Predicting solar energetic proton events (E > 10 MeV)

Abstract

A high level of proton radiation exposure can be dangerous to astronauts, satellite equipment, and air passengers/crew flying along polar routes. The presented solar energetic proton (SEP) event forecaster is based on a dual‐model approach for predicting the time interval within which the integral proton flux is expected to meet or surpass the Space Weather Prediction Center threshold of J (E > 10 MeV) = 10 pr cm−2 sr−1 s−1 and the intensity of the first hours of well‐ and poorly connected SEP events. This forecaster analyzes flare and near‐Earth space environment data (soft X‐ray, differential and integral proton fluxes). The purpose of the first model is to identify precursors of well‐connected events by empirically estimating the magnetic connectivity from the associated CME/flare process zone to the near‐Earth environment and identifying the flare temporally associated with the phenomenon. The goal of the second model is to identify precursors of poorly connected events by using a regression model that checks whether the differential proton flux behavior is similar to that in the beginning phases of previous historically poorly connected SEP events and thus deduce similar consequences. An additional module applies a higher‐level analysis for inferring additional information about the situation by filtering out inconsistent preliminary forecasts and estimating the intensity of the first hours of the predicted SEP events. The high‐level module periodically retrieves solar data and, in the case of well‐connected events, automatically identifies the associated flare and active region. For the events of solar cycles 22 and 23 of the NOAA/SWPC SEP list, the presented dual‐model system, called UMASEP, has a probability of detection of all well‐ and poorly connected events of 80.72% (134/166) and a false alarm rate of 33.99% (69/203), which outperforms current automatic forecasters in predicting >10 MeV SEP events. The presented forecaster has an average warning time of 5 h 10 min for the successfully predicted events, 1 h 5 min for well‐connected events and 8 h 28 min for poorly connected events, with a maximum warning time of 24 h for very gradual SEP events.