Ionosphere TEC prediction based on Chaos

This paper proved the chaotic property of ionosphere total electron content (TEC) by using the data in 1997 and 2001 at Wuhan station, based on Chaos theory. The time delay was calculated by using mutual information method, the best embedding dimension was calculated by using G-P method and the largest Lyapunov exponent was calculated by using small data method. Based upon this, it constructed the forecasting network with an advanced time of one or twenty-four hours by means of BP Neural Network and using the data of 1998 and 2002 to test the network. Analysis of the forecasting error indicates that one hour ahead prediction network which can reach high accuracy is quite suitable for quasi real-time prediction of ionosphere TEC over this region, twenty-four hours ahead prediction can controlled the relative error to be about 10%. The average absolute error is about 1.6256TECU during solar minimum, far less than 6.4575 TECU predicted during solar maximum. The forecasting error is relative high when magnetic storm happens on the morning, noon and sunset.