Spatial Analysis of Meteorological Drought Return Periods in China Using Copulas

Xiong-Fei Liu,Shi-Xin Wang,Yi Zhou,Fu-Tao Wang,Guang Yang,Wen-Liang Liu

DOI: https://doi.org/10.1007/s11069-015-1972-7

IF: 3.158

2015-01-01

Natural Hazards

Abstract:China is considered to be one of the most vulnerable drought-prone countries in the world, and it has recently suffered many severe droughts with large economic and societal losses. Drought events in China have been extracted using run theory based on the Standardized Precipitation Evapotranspiration Index, which covers the period 1961–2013 across 810 stations. The drought events are characterized by three variables: duration, severity and peak. Exponential, Weibull and Pareto functions are then selected to describe the marginal distribution of duration, severity and peak, respectively. The Gumbel–Hougaard Copula was used to construct the joint distribution of Duration–Severity and Duration–Peak, while the Clayton Copula and the Gaussian Copula are used to construct the joint distribution of Severity–Peak and Duration–Severity–Peak, respectively. The results indicate that the return period is dependent on spatial location, variable type and the combination of variables. For extreme droughts, trivariate ‘and’ return periods are longer, with an average of 42.1 years. The short return period is mainly distributed in southern China, especially on the border between Sichuan and Yunnan, the coastal regions of Guangdong, western Hunan and northern Jiangxi. Studies on the identification of spatial distributions of drought return periods across China have therefore been undertaken for drought mitigation and strategy planning.