基于小波与游程耦合的时序模型在降水预测中的应用  

Application of the Coupling Time Series Model Based on Wavelet and Runs in Precipitation Forecast

作　　者：彭高辉[1] 韦保磊 马建琴[2] 

机构地区：[1]华北水利水电大学数学与信息科学学院,郑州450045 [2]华北水利水电大学水利学院,郑州450045

出　　处：《长江科学院院报》2014年第8期18-22,28,共6页Journal of Changjiang River Scientific Research Institute

基　　金：国家自然基金项目资助(41071025);国家级大学生创新创业训练计划项目(201210078060);河南省教育厅自然科学研究计划项目(2010B120007)

摘　　要：针对降水量时间序列具有多尺度非平稳性特点、长时段预测精度不高的问题,选择db3小波,运用Mallat算法2尺度分解降水序列,利用游程分析检验分解序列间的独立性,建立了二者耦合的时序模型,并应用于郑州市2008—2012年的降水预测,得到了相对于传统时序模型更好的结果。基于耦合模型预测了郑州市2013—2015年月均降水量,以期为决策提供依据。According to the multi-scale and non-stationary features of precipitation time series, we choose db3 wavelet and use Mallat algorithm to decompose the precipitation series in two types of scales to solve the problem of longtime forecast with inadequate precision. After independent Runs tests between the decomposing consequences, we established the coupling ARIMA models based on Wavelet and Runs to forecast precipitation in the city of Zheng- zhou from 2008 to 2012. It is proved that the accuracy of the coupling model is higher than that of traditional model. We predict the precipitation per month from 2013 to 2015 in order to provide a basis for decision-making. This research is of theoretical significance and application value.

关 键 词：降水预测 小波变换 游程检验 耦合模型 ARIMA模型 

分 类 号：P457.6[天文地球—大气科学及气象学]