摘要
洪水地区组成随机模型的核心问题是构建联合分布或条件概率分布.从白噪声项切入,建立了基于Copula函数的多元季节性一阶自回归模型,即Copula-SAR(1)模型,并与常用的SAR(1)模型作比较.清江流域的应用结果表明,Copula-SAR(1)模型能较好地保持地区洪水组成的统计特征,非线性自相关系数的均方根误差RMSE减少了0.071,各项统计特征值的均方根误差RMSE的平均值减少了0.012.该模型显著地降低了白噪声项的模拟误差,为水文水资源的随机模拟提供了一种新的途径.
The key issue of stochastic simulation model for regional flood composition is to establish joint distribution or conditional probability distribution. Taking white noise random item into account, a new Copula-SAR model was proposed based on Copula function and the first-order seasonal autoregressive model. The Qingjiang River basin was selected as a case study to test the models. Results show that the Copula-SAR model can preserve the statistical properties of the observed flood series better than that of the first-order seasonal autoregressive model; the root mean square error （RMSE） of the nonlinear autocorre- lation coefficient and the mean value of RMSE are decreased 0. 071 and 0. 012 respectively. The proposed model can evidently reduce the simulation errors of white noise random item and provide a new approach for hydrological stochastic simulation.
出处
《武汉大学学报（工学版）》
CAS
CSCD
北大核心
2013年第2期137-142,共6页
Engineering Journal of Wuhan University
基金
国家自然科学基金项目(编号:51079100)
水利部公益性项目(编号:201001002)
