Estimating generalised flood sew ceofficients for penisular Malaysia

Abstract

Flood estimations based on fitting the frequency of occurrence of annual peak discharges using the Log-Pearson Type 3 distribution are commonly used but they are sensitive to the skew coefficients of the gauging stations. The estimation accuracy can be improved by using a weighted average population skew coefficient calculated from the sample station skew and the generalised unbiased skew. The U.S. Water Resources Council (WRC) has documented guidelines for estimating the generalised skew coefficients and published a map of generalised skew values for the United States. The map shows isolines of skew coefficient values and the average skew coefficient for each 1-degree quadrangle of latitude and longitude for the United States. Following the WRC guidelines, many of the state authorities in the US have developed the generalised skew coefficients separately on a state/regional basis. In Malaysia, the Log Pearson Type 3 distribution has been widely used for flood peak estimation but there are no guidelines available for estimating the generalised skew coefficients for use in conjunction with the distribution and as more accurate results can be obtained if these data are available, it is clear that a regional flood skew study is needed. With the regional skew data available, the peak flow can be simply and easily calculated with the aid of a software such as HEC-SSP. The aim of this paper is to use the WRC guidelines to derive the generalised skew coefficients using the peak annual discharge data of Peninsular Malaysia for general use. The WRC recommended several techniques for estimating and evaluating generalised skew of the Log-Pearson Type 3 distribution for the annual peak discharges. Station skews (skew coefficients computed from gauging station records) and unbiased and weighted skews derived from these station skews are to be used to develop these techniques. In this study, peak discharge records at 66 gauging stations having 16 or more annual peak discharges in Peninsular Malaysia were selected for computing station skews. Station skew values ranged from -0.831(log10 unit) to 1.475 (log10 unit). The three techniques recommended by WRC used for estimating the generalised skew of annual peak discharges were adopted for this study. These methods are: (1) An isoline map, (2) a prediction equation (3) a regional mean skew. Attempts to develop a prediction equation were unsuccessful. An error analysis showed that the regional mean skew method has a lower MSE (mean square error) than that obtained from the state wide generalised skew coefficient contour map. As a result, the mean station skew for the selected gauging stations can be used to estimate the generalised skew for any gauging site in the peninsula. The mean skew is -0.022 (log10 unit) and the associated mean square error is 0.05 (log10 unit).