Application of arr flike for at-site flood frequency analysis: A case study in New South Wales, Australia

Design flood estimation is needed to size hydraulic structures and for many other water resources management tasks. Flood frequency analysis is the preferred method for design flood estimation when reliable and longer period of streamflow data is available. In the new Australian Rainfall and Runoff (ARR) 2016, for at-Station flood frequency analysis, no particular probability distribution has been recommended; however, a greater emphasis has been provided on the log Pearson Type 3 (LP3) and Generalized Extreme Value (GEV) distributions. In this paper, 20 catchment stations (with high quality recorded streamflow data) have been selected from New South Wales (NSW) State in Australia and flood frequency analyses have been undertaken using ARR FLIKE software. Five different distributions (LP3, Log-Normal, GEV, Gumbel and Generalised Pareto) were fitted to the annual maximum flood data. It was found that there is no common distribution providing the best fit for all the selected stations. Overall, LP3 was found to be the best performer. It was also found that censoring of low flood values should be done properly in flood frequency analysis as this could make a remarkable difference in quantile estimates for smaller return periods in particular for LP3 and Log-normal distributions. For GEV distribution, censoring did not make a notable difference in quantile estimation as GEV uses L moments, which are less sensitive to outliers.