Research On Characteristics Of X-Class Solar Proton Flare

Solar proton events(SPE) is one of the major hazardous space weather events, bringing in energetic particles which may damage satellite fragile device or threaten the safety of astronaut. Therefore, forecasting SPEs, especially short term forecast, is essential for practical application in space weather prediction.Solar proton events are often associated with solar flares. The probability that a solar proton event may occur is larger after X-class flares than other types of flares. Many phenomena are easier to observe during X-class flare occurrence. Thus, we select X-class flares as our research sample. In order to improve the accuracy in the forecast of solar proton events triggered by X-class flares, we analyze statistically the relation between proton/non-proton flares and various space environmental parameters, including X-ray integrated flux, source region, CME speed, CME angle width, ambient solar wind speed, and ambient X-ray flux. We finds that the distribution of X-ray integrated flux, heliolongitude, CME speed, CME angle width are different for non-proton flare and proton flare. Non-proton flares are mostly located at eastern and western part, side with X-ray integrated flux less than 0.3J/m2 and CME speed less than 1300km/s, meanwhile proton flares are mostly located at region with central and western part, side with X-ray integrated flux greater than 0.3J/m2,and CME speed greater than 1300km/s. Furthermore, proton flares are often associated with halo CMEs, and while non-proton flares with CMEs whose angle width has a decentralized distribution. As for the ambient solar wind speed and the ambient X-ray flux, the distribution trends of proton flares and non-proton flares are similar. Adopting the different characteristics of the SPEs for proton flares and the non-proton flare, We can improve the accuracy of SPE forecast.We choose X-ray integrated flux, heliolongitude, CME speed, CME angle width and 1MHZ radio time integrated flux as a set of predictors. For our purpose, We adopt SVM(support vector machine) since SVM is optimal for small sample with high confidence. Based on our analysis, the success rate is 84.7% and the false alarm rate 15.4%。Moreover, the false alarm rate can be reduced by considering more parameters that are related to solar corona or interplanetary medium.