Application of bivariate frequency analysis for estimating design rainfalls

Abstract

Univariate frequency analyses are widely used in practical hydrologic design. However, a storm event is characterized by storm amount, peak intensity, and storm duration. To fully understand these characteristics and to use them appropriately in hydrologic design, a multivariate statistical approach is necessary. This study applied Gumbel mixed model to bivatiate storm frequency analysis using hourly rainfall data collected for 34 years at the Jecheon rainfall gauge station in Korea. This study estimated bivariate return periods of a storm such as joint return periods and conditional return periods based on the estimation of joint cumulative distribution functions of storm characteristics.