Estimation of Major Floods: Applicability of a Simple Probabilistic Model

Abstract

Estimation of major flood flows is needed in the design and operation of large water infrastructure. This paper presents a simple probabilistic model (PM) that can be used to derive “easy to apply” prediction equations for estimation of major flood flows. The proposed method assumes that the maximum observed flood data over a large number of sites in a region can be pooled together by accounting for the across-site variations in the mean and standard deviation values. The method is developed and tested in this paper using data from 227 catchments across Victoria and NSW. The application to ungauged catchments involves the development of prediction equations using generalised least squares regression for the mean and coefficient of variation of the annual maximum flood series as a function of catchment characteristics. An independent test shows that the PM provides quite reasonable design flood estimation in the study area for average recurrence intervals in the range of 20 to 200 years, with median relative error values (compared to at-site flood frequency estimates) in the range of 10% to 35%. The method is under further development, eg. consideration of the inter-station correlation in pooling the standardised data from the nearby stations and comparison of results with the rainfall-based methods. The method has the potential to derive regional flood prediction equations for major floods by pooling the flood data from all the Australian states.