Changes in Australian rainfall runoff and its implication for estimating design rainfall

Recently, Geoscience Australia has released updated national guidelines for the estimation of design floods in Australia, commonly known as Australian Rainfall Runoff (ARR). The methodologies and guidelines proposed in ARR are crucial for an accurate estimation of flood risk and for the design of safe and sustainable infrastructures in Australia. The newly proposed ARR (ARR2016) has adopted new methods and data compared to old ARR (ARR1987), which resulted differences in the estimation of design rainfall across Australia. For example, in ARR2016 additional 30 years' rainfall data and 2300 extra rainfall stations have been included compared to ARR1987. Rainfall frequency analysis has been conducted by Generalised Extreme Value (GEV) distribution compared to Log-person Type III (LPIII), previously adopted in ARR1987. Bayesian Generalised Least Square Regression (BGLSR) is adopted in ARR2016 to predict daily rainfall from sub-daily rainfall statistics, which was previously done by Principal Component Analysis (PCA). In this paper, we review the guidelines for both ARR2016 and ARR1987 for design rainfall estimation and evaluate changes in design rainfall for selected stations in 8 major cities in Australia including Adelaide, Brisbane, Canberra, Darwin, Hobart, Melbourne, Sydney and Perth. Results presented in the paper will help Engineers and Managers of local governments to understand and implement new regulations proposed in ARR2016 for the estimation of design rainfall.