参考作物腾发量的混沌性识别及预测

本文应用饱和关联维数法对海河流域张北站从1966~2005年50年的参考作物腾发量序列进行混沌性识别,结果表明该序列存在一定的混沌特性。同时,运用自相关函数法和饱和关联维数法确定了该序列重构相空间的嵌入维数和延迟时间,并在此基础上进行了相空间的重构。建立了混沌局域法预测模型对相空间的演化进行了计算,实现了参考作物腾发量的预测,并与时间序列自回归(AR)模型和基于气象资料的BP神经网络模型预测结果进行了比较。结果表明,预测效果比BP网络模型稍差,但明显优于AR模型。这为解决缺乏气象资料地区参考作物腾发量预测问题提供了新的思路。

Chaotic behavior analysis and prediction of reference crop evapotransporation

The correlation dimension method is applied to investigate the presence of chaos in the evapotransporation ET_0 time series obtained from the Zangbei Station,located at the Haihe River basin.The time series is calculated by using FAO56 Penman-Monteith model according to the daily weather data.The finite and low correlation dimension characteristic of the ET_0 indicates the existence of chaos and a chaos model is established.The embedded dimension and time delay are calculated by using auto correlation method and correlation dimension method respectively.On this basis the phase-space can be reconstructed.A local approximate method is established to carry out the chaotic prediction of the variation of evaprotranspiration.It is found that the result is not as good as that calculated by BP neural network model based on weather data,but it is better than that simulated by AR model.This encouraging result indicates that the chaos theory provides a new method for ET_0 prediction without weather data.