Datensatz

Zusammenfassung

The electric field in the polar ionosphere is imposed as a result of physical processes in the magnetosphere. The modeling of the magnetospheric processes is thus essential to reproduce physical phenomena in the polar ionosphere. Nowadays, magneto-hydrodynamic (MHD) models of the magnetosphere produce a realistic ionospheric potential pattern. Therefore, data assimilation into an MHD model would be a promising approach to reproduce ionospheric phenomena with high accuracy. However, the realistic MHD model is too computationally expensive to apply data assimilation. To overcome the problem of the computational cost, we employ a machine learning-based emulator of the global MHD model. The emulator is based on an echo state network model and efficiently mimics the MHD model to reproduce an ionospheric potential pattern under a give solar wind condition. As a pilot study, we assimilate the SuperDARN data into this emulator and obtain the global potential map. We will demonstrate the electric potential maps as a result of data assimilation into the emulator.