Flood frequency analysis for the Brisbane River catchment

Flood is one of the worst natural hazards worldwide. During 2010-2011 Brisbane, Queensland, experienced one of the most damaging flood events in Australia's history. To minimise the flood risk, impact and damage, it is important to understand the cause of such floods and their frequency of occurrence. In this study, a flood frequency study is presented for the Brisbane River catchment of Queensland with the aim of finding suitable flood frequency distribution models that can provide accurate design flood estimates. A total of 26 stream gauging stations are selected with annual maximum flood series data ranging from 20 years to 91 years. In this study, a probability distribution model is selected using statistical tests and by graphical methods. Five probability distribution models – Log Normal, Log Pearson type III, Gumbel, generalised Pareto and generalized extreme value – are evaluated. EasyFit statistical software is used to assess the statistical goodness-of-fit of the selected probability distributions and to select the best-fit probability distribution. FLIKE software is used for quantile estimation based on the fitted distribution. It is found that Log Pearson type III is the most suitable probability distribution model, followed by Generalised Perato, for this study area