WWW Homepage of Real-time Space Weather Prediction

Real-time prediction of geomagnetic storms
from ACE solar wind data

Geomagnetic storms, characterized by Dst index, are predicted 1 hour ahead from solar wind data using Elman recurrent neural networks. Real-time ACE solar wind plamsa and IMF data are taken from the Space Environment Center. The neural network models used here have been developed by Wu [1997] and Wu and Lundstedt [1996, 1997a, 1997b]. The models' performance has further been verified from their capability to accurately predict a major magnetic storm triggered by the 1997 January halo-CME [Wu et al., 1998] August 1, 1998 Issue 14 - Sun-Earth Special Section Part II. Provided by the World Data Center C2 (WDC-C2 for Geomagnetism, Kyoto University, Japan), the provisional Dst index (available on the previous day) is used for comparison with prediction. The prediction given below is 1 hour in advance, made from the solar wind input parameters (B, By, Bz, N, V). To see the prediction from the input parameters (B, Bs, N, V), please click here.

The prediction presented here is still in a test phase. The staff at NOAA's Space Environment Center in Boulder and the ACE project teams are gratefully acknowledged for providing real-time solar-wind data. The WDC-C2 for Geomagnetism, Kyoto University, is gratefully acknowledged for providing the provisional Dst.

References

(1) Prediction of geomagnetic storms from solar wind data using Elman recurrent neural networks,: Wu, J.-G., and H. Lundstedt, Geophys. Res. Lett., 23, 319, 1996.
(2) Space Weather Physics: Dynamic Neural Network Studies of Solar Wind-Magnetosphere Coupling, Ph.D thesis,: Wu, J.-G., LUNFD6/(NFAS-1016)/1-136/(1997), Lund University, Lund, Sweden, 1997.
(3) Geomagnetic storm predictions from solar wind data with the use of dynamic neural networks,: Wu, J.-G., and H. Lundstedt, J. Geophys. Res., 102, 14255-14268, 1997.
(4) Neural network modeling of solar wind-magnetosphere interaction,: Wu, J.-G., and H. Lundstedt, J. Geophys. Res., 102, 14457-14466, 1997.
(5) Neural network models predicting the magnetospheric response to the 1997 January halo-CME event,: Wu, J.-G., H. Lundstedt, P. Wintoft, and T.R. Detman, Geophys. Res. Lett., 25, 3031, 1998.