摘要
以闽江建溪流域为研究区,建立建溪流域洪水预报模型,比较了误差自回归校正方法、卡尔曼滤波校正方法以及以贝叶斯实时校正技术对流域洪水预报的校正效果。结果表明,3种实时校正方法都能有效提高预报精度,在充分利用实时观测数据对子流域进行实时校正后,七里街的洪水预报精度显著提高。在水量误差方面,贝叶斯方法较好;对于洪峰流量的校正,卡尔曼滤波方法表现较好;对于纳西效率系数,贝叶斯方法适合短预见期预报,而卡尔曼滤波方法对于长预见期的预报较为稳定且能保持高值。
In order to learn the correction effect of different real-time correction technologies,this research takes Jianxi River Basin in Fujian Province as an example. Three methods,the auto-regressive error correction method,Kalman filter and the Bayesian method,were adopted.As the results showed,all the three methods function well compared with the forecast result before error correction. The Bayesian method correction results has less relative error（ RE） while the Kalman filter method function better on the peak discharge. As for the Nash Sutcliffe coefficient（ NSE）,the comparison results show that the Bayesian method suits short prediction period while the Kalman filter is better for the long prediction period which can keep high value.
作者
周梦
陈华
郭富强
许崇育
ZHOU Meng;CHEN Hua;GUO Fu-qiang;XU Chong-yu(State key Laboratory of Water Resources and Hydropower Engineering Science,Wuhan University,Wuhan 430072,Chin)
出处
《中国农村水利水电》
北大核心
2018年第7期90-95,共6页
China Rural Water and Hydropower
关键词
洪水预报
实时校正
自回归修正
卡尔曼滤波
贝叶斯方法
flood forecasting
real-time updating
auto-regressive error correction
Kalman filter method
Bayesian method
