论文部分内容阅读
利用黄河高寒区吉迈(达日县)水文站、国家气象局达日、玛多观测站1959—2007年,共计49 a水文、气象及部分冻土监测资料,采用峰型度及丰枯率指数、Mann-Kendall时间序列突变点及趋势检验方法、小波分析法、灰色关联分析等统计方法,分析径流年际、年内变化特征,探讨径流长期变化主要影响因素,并建立核主成分支持向量机(KPCA-LSSVM)预测模型,对未来2011—2014年径流情势进行分析。结果表明,由于区域气候、下垫面条件变化及人为因素影响,不同时期作用于水资源的主导因素存在差异,径流时间序列存在明显突变点,为1961、1999和2005年;黄河源区径流2002—2006年间存在不显著减少趋势,小波分析表明未来3～9 a内,年径流量将略高于1990—2007年间的平均值;1983年后冻结期气温升高,季节性冻土退化较为明显,对于径流的补给减少,是地区河道水资源减少的原因之一;模型预测结果显示,由于降水对于径流的延迟作用和气温升高引起冻融关系变化等原因,2011—2014年水源区河川径流量与多年平均值相比减少25.3%。
Using the hydrological station of Jimai (Dari County) in the alpine region of the Yellow River and the monitoring stations of the Dari and Maduo National Meteorological Bureau from 1959 to 2007 for a total of 49 years of hydrology, meteorology and some frozen soil monitoring data, Index, Mann-Kendall time series mutation points and trend test methods, wavelet analysis and gray relational analysis, the paper analyzes the interannual and annual variation characteristics of runoff and discusses the main influencing factors of long-term runoff variation and establishes the principal component support vector machine (KPCA-LSSVM) prediction model to analyze the future runoff pattern in 2011-2014. The results show that there are differences in the dominant factors acting on water resources at different periods due to the changes of regional climate, underlying surface conditions and human factors, and there are obvious change points in runoff time series between 1961, 1999 and 2005; - There was an insignificant trend of decrease in 2006. The wavelet analysis shows that the annual runoff will be slightly higher than the average of 1990-2007 in the next 3 ~ 9 years. After 1983, the temperature in the freezing period increases and the seasonal degradation of frozen soil is obvious , And the reduction of runoff is one of the reasons for the decrease of water resources in the river channels. The model predictions show that due to the delay effect of precipitation on runoff and the change of freeze-thaw relationship caused by the temperature increase, Traffic decreased 25.3% from the multi-year average.