A monte carlo comparison of parametric and nonparametric estimation of low flow frequencies

Abstract

Low streamflow, which u one aspect of drought, constitutes one of the extremes of.the hydrological regime. Among the low flow characteristics of rivers, lowflow frequency analysis, that isfundamental to a wide range of design and operational problems in area of both water quality and quantity, is the one. Thu paper deals with the introduction of the nonpaiame~c methods in the low flow frequency analysis and then make comparative evaluation on the magnitude of low flow quantile corresponding to a ghien return period with the parametric statistics. A currently used approach to low flow frequency analysu is based on the assumption made that the distribution function describing the anntlal minimum lowflow data is known, which is never known exactly. Recently, nonparametric method of estimating probability distribution functions have been developed, which doesn't require a distributional assumption. This involves the use of a suitable smoothing function known as a kernel. The fIXed kernel nonparametric method is proposed and developed for estimating low flow quantiles. Based on annual minimum low flow data and Monte Carlo  SimulaJion Experiments, the proposed model is compared with Weibull models both for its descriptive and predicthie ability. Computed reSftlts showed that the fIXed kernel estimator has small bias and root mean square error in low '.flow quantile estimates. Applicalion of the model to data from the Blue Nile at E/deim (Sudan) and Komati (South Africa) rivers have shown that the nonparametric approach is viable alternathie to the Weibul/ modeu. It is, therefore, concluded that the nonparametric method u accurate, uniform, and particularly suitable for the muJtimodal data.