新疆气象干旱过程识别及多维变量联合分布特征分析

Identification of drought process and analysis of multi-dimensional variable joint distribution in Xinjiang

【目的】新疆是我国典型的干旱区,境内干旱灾害频发,缓解其严峻的干旱形势迫在眉睫;为揭示新疆气象干旱特征并探究其干旱重现期,【方法】采用1960-2021年新疆60个气象站点逐日气象数据,选取气象干旱综合指数(MCI)作为干旱评价指标进行干旱等级的划分,基于游程理论识别干旱过程并提取3个典型干旱特征变量,分析新疆近62年来发生干旱过程的时间变化趋势和各等级干旱日数的空间分布特征,通过优选得出的Copula函数构建干旱特征变量的二维、三维联合分布函数模型,推求干旱重现期。【结果】结果表明:在年际变化上新疆干旱过程数量呈现减少趋势,1960-2021年北疆发生干旱过程总数较南疆高,在春季新疆易发生干旱过程,且易发生春夏连旱;随着干旱等级的增加,干旱日数显著减少,旱情呈现出北疆重于南疆的空间分布特征;广义极值分布函数能较好的拟合干旱强度和干旱历时,威布尔分布函数拟合干旱峰值的效果最佳,Gumbel Copula函数拟合干旱强度-历时的二维联合分布时效果较好,Clayton Copula函数在干旱历时-峰值和强度-峰值的二维联合分布及三维联合分布中拟合效果最佳。【结论】Copula函数能较好地描述新疆多维干旱特征变量的联合分布,新疆发生长历时、高强度的极端干旱事件的概率相对较小,发生短历时、低强度的干旱过程概率相对较大,研究成果可为干旱风险评估提供技术支持。

[Objective]As a typical arid area in China, Xinjiang has frequent drought disasters and it is urgent to alleviate the severe drought situation. In order to reveal the characteristics of meteorological drought in Xinjiang and explore the recurrence period of drought, [Methods]Using the daily meteorological data of 60 meteorological stations in Xinjiang from 1960 to 2021, the meteorological drought composite index(MCI) was selected as the drought evaluation index to classify the drought grade. The drought process was identified based on the run theory and three typical drought characteristic variables were extracted. The temporal variation trend of drought process and the spatial distribution characteristics of drought days of different classes were analyzed in Xinjiang from the last 62 years. The two dimensional and three dimensional joint distribution function model of drought characteristic variables was constructed by the optimized Copula function and the return period of drought was deduced.[Results]The result showed that the number of drought processes in Xinjiang showed a decreasing trend in the interannual variation. During 1960-2021, the total number of drought processes in northern Xinjiang was higher than that in southern Xinjiang. Xinjiang is prone to drought in spring and it is prone to continuous drought in spring and summer. With the increase of drought grade, the number of dry days decreased significantly and the drought was heavier in northern Xinjiang than in southern Xinjiang. The Generalized Extreme Value distribution function can better fit the drought intensity and duration, Weibull distribution function can best fit the drought peak, Gumbel Copula function can better fit the two-dimensional joint distribution of drought intensity and duration. The Clayton Copula function fits best in the two dimensional joint distributions of drought duration-peak and intensity-peak and three dimensional joint distributions.[Conclusion]The Copula function can well describe the joint distribution of multi-dimensional drought characteristic variables in Xinjiang, the probability of long-duration and high-intensity extreme drought events is relatively small, while the probability of short-duration and low-intensity drought processes is relatively large. The research result can provide technical support for drought risk assessment.