摘要
针对松花江干流汛期洪水的特点以及松花江流域防洪减灾的需求,采用多元门限回归模型建立了松花江干流肇源、三家子、涝洲、木兰、富锦5个水位站的水位预报模型;在多元门限回归模型的基础上进行改进,得到混合门限回归模型,并以此建立松花江干流5个站的水位预报模型。两种模型的预报因子均通过AIC准则和DW检验法筛选确定,并用最小二乘法估算模型的参数。选取各水位站2008—2012年汛期的水位资料分别率定相应的水位预报模型,选取2013年汛期的水位资料对各个率定的模型进行验证。率定和验证的结果表明:多元门限回归模型的预报精度偏低,而混合门限回归模型的预报精度高,且有一定的通用性,适用于水位预报。
According to hydrologic characteristics of the main stream of Songhua River and the needs of flood control and calamity reducing, for some hydrological observation stations in the river basin,water-level forecasting models were developed with black-box model based on multiple threshold regression model, including the stations of Zhaoyuan,Sanjiazi, Laozhou, Mulan and Fujin. On the basis of multiple threshold regression model,it was improved as mixed threshold regression model and water-level forecasting models of these five stations were developed by it. The two model structure was deter^nined by AIC and DW method and parameters of models were estimated by the least square method. The models were calibrated with hydrologic data of 2008-2012 and verified with hydrologic data of 2013 in the flood season. The results show that the accuracy of multiple threshold regression model is low, but the mixed threshold regression model is accurate within acceptable range and it has certain universality,so it can be used for water-level forecasting.
出处
《人民黄河》
CAS
北大核心
2017年第5期36-40,44,共6页
Yellow River
基金
国家重点研发计划项目(2016YFC0402704)
水文水资源与水利工程科学国家重点实验室专项经费项目(1069-514031112)
关键词
水位预报
多元门限回归
混合门限回归
AIC准则
DW检验法
松花江干流
water-level forecasting
multiple threshold regression
mixed threshold regression
AIC criterion
DW method
Songhuajiang mainstream
main stream of songhua River
