基于奇异谱分析的闽江流域径流长期预报研究  

Long term forecast of runoff in the Minjiang River Basin based on singular spectrum analysis

作　　者：陈莹[1] 陈兴伟[1] 

机构地区：[1]福建师范大学地理科学学院,福建福州350007

出　　处：《水资源与水工程学报》2011年第5期16-19,共4页Journal of Water Resources and Water Engineering

基　　金：教育部博士点基金(20103503110002)资助;福建省自然科学基金(2010J05092)资助;福建省教育厅项目(JA10081)资助

摘　　要：以闽江流域1950-2007年月径流时间序列为例,采用毛毛虫-奇异谱分析方法进行长期径流预报研究。经过嵌套、奇异值分解、聚合、对角线平均等步骤,识别序列的趋势、周期等特征,得到趋势成分,以及3个月、4个月、6个月、12个月、2.6年和6.7年的周期成分。进一步基于获得的趋势、周期成分,采用线性递归公式检验验证期(2006-2007年)预测效果,并对未来24个月份(2008-2009年)的径流进行预测。结果表明:毛毛虫-奇异谱分析对水文时间序列具有较好的长期预测能力。This paper applied caterpillar-singular spectrum analysis to the analysis and long term runoff forecast of hydrologic time series.Based on the monthly runoff data from 1950 to 2007 in the Minjiang River Basin,the paper revealed the slowly varying（trend） component and oscillatory components of the runoff series through embedding step,SVD step,grouping step and diagonal averaging step of caterpillar-singular spectrum analysis.The oscillatory components of the runoff series include components with periods of 3 months,4 months,6 months and 12 months,2.6 years and 6.7 years.The prediction results of the validation period（2006-2007） were validated and the runoff of future 24 months（2008-2009） was forecasted by using linear recurrent formula based on the obtained trend and oscillatory components.The results indicated that caterpillar-singular spectrum analysis provides a proficient way for the long term forecast of hydrologic time series.

关 键 词：奇异谱分析 长期预测 径流 闽江流域 

分 类 号：TV121[水利工程—水文学及水资源] P338[天文地球—水文科学]