Estimation of Extreme Value Distribution and Return Levels for Annual Maximum of Daily Rainfall Amounts

• 발행기관: 한국자료분석학회
• 저자명: Xianglong Liang Keon Tae Sohn
• 간행물 정보: 『Journal of The Korean Data Analysis Society (JKDAS)』Vol.15 No.5, 2297~2305쪽, 전체 9쪽
• 주제분류: 자연과학 > 통계학
• 파일형태: PDF
• 발행일자: 2013.10.30

영문 초록

This study aims to estimate parameters and return levels of extreme value distributions and to examine dynamic features of return levels for annual maxima of daily rainfall amounts in South Korea. For this study, we used daily rainfall amounts in Seoul and Busan. Four kinds of extreme value distributions (GEVD, GPD, POT and Poisson-GPD) were applied to historical data and compared for each city separately. According to goodness-of-fit test and the estimation results of return levels, Gumbel-type GEVD was selected as the optimal distribution. Dynamic features of parameters and return levels of Gumbel-type GEVD were examined based on the moving-window method with shifting 30-year data period by one year separately. The estimates of parameters and return levels for Busan seem to be stable but those for Seoul are increasing.

목차

1. Introduction

2. Extreme value distributions and their return levels

3. Data

4. Estimation of return levels of daily rainfall amounts

5. Dynamic features of Return Levels

6. Concluding Remarks

References

키워드

daily rainfall amounts GEVD return level moving-window