Development of an improved flood frequency curve applying Bulletin 17B guidelines

Abstract

Flood frequency curves for gauging stations are developed by fitting the logarithms of annual peak flows to a Pearson Type 3 distribution. Such a curve is sensitive to the mean, standard deviation and skew coefficient computed from the logarithms of the annual peak flows. If a gauging station is identified as a short record site then adjustments of the mean, standard deviation and skewness are required as suggested in Bulletin 17B, developed by the Hydrology Subcommittee, Office of Water Data Coordination, U.S. Geological Survey. The mean and standard deviation can be improved by undertaking a regression analysis for the short record site with a nearby long record site. The estimate of the station's skew coefficient can be improved if it is weighted by the regional skew coefficient. In this study, annual peak flow data of 80 gauging stations in Western Australia are analysed to estimate the regional skew coefficient for the area of interest. The regional skew coefficient estimation technique applied in this study is rudimentary which does not follow the complete instruction of the Bulletin 17B. The improvement of the mean and standard deviation is done by following the complete instruction provided in Bulletin 17B. A software tool is developed using MATLAB to carry out the current study. This tool is GUI-enabled which takes flood data from the users and generates flood frequency curves based on the two station method mentioned in the Bulletin 17B. A flood frequency curve is developed for the Kentish Farm gauging station (614005) using this software tool. The Mundlimup gauging station (614073) is used as a long record site for the adjustments of the mean and standard deviation to improve the flood frequency estimates.