Enhanced fluxes of solar energetic particles (SEP) - so-called SEP events - result from processes of explosive energy conversion in the corona, notably flares and coronal mass ejections. The highest energies of solar energetic nucleons detected in space or through gamma-ray emission in the solar atmosphere are in the GeV range. Where and how the particles are accelerated is still controversial. The most debated candidate processes are related to magnetic reconnection and to the shock wave driven by a fast coronal mass ejection (CME).
We search for observational indications on the acceleration site of relativistic protons in SEP events, via a comparative analyses of their timing, observed by neutron monitors on the Earth, and electromagnetic emissions of the associated eruptive solar activity. This is usually a difficult task, because the time profiles of the SEPs at 1 AU have been strongly smeared-out by the propagation in the turbulent interplanetary magnetic field. Since the scattering cross section is smaller at relativistic than at lower energies, relativistic proton events, also called ground-level enhancements (GLEs), give the best chance to observe a temporal relationship. It is probably for this reason that two successive pulses could be distinguished in the proton time profile of the 20 Jan 2005 GLE. We studied the first pulse in Masson et al. (2009), and the second in a recent publication (Klein et al., 2014). Here we give a brief outline of the results.
Figure 1 shows the intensity-time profiles of the relativistic protons (Fig. 1.e) derived from the observations of the worldwide network of neutron monitors (Bütikofer et al. 2006, Masson et al. 2009), in comparison with soft X-ray (1.a) and broadband radio emission (1.b-d). The proton profiles display two successive peaks, related to different features of the radio emission. Different acceleration episodes are labelled 0 to 6 in the figure.
1 - Relativistic proton release during the impulsive flare phase: intervals 2 to 4
The time intervals labelled 0-4 indicate the pre-flare and impulsive phase of the event, shown in this Figure by the microwave flux density profile at 15.4 and 17 GHz (red curve, Fig. 1.b). Type III bursts below 14 MHz (Fig. 1.d) show that electrons get access to interplanetary space. The proton profile (Fig. 1.e) was shifted backward in time by 216 s, such as to make its rise coincide with the onset of the strong type III bursts. When this is done, the release of the first relativistic protons detected by neutron monitors on Earth is found to coincide with the start of gamma-ray emission at photon energies above 60 MeV. This emission results from the decay of pions, produced by the interaction of energetic protons above 300 MeV with chromospheric material. The pion-decay emission hence demonstrates the acceleration of protons to relativistic energies during the impulsive flare phase, and the type III bursts demonstrate that particles accelerated during this flare phase have access to field lines open to the interplanetary space. The set of complementary observations suggests that the first proton pulse detected on the Earth comes from particles accelerated during the impulsive flare phase. The backward time shift of the relativistic proton profiles at Earth by 216 s is consistent with an interplanetary path length of about 1.5 AU. This is longer than the nominal Parker spiral, but it is consistent with field line lengths during disturbed interplanetary conditions, when the magnetic field is distorted by CMEs from previous solar events. Masson et al. (2012) showed that this was actually the case on 20 Jan 2005 and in other GLEs. The detailed observations and arguments on the impulsive phase acceleration are given in Masson et al. (2009).
2 - Late relativistic proton release: interval 5
The time intervals 5-6 (see Fig 1) comprise the second peak of the relativistic proton profile. The radio emissions near its onset display a new peak of low-frequency microwaves (2.7 GHz), with only a weak counterpart at 17 and 15.4 GHz, a slowly drifting burst with narrow bandwidth between 700 and 100 MHz, and another decametric-to-kilometric type III burst below 14 MHz. These emissions define the acceleration episode labelled 5. Of particular interest is the drifting radio burst in the 700-100 MHz range. Its bandwidth and drift rate are reminiscent of a type II burst, and hence of a coronal shock wave. The time correspondence of this signature with the second pulse of the relativistic proton time profile could hence be understood as evidence of the acceleration at the bow shock of the fast CME that was associated with this event (Pohjolainen et al. 2007, Grechnev et al. 2008).
A detailed analysis of the dynamic spectrum observed by the ARTEMIS spectrograph of the University of Athens contradicts this interpretation. Figure 2 shows the dynamic spectrum in more detail: the two panels on the lower right display the flux density spectrum of the drifting feature (top frame) and the difference spectrum, where fine structure is more clearly seen. The drifting burst has none of the characteristic fine structures of type II bursts, such as the presence of a fundamental and harmonic band or the splitting of the band in two or more sub-bands. Instead the high-resolution spectrum taken with the acousto-optical spectrograph shown in the upper left part of Fig. 2 displays fine structure characteristic of a type IV burst. The two frames result from analysing the same spectrum using different filtering techniques. The upper spectrum shows fiber bursts in the 280-380 MHz band and zebra pattern (e.g., 360-420 MHz, 6:55:20-6:55:35 UT), the lower spectrum broadband pulsations. The drifting feature is hence not a type II burst, but a type IV burst. This changes the interpretation of the origin of the second relativistic proton release.
3 - Interpretation of the late relativistic proton acceleration
Type IV bursts are a typical radio counterpart of CMEs. They are ascribed to electron acceleration during the CME development, more specifically to acceleration in the post-eruptive current sheet that develops in the aftermath of the CME. We have no imaging observations of the type IV burst on 20 Jan 2005. But a number of studies of other events show that the frequency-drifting emission comes from electrons confined in a rising magnetic structure, likely in the flux rope of the CME. In the 20 Jan 2005 event, this acceleration episode is accompanied by a new rise of UV emission in flare kernels, emphasising that the region of energy release conserves a connection with the low solar atmosphere.
The time coincidence between the type IV radio emission and the second pulse of relativistic protons suggests that these protons are accelerated in the same or closely related regions as the radio-emitting electrons. This points to the importance of magnetic reconnection and/or turbulence in the post-CME corona for post-impulsive relativistic particle acceleration. The idea dates back to Carmichael's early flare model, and was developed by Litvinenko and Somov (1995), Ryan (2000), and others. It shows that a fast CME may be crucial for relativistic SEP acceleration even if it is not the shock wave that accelerates the particles. There is indeed no suggestive time coincidence of the proton releases with type II emission during the 20 Jan 2005 event, although a type II burst is observed by Wind/WAVES during episode 6 (cf. Fig. 1.d).
Figure 1 : X-ray and radio emission and the relativistic proton profile of the 2005 Jan 20 event. From bottom to top: (a) soft X-rays λ = 0.1 - 0.8 nm (dark line) and 0.05 - 0.4 nm (light; red in the colour plot of the online version); (b) microwaves (dark line 2.7 GHz, light - red in the colour display - a combination of 17 GHz (NoRP) before and 15.4 GHz (LEAR) after 06:55 UT); (c) dynamic radio spectrum at dm-m waves (ARTEMIS-IV; inverse colour scale; 1 s integration time); (d) decametre-kilometre wave radio emission (Wind/WAVES; inverse colour scale; 1 min integration); (e) proton flux time history at 2 GV (dark curve) and 5 GV (red) rigidity (kinetic energy 1.27 and 4.15 GeV, respectively), time axis shifted back by 216 s. The intervals delimited by vertical lines and numbered 0 to 6 are different episodes of particle acceleration.
Figure 2: Details of the dynamic spectrum of the drifting burst in episode 5 (ARTEMIS IV). The two spectra in the lower right corner show the flux density and its time derivative, respectively. The tow spectra on the upper left show a detailed view, at high time resolution, during part of the drifting feature (as delimited by the red rectangle). The two spectra result from different filtering procedures, showing in the top panel fiber bursts and zebra pattern, and in the bottom panel broadband pulsations.