Charles Explorer logo
🇨🇿

Machine Learning Based Models of the Magnetopause Location

Publikace na Matematicko-fyzikální fakulta |
2023

Tento text není v aktuálním jazyce dostupný. Zobrazuje se verze "en".Abstrakt

Empirical models of the magnetopause location are typically based on the fitting of a large set of identified magnetopause crossings by a predefined function dependent on relevant parameters (solar wind dynamic pressure, interplanetary magnetic field Bz component, possibly also the tilt angle others. We remove the assumption of the predefined boundary shape by applying a more general approach based on the artificial neural network.

This is used, along with a set of about 40,000 dayside magnetopause crossings identified in the THEMIS A-E, Magion 4, Geotail, and Interball-1 satellite data to model the magnetopause location as a function of relevant parameters and to discuss their influence and the boundary shape. An alternative approach that we consider is an automated identification of the region (magnetosheath/magnetosphere) where the spacecraft occurs at a given location under given conditions.

The neural network is then used not to predict the magnetopause radial distance, but rather only the particular plasma region at a given location. The magnetopause boundary is then found as a boundary separating the two regions, i.e., as the boundary where the model region classification changes from the magnetosphere to the magnetosheath.

The performance of the obtained models is compared with other empirical magnetopause models.