Efficient flood estimation using ARR 2016 temporal patterns in South East Queensland

Abstract

With the introduction of a range of temporal patterns in the 2016 edition of the Australian Rainfall and Runoff: A Guide to Flood Estimation (ARR), comes a challenge for modellers in the industry to define and implement the best methodology for determining the critical duration, peak discharge and temporal patterns for design storms. Previously for design storms, the ARR guideline recommended the use of a single temporal pattern for each of the 8 climate zones in Australia for a given average recurrence interval (ARI) and storm duration. As part of the ARR 2016 revision project, a national database for storm events was created using the Bureau of Meteorology pluviograph database that recommends ten temporal patterns for a given annual exceedance probability (AEP) and duration. This paper explores the opportunities and challenges facing the industry in terms of implementing the ARR 2016 temporal patterns and evaluates various methods for applying them for flood estimation. An evaluation of the accuracy and reliability of 1) the Watershed Bounded Network Model (WBNM) using ARR 1987 and 2016, 2) the Rational Method and 3) the Regional Flood Frequency Estimation (RFFE) model in comparison with the gauged data was undertaken to gain an understanding of which method, or combination of methods, would produce estimations that were most reliable. From this evaluation, the difference in the magnitude of peak discharge estimation between ARR 1987 and 2016 was found to be significantly large and primarily caused by the differences in the Intensity-Frequency-Duration (IFD) curves from ARR 1987 and 2016. Gauged and ungauged catchments in rural south east Queensland were selected for this study. A combination of iWBNM and the beta release of Storm Injector was used to run the WBNM. This method was efficient and allowed models to be run using both the ARR 1987 and 2016 temporal patterns and IFD data. Comparisons of the WBNM models using ARR 1987 and 2016 data showed that for ungauged catchments, 1987 data estimated peak discharges up to 48% higher than those estimated using 2016 data. Further, estimated peak discharge using the Rational Method was on average 80% higher than ARR 2016 estimations. For the gauged catchment, peak flow estimates using the WBNM were mixed, with results showing increases of up to 17% and decreases of 5% between ARR 1987 and ARR 2016 estimations for different AEPs. RFFE results varied greatly among the catchments, and estimated peak discharges were 12 and 71% higher than those using the WBNM and ARR 2016 data.