2. Solar energetic particles
The 1996 July 9, 9:10 UT flare occurred in the NOAA region 7978 on the western hemisphere of the Sun, S W, about eastward of the nominal interplanetary magnetic flux tube connected to the SoHO spacecraft (solar wind speed was km s-1 according to SWE/Wind observations). Gamma-ray signatures of energetic ions and relativistic electrons at the Sun were recorded with the Compton Telescope instrument (Schönfelder et al. 1993) aboard the Compton Gamma-Ray Observatory spacecraft (COMPTEL/CGRO).
2.1. Gamma-ray observation of the flare impulsive phase
Observations at HXR/-ray energies of the 1996 July 9 flare event could be obtained by COMPTEL/CGRO until about 9:16 UT when CGRO entered orbital night. The COMPTEL instrument measured a signal in the Compton-telescope configuration which is sensitive to -ray photons of 0.75-30 MeV, as well as in the two single burst detectors which detect photons in the energy range 100-600 keV and 600-11000 keV, respectively. Fig. 1 gives the measured time profiles and allows to compare the timing of different energy ranges.
Both burst detectors register the maximum of the emission at about 9:09:40 UT followed by two smaller emission peaks. The times agree very well with the peak times measured by the BATSE large area detectors for energies 25-100 keV (denoted t1, t2, and t3 in Fig. 1). All three emission curves originate from bremsstrahlung of electrons with an energy in the range of roughly 50 to a few 100 keV.
The signal in the 3-7 MeV channel originates from nuclear line emission which is emitted when protons and/or ions with an energy of about 10-20 MeV/nucleon interact with ambient material in the chromosphere. As can be seen in Fig. 1 the main emission in this energy range coincides with the third peak at t3 of the electronic emission. A faint signal can be seen also for the first peak at t1, but no significant emission is seen during the second peak at t2. A signal of the 2.2 MeV line was detected well before t3. The emission in the 2.2 MeV neutron capture line is produced with a time delay of about one minute and requires reactions of protons and/or ions with an energy in the range of about 10-100 MeV/nucleon. The detection of this signal confirms that the emission during the first peak stems indeed from nuclear lines, and that protons and/or ions accelerated up to energies of at least 10 MeV have to be present.
It is known that the ratio of accelerated electrons to protons can vary from flare to flare, or within flares (e.g., Chupp et al. 1993; Marschhäuser et al. 1994; Miller et al. 1997). Here, the ratio is highly variable for the three observed peaks: the emission at t1 and t3 includes nucleonic signatures, while during the peak at t2 it appears to be dominated by the electron bremsstrahlung.
From the perspective of the particle injection times the COMPTEL measurements show that the main HXR emission caused by 100 keV electrons occurred at 9:09:40 UT, while the main -ray line emission caused by 10 MeV protons appeared somewhat later at 9:10:20 UT. However, high-energetic protons are present from the beginning of the event.
2.2. Overview of the SEP event
The solar energetic particle event was observed with the energetic particle detectors ERNE (Torsti et al. 1995) and COSTEP (Müller-Mellin et al. 1995) aboard SoHO. The interplanetary proton event was weak enough for ERNE to be able to record nearly every individual particle as pulse height data. For the goals of present analysis, the ERNE recorded protons have been divided into six energy channels: 1.6-3 MeV, 3-6 MeV, 6-12 MeV, 12-15 MeV, 15-20 MeV and 20-30 MeV. The lowest channel has a geometrical factor of 0.260 and the next two 0.915 , thus the statistics are lower on these channels compared to the three highest energy channels possessing geometric factors of .
The time versus inverse-velocity scatter plot is shown in the top panel of Fig. 2. The horizontal lines mark the energy channel limits. Velocity dispersion is seen in the beginning of the event indicating arrival of particles from the Sun. The observed solar wind velocity of nearly 400 km s-1 implies the interplanetary magnetic line length of . The oblique line A in the figure corresponds to this distance traveled from the reference time 9:12 UT - 500 s, the soft X-ray maximum time. We have attempted to estimate first injection time and the distance traveled also from our proton data using a kind of velocity-dispersion technique described by Torsti et al. (1999). However, in 3-12 MeV channels there was also contribution of the previous flare observed on the same day at 7:58 UT (SGD 1996). This adds uncertainty to the estimation. With data in hand we can rule out injection of first protons before 9:08 UT - 500 s and after 9:28 UT - 500 s (with a view to comparison with results of radio observations, we artificially add to and then explicitly subtract from the particle injection time the 500 s value required for radio waves to travel from near-Sun to Earth). Thus based on velocity dispersion, we can estimate the first proton injection time as (9:18 UT - 500 s) 10 min.
Fig. 2 also illustrates an abrupt decrease of proton flux, simultaneously seen in all energy channels at 12:50 UT. This fall of intensity indicates entering a new magnetic flux tube with very different proton transport conditions (Kocharov et al. 1997). The tube named a slow transport channel (STC) had been traversed at 16:00 UT on July 9, 1996. In our present study, we do not consider observations after entering STC. The employed period is limited by vertical lines and in Fig. 2.
On the middle panel of Fig. 2 we present the 5-minute averaged proton intensity in the energy channel of 12-20 MeV employed for the anisotropy measurement. In order to gain better statistics, the channel is wider than in the anisotropy study by Torsti et al. (1997), but otherwise the method is similar. A high anisotropy observed during the entire event (excluding STC) implies a prolonged injection of protons at the Sun. There were two peaks in the injection. The first peak of the 12-20 MeV proton flux was about 90 min long, while only a portion of the second peak was observed before entering a new flux tube at 12:50 UT.
Relativistic electron flux was recorded with the COSTEP instrument (Bothmer et al. 1997). The electron event start corresponds to the injection of first electrons several minutes after the soft X-ray emission maximum, 9:12 UT - 500 s (cf. vertical line A in Fig. 3). Two sub-peaks can be seen at the top of the 0.25-0.7 MeV electron profile (Fig. 3), but dip between them is rather shallow. Entering a new transport channel (STC), observed with protons at 12:50 UT, is only marginally observable in the electron profile (marked with line B in Fig. 3). The STC effect in electrons is much weaker than in protons.
2.3. Scenario of electron injection
The electron intensity maximum (Fig. 3) is too flat and decays too fast to be fitted with an exactly impulsive injection, but the introduction of minor additions before and after the main injection pulse produces a satisfactory fit. We have fitted the SoHO-observed electron event with three impulsive injections, using the injection time and injection strength as free parameters. The transport model parameters have been adopted as explained in Appendix. The best-fit curve in the case of the three-pulse injection is shown in Fig. 3. It fits well the observed intensity-time profile. The deduced injection times at the Sun are 9:17 UT - 500 s (the first minor pulse of injection), 9:26 UT - 500 s (the major injection peak) and 9:58 UT - 500 s (the last minor peak). Contribution of the injections to the total 0.25-0.70 MeV fluence is 9%, 76% and 15% for the E1, E2 and E3 pulses, respectively. Note that the decay of the first minor injection, E1, and also the rise of the last minor injection, E3, were essentially overlaid by the major peak, E2. For this reason, time profiles of the minor injections could not be precisely deduced, but should be regarded as the first electron injection time, while gives a min estimate of the last injection time. Uncertainties in the determination of and do not exceed min. Thus, the major electron injection is impulsive and occurs not earlier than 15 min after the flare start.
An analysis of the pulse-height data for the period 9:18-10:18 UT within the 0.25-1.5 MeV energy range indicates nearly power law differential energy spectrum: (H. Sierks, pers. comm. 1999). The total number of MeV electrons injected at the Sun is per sr of the heliocentric solid angle (i.e. per the area of solar surface at the root of the Earth-connected interplanetary magnetic field line).
The mildly relativistic electrons seen by COSTEP are associated with an electron event observable also at lower energies, from about 1 keV to hundreds of keV, as detected by the 3-D Plasma and Energetic Particle experiment on the Wind spacecraft (Lin et al. 1995). Determining the solar release time from the energy dispersion of the electrons at the spacecraft (Krucker et al. 1999), S. Krucker (pers. comm. 1999) finds that the first electrons are released at the Sun at (9:21 UT - 500 s) 3 min. We have deduced with the COSTEP data the first injection time (9:17 UT - 500 s) 2 min. Given the different techniques of analysis, the onset times of the electrons at Wind and of the mildly relativistic electrons at SoHO are consistent. A mean estimate for the first electron injection time finally is (9:18-9:19) UT - 500 s. That implies that the electron injection into interplanetary space occurs nearly ten minutes after the onset of the radiative signatures of the flare.
2.4. Scenario of proton injection
Proton injection functions are deduced by a careful fitting of the observed anisotropy and intensity-time profiles. A choice of proton transport model is described in the Appendix. Several injection scenarios have been studied. In particular, we attempted to fit the proton intensity-time profile with an injection scenario deduced for relativistic electrons, but it became evident that the proton injection is prolonged and the impulsive injection scenario is not applicable. Several other scenarios were also tested before we decided to use a double-exponential injection profile (the alternative scenarios are described more thoroughly in Sect. 4.2.3). The injection rate is finally approximated in the following form
The fitting parameters, , and , dominate the injection rise, decay and the maximum injection time, respectively (Fig. 4). Those parameters were allowed to be energy dependent and adjusted to get a best fit for each energy channel. The normalization factors determine injection energy spectra for the first and second injection components, . Similar to the 1990 May 24 event study (Torsti et al. 1996), we call the 1st and 2nd injections p-component and d-component injections, respectively.
SoHO is a three-axis stabilized spacecraft and thus ERNE's particle detectors do not cover the whole solid angle, and the magnetic field direction is not stable. In order to take this into account we calculate the differential acceptance of the detector as a function of time and pitch angle and convolve it with the interplanetary transport function to get a value of the proton intensity averaged over the instrument acceptance cone. The resulting intensity curve is directly comparable with the experimental intensities (Fig. 5).
Highest uncertainties are expected in the low energy channels, 1.6-3 MeV and 3-6 MeV. In the lowest energy channel the instrument consists of two detectors with a narrow, , field of view. This results in a high sensitivity to the magnetic field direction, if proton anisotropy is large. In addition, low energy protons arrived too close to the slow transport channel. Also the influence of the earlier flare contaminates the 3-6 MeV channel (Fig. 2). In this view, we had decided for the two lowest energy channels not to vary the maximum injection time, , but to set it equal to the time obtained from the four higher energy channels. Furthermore in the lowest, 1.6-3 MeV, channel, we do not attempt to fit the d-component portion of the intensity profile, i.e. the last 3-4 points in the upper left frame of Fig. 5. Those data points certainly indicate a new rise, but uncertainties of fitting few points would be very large.
The deduced injection functions are presented in Fig. 4. In this figure, we do not show initial portions of injection profiles before the maximum cumulative effect of the injection exceeds the level above background. The portion of injection curves from which the particles would not reach SoHO before entering the new magnetic tube (STC) is also left out. It is seen from Fig. 4 that the maximum injection of the p-component protons occurred around 9:50 UT (the corresponding near Earth light-arrival time is 9:58 UT). No energy dependent trend in the maximum injection time, , is seen. The variance of illustrates statistical uncertainties, min. It is noteworthy that the maximum time of the p-component proton injection, , is close to the deduced time of the last electron injection, . On the other hand, the earliest observable injection of protons occurred close to the time of the first electron injection (see the 12-15 MeV proton injection profile in Fig. 4). The first and the last electron injection times, and , bracket the rise phase of the p-component proton production. Those minor electron injections might be related to protons, and a kind of continual minor production of electrons between and cannot be ruled out.
Amplitudes and (Eqs. 1, 2) give the proton injection spectra at the peak injection time for the p-component and the d-component, respectively. Best power law fits to the deduced differential energy spectra are obtained for the spectral indexes and . Total numbers of the p- and d- component protons injected at the Sun are respectively and per sr of the heliocentric solid angle. Note that the d-component is dominant at low energies and correspondingly in the total proton energetics.
© European Southern Observatory (ESO) 2000
Online publication: August 17, 2000