Assessing Flood Frequency Analysis Changes in ARR 2016 and the Impact on Creek Hydrology

Abstract

The 2016 revision of the Australian Rainfall and Runoff (ARR) guideline comprised major changes when compared to the 1987 revision. ARR 1987 was a staple guideline for the Australian engineering community and has been adopted by practitioners for stormwater and flood assessment over the past 30 years. The 2016 update therefore has the potential to impact previous, current and future flood study results. This study has focused on a specific change in ARR 2016 relating to flood frequency analysis. The flood frequency changes recommend applying the Generalised Extreme Value method with L-Moments, as it has been found to suit Australian catchments. To investigate this flood frequency method, a continuous simulation was developed for two Sunshine Coast catchments spanning a 130 year period in order to obtain peak flows. An analysis was then conducted with different distribution methods on both modelled data and observed streamflow data. The Log Pearson Type III method with Bayesian parameters was found to be the best overall fit for the modelled data, however, the Generalised Extreme Value method performed best for the observed streamflow data, and performed very well for the modelled data. The findings support the ARR 2016 recommendations for flood frequency distribution methods.