Estimation of Extreme Floods with Particular Reference to Southern Africa

Abstract

The prediction of extreme floods associated with different recurrence intervals has always been a difficult task. Even the alternative of estimating maximum probable floods is fraught with uncertainty. The methodology is not yet standardized and the datums and factors to consider are not all incorporated in any single procedure for estimating these floods. The estimates are even more difficult in areas with limited data. Developing areas in particular have sparse hydrological networks and the spectrums of floods gauged in these areas provide limited, if not at times misleading, information. Rain data are often more reliable and more extensive and therefore many methods for estimating floods use rainfall data. The paper compares alternative methods such as the rational method, unit hydrograph method, soil conservation method, and empirical methods. Owing to wide variations in topography and climate over large catchments it is demonstrated that many of these methods cannot be scaled up or extrapolated. It appears catchment modelling using hydraulic techniques is the most reliable where extrapolation is required. Various empirical methods of estimating extreme floods are compared and it is shown that for large catchments available data illustrates the floods converge to common figures whichever data are used.