中文摘要：

选用Morlet连续复小波变换（cmor）分析1959-2012年云南全省125个气象站的逐年站平均降水时间序列的多时间尺度周期分布特征，解析云南2009年以来降水异常特少导致连续特大干旱的成因，用各主周期的小波系数实部数据建立虚拟变量回归模型，模拟年降水距平的时间演变，预估未来5a逐年降水趋势．结果表明：云南年降水存在明显的26、16—17、9～10、4a4个时间尺度不同的振荡主周期；其中26a周期最为明显，且最持续稳定，次明显的是9～10a周期，也有持续稳定的特点；1992年前、后的振荡主周期差异明显；自2009年以来，4个不同时间尺度的主周期均进入或处于谷期，4个主周期谷底效应的叠加，是云南降水连续4a异常偏少导致连续特大干旱的重要成因之一；云南的降水偏少时段可能在2014--2015年结束，以持续年降水异常偏少为特征的云南连续特大气象干旱也将可能随之结束．

英文摘要：

This paper has applied continuous Morlet Complex Wavelet Transformation （comr） to analyze cycle distribution characteristics with multiple time scales of time series of Yunnan Meteorological Stations＇ data of annual average precipitation from 1959--2012, aiming to explain why Yunnan＇s precipitation has been particular- ly rare since 2009. By using real parts of wavelet coefficients＇ data of each main cycle to build a dummy variables regression model,we＇ve tried to simulate the time variation of annual precipitation anomaly,and to predict the precipitation of next five years. The results show that there exist four main oscillation cycle patterns with different timescales in Yunnan＇s annual precipitation： 26years,9-10 -years,4 -years and 16-17 -years. The 26years cycle is the most distinct and stable cycle,followed by 9-10- years cycle,which is also stable. There are significant differences of the main oscillation cycles before and after 1992. Since 2009, these four main cycles have all been stepping into or staying at the bottom of the cycle, and superposition of these bottom effects of these main cycles has contributed to abnormally less precipitation resulting in 4 years＇ continuous severe drought in Yunnan Province. The period of less precipitation in Yunnan will end in 2014--2015, the same as Yunnan con- tinues severe drought characterized as continuously abnormally less annual precipitation.