A Study on Estimation of Quantile Using Regional Scaling Factors

Tuesday, 16 December 2014
Younghun Jung, Sunghun Kim, Hanjin Jang and Jun-Haeng Heo, Yonsei University, Seoul, South Korea
The estimation of design rainfall is generally done by applying a regional frequency analysis that uses hydrologic data from a large number of rainfall sites in the region. In this study, index flood method was applied to estimate the quantile of annual maximum rainfall data in Han River basin using cluster analysis. It is important to determine hydrologically homogeneous regions using cluster analyses in regional frequency analysis. The cluster analysis is performed by using precipitation related variables at 104 rainfall observation stations in Han-River basin and regional frequency analysis is applied to estimate rainfall quantile using annual maximum rainfall data of Han River Flood Control Office. Many variables relevant to precipitation can be used for grouping a region in regional frequency analysis. For the regionalization of Han River basin, 4 cluster analyses such as average linked method, WARD method, two-step method, and k-means method are applied for grouping regions by using the results of factor analysis and variables of meteorology and geomorphology. The results of cluster analysis showed that mean annual precipitation, seasonal mean precipitation and geographic information were the main variables affecting to decide the precipitation region. The study area was divided into four regions based on the k-means method of clustering approach. In the final step of the regional scaling-factors analysis, scaling is used to evaluate the accuracy of regional scaling factors. And then, rainfall quantiles by regional scaling factors are compared based on the applied index flood method for rainfall data of Han River basin in South Korea. The relative root mean square error (RRMSE) measure was used to evaluate the accuracy of index flood method and estimation of quantile using regional scaling factors method.