A comparison of methods for forecasting total electron content

Cathryn Mitchell, Tamsyn Baxter and Steve Musman

Abstract:	The application of artificial neural network (NN) techniques to the prediction of the ionospheric parameter total electron content (TEC) is demonstrated and compared with other methods of forecasting. The NN method can be used to predict non-linear systems and has been recently applied to the problem of forecasting electron density at the F-layer peak of the ionosphere. However, it has not previously been used extensively for TEC prediction. In this paper a time-series of vertical TEC values monitored during 1997 have been partitioned for training and testing a number of different prediction algorithms. The data were obtained from a GPS receiver located in Keweenaw (47.2°N, 88.6°W), in the upper peninsula of Michigan. The predictions have been performed for half, one and three hours ahead and the accuracy of each of the different forecasting methods has been assessed. The results have revealed some of the key factors in TEC prediction. The greatest difficulty for the NN was to predict the TEC throughout a geomagnetic storm and this event is examined in some detail. An unexpected result was the differences in the capabilities of each of prediction method to cope with the changes in TEC throughout this event. The application of these various techniques to predicting TEC during other times in the solar cycle is also discussed.
Published in:	Proceedings of the 13th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GPS 2000) September 19 - 22, 2000 Salt Palace Convention Center Salt Lake City, UT
Pages:	680 - 687
Cite this article:	Mitchell, Cathryn, Baxter, Tamsyn, Musman, Steve, "A comparison of methods for forecasting total electron content," Proceedings of the 13th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GPS 2000), Salt Lake City, UT, September 2000, pp. 680-687.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In