4. SEP injection and its relationship with coronal processes
The remote sensing data show different signatures of rapidly and slowly evolving energetic particle populations in the corona: -ray, hard X-ray and centimetric radio emission from the initial impulsive acceleration of electrons and protons in the low atmosphere, long lasting acceleration of the electrons emitting the broadband continuum, repeated episodes of impulsive acceleration in the corona leading to the broadband enhancements, and shock acceleration of electrons in two different altitude ranges producing type emission. Continued changes in the structure of metric radio sources show the persistence of electron acceleration until at least five hours after the start of the flare. Both impulsive and long lasting processes of particle acceleration were independently inferred for the electrons and protons of the SEP event. We now discuss the possible relationship between the coronal processes and the particle populations seen at 1 AU, referring especially to Fig. 6.
4.1. Impulsive phase acceleration
We identify the impulsive phase with the first minutes of the event (9:09-9:11 UT), when bright microwave and moderately intense hard X-ray and -ray emissions are observed. Although no -ray and hard X-ray data are available after 9:16 UT, because CGRO is in the Earth's shadow, the microwave profile and the continuous decrease of the soft X-ray flux suggest that the impulsive phase is finished at that time. From the absence of emission at metric and decametric wavelengths during the impulsive phase we conclude that the accelerated particles are initially injected into closed magnetic structures of the low atmosphere. This would also explain why no signature of the mildly relativistic electrons and the energetic protons accelerated during the impulsive phase is observed at 1 AU, although the flare occurs close to the nominal Earth-connected magnetic field line. The subsequent drift of the low-frequency limit of the radio emission towards lower frequencies shows that particles (at least electrons) have successively access to more dilute plasmas, i.e. higher coronal altitudes. The association with a CME probably means that the low lying magnetic structures are gradually opened and expand.
Similar events where no interplanetary counterpart of impulsive phase acceleration was detected, although the satellite was nearly well connected to the flare site, were reported in the literature. For instance, during the large SEP events on 1990 May 24 at W (Torsti et al. 1996; Debrunner et al. 1997, Kocharov et al. 1999a) and 1991 June 15 at W (Kocharov et al. 1994; Akimov et al. 1996) the comparison of -ray and neutron observations with particle measurements in space showed that protons started to be injected into flux tubes connected with the satellite minutes after interacting protons with comparable energies were detected in the low solar atmosphere. The radio spectrum of the 1991 June 15 event has a similar drift towards lower frequencies as 1996 July 9 (Akimov et al. 1996). No dynamic spectrum at meter waves is available for the 1990 May 24 event, but Lee et al. (1994) show that the microwave spectrum displays a systematic gradual drift from 10 GHz to 1 GHz within 3 min. However, many impulsive flares with prominent hard X-ray and -ray emission do not display such delays (Raoult et al. 1985), and interplanetary electrons were observed to be injected during the impulsive phase of solar flares (Bieber et al. 1980; Kallenrode & Wibberenz 1991). We surmise that the access of particles from coronal acceleration sites to interplanetary space is different in different events. This could be ascribed to different configurations and evolutions of the magnetic field in the vicinity of the acceleration site(s). A key role of a CME is that it will open magnetic fields (cf. also Manoharan et al. 1996). The CME is then an essential ingredient of the energetic particle event, irrespective of whether it contributes to the acceleration or not.
4.2. Post-impulsive acceleration
4.2.1. Shock acceleration in the middle corona
The type II burst emission at decimetric-to-hectometric wavelengths is evidence for electron acceleration at shock waves in the middle and high corona. If the shock speed derived from meter wave observations is 1000 km s-1 (Klassen et al. 1999), a single large-scale shock cannot generate the type II bursts in the two frequency ranges. The hectometric type II burst suggests a velocity 600 km s-1. This is comparable to the CME leading-edge velocity, 400-450 km s-1, observed at a later time when the leading-edge was already at (Pick et al. 1998, Fig. 8). The hectometric type II emission is probably associated with the CME when it was between 1 and 2 above the photosphere.
The time coincidence of the first hectometric-to-kilometric type burst and the decimetric-metric type signature of a coronal shock strongly suggests that at least the electron beams that generate the initial part of the hectometric burst are accelerated at the coronal shock wave, especially since no simultaneous decimetric type emission is seen in the low corona. Escape of electron beams from the shock front is also directly indicated by the observed decametric type bursts. This and a possible acceleration mechanism were discussed by Mann et al. (1997) for the 1996 July 9 event, while Bougeret et al. (1998) presented similar evidence for a different event. On 1996 July 9 the electron beams producing the first hectometric type burst group are accelerated about 6 minutes before the first identified injection of mildly relativistic electrons measured by COSTEP () and also before the injection of suprathermal electrons detected onboard Wind.
Instead of inferring from the particle data and comparing it with the radio timing, we can use the radio time profile at frequencies where the type burst is observed and compute the expected time evolution of the electrons at SoHO. Such a computation for the frequency of 327 MHz results in the expected rise of the electron intensities at 1 AU more than 5 minutes earlier than observed. This confirms that acceleration at the shock wave in the middle corona does not contribute significantly to the mildly relativistic electrons seen by COSTEP, in disagreement with the claim of Mann et al. (1997). Similarly, no significant amount of MeV protons from the shock is detected.
A possible interpretation of this result is that the energy of the electrons accelerated by the shock wave producing the decimetric-metric type burst does not exceed a few tens of keV. Electrons in this energy range may be hidden in the Wind data by a preceding event (S. Krucker, pers. comm.). The result is also consistent with the statistical study performed by Hucke et al. (1992), which indicates that the presence of a type burst enhances the flux density of an associated hectometric type burst without changing significantly the intensity of mildly relativistic electrons. We cannot exclude, however, that particles accelerated by the shock remained undetected because of a poor magnetic connection to Wind and SoHO.
4.2.2. Acceleration of mildly relativistic interplanetary electrons
The COSTEP electron observations were ascribed in Sect. 2 to the mainly impulsive acceleration peaking at . Around this time a broadband enhancement is observed from microwaves down to the low-frequency limit of the Tremsdorf spectrograph (Fig. 6e), with evidence for the acceleration of mildly relativistic electrons. The simultaneously occurring second hectometric type group shows electrons escaping from the corona. Therefore the processes which accelerated the radio emitting electrons in the low and middle corona are the prime candidate for producing the main injection of mildly relativistic electrons into interplanetary space. The hectometric type emission corroborates the impulsive character of the main injection of mildly relativistic electrons into interplanetary space.
Coronal shock waves of probably different origins are seen through their radio emission before and after the main injection of relativistic electrons. If one of these shocks is to be invoked as the accelerator, ad hoc explanation is required for the short duration of the electron acceleration to mildly relativistic energies. In the high corona, the shock is expected to accelerate more or less continuously. This is different from the middle corona, where shock acceleration can have different consequences, depending on the magnetic structures encountered (cf., e.g., scenarios proposed by Stewart & Magun 1980; Aurass et al. 1998; Bougeret et al. 1998). The temporal coincidence of the injection E2 with the broadband brightening from the low and middle corona argues in any case for acceleration behind the shock, and independent of it.
There is no decimetric-to-metric counterpart of the last minor pulse of mildly relativistic electron injection, E3. It occurs not far from the last brightening of the hectometric type burst, but other, similar brightenings, occurred before. A more or less continual minor acceleration of electrons by the shock or another proton accelerator, starting at and ending at , cannot be excluded. An alternative is acceleration in the course of the coronal processes which give rise to the slowly drifting south-western radio source illustrated in Fig. 7. At 410 MHz, the source peaked at 9:45 UT. Using the source intensity-time profile at 410 MHz as the injection function, we find a good correspondence with the last phase of the electron event observed by COSTEP (Fig. 8, top center panel).
4.2.3. Acceleration of interplanetary protons
The proton injections have a fundamentally different character from those of the electrons. The time evolution is very slow, inconsistent with the impulsive behavior of the mildly relativistic electrons (see the left column of Fig. 8). The slow evolution of the metric-decametric continuum is similar to the p-component injection profile (cf. Fig. 6e, especially in the range 40-90 MHz). As an illustration, in the right lower panel of Fig. 8 we compute the proton intensity profile assuming that the proton injection is proportional to the logarithm of the whole Sun emission at 90 MHz. While there is not exact coincidence between the calculated and observed profiles, a similarity could be seen. The p-component of proton production is correlated with a radio continuum in the 40-90 MHz range. The corresponding height above the photosphere is 0.3-1 (e.g., Mann et al. 1999). Thus the time evolution of the p-component proton production accompanies two signatures of coronal acceleration:
The CME is important for the p-component proton acceleration, because it is related to both processes. However Kahler (1982) concluded from a statistical comparison of SEP associations with type bursts and broadband continua that there is no evidence that the type shock accelerates the protons detected in space. The type shock signature ceases at about 10 UT. Even if this does not prove that the shock decays in the high corona, it is in line with the measured low speed of the CME and the absence of kilometric type emission, which Cane et al. (1987) showed to be an unambiguous signature of fast CMEs in interplanetary space. Indeed no shock signature was detected in interplanetary space despite dedicated research (Dryer et al. 1998).
The p-component injection was followed by the delayed rise of the second (d-component) proton acceleration. The d-component proton injection has no suggestive coronal counterpart. This acceleration produces a steeper spectrum and a higher particle flux at energies below 6 MeV than the first acceleration. In terms of the energetics of the proton population the delayed injection is more important than the p-component injection, but it is not dominant at high energies. Although the radio and soft X-ray observations show that energy is continuously released during several hours in and around the active region, nothing indicates that the amount of energy release increases until the maximum of the delayed proton injection near 11 UT. An earlier onset of the second injection at lower energies (Fig. 4) indicates a new, slow acceleration of protons after the first production has decayed. These observations support the idea that the proton acceleration is driven by the CME. In this case the acceleration of the delayed proton component would be maximum at projected heights of about (cf. Fig. 8 of Pick et al. 1998and Fig. 10 of Dryer et al. 1998).
4.3. Event classification
Conventional classification of SEP events had its origins in the impulsive/gradual distinction of flare microwave and soft X-ray emissions being related to parameters of corresponding energetic particle events (Cane et al. 1986, Cliver 1996, and references therein). The "two-class" picture has been recently extended by separation of each class into two subclasses (Kallenrode et al. 1992, Cliver 1996). According to the expanded classification system, there are two types of impulsive events: (Ia) impulsive events enhanced in , and (Ib) impulsive events associated with CMEs (see "Pure Impulsive" and "Mixed-Impulsive" columns in Table 2 of Cliver 1996).
How well does the 1996 July 9 event fit into this classification? The SEP event was weak, maximum proton intensity . For this reason, only few Fe ions were detected by ERNE, but the He intensity was sufficient to deduce the helium-to-proton ratio and use it for the event classification. The time integrated He/p ratio is high, , so that the event might be (cf. Kocharov & Kocharov 1984, their Figs. 7 and 8). However, a search for led us to conclude that the event was not very rich in , the time-integrated ratio . We have searched for a temporal evolution of the -to-proton ratio. However, unlike the electron-to-proton ratio, no significant difference between the helium abundances of first and second intensity-peaks (p- and d- components) is found. It is important that the ratio of the numbers of mildly relativistic electrons to energetic protons significantly exceeds 100, the value used to distinguish impulsive events from gradual events: . All these facts, the impulsiveness of associated soft X-ray flare, presence of CME, high electron-to-proton and helium-to-proton ratios, but no high enhancement in , fit well properties of the subclass Ib ("Mixed-Impulsive" events according to the terminology of Cliver 1996).
We do not observe significant injection of solar energetic particles during the flare impulsive phase as well as after the CME arrival at distances . Therefore, the mixed characteristics of the event do not come from the plain summation of flare accelerated and interplanetary-CME accelerated particles in consequence of their transport and registration in the interplanetary space. The properties of particle populations observed in space during this event are provided by one or several impulsive and two time-extended particle injections. Probably for this reason and also in this sense the event has mixed characteristics.
© European Southern Observatory (ESO) 2000
Online publication: August 17, 2000