Abstract:
In this study, two methods of regional low-flow frequency analysis for Ping Wang Yom and Nanriver basins were presented. Flow records, ranging from 7 to 65-year length (up to year 1999), of 80selected gauging stations (catchment areas ranging from 11.6 to 45,851 square kilometers) wereused in the analysis. Initially, a comparison test of goodness of fit (by the Chi-Square, Kolmogorov-Smirnov, and Least Square methods) was made between the Log-Pearson Type III, Three-Parameter Weibull, Three-Parameter Log-Normal, Two-Parameter Weibull, Two-ParameterGamma and Gumbel distribution functions to identify the suitable type of distribution for low-flowanalysis in the studied basins. The Gamma theory with the Maximum Likelihood parameter estimation was found to give best fit with the observed data. The studied river basins were divided into 10 subregions of similar meteohydrological conditions. The regional low-flow frequency equations and the regional low-flow frequency curves were constructed for each subregion to estimate the 7-, 14-, 30-, 60-, 90- and 180-day low-flows for any return period at any location in the studied basins. It was found that the regional low-flow frequency curves for all durations in any subregion could be represented by a single dimensionless curve. From comparative tests, it was found that both proposed methods of low-flow regional frequency analysis gave comparable accurate result. For low-flow periods not greater than 30 days, the predictions were found to be reliable. For 7-day-10-year low-flow (7Q10), both methods gave acceptable results with standard errors less than one time of the actual results.