Integrating synthetic flood data for selection of regional frequency distribution

Abstract

Regional frequency distributions are estimated using respective regional average coefficients of variance, skewness and kurtosis. The regional average coefficients may be significantly biased for a region with inadequate hydrometric data. The objective of this study is to select suitable regional frequency distributions in previously identified five hydrologic homogeneous regions of Nepalese territory with larger focus on data-scarce regions. Incorporation of synthetic flood data has been proposed in two hydrologic regions which have insufficient hydrometric data. A SimHyd rainfall runoff model was employed for the generation of synthetic flood data to augment hydrometric data inadequacy. Selection of best-fit regional frequency distribution was identified using L-moment ratio diagrams and goodness-of-fit measure. The selected distributions were tested by comparing observation and regional distribution factors at eight observation hydrometric sites. Incorporation of synthetic flood data enabled to reduce regional distribution factor error to 6% from that of 10%.