A regional frequency analysis of precipitation extremes in Mainland China with fuzzy c‐means and L‐moments approaches. (7th March 2017)

• Record Type: Journal Article
• Authors: Wang, Zhaoli; Zeng, Zhaoyang; Lai, Chengguang; Lin, Wenxin; Wu, Xushu; Chen, Xiaohong
• Is Part Of: International journal of climatology. Volume 37(2017)Supplement 1
• Page Start: 429
• Page End: 444
• Publication Date: 2017-03-07
• Subjects: precipitation extremes -- regional frequency analysis -- L‐moments method -- fuzzy c‐means -- Generalized Extreme Value distribution -- Mainland China; Climatology -- Periodicals; Climat -- Périodiques; Climatologie -- Périodiques; 551.605
• DOI: 10.1002/joc.5013
• Languages: English
• ISSNs: 0899-8418

Abstract

ABSTRACT: Owing to their essential influence on human society and natural environment exerted by inducing disasters, such as floods and droughts, further studies on precipitation extremes in China are needed. This study presents the regional frequency and spatial‐temporal patterns of precipitation extremes in China based on a high‐resolution (0.5° × 0.5°) daily precipitation dataset from 1961 to 2013. With fuzzy c‐means, L‐moments methods and other scientific statistical tests, a regional frequency analysis (RFA) is conducted, aiming to further understand the regional and spatial distribution of precipitation extremes across China. The results show that: (1) the whole Mainland China can be divided into 50 homogeneous regions on the basis of the characteristics of mean annual precipitation and location indices. (2) For most of the regions, Generalized Extreme Value (GEV), Generalized Normal (GNO) and Pearson type III (PE3) distributions of precipitation extremes fit well, according to the results of goodness‐of‐fit (GOF) test. (3) For RX1DAY, GEV has the best‐fit distribution in the east, northeast and southwest of China, whereas GNO distribution mostly fits the northern and parts of southwest and southeast; in addition, regions which fit PE3 and Generalized Logistic (GLO) distribute dispersedly across the country. (4) For RX5DAY, GEV mainly fits in the middle, southwestern and southern; GNO and PE3 apply best to the northeastern and northern, respectively. (5) Return periods of 20, 50 and 100 years for their best‐fit distributions decrease gradually from southeastern China to northwestern China. Compared with the results of GEV distribution fitted to each grid, RFA may provide more accurate estimates of rainfall quantiles. Definitely, the study results will not only benefit further understanding of the unique and complex features of extreme precipitation in the whole Mainland China but also contribute to the nation‐scale flood prevention, control and management in the backdrop of the changing climate.