Empirical forecasting models for peak intensities of energetic storm particles at 1 AU

Abstract

We have investigated the dependence of the peak intensities of energetic storm particles (ESPs) on various parameters characterising the coronal mass ejections (CMEs) and associated phenomena. The aim of this study is to suggest empirical models for forecasting the peak intensities of ESP events at 1 AU based on solar and interplanetary (IP) space observations.For this study we searched for the associations of front-side full and partial halo CMEs with linear speeds >400 km s−1during the years 1996–2015 with IP shocks at 1 AU and ESP events observed near the time when the shock passes the observer. We found 88 CME-driven IP shocks associated with ESP events at proton energy range 5.0–7.2 MeV (nominal energy 6.0 MeV) and 59 shocks at the energy range 15.1–21.9 MeV (nominal energy 18.2 MeV). At these two energies 71% and 68% of the ESP events were associated with solar energetic particle (SEP) events, 85% and 84% were associated with decametric–hectometric (DH) type II radio bursts while 67% and 66% were associated with both.For each CME - shock pair we calculated the predicted shock transit speed (VTR) by using the method of Belov et al. (2022) and used this as the primary parameter in the investigation. We performed correlation analyses between the logarithm of the peak intensities of the ESP events (log10 [IESPpeak]) and the solar parameters related to the CMEs, solar flares, IP shocks, SEP events, and type II radio bursts. When using a single explanatory variable, we found best correlation coefficients for VTR(0.68 ±0.05 and 0.71 ±0.06), the CME space speed (VCMEspace) (0.59 ± 0.05 and 0.68 ± 0.07), and the logarithm of SEP peak intensity (log10 [ISEPpeak]) (0.55 ± 0.08 and 0.70 ±0.08) at 6.0 and 18.2 MeV, respectively. Weak to moderate correlations were found for the logarithm of the soft X-ray flux (log10 [SXRF]) and the logarithm of the duration of DH type II radio burst (log10 [DTII]).Using linear combinations of two or more variables improved the correlations. The best two-variable combination explaining log10 [IESPpeak] was VTRcombined with log10 [ISEPpeak] and the best three- and four-variable combinations also included these two parameters. We found two methods for forecasting ESP peak intensities, one of which can be used for long lead time and the other for medium lead time forecasting. For long lead time forecasting VTR, VCMEspaceand log10 [SXRF] are used. The correlation coefficients between the calculated and observed log10 [IESPpeak] were 0.71 ±0.05 at 6.0 MeV and 0.74 ±0.06 at 18.2 MeV. This method only depends on the coronagraph and X-ray observations at the Sun. For medium lead time forecasting the four parameters used are VTR, log10 [ISEPpeak], VCMEspace(or log10 [SXRF]), and log10 [DTII]. The correlation coefficients were 0.80 ±0.04 at 6.0 MeV and 0.84 ±0.05 at 18.2 MeV. Coronagraph observations at the Sun and solar energetic particle and DH type II burst measurements in IP space are required for this method. The medium lead time forecasting provides an average warning time of 30 ±16 h.