[关键词]
[摘要]
为了合理评估区域干旱灾害风险, 以岷江流域 14 个气象站 1961- 2012 年逐月降水数据为研究对象, 以阈值 为变量设定 3 种游程截取水平识别干旱, 然后利用经拟合优选的 Co pula 函数构建干旱历时与干旱烈度的联合分 布, 并进行概率分析和重现期计算。结果表明: Gumbel Co pula 函数对整个流域的拟合效果最优; 利用 Gumbel Cop2 ula 函数分析二维条件概率 P( S [ 60 mm | D\3 月) , 3 种截取水平的发生概率从流域东南到西北递增; 联合重现期 结果显示流域东北部和南部一带的旱情比中西部严重, 而同现重现期结果则显示流域西部的旱情比东部严重; 联合 重现期比同现重现期的空间分布更稳定, 任意一种截取水平下的联合重现期都比较接近。研究结果可为游程理论 截取水平设定、Copula 函数优选以及多变量干旱分析提供依据。
[Key word]
[Abstract]
In or der t o assess the risk of r egional droug ht disaster r easo nably, w e too k the monthly precipitatio n data of 14 mete2 o ro log ical stat ions in Minjiang Riv er Basin fr om 1961 to 2012 as the r esear ch o bject, and, with thr esho ld as a variable, set thr ee cutting lev els acco rding to the t heo ry o f runs to identify dro ug ht events. Then w e used the best fit Co pula function t o co nstr uct the joint distribution of droug ht durat ion and severity and analy zed the pr obabilistic char acteristics and r et ur n perio d. The re2 sults show ed that Gumbel Co pula funct ion demonstrated the best fitness for M injiang River Basin. Aft er the analy sis of the spa2 tial distribution of conditional pr obability P( S [ 60 mm| D \3 mo nths) , t he results indicat ed that the pro babilit y of occurrence on t he three cut ting lev els increased from the southeast to the northwest of the basin. T he r esults o f joint r et ur n period demon2 str ated that the dr ought in the no rtheaster n and southern parts o f the basin w as mo re serious than that in the central and w estern re2 g ions, while the results of co2occurrence retur n period showed that the drought in the west of the basin was more serio us than that in the east. The spatial dist ribution cha racterist ics o f joint r eturn period wer e mo re stable than tho se o f co2occurrence r eturn peri2 o d. The jo int return perio ds of the w ho le basin on each cut ting level wer e r elativ ely clo se. This study could pr ovide a basis for sett ing the cutting lev el o f the theor y o f runs, cho osing o pt imal Co pula funct ion, and co nduct ing multivariate dr ought analy sis.
[中图分类号]
[基金项目]
国家自然科学基金项目( 41272249; 41301021)