WCD-25. Title: Validation of Machine Learning Models for Classification of Solar Wind

Abstract
The Space Weather Prediction Center (SWPC) at NOAA plans to use a machine learning (ML) model with Space Weather Follow-On Lagrange 1 (SWFO-L1) data to improve forecaster situational awareness. The SWFO-L1 mission will operate in a Lissajous orbit at the Sun-Earth Lagrange 1 (L1) point, allowing for measurements of solar wind disturbances before they reach Earth. The mission will provide continuous measurements of the sun's corona and solar wind at the L1 point, transmitting real-time data to Earth. The mission is scheduled to launch in the second half of 2025 and will replace ACE's and DSCOVR's monitoring of solar wind, energetic particles, and the interplanetary magnetic field. As part of this effort, a Gaussian process ML model developed by Camporeale et al. (2017) was validated using data from DSCOVR and ACE. This ML model is a four-category classification algorithm for the solar wind, previously adopted in Xu and Borovsky (2015): ejecta, coronal hole origin plasma, streamer belt origin plasma, and sector reversal origin plasma. The algorithm was trained and tested on a labeled portion of the OMNI dataset, identifying the wind regime based on in-situ observations of wind speed, proton temperature standard deviation, temperature ratio, proton specific entropy, Alfven speed, sunspot number, and solar radio flux. Validation of machine learning models for the classification of solar wind is an important task in space weather forecasting. In order to accurately classify solar wind, machine learning models need to be trained on labeled datasets and validated on new data. The use of machine learning models with SWFO-L1 data has the potential to significantly improve forecasting capabilities for space weather events, allowing for more accurate predictions of space weather impacts on Earth.