Skip to main content
NASA Advanced Supercomputing (NAS) Division World-class modeling, simulation, and supercomputing in support of the NASA mission

Machine Learning Tools for Predicting Solar Energetic Particle Hazards: Progress and Plans

AMS Seminar Series

Speaker: Viacheslav Sadykov, BAERI / NASA Ames Research Center

Presentation

Seminar Slide Deck (PDF-42MB)

Abstract

Solar Energetic Particles (SEPs) are among the most dangerous transient phenomena of solar activity. Representing hazardous radiation, SEPs may affect the health of astronauts in the open space and create difficulties for current and future space exploration. In this seminar, the first-year progress on the “Machine Learning Tools for Predicting Solar Energetic Particle Hazards” project under the NASA Early Stage Innovation STMD program will be presented. The primary objective of the project is to enhance predictions of SEPs by implementing automatic data characterization and Machine Learning (ML) tools. The first part of the seminar highlights the development of an online-accessible database that integrates a variety of solar and heliospheric data, metadata, and descriptors related to Solar Proton Events (SPEs). The database is supported by the Application Programming Interface (API) and a web application for data search and retrieval. The second part of the seminar is dedicated to the development of “all-clear” forecasts of SEP events. In particular, we consider a problem of the daily forecasts of SPEs based on the properties of the magnetic fields in solar Active Regions (ARs), preceding soft X-ray and proton fluxes, and statistics of the solar radio bursts. The predictions of the ML model are compared with the SWPC NOAA operational forecasts of SPEs. Our preliminary results indicate that 1) it is necessary to account for the ARs at the Western limb and the far side of the Sun for the AR-based predictions, 2) characteristics of the preceding proton flux are the most important for the prediction, 3) daily median properties of the ARs may be excluded from the forecast without the significant performance loss, 4) ML-based forecasts outperform SWPC NOAA forecasts in the situations when missed SPEs are very undesirable. The introduced approach indicates a possibility to develop a robust “all-clear” SPE forecasts by employing machine-learning methods.

Biography

Photo of Viacheslav Sadykov
Viacheslav Sadykov is a Research Scientist at the Bay Area Environmental Research Institute (BARI) affiliated with NASA's Ames Research Center. He obtained his Ph.D. in 2019 from the New Jersey Institute of Technology, and his master’s degree in Applied Mathematics and Physics in 2015 from the Moscow Institute of Physics and Technology (MIPT). His research interests include problems of prediction of solar transient events (solar flares and solar energetic particles); development and maintenance of databases of solar flares and flight radiation measurements; machine learning-aided analysis of multiline spectral observations by the NASA’s IRIS satellite and other facilities; modeling of solar spectral lines and EUV emission (radiation transfer problem) observed by IRIS, HMI/SDO, and AIA/SDO; analysis of the realistic radiative MHD simulations of the quiet Sun.