Solar Wind Classifications at Mars using Machine Learning Techniques

TL;DR

This paper uses machine learning to classify different solar wind regimes at Mars, revealing how their occurrence varies with solar activity based on MAVEN data.

Contribution

It introduces an unsupervised machine learning framework combining PCA and K-Means to identify and interpret solar wind regimes at Mars.

Findings

Identified distinct slow, fast, intermediate, and compressed solar wind regimes.

Revealed that the occurrence of these regimes is modulated by solar activity.

Provided a physically interpretable classification of solar wind at Mars.

Abstract

Understanding solar wind variability throughout the heliosphere is essential for fundamental space physics and future exploration of the Moon and Mars. The Mars Atmosphere and Volatile EvolutioN (MAVEN) spacecraft has provided upstream solar wind measurements at Mars spanning Solar Cycles 24 and 25, enabling a statistical investigation of solar wind regimes at this heliocentric distance. In this work, we apply an unsupervised machine-learning framework combining Principal Component Analysis and K-Means clustering to a normalized, multi-dimensional solar wind dataset to identify recurrent solar wind regimes in a physically interpretable, data-driven manner. The resulting classification reveals distinct slow, fast, intermediate, and compressed solar wind regimes whose relative occurrence and temporal organization are strongly modulated by solar activity. This manuscript is part of the Heliophysics Summer School Machine Learning Special Collection.