A Machine Learning Approach to Forecasting Solar Radiation Storms at Earth
Abstract
Providing advance warning of solar radiation storms, consisting of solar energetic particles (SEPs), remains a challenge for space weather forecasting. While physics-based numerical models can be used to improve our understanding of particle acceleration and transport, these models are not yet capable of providing robust, real-time proton event forecasts required in an operational setting. As an alternative approach, we have investigated the application of machine learning classification algorithms (Logistic Regression, Support Vector Machines, Adaboost) to solar observations to address this problem. These supervised learning algorithms were trained using a historical dataset of flares, coronal mass ejections and associated phenomena from 1986 to 2017 with the goal of predicting whether an SEP event would be observed at Earth. Performance of the models was measured against the statistical PROTONS prediction model currently used in operations at NOAA Space Weather Prediction Center.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFMNG21A..03B
- Keywords:
-
- 1914 Data mining;
- INFORMATICS;
- 1942 Machine learning;
- INFORMATICS;
- 7599 General or miscellaneous;
- SOLAR PHYSICS;
- ASTROPHYSICS;
- AND ASTRONOMY;
- 7999 General or miscellaneous;
- SPACE WEATHER