Prediction of Solar Proton Events with Machine Learning: Comparison with Operational Forecasts and "All-Clear" Perspectives

Abstract

Solar Energetic Particle events (SEPs) are among the most dangerous transient phenomena of solar activity. As hazardous radiation, SEPs may affect the health of astronauts in outer space and adversely impact current and future space exploration. In this paper, we consider the problem of daily prediction of Solar Proton Events (SPEs) based on the characteristics of the magnetic fields in solar Active Regions (ARs), preceding soft X-ray and proton fluxes, and statistics of solar radio bursts. The machine learning (ML) algorithm uses an artificial neural network of custom architecture designed for whole-Sun input. The predictions of the ML model are compared with the SWPC NOAA operational forecasts of SPEs. Our preliminary results indicate that 1) for the AR-based predictions, it is necessary to take into account ARs at the western limb and on the far side of the Sun; 2) characteristics of the preceding proton flux represent the most valuable input for prediction; 3) daily median characteristics of ARs and the counts of type II, III, and IV radio bursts may be excluded from the forecast without performance loss; and 4) ML-based forecasts outperform SWPC NOAA forecasts in situations in which missing SPE events is very undesirable. The introduced approach indicates the possibility of developing robust "all-clear" SPE forecasts by employing machine learning methods.

Interpretable Machine Learning to Forecast SEP Events for Solar Cycle 23

“…In their work they conclude that random forests (RF) could be the prediction technique of choice for an optimal sample comprised by both flares and CMEs while proving that the most important features are the CME speed, width and flare soft X-ray (SXR) fluence. Lastly, Sadykov et al (2021) recently indicated the possibility of developing robust "all-clear" SPE forecasts by employing machine learning methods. Their approach indicates that for AR-based predictions, it is necessary to take into account western limb and far-side ARs, characteristics of the preceding proton flux represent the most valuable input for prediction, daily median characteristics of ARs and the counts of type II, III, and IV radio bursts may be excluded from the forecast and that ML-based forecasts outperform SWPC NOAA forecasts.…”

Predicting Solar Proton Events of Solar Cycles 22-24 using GOES Proton & Soft X-Ray Flux Statistics

“…In recent years, there has been a plethora of research projects contributing to the effort of predicting SPEs in an attempt to mitigate their detrimental effects. In our previous study (Sadykov et al 2021) we use SXR wavelength ranges (long (0.1 -0.8 nm) and short (0.05 -0.4 nm)), along with ≥ 10 MeV proton flux data observed by the Geostationary Operational Environmental Satellite (GOES) series. From the various products obtained by GOES, we retrieve and use SXR flux data with 1-minute cadences, and ≥ 10 MeV proton fluxes with a 5-minute cadence.…”

“…The success of utilizing derivatives of these data products alone-specifically an event's preceding proton flux, presents the most valuable input for prediction, allowing us to continue exploring these parameters in depth in this work. Sadykov et al (2021) also discuss the lack of performance loss when excluding daily regional, ground, and space-based radio characteristics of Active Regions (ARs) and type II, III, IV radio bursts when generating predictive algorithms, although acknowledging the brevity of the considered data set. Proton and SXR flux characteristics alone were comparable with predictions based on the addition of both AR characteristics, and , 1986to August 1996, SC 23 from August, 1996to December, 2008, and SC 24 from December, 2008to December, 2019 the inclusion of radio burst counts.…”

“…This is further discussed in Section 3.2. Sadykov et al (2021) also discuss how GOES' recently deployed Energetic Proton, Electron and Alpha Detector (EPEAD) East and West detectors capture different populations of protons given their orientation and alignment with the interplanetary magnetic field structures (Parker 1958). He & Rodriguez (2018) further discuss this, concluding that differences in proton flux measurements between the different GOES detectors (when applicable) are due to the effect of magnetic field configuration variation with geomagnetic longitude.…”

“…Garnering knowledge leading up to such events by studying proton and soft X-ray (SXR) flux data to alleviate the burdens they cause is therefore critical for their forecasting. Our previous SEP prediction study (Sadykov et al 2021) indicated that it may be sufficient to utilize only proton and SXR parameters for SPE forecasts considering a limited data set from Solar Cycle (SC) 24. In this work we report the completion of a catalog of ≥10 MeV ≥10 particle flux unit (pfu) SPEs observed by Geostationary Operational Environmental Satellite (GOES) detectors operated by the National Oceanic and Atmospheric Administration (NOAA), with records of their properties spanning through SCs 22-24.…”