摘要
以1960—2010年的潘家口水库流域的入库径流序列,计算得到12个时间尺度的水文干旱指数SRI序列。基于水文干旱指数SRI时间序列,分别采用传统的极大似然法和改进的贝叶斯方法构建时间序列模型ARMA,并对比基于极大似然法和贝叶斯方法构建的两种ARMA模型的模拟和预测精度。结果表明,基于贝叶斯方法构建的模型在模拟和预测期的模拟效果较优。随着预测期的延长,两种模型的模拟精度均呈现降低趋势。
Based on the inflow runoff series of Panjiakou Reservoir basin from 1960 to 2010,this paper calculated the SRI series of hydrologic drought index at 12 time scales.Based on the SRI time series of hydrological drought index,the traditional maximum likelihood method and the improved Bayesian method were used to construct the time series model ARMA,and the simulation and prediction accuracy of the two ARMA models were compared.The results show that the simulation of the model based on Bayesian method performs better in both simulation and prediction periods,and the simulation accuracy of the two models decreases with the extension of prediction period.
作者
李敏
张铭锋
朱黎明
彭卓越
LI Min;ZHANG Mingfeng;ZHU Liming;PENG Zhuoyue(Yangzhou University,Yangzhou 225000,China)
出处
《水文》
CSCD
北大核心
2023年第3期34-39,共6页
Journal of China Hydrology
基金
江苏省“双创博士项目”项目(JSSCBS20211029)
江苏省自然科学基金项目(BK20200942)。
作者简介
李敏(1992-),女,山西人,博士,讲师,主要从事水文学及水资源方面的工作。E-mail:limintju@126.com。
