New approach for choice of time delay in nonlinear time series of satellite remote sensing of rainstorms

In this paper, to develop novel methods for satellite optical remote sensing of severe storms, chaotic time-series analysis is carried out and the time delay embedding technique is used for phase space reconstruction, which is relied strongly on a choice of good time delay and the embedding dimension. A new approach for calculations of the mutual information for the choice of time delay for a time series with any probability distribution is proposed. To confirm the validity of the approach developed, the tests using simulated nonlinear time series for some famous chaotic attractors are performed. Then, application of the approach in the time series of GMS-5 11 mu m 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 method proposed is a good tool for the best choice of time delay in time series analysis.