Currently the disasters of climate changes have increased. Frequency analysis has been performed to prevent the disaster that might come from certain climate change. The main purpose of frequency analysis is to estimate quantiles for a given data. In ...
Currently the disasters of climate changes have increased. Frequency analysis has been performed to prevent the disaster that might come from certain climate change. The main purpose of frequency analysis is to estimate quantiles for a given data. In this study, a comparative study was made on parametric and nonparametric methods of frequency analyses of rainfall quantile estimation. The most important thing for frequency analysis is to gather good data. The observation data are based on annual maximum rainfall series at substations in the Korea Methodology Adminstration(KMA). The frequency analysis can be mainly divided into two approaches; parametric and nonparametric methods. First, the method of parametric is to estimate quantiles by using any probability distributions. Second, the nonparametric method is not making the assumption of any distribution, in which quantile is estimated by using kernel density function.In this research, a comparative of analysis is performed to compare quantiles based on parametric and nonparametric methods. For parametric methods, both Gumble and GEV distributions are selected for the appropriate distributions. But the result of goodness-of-fit test shows that the GEV distribution is better than Gumbel one for annual maximum rainfall data. Although quantile estimation of the nonparametric method does not need the assumption of any distribution, the quantiles can be affected by choice of kernel function and bandwidth. As a result, quantile estimation of the nonparameric method is better than that of the parametric method with in interpolation range. On other hand quantile estimation of the nonparametric method can be underestimated in extrapolation. For further investigation, we would be interested more in the semiparametric method.