| Title |
Applicability of kappa Probability Distribution for Rainfall Regional Frequency Analysis of Extreme Rainfall in South Korea |
| Authors |
이가영(Lee, Gayoung) ; 신주영(Shin, Ju-young) ; 장상범(Jang, Sangbeom) ; 박지연(Park, Jiyeon) |
| DOI |
https://doi.org/10.12652/Ksce.2026.46.2.0127 |
| Keywords |
적합성 척도; 극치강우분석; 확률강우량; 확률분포형; 기후변화 Extreme rainfall; Goodness-of-fit measures; Rainfall quantiles; Probability distribution; Climate change |
| Abstract |
Regional frequency analysis (RFA) improves rainfall quantile estimation by pooling data from hydrologically homogeneous regions, providing higher reliability than at-site analysis in regions with a shortage of observed data. While the generalized extreme value (GEV) distribution is recommended as the preferred candidate distribution in the 2019 Ministry of Environment guidelines, a limited capacity to reproduce tail behavior has been noted due to the increasing intensity and localization of rainfall caused by climate change. This study evaluated the applicability of the four-parameter kappa distribution, which offers greater flexibility in presenting extreme tail behavior, as an alternative to the three-parameter GEV distribution. Among the 26 homogeneous regions, this study focused on Regions #12 (Gyeongbok) and #15 (Northern Gangwon), where all candidate distributions in the current version of the guideline were rejected. In Region #15, the kappa distribution satisfied the goodness-of-fit measures and adequately represented the statistical characteristics. However, parameters of the kappa distribution could not be estimated for some durations in Region #12. Furthermore, for the 50- and 100-year return periods, the kappa distribution yielded rainfall quantiles 20-30 % lower than GEV, suggesting a potential conservative bias in current GEV-based designs. These findings demonstrate that the kappa distribution is a viable alternative for responding to the statistical shifts in extreme precipitation. |