基于SPEI和信息扩散加权一阶局域法的旱涝特征分析

Characteristics Analysis of Drought and Flood Based on Information Diffusion Adding-Weight One-Rank Local-Region Method

为分析区域旱涝变化规律,在用标准化降水蒸散指数(SPEI)定量计算旱涝程度的基础上,利用小样本数据分析的信息扩散技术,对旱涝发生的频率和趋势特性进行研究,并新提出信息扩散加权一阶局域法的混沌时间序列旱涝预测模型。以西安市为典型研究区,根据1951—2015年逐月气温和降水资料,选取该区域玉米成熟期9月份为研究时段,分析该区域不同程度旱涝发生的频率分布规律、趋势变化规律及趋势预测情况等。研究表明:(1)发生中旱及以上程度的干旱频率约19.5%,发生中涝及以上程度的渍涝频率约17.5%;(2)1951—1962年呈偏旱趋势,1963—1986年呈偏涝趋势,1987—2002年又呈偏旱趋势,2003年后再次呈偏涝趋势;(3)旱涝序列的最大李雅普诺夫指数为0.1492。利用新的混沌预测模型对1985—2002年旱涝趋势逐年滚动预测,与实际情况吻合较好,验证了信息扩散加权一阶局域法的有效性。

In order to analyze the change regulation of regional drought and flood,based on the standard precipitation evapotranspiration index(SPEI)that can effectively evaluate the drought and flood degree quantitatively,the frequency of drought and flood was studied by using information diffusion technology which was good at small sample data analysis,and the chaotic time series prediction model based on information diffusion adding-weight one-rank local-region method was proposed to analyze the trend of drought and flood time series.Taking Xi’an as a typical region and September as the research period,the method suggested in this paper was used for analysis,according to the temperature and precipitation data from 1951 to 2015.The results show that a)the probability of moderate drought or above is about 19.5%,while the probability of moderate wet or above is about 17.5%.b)During 1951-1962,the trend is drought.During 1963-1986,there is a trend of wet.From 1987 to 2002,there is a trend of drought.After 2003,the trend of wet appears again.c)The maximum Lyapunov exponent of drought and flood series is 0.1492,which shows that the time series of drought and wet have chaotic characteristics.The new chaotic prediction model is used to predict the trend of drought and flood from 1985 to 2002,which is in good agreement with the actual situation and verifies the effectiveness of the method.