投影寻踪自回归和多维混合回归模型及其在大河长河段洪水预报中的应用

Application of the PPAR and PPM R in Flood Forecasting for Long Reaches of Large Rivers

投影寻踪自回归和多维混合回归模型，是将投影寻踪技术应用于时间序列分析中的一种新型的非线性时序分析模型。本文将此模型用于长江寸滩～宜昌河段的洪水演算，不必使用区间雨量资料，仅仅根据上下游站的流量资料，就可做出预见期为４８～７２ｈ的复杂长河段洪水预报，并取得了令人满意的研究成果。

As the analysis models of non-linear time series,the new developed Projection PursuitAuto-regression Model(PPAR) and Projection Pursuit Multivariate Mixing Regression Model(PPMR) can beused in the time series analysing The models introduced in this paper have been used in the flood routing for theCuntan-Yichang Reach on the Yangtze River The results indicate that flood forecasting for long reach can bemade within the lead time of 48～72hours only according to the discharge data from the hydrometric stations onlower and upper reaches, and the rainfall data for the reach are not needed.