ニューラルネットワークによる太陽風物理量を用いた地磁気擾乱指数の予測

Abstract

We try to determine what solar wind parameter has an impact on the geomagnetic disturbance based on neural computational method “potential learning (PL)”. In our previous study, using the PL method, the clock angle, which is the angle between the dawn — dusk (By) and north — south (Bz) components of solar wind magnetic field, was extracted as the most significant parameter to disturb the geomagnetic field. However, that result included a little uncertainty due to redundancy of parameters. Therefore, we scrutinize the solar wind parameters to evaluate with the PL method, and retry to extract the parameters to provide an impact on geomagnetic disturbances. We could find that the clock angle, solar wind dynamic pressure and number density impinged on magnetospheric disturbances. These parameters are important in considering geomagnetic disturbances. Our result suggests that the potential learning method is very useful tool to study the solar wind-driven geomagnetic disturbances.