New approach for studies of time series of satellite observations of rainstorms: I. Mutual information calculations

摘要

To develop a new and modern satellite retrieval method to study the severe storm process, satellite remote sensing is combined with nonlinear chaos dynamics. In this paper, the choice of time delay for the phase space reconstruction for nonlinear time series analyses is carried out. The method based on the information generation is employed. A new algorithm for calculations of the mutual information for the choice of time delay for phase space reconstruction for any probability distribution is developed. To confirm the validity of the algorithm proposed, the tests using simulated time series for some famous chaotic attractors and other nonlinear processes are performed. Finally application of the algorithm in the time series of GMS-5 11μm IR channel brightness temperature observations of rainstorm occurred in Wuhan area in China on 21-27 July 1998 is discussed. The results show that the new algorithm proposed is a good tool for the best choice of time delay in time series.