干旱历时拟合方法对基于Copula函数的干旱分析影响研究

Effect of Drought Duration Fitting Methods on Copula-based Drought Analysis

研究干旱事件的演变规律对于保障水资源安全、促进经济社会发展具有重要意义。通常采用Copula函数分析干旱历时强度等多变量规律,而不同干旱历时拟合方法会对结果造成影响。使用北京和上海近百年逐月降水资料比较了参数拟合方法(指数分布)和非参数拟合方法(游程方法)。结果表明,基于不同历时拟合方法的计算结果均有其合理性,但非参数方法处理极端干旱历时需要外延,因而在这方面不如参数方法。不论以哪种双变量重现期为标准,游程理论拟合历时评估出的风险总小于指数分布拟合历时的评估结果。对于“且”和“或”重现期,历时拟合方法对等值线图的影响与强度拟合的影响相互独立;而对于条件重现期,两者的影响不独立。

Researches on evolution of drought play an important role in ensuring the safety of water resources and economic and social development.Nowadays,Copulas are used in the multivariate drought analysis considering duration and severity,but duration fitting methods could affect the results.In this paper,about 100-years-long monthly precipitation data of Beijing and Shanghai was used to compare parametric(exponential distribution)and nonparametric(run-length theory)fitting methods.The results indicate that these fitting methods are all reasonable,but non-parametric methods need extrapolation when dealing with extreme drought duration,so parameter methods are better in this case.The risk evaluated by using the run-length theory is always lower than the one using exponential distribution no matter which bivariate return period is used as the standard.For“and”and“or”return periods,the influences of duration fitting and severity fitting on the isolines are independent,whereas for conditional return periods,the effects are not independent.