Enhancing the new intensity-frequency-duration (IFD) design rainfalls – sub-annual IFDs
Green, J and Xuereb, K and Jolly, C, Enhancing the new intensity-frequency-duration (IFD) design rainfalls - sub-annual IFDs, Proceedings of Hydrology and Water Resources Symposium 2014, HWRS 2014, 24-27 February 2014, Perth, Australia, pp. 605-612. ISBN 978-192210719-0 (2014) [Refereed Conference Paper]
Copyright 2014 Engineers Australia.
The Intensity-Frequency-Duration (IFD) revision project, part of the Australian Rainfall and Runoff (AR&R) Revision, has made available new IFDs for durations from one minute to seven days and for probabilities from 1 Exceedance per Year (EY) to 1% Annual Exceedance Probability (AEP). The new IFDs meet the design probability requirements of most infrastructure, however, for Water Sensitive Urban Design and some stormwater applications, IFDs for probabilities more frequent than 1 EY are required. In order to address this need, enhancements to the new IFDs to be provided over the next 12 to 18 months include the provision of estimates of sub-annual IFDs for probabilities of 2, 3, 4, 6, and 12 EY. However, as sub-annual IFDs have not been previously derived, it will be necessary to first develop the method to be adopted when deriving the sub-annual IFDs. To ensure consistency, the overall approach adopted for the sub-annual IFDs will be similar to that used for the new IFDs. However some changes will be necessary because of the increased frequency of occurrence. One difference will be the necessity to extract the Partial Duration Series (PDS) rather than the Annual Maximum Series (AMS). While the new IFDs used stations with at least nine values in the AMS, the number of exceedances per year to be used for IFDs more frequent than 1EY is expected to be at least 12. The PDS approach was explored previously and the Generalised Pareto distribution was found to provide the best fit for an average of three exceedances per year. However, the appropriateness of this distribution for 12 exceedances or more will need to be explored. Consistent with the new IFDs, the Bayesian Generalised Least Squares Regression (BGLSR) method will be applied to the sub-daily data to generate L-moments for sub-daily durations at the daily read stations. Regionalised L-moments will be derived using a similar approach to the regionalisation that was carried out for the new IFDs. These regionalised L-moments will then be used to estimate the parameters of the appropriate distribution to be fitted to the data. The parameters of the distribution will be gridded using ANUSPLIN and rainfall quantile estimates will be calculated for standard durations from one minute to seven days and probabilities of 2, 3, 4, 6 and 12 EY.
Refereed Conference Paper
design, least squares approximations, pareto principle, probability, rain, water resources