黄河源区径流长期演变特征与趋势预测模型研究  

Annual Runoff in the Source Regions of Yellow River:Long-Term Variation Features and Trend Forecast

作　　者：赵国辉[1] 张耀南[1,2] 蓝永超[1] 

机构地区：[1]中国科学院寒区旱区环境与工程研究所,甘肃兰州730000 [2]甘肃省高性能网格计算中心,甘肃兰州730000

出　　处：《冰川冻土》2010年第1期189-195,共7页Journal of Glaciology and Geocryology

基　　金：中国科学院黑河流域生态水文模型集成研究的e-science环境项目(INFO-115-D01-Z007);中国科学院信息化专项冰雪冻土环境本底专题数据库项目(INFO-115-C01-SDB2-03);科技部国家基础科学人才培养科学基金项目(J0630966);中国科学院知识创新工程重要方向项目(KZCX2-YW-328)资助

摘　　要：利用小波分析方法对黄河源区径流数据系列的多尺度变化特征、突变点及变化趋势进行了分析.结果表明:黄河源区年径流量具有8a、15a、22a和36a左右的变化周期,其中8a、36a左右的周期变化最为显著.这些周期变化表明,2007年后流量将呈增加的趋势;1928、1982年和1985年是径流变化趋势重要的转变点.在小波分解的基础上,基于BP神经网络模型构建了黄河源区年径流量的长期动态预报模型,利用该模型对未来10a的流量变化进行了预测,并对其预报结果进行了分析.The Yellow River is the maximum water source in Northwest China, especially in the upper reaches above Tangnag, which is called ＂ water tower＂. In the background of global warming, the variation in runoff is responsible for social and eco-nomic sustainable development in the whole Yellow River basin. To assess the long-term run-off changing tendency and characteristics, the complex Morlet continuous wavelet function was applied to analyzing the annual runoff time series in the source regions of Yellow River above Tangnag for the period of 1920--2007. Four major changing periods （8 a, 15 a, 23 a and 36 a） are found. Among them, the period of 36 a is the most noticeable one. By decomposing the long-term runoff time series, the runoff in the upper Yellow River had an increasing tendency before 1985 apparently and has a decreasing tendency after 1985. In order tO find out the future changing tendency of the annual runoff, a combined model is developed by combining wavelet decomposition with artificial neural network. According the stochastic simulation, it is found that the runoff will slightly increase over the next decade, but still has had a significant decrease as a whole since 1956. The results also indicate that the combined model is feasible and efficient.

关 键 词：黄河源区 径流变化 小波分析 人工神经网络 趋势预测 

分 类 号：P338.1[天文地球—水文科学]