JP Calitz
PhD Paper 2
Selection of a suitable parent distribution for flood frequency analysis in South Africa
ABSTRACT
In South Africa flooding has a significant social and economic impact, as has been highlighted in recent flooding events. To minimise the impact of floods designers of hydraulic structures are required to estimate the risk associated for a given magnitude of a flood event. One method of estimation is Flood Frequency Analysis (FFA), which entails statistical analysis of historical streamflow data. Outliers and the selection of the Probability Distribution (PD) in FFA significantly impacts the magnitude of the estimated events. Several guidelines are available which recommend a number of PD’s, however, no detailed study on the selection of the best PD to use for design flood estimation has been published for for South Africa. To assess the suitability of the PD’s recommended in literature, Goodness-of-Fit (GoF) testing is undertaken on 411 Department of Water and Sanitation (DWS) flow gauges. This process entailed the collection and screening of flow data, Flood Frequency Analysis (FFA) and GoF testing. The results of the investigation identified that the General Pareto and Kappa distributions are well suited for design flood estimation in South Africa.