projects such as bridge and embankment require design flood estimation. Ideally, design flood estimation is done using at-site flood frequency analysis. However, in many cases, the flood estimation needs to be done at ungauged locations. Methods often adopted for this task such as index flood, rational and regression-based methods are limited to peak flow estimation. As a result, these methods are not suitable in the estimation of complete streamflow hydrograph. At present, Design Event Approach (DEA) is the method currently recommended in Australia for the estimation of design flood hydrograph. However, this method has serious limitations as it treats the rainfall depth as the only random variable in modelling, while other inputs (e.g. temporal patterns and losses) are kept fixed. Joint Probability Approach (JPA)/Monte Carlo Simulation Technique (MCST) is a holistic approach that has proven to overcome some of the limitations associated with the DEA. However, this method has not been tested to a wider hydrologic and catchment conditions. For wider application of the JPA/MCST, one needs readily available regional design data such as stochastic rainfall duration, temporal patterns and losses in the rainfall runoff modelling. This paper presents regionalisation of the model inputs/parameters (i.e. rainfall duration, inter-event duration, intensity-frequency-duration, temporal patterns, initial loss, continuing loss and runoff routing model storage delay parameter) to the JPA/MCST for the State of New South Wales (NSW). This study uses data from 86 pluviograph stations and six catchments from NSW to regionalise the distributions of the input variables and runoff routing model storage delay parameter for the application with the JPA/MCST. A test catchment (the Orara River) is used to test the applicability of the regionalised JPA/MCST in flood estimation. In this study, complete storm events are selected from the selected pluviograph stations. The selected complete storms are then analysed to derive rainfall complete storm duration (DCS), inter-event duration (IED), intensity-frequency-duration (IFD) data and temporal patterns (TP). In addition, concurrent rainfall and streamflow events data are used to derive values of initial loss (IL), continuing loss (CL) and runoff routing model storage delay parameter (k). The DCS, IED, IL, CL and k data are described by probability distributions, in that three goodness-of-fit tests are applied (i.e. the Chi-Squared test, Kolmogorov-Smirnov test and Anderson-Darling test). The fitted probability distributions for these variables are then used to specify the regional stochastic inputs in the application of JPA/MCST in NSW State. The spatial proximity method is adopted in the regionalisation of the DCS, IED, IFD and TP data, i.e. a distribution at an arbitrary location is determined by using a number of nearby gauged stations’ data. To regionalise DCS, IED and IFD, data from pluviograph stations within 30 km from the catchments’ centre were considered with the aid of inverse distance weighted averaging method. For the TP data regionalisation, a maximum of 20 nearest pluviograph stations within 200 km were used. Here, it has been found that DCS, IED, IL and k data can be approximated by gamma distribution and the CL data by exponential distribution. These regionalised stochastic inputs have been applied to the Orara River catchment in NSW. The derived flood frequency curves from the regionalised JPA/MCST method have been compared with the results from the DEA; it has been found the adopted method showed better results than the DEA generally. However, it should also be noted here that the applicability of the regionalised JPA/MCST needs to be tested to some additional catchments, which is being carried out as part of the on-going research and will be reported in future publications. Although, the method and design data developed in this study are primarily applicable to the eastern part of NSW, it can be adapted to other States of Australia and other countries.]]