多变量干旱事件识别与频率计算方法

Drought identification and drought frequency analysis based on multiple variables

干旱持续时间久、影响范围大,时空连续性是干旱的基本特征,以往研究大多考虑单变量或双变量。通过给定阈值识别干旱斑块和判断两相邻时间干旱的连续性,提出了时空连续的干旱事件三维识别方法,用干旱历时、干旱面积、干旱烈度、干旱强度和干旱中心位置5个特征变量对一场干旱事件进行度量;提出了基于Copula函数的干旱历时-面积-烈度三变量频率分析方法。以中国西南地区为例,采用SPI(Standard Precipitation Index)干旱指标识别了近52年发生历时等于或大于3个月的干旱事件,一共78场,其中2009年8月至2010年6月最严重干旱事件的重现期为94年一遇。通过比较概率分布函数和Copula函数,表明在干旱频率分析时需要考虑干旱历时、面积、烈度3个特征变量。

Droughts have such essential features as long duration,large scope of influence,and spatio-temporal continuity,and a single variable or double variables were mostly considered in previous studies. By identifying arid patches and judging the continuity between two adjacent droughts based on the given threshold,a spatio-temporally continuous 3D identification method has been proposed for drought events,measuring a drought event in terms of 5 characteristic variables of duration,area,severity,intensity,and central location;the duration-area-severity frequency analysis method based on Copula function has been proposed. Taking Southwest China as an example,SPI drought indices are used to identify the drought events that lasted 3 months or longer in recent 52 years,which totaled 78. Among them,the most serious event from Aug. 2009 to Jun. 2010 has a recurrence period of 94 years. The comparison between the probability distribution function and the Copula function indicates that the 3 characteristic variables of duration,area and severity shall be simultaneously considered upon frequency analysis of drought events.