摘要
中长期径流预报是水文预报中的经典难题之一,其在防洪、水库调度及水资源管理中起着十分重要的作用。由于缺乏相应预见期的可靠气象预报资料,中长期径流预报一般采用统计方法。超越概率贝叶斯判别分析方法是一种数据驱动的非参数贝叶斯经验统计方法,通过设置不同的流量等级反复进行贝叶斯判别分析,对未来径流超过某一流量等级的概率(超越概率)进行预报。本文运用该方法对长江宜昌站、大通站的月、季径流预报进行了研究,其结果表明,超越概率贝叶斯判别分析方法能够有效实现宜昌站和大通站非汛期径流预报;对于汛期径流预报,采用厄尔尼诺和南方涛动等气象水文指标变量作为预报因子,是提高预报精度的可行途径。
Mid-term and long-term streamflow prediction is of vital importance in flood control, reservoir operation and water resources management. Due to the lack of reliable meteorological inputs, statistical method is commonly used to mid-term and long-term streamflow prediction. The Exceedance Probability Bayesian Discriminant Analysis Method is a Bayesian statistical method which relies on discriminant analysis to predict the probability of ＂future streamflow goes beyond a certain value＂ （Exceedance probability） by the historical predictand and predictor samples and real-time predictor information. This method has been used to predict the streamflow of Yichang and Datong Gauge Station, Yangtze River. The results show that the exceedance probability method is effective in predicting the non-flood season streamflow and is a viable approach for adopting hydro-meteorological indexes as predictor to promot prediction accuracy of flood season streamflow.
出处
《水利学报》
EI
CSCD
北大核心
2011年第6期692-699,共8页
Journal of Hydraulic Engineering
基金
“十一五”国家科技支撑计划资助项目(2008BAB29B09-2
2008BAB29B08)
关键词
中长期径流预报
超越概率方法
贝叶斯判别分析
长江径流预报
mid-term and long-term streamflow prediction
exceedance-probability method
Bayesian Discriminant Analysis
Yangtze River streamflow prediction
