Design flood peak determination in the rural catchments of the Eastern Cape, South Africa

Includes bibliographical references.Rainfall is a natural occurring phenomenon, and is usually a welcome event, nourishing the earth and with it the fauna and flora. When the runoff is high, flooding occurs, leading to damage to the environment, property and even to loss of life. Flooding is becoming more common. The reasons for this are complex, and include social expansion, urbanization and may also result from global warming. These flooding events have significant implications to the engineering profession and the affected communities. The estimation of peak design floods is necessary for the planning and design of civil engineering projects. Over the past century standard methods for flood peak estimation have been developed for most countries, and are usually categorized in the literature as direct statistical analyses, regional statistical analyses, empirical methods and deterministic methods. Some of these methods are easy to apply, while others require an in-depth analysis of the catchment and other parameters. Each method has its limitations. In rural gauged catchments, design engineers in the workplace typically use statistical methods while in rural un-gauged catchments, they use empirical or deterministic methods, even although the reliability of these methods to estimate the design flood peak have never been verified in South Africa. The objective of this study was to identify the most reliable statistical, deterministic and empirical method(s) of flood peak determination in the rural catchments of the Eastern Cape, South Africa. In this investigation the recorded annual peak runoff from 18 river flow gauging stations in the Eastern Cape were statistically analysed using the statistical distributions commonly used in South Africa. These statistical analyses were used to establish a benchmark for evaluating the deterministic and empirical methods. The catchments of all the stations were then analysed using the deterministic and empirical methods. Finally, the empirical and deterministic methods were compared against the best-fit statistical method. This highlighted which empirical and deterministic method(s) under- and over-estimated peak floods when compared with the statistical analyses of recorded annual peak runoff. The finding from the statistical analyses was that the Log Pearson Type 3 (LP3) distribution performed the best, generally fitting the recorded data well. In the comparison of deterministic methods it was found that the Standard Design Flood (SDF) method was the most conservative deterministic method at the higher Recurrence Intervals (RIs) while the Rational Method-Alternative was the most conservative at the lower RIs. In the final comparison between the LP 3 distribution and the empirical and deterministic methods, it was found that in the higher RI range, the SDF estimated runoff values similar to that estimated by the LP3 distribution, while in the lower RI range, the Rational Method-Alternative variation proved to be the most consistent. The other deterministic methods generally under-estimated runoff values when compared to the LP3 distribution. Generally, the Regional Maximum Flood method appeared to have a RI about 1000 years, although it was as low as 1 :200 years in some of the smaller sized catchments. In rural catchments of all sizes in the Eastern Cape of SA, design engineers in the workplace should analyse a catchment using all of the statistical, deterministic and empirical methods available and then select the most conservative result