Statistical Distribution of Magnetic Perturbations and Solar Wind Drivers Using SWARM Data and Machine Learning

Abstract:

We investigate the statistical distributions of magnetic perturbations measured by low-Earth orbit (LEO) satellites under different levels of geomagnetic activity and apply a machine learning algorithm to classify the relationship between solar wind drivers and the range of geomagnetic disturbances using LEO satellite magnetic field measurements and solar wind data.

In this project, magnetic perturbations are calculated as the residual between Swarm level 1b magnetic field vector and intensity data provided by the European Space Agency (ESA) and the Earth’s main magnetic field. Six and a half years of available Swarm data are used, and magnetic perturbations calculated are binned into grids globally in Corrected GeoMagnetic (GCM) coordinates. The distribution of perturbation amplitudes are determined spatially (geomagnetic latitude, longitude and local time) and with respect to geomagnetic indices and solar wind parameters.

Magnetic perturbation data are divided into groups based on the value ranges for geomagnetic indices such as Kp, Dst and AE at the time of measurement, which indicate different geomagnetic activity levels. Mean and variation in the magnetic perturbation patterns are created for different activity levels to characterize the global magnetic field status under different activity levels. Swarm magnetic field data and related solar wind data are then applied to a machine learning algorithm to predict the magnetic perturbation.