An investigation into the efficacy of Australian rainfall and runoff 2016 procedures in the mount lofty ranges, South Australia

This paper presents the results of an application of the 2016 Australian Rainfall and Runoff (ARR) procedures in 25 of the catchments in the Mount Lofty Ranges of South Australia that were used to derive those procedures. The paper compares the design peak flows derived from a RORB model with normal sub-area division to one that has a standard number (10) of equal area sub-areas, as was used in the LRRM model, and is used in the xprafts model to derive node flows. This showed that there was very little difference in peak flow estimates. The use of the simplified RORB model on the 25 catchments resulted in a significant time-saving, with little difference in predicted flows. The RORB model flows were compared with at-site flood frequency flows used to derive the regional flood frequency estimate (RFFE), and the RFFE itself. If the at-site flood frequency estimate is assumed to be the best estimate of flow then the RFFE estimate was reasonable, but the RORB estimate is very poor, particularly for frequent events. The RORB model estimates are worse also than those estimated using the 1987 version of Australian Rainfall and Runoff. The reasons for this are discussed, but it could be due to the application of the Monte Carlo approach to catchments producing small flows in frequent events. It is also possible that the RORB model cannot adequately describe the catchment response for the full range of Annual Exceedance Probabilities (AEPs) and that baseflow extraction and estimation for the range of AEPs is inadequate.