A neural network-based local model for prediction of geomagnetic disturbances

This study shows how locally observed geomagnetic disturbances can bepredicted from solar wind data with artificial neural network (ANN)techniques. After subtraction of a secularly varying base level, thehorizontal components XSq and YSq of the quiettime daily variations are modeled with radial basis function networkstaking into account seasonal and solar activity modulations. Theremaining horizontal disturbance components DeltaX and DeltaY aremodeled with gated time delay networks taking local time and solar winddata as input. The observed geomagnetic field is not used as input tothe networks, which thus constitute explicit nonlinear mappings from thesolar wind to the locally observed geomagnetic disturbances. The ANNsare applied to data from Sodankyla Geomagnetic Observatory locatednear the peak of the auroral zone. It is shown that 73% of the DeltaXvariance, but only 34% of the DeltaY variance, is predicted from asequence of solar wind data. The corresponding results for prediction ofall transient variations XSq+DeltaX andYSq+DeltaY are 74% and 51%, respectively. The local timemodulations of the prediction accuracies are shown, and the qualitativeagreement between observed and predicted values are discussed. If drivenby real-time data measured upstream in the solar wind, the ANNs heredeveloped can be used for short-term forecasting of the locally observedgeomagnetic activity. (Less)

[1]  Judith L. Lean,et al.  Solar ultraviolet irradiance variations: A review , 1987 .

[2]  Toshiki Tajima,et al.  Forecasting auroral electrojet activity from solar wind input with neural networks , 1999 .

[3]  D. Vassiliadis,et al.  Modeling the spatial structure of the high latitude magnetic perturbations and the related current systems , 1999 .

[4]  Joydeep Ghosh,et al.  Use of localized gating in mixture of experts networks , 1998, Defense, Security, and Sensing.

[5]  D. M. Schlapp,et al.  A study of day-to-day variability of Sq over Europe , 1979 .

[6]  D. Baker,et al.  IMF control of geomagnetic activity , 1988 .

[7]  Shang-Liang Chen,et al.  Orthogonal least squares learning algorithm for radial basis function networks , 1991, IEEE Trans. Neural Networks.

[8]  D. Weimer,et al.  A flexible, IMF dependent model of high-latitude electric potentials having “Space Weather” applications , 1996 .

[9]  P. Sutcliffe.  The development of a regional geomagnetic daily variation model using neural networks , 2000 .

[10]  Wolfgang Baumjohann,et al.  Studies of polar current systems using the IMS Scandinavian magnetometer array , 1993 .

[11]  Daniel N. Baker,et al.  Magnetospheric Impulse Response for Many Levels of Geomagnetic Activity , 1985 .

[12]  Daniel N. Baker,et al.  A description of the solar wind-magnetosphere coupling based on nonlinear filters , 1995 .

[13]  Peter Wintoft,et al.  A neural network study of the mapping from solar magnetic fields to the daily average solar wind velocity , 1999 .

[14]  Y. Feldstein,et al.  Electric potential patterns in the northern and southern polar regions parameterized by the interplanetary magnetic field , 1994 .

[15]  H. Gleisner,et al.  Response of the auroral electrojets to the solar wind modeled with neural networks , 1997 .

[16]  Mohammad Bagher Menhaj,et al.  Training feedforward networks with the Marquardt algorithm , 1994, IEEE Trans. Neural Networks.

[17]  Y. Feldstein,et al.  Solar Wind Control of Electric Fields and Currents in the Ionosphere , 1986 .