Multivariable State-Space Method And Its Application In Climatic Time Series Prediction

In practical problems, complex systems exist everywhere。They usually contain many variables, i。e。climatic system。In realistic situations, it is hard to build up exact analytic models for complex systems because their constructions are very intricate and the information available is also incomplete and inaccurate。Complete systems are usually analyzed by time series observed or measured from the systems。In order to make full use of the accumulative huge climatic data, this paper presents a new model of short-range climatic forecast with multivariate time series。combining the two streams of thoughts, i。e。the state-space reconstructed from single variable time series and traditional multivariate analysis, a multivariate state space forecasting method was developed。In the paper, through the technique of the reconstructing phase space and the implementation of the G-P algorithm in Matlab, the correlation dimension and the Lyapunov Exponent of climatic time series are worked out。Systematically study two nonlinear prediction methods for multivariate time series, including prediction method of local linear analogy model and Lyapunov Exponent analysis。Through the prediction errors of univariate time series and multivariate ones are compared by climatic time series prediction,the result of the research also shows that reconstructing phase space via multivariate time can forecast much better than via univariate time series。Finally, the paper points out the applied value and prospect of this method in forecasting the climate。...