The drought frequency analysis mainly used the method of deriving and analyzing the return period through the univariate probability distribution of each drought duration and severity. To compensate for the shortcomings of the univariate drought frequency analysis and to determine the degree of drought sensation based on water demand, a bivariate drought frequency analysis was performed using the Copula model that considers the duration and severity at the same time. For the target areas, Soyanggang Dam and Chungju Dam, which are multi-purpose dams in Korea, were selected and the daily inflow was used. The threshold was set using the percentage of the cumulative normal distribution of the dam inflow, and the duration of the drought and the accumulated water deficit were calculated accordingly. A univariate probability distribution function, which is a log-normal distribution and GEV distribution for each reference flow rate, was applied through the fit test of the duration and accumulated water deficit. Then, a bivariate frequency analysis using the Copula model was performed. As a major result, the most severe drought was evaluated to have a return period of 141.19 years in ’96 based on the ordinary water level in Soyanggang Dam and 184.19 years in ’14 based on the ordinary water level in Chungju Dam.
.jpg "Boosik Kang, PhD photo")