Entropy-based derivation of generalized distributions for hydrometeorological frequency analysis

JOURNAL OF HYDROLOGY

卷:557

Article

Chen, Lu¹  Singh, Vijay P.^2,3 

[1] Huazhong Univ Sci & Technol, Coll Hydropower & Informat Engn, Wuhan 430074, Hubei, Peoples R China

[2] Texas A&M Univ, TAMU, Dept Biol & Agr Engn, College Stn, TX 77843 USA

[3] Texas A&M Univ, TAMU, Dept Civil & Environm Engn, College Stn, TX 77843 USA

关键词: Entropy theory;    Generalized distributions;    Frequency analysis;    Hydrometeorological extremes;    Rainfall;   

DOI  :  10.1016/j.jhydrol.2017.12.066

来源: Elsevier

【 摘 要 】

Frequency analysis of hydrometeorological and hydrological extremes is needed for the design of hydraulic and civil infrastructure facilities as well as water resources management. A multitude of distributions have been employed for frequency analysis of these extremes. However, no single distribution has been accepted as a global standard. Employing the entropy theory, this study derived five generalized distributions for frequency analysis that used different kinds of information encoded as constraints. These distributions were the generalized gamma (GG), the generalized beta distribution of the second kind (GB2), and the Halphen type A distribution (Hal-A), Halphen type B distribution (Hal-B) and Halphen type inverse B distribution (Hal-IB), among which the GG and GB2 distribution were previously derived by Papalexiou and Koutsoyiannis (2012) and the Halphen family was first derived using entropy theory in this paper. The entropy theory allowed to estimate parameters of the distributions in terms of the constraints used for their derivation. The distributions were tested using extreme daily and hourly rainfall data. Results show that the root mean square error (RMSE) values were very small, which indicated that the five generalized distributions fitted the extreme rainfall data well. Among them, according to the Akaike information criterion (AIC) values, generally the GB2 and Halphen family gave a better fit. Therefore, those general distributions are one of the best choices for frequency analysis. The entropy based derivation led to a new way for frequency analysis of hydrometeorological extremes. (C) 2017 Elsevier B.V. All rights reserved.

【 授权许可】

Free