Viola Sembiring
Research Mentor: Liang Zhao
Mentor Department: CLaSP, Engineering
Author(s): Not Available
Session: Session 6 (3:00 PM – 3:50 PM)
Presentation Type: Poster 9
Abstract
Although the solar wind has been studied for decades, understanding its complex and highly variable behavior remains challenging. In-situ solar wind measurements contain large volumes of noisy, high-dimensional data, and traditional analysis methods often rely on predefined thresholds or assumptions that may overlook subtle patterns. As a result, potentially meaningful structures in the data can remain hidden. This research explores how machine learning and artificial intelligence can be used as complementary tools to better understand in-situ solar wind measurements. Using time-series plasma and magnetic field data collected by spacecraft, the project combines data preprocessing, visualization, and machine-learning techniques to examine patterns and variations in solar wind behavior. Rather than replacing physical intuition, these methods are used to assist exploratory analysis by identifying recurring trends, anomalous events, and relationships that may not be immediately apparent through conventional approaches. Preliminary analysis indicates that machine learning models are capable of capturing meaningful structure within the data, distinguishing between different solar wind conditions and highlighting periods of unusual behavior. Visualization plays a key role in this process by helping relate algorithmic outputs back to known physical phenomena, improving interpretability and guiding further investigation. By integrating data-driven methods with domain knowledge in space physics, this study demonstrates the potential of machine learning to enhance the analysis of solar wind measurements. The results contribute to ongoing efforts to better characterize solar wind dynamics and may support future research in space weather forecasting and real-time monitoring.
