LARGE SOLAR ENERGETIC PARTICLE EVENTS ASSOCIATED WITH FILAMENT ERUPTIONS OUTSIDE ACTIVE REGIONS

In this paper, we consider only large SEP events defined as those with a proton intensity in the >10 MeV GOES energy channel ≥10 pfu (pfu = particle flux unit; 1 pfu = 1 particle cm⁻².s⁻¹.sr⁻¹). In addition, we required that SEPs were detected in >50 MeV channel for a fair comparison with previous events reported by Kahler (2000). Table 1 provides an overview of the four events. All the four events have been listed in various papers involving statistical properties of SEP events or type II radio bursts (Gopalswamy et al. 2002, 2004, 2005, 2014; Gopalswamy 2012). The filament in the 2002 May 22 event was also studied extensively as it obscured the thermal emission from a neighboring flare (Gopalswamy & Yashiro 2013). The first two columns of Table 1 give the approximate onset time and the peak intensity (Ip) of the SEP events in the GOES >10 MeV channel. Column 3 gives the SEP intensity in the >50 MeV channel to check that these events were similar to the three events discussed by Kahler (2000). The solar source of the eruption is given in column 4 as the heliographic coordinates of the filament centroid. The size of the soft X-ray flare is given in column 5 either from the online Solar Geophysical Data or from our estimate based on the GOES soft X-ray time profile in the 1–8 Å channel. The first appearance time of the associated CME is given in column 6 as obtained from the CME catalog (http://cdaw.gsfc.nasa.gov; Yashiro et al. 2004; Gopalswamy et al. 2009) compiled from SOHO's Large Angle and Spectrometric Coronagraph (LASCO, Brueckner et al. 1995). The sky-plane speed of the CMEs (V_CME) from SOHO/LASCO observations is given in column 7 along with the space speeds in parentheses. The space speeds of the cycle-23 events were obtained from the cone model (Xie et al. 2004). For the last two events, space speeds were obtained using the combined SOHO–STEREO observations fit to the Thernisien (2011) graduated cylindrical shell model (Gopalswamy et al. 2014). The entry “H” in column 8 for CME widths (W) indicates that all CMEs were halos. Column 9 gives the average acceleration (a_r) in the coronagraph field of view (FOV). The wavelength range of the type II burst is given in column 10; this range is a good indicator of where the shock forms and how long it survives (Gopalswamy et al. 2005). Only one event had a type II burst starting from metric (m) wavelengths (starting frequency was 24 MHz). In all other cases, the type II burst started in the decameter-hectometric (DH) domain (below 13 MHz) as observed by the Radio and Plasma Wave Experiment (WAVES; Bougeret et al. 1995) on board the Wind spacecraft. In all cases, the radio emission continued down to kilometric (k) wavelengths, indicating that the shocks survived at least up to 1 AU. This was further confirmed by the shock detections at L1 by the SOHO MTOF proton monitor (Hovestadt et al. 1995); the shock arrival times at L1 are given in column 11. The final column (12) gives references where related information on these events can be found.

The composite plots of the four events in Figure 1 show the proton intensities in three GOES energy channels (>10, >50, and >100 MeV), CME height–time plots with the primary CME of interest along with other CMEs in the time intervals, and the Wind/WAVES dynamic spectra. The plots cover long time periods to show the full evolution of SEP intensity even after the shocks arrived at Earth (marked by the vertical lines in the radio dynamic spectra and the SEP intensity plots). The overview plots illustrate that these are Sun-to-Earth events. The shocks can be readily identified from the jumps in the local plasma frequency (fp) at Sun–Earth L1, where the Wind spacecraft is located. The shocks were also observed by the SOHO/MTOF instrument and listed in http://umtof.umd.edu/pm/FiguresHTML. There was also some indication of the shocks in the particle intensity data as ESP events. The 2002 May 22 event had the largest spike due to the ESP event, with an increase in the >10 MeV channel by an order magnitude. The 2000 September 12 and the 2011 November 26 events showed weak enhancements around the times of shock passage. On the other hand, the 2013 September 29 event showed a sudden drop in the SEP intensity, probably corresponding to the entrance of Earth into the associated ICME. A sudden impulse at Earth was found in all four events in the form of positive values of the Dst index (not shown, but available at http://wdc.kugi.kyoto-u.ac.jp/dstdir/index.html) shortly after the shock arrival times at L1. The impulse on 2002 May 23 was a sudden commencement, followed by a major storm (Dst = −109 nT). The sudden commencement on 2013 October 2 at 3:00 UT was followed by a moderate storm (Dst = −67 nT). The other two events did not produce significant magnetic storms.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In Figure 2, the proton intensity is plotted in several GOES differential energy channels that show the highest energy channel in which there is a discernible enhancement above the background. There was an enhancement above the background in the lowest 3–4 channels. The first and last events had increases in the 80–165 MeV channels, while the remaining two events had an increase only up to the 40–80 MeV energy channel. These observations are consistent with the plots in Figure 1, which show integral fluxes. The 2011 November 26 event shows a tiny enhancement in the >100 MeV (integral) channel, although no discernible increase was observed in the 80–165 MeV differential channel.

Zoom In Zoom Out Reset image size

We also examined the proton intensity observed by the Energetic and Relativistic Nuclei and Electron (ERNE; Torsti et al. 1995) instrument on board SOHO. The intensities are generally consistent with the GOES observations. In order to get the energy spectra, we used the following ERNE energy channels: (1) 14.6–16.9 (15.8) MeV, (2) 16.9–30.0 (23.0) MeV, (3) 30.0–40.5 (35.2) MeV, (4) 40.5–47.0 (43.8) MeV, (5) 47.0–68.8 (58.2) MeV. The numbers in parentheses correspond to the effective energy of the channel. There were two other higher energy channels (effective energy 77.6 and 97.5 MeV) but the background level was too high to measure the SEP intensity, so we did not use them. Figure 3 shows the background-subtracted time-of-maximum differential energy spectra of the four SEP events in the energy range 10 MeV < E ≤ 100 MeV. The fitted spectra have the form dN / dE = kE ^− γ where N and E are the energetic particle density and energy, respectively, k is a constant, and γ is the spectral index. In our estimation of the SEP spectra, we have excluded periods where the proton intensities peaked due to the ESP event. The power-law spectral index γ ranges from 4.15 to 4.69, comparable to the 4.3 value derived for the 1981 December 5 SEP event by Kahler et al. (1986). These values also lie in the high end of the approximate range of 2 < γ < 4.5 for well-connected >10 pfu SEP events of the later survey by Kahler (2001). Those differential spectral fits were determined from the two maximal event values of GOES E > 10 MeV and E > 60 MeV intensities.

Zoom In Zoom Out Reset image size

A variety of observations are available to pinpoint the solar sources of the CMEs. The filaments and flare ribbons were observed in H-alpha by ground-based observatories (Big Bear Solar Observatory (BBSO) in the United States and Hida Observatory in Japan). The photospheric magnetograms are from the SOHO Michelson Doppler Imager (Scherrer et al. 1995) and the SDO Helioseismic and Magnetic Imager (Scherrer et al. 2012). The post-eruption arcades were observed by SOHO Extreme Ultraviolet Imaging Telescope (EIT, Delaboudinière et al. 1995) and SDO Atmospheric Imaging Assembly (Lemen et al. 2012). The soft X-ray light curves are from the GOES.

In the pre-eruption phase, the source regions were large-scale bipolar regions with filaments marking the polarity inversion lines of the magnetic structures. The filaments were best observed in H-alpha filtergrams, as shown in the top row of Figure 4. The FE was accompanied by a two-ribbon flare in each case, with the ribbons located on either side of the original position of the filament. The photospheric magnetograms show that the flare ribbons were located on opposite polarity regions, while the filaments were located on the polarity inversion line. The ribbons in the 2011 November 26 events were confined to the stronger magnetic field region, where the pre-eruption filament was very thin. There was another section of the filament that erupted from the southern end, but no ribbons were found around this section. The post-eruption arcades were also very extended, mostly in the north-south direction.

Zoom In Zoom Out Reset image size

The bottom panels in Figure 4 show that the soft X-ray flares associated with the four SEP events were generally weak, with the flare size ranging from C1.2 to M1.0 and were all listed in the Solar Geophysical Data. The M1.0 flare was associated with the 2000 September 12 event, which also had a weak impulsive phase: microwave emission was observed at the San Vito radio station of the Radio Solar Telescope Network (RSTN) up to 15.4 GHz. The microwave intensity was ∼46 solar flux units (sfu; 1sfu = 10⁻²² Wm⁻² Hz⁻¹). There were no hard X-ray data from Yohkoh/HXT for this event. None of the other flares had impulsive phase microwave emission.

The eruptive filaments can be traced as cores of the associated white-light CMEs (see Figure 5). The CMEs were fast and wide in all four cases. Figure 5 shows snapshots of the CMEs within the FOV of the outer telescope (C3) of LASCO on board SOHO. The images were taken about 2–3 hr after the first appearances. The morphologies of the prominence cores are consistent with the polarity inversion lines and the post-eruption arcades in Figure 4. The CME speeds in the sky plane, as measured from the LASCO images, range from 933 to 1550 km s⁻¹. For the halo CMEs in cycle 23, a cone model has been used to correct for projection effects (Xie et al. 2004; Gopalswamy et al. 2010). The deprojected space speeds were higher: 976–1946 km s⁻¹. For the last two events, Gopalswamy et al. (2014) obtained the space speed from the Thernisien (2011) flux rope model by combining the SOHO and STEREO observations. The space speeds were not too different from the cone model values (1187 versus 976 km s⁻¹ and 1864 versus 1543 km s⁻¹). Since the CMEs were all halos, they are expected to be wide (also clear from Figure 5). Therefore, all CMEs in Table 1 meet the criterion of fast (≥900 km s⁻¹) and wide (≥60°) CMEs to be driving shocks that accelerate particles. The speeds of the CMEs considered in this paper were somewhat higher than the sky plane speed of 840 km s⁻¹ for the 1981 December 5 CME reported by Kahler et al. (1986).

Zoom In Zoom Out Reset image size

We were able to measure the height–time history of three of the four eruptive filaments using EUV images in the early phases and LASCO images in the later phases when the filaments became white light cores inside CMEs. In the case of the 2011 November 26 eruption, the eruptive filament was not clearly visible to make the height–time measurements, so we used the initial location of the filament in H-alpha and followed the filament as the prominence core in white light. We also used the CME leading edge (LE) measurements available at the SOHO/LASCO CME catalog.

The height–time histories of the four events are shown in Figure 6. The speeds of the prominences in the three events were roughly ∼275 km s⁻¹ by the time they reached about 0.5 Rs above the surface. This value is similar to the prominence speed of 305 km s⁻¹ reported by Kahler et al. (1986). The height–time plots show that the separations between the LEs of the CMEs and the corresponding prominence cores steadily increased because of the different speeds, which is well known (Gopalswamy et al. 2003). The core-LE separation was the smallest for the November 26 event and the final speeds differed only by a factor of 1.2. The filaments accelerated to much higher speeds when they became eruptive prominence cores of the white-light CMEs (by more than a factor of 2—see Table 2). However, the CME LE speeds were higher by a factor of up to four. The initial average acceleration of the filaments obtained from the EUV measurements are at the higher end of the values obtained for the general population eruptive prominences (Gopalswamy 2015).

Zoom In Zoom Out Reset image size

Soft X-ray flares were well observed in all four events, so we estimated the initial acceleration of CMEs from the flare rise time and the average CME speed in the LASCO FOV assuming that the CMEs finished accelerating by the time they reached the LASCO FOV (Zhang & Dere 2006; Gopalswamy et al. 2012). The average initial acceleration of the CMEs was in the range from 0.28 to 0.80 km s⁻², much smaller than that of CMEs producing GLE events (average ∼2 km s⁻² see Gopalswamy et al. 2012). The initial accelerations were also generally lower than that (∼1 km s⁻²) of limb CMEs associated with soft X-ray flares of importance ≥C3.0. On the other hand, the accelerations were significantly higher than those of the general population of FE events (≤50 m s⁻²—see Gopalswamy 2015). The accelerations of the eruptive prominences obtained from the height–time plots in Figure 6 are also lower than those derived from the flare rise time and CME LE speed in the LASCO FOV.

Type II radio bursts in the metric and longer wavelengths are of primary interest because they are indicative of shock-accelerated particles. As noted in Table 1, a type II burst was associated with each of the SEP events. A metric type II burst was observed only in the 2000 September 12 event. Type II bursts were observed in the DH and kilometric domains in all four events. Radio observations were not available in the DH domain for the previous EP-associated SEP events, so the shock formation height could not be determined. The highest frequency of the ISEE-3 radio instrument was 1.98 MHz (Knoll et al. 1978), compared to 13.8 MHz in Wind/WAVES and 16 MHz in STEREO/WAVES. The frequency range 2–14 MHz corresponds to plasma frequencies in the 3–10 Rs heliocentric range and hence provides radio observations filling the gap between metric and kilometric observations. Kahler et al. (1986) reported that the type II burst during the 1981 December 5 event was barely discernible at 1.98 MHz. In the present events, the type II bursts were observed from 13 MHz and below.

Figure 7 shows composite dynamic spectra constructed from ground based RSTN observations and Wind/WAVES observations. All the events show fundamental harmonic structure at frequencies below 1 MHz, except the May 22 event, which had an accompanying continuum and a radio data gap. A metric type II burst was observed only during the September 12 event. The metric type II started at ∼48 MHz. There was a weak type II emission between 10 and 5 MHz in the DH domain. Finally the type II appeared with fundamental-harmonic structure at frequencies below 2 MHz. A line-up of the type II features at various wavelengths suggests that the metric type II burst is the harmonic component, which means the type II burst started at a plasma frequency of 24 MHz.

Zoom In Zoom Out Reset image size

The fundamental component at frequencies below about 2 MHz intensified in all four cases. In all events, the type II bursts were weak and of narrow band in the DH domain (the sensitivity of the Wind/WAVES antenna in the 1–14 MHz range is lower than that at lower frequencies), but intensified around 1 MHz. The starting frequency of the November 26 and September 29 events were at 10 MHz with no metric component. The only case with much lower starting frequency is the May 22 event, in which the type II burst started at 0.5 MHz, again with no metric component. In this event, there was an ongoing intense continuum emission between 5 and 0.4 MHz, so it is difficult to conclude on the type II emission in this range. However, the starting of the type II burst at 0.5 MHz is quite clear, so we think there was no type II emission at higher frequencies. The circumstances of the May 22 event are somewhat different, which we discuss later (see below).

The starting frequency of a type II burst is of particular interest because it is indicative of the height of shock formation ahead of the CME. By studying a large number of type II bursts associated with EUV waves or white-light CMEs, Gopalswamy et al. (2013) were able to determine the heights of shock formation. They found that the starting frequency (f) of a type II is smaller when the shock forms at a greater heliocentric distance according to the empirical relation, f = 308.17r^−3.78–0.14, where f is in MHz and r is the heliocentric distance in solar radii. For the 2000 September 12 event, f = 24 MHz, so the shock formation height was r = 1.97 Rs, corresponding to the outer corona. This is consistent with the estimate of ∼1.92 Rs made from the flare acceleration method (Mäkelä et al. 2015). For the November 26 and September 29 events, the shock formation heights were at 2.5 Rs because the starting frequencies of the type II bursts were 10 MHz. For the May 22 event, f = 0.5 MHz, so we get r = 5.1 Rs. This distance is well below the measured CME height (8.48 Rs) in the LASCO FOV. If we use the cone model result, the space speed is larger than the sky plane speed by a factor of 1.15, and we get a CME height of 9.75 Rs. This is almost twice the height implied by the empirical relation. The discrepancy can be readily understood from Figure 5, which shows that the May 22 CME is entering into the region depleted by the preceding CME. By comparing the actual CME height at the type II start with the expected height from statistics, we see that the plasma frequency was lower by a factor of ∼1.9. This means the density was depleted by a factor ∼3.7. An alternative explanation would be that the radio emission originated from the flanks of the shock. However, the Alfvén speed at the flanks is expected to be higher because the plasma level corresponds to the deceasing side of the Alfvén speed profile (Mann et al. 1999; Gopalswamy et al. 2001). We conclude that the severe deviation from the empirical formula is due to the large depletion caused by the preceding CME. This is further supported by the fact that the type II burst at later times (beyond 14 UT after a data gap of 8 hr) was observed at a frequency of 0.2 MHz. At this time, the CME was expected to be at a heliocentric distance of >50 Rs. The type II burst frequency at this time appears normal, which means the CME merged with the preceding one and the resultant CME moved through the normal interplanetary medium (no depletion).

Table 3 shows the shock speeds derived from the drift rates of type II bursts, and CME space speeds are given for comparison. The drift rates were obtained in the 1–13 MHz and <1 MHz bands. These roughly correspond to the outer corona and IP medium, respectively (see e.g., Gopalswamy 2011). In deriving the density scale heights, we assumed that the density n varied as n_or^−α , where r is the heliocentric distance, α is the power law index and n_o is the coronal base density. This form gives a simple estimate of the scale height as r/α. We use the heliocentric distance of the CME nose for r. For cycle-23 events, we use the cone model to estimate the projection effects. For cycle-24 events, the heliocentric distance is obtained by the forward-fitting technique to the STEREO and SOHO coronagraph images. In the IP medium, α = 2, whereas it is higher in the corona. Table 3 gives α values that yielded shock speeds close to the CME speeds. In the 1–13 MHz band, α = 4 was suitable except for the September 29 event, where α = 2 was suitable. Type II bursts were fragmented especially in the 1–13 MHz band, so some of the estimates are uncertain. Similarly, α = 3 was suitable at 0.6 MHz, which corresponds to the transition from the outer corona to IP medium. The main purpose of this analysis is to show that both the shock speed and the CME speed obtained from independent techniques indicate that the CME speeds were very high, suggesting that the SEPs are consistent with shock acceleration.

There were intense type III bursts in all events, but only the cycle-23 events were of long duration at 1 MHz (35 and 40 minutes for the September 12 and May 22 events). The November 26 event had a duration of only 15 minutes, while the September 29 event had the shortest duration (∼10 minutes). The 14 MHz type III burst was either non-existent or of extremely short duration. One interesting point is that the onset of the type II burst preceded that of the type III bursts in all cases, except in the May 22 event. It has been shown before that the existence of a long duration type III burst is not a sufficient condition for SEP association (Gopalswamy & Mäkelä 2010). The two cycle-24 SEP events show that the long-duration type III burst is also not necessary. The soft X-ray flares are the weakest in the cycle-24 events and have shorter-duration type III bursts, suggesting that they may not be related to the CME-driven shock.

The CME leading-edge accelerations derived from the flare rise times and CME speeds in the outer corona were within the range of values obtained for non-GLE large SEP events (Mäkelä et al. 2015). Figure 8 shows the distribution of CME accelerations for 59 large SEP events from cycles 23 and 24 (1997 November–2014 January). The four FE-SEP event values are shown by arrows. Note that all the four CME accelerations are below the mean and median values of the distribution. However, two of the CME accelerations in FE-SEP events (September 12 and May 22) are in the same bin as the most probable value of the distribution. The September 29 event is in the lowest bin. Only the 2011 November 26 event is the least impulsive event with an acceleration of 0.28 km s⁻², which is lower than the lowest bin of the distribution.

Zoom In Zoom Out Reset image size

The peak >10 MeV proton intensities ranged from 80 to 820 pfu in the four FE-SEP events studied in this paper. It must be noted that there are many large SEP events with >10 MeV proton intensities smaller than those of the FE-SEP events. The 1981 December 5 FE-SEP event (Kahler et al. 1986) was not found in the NOAA proton events list compiled since 1976. However, we examined the ISEE-3 plots available in the Solar Geophysical Data (comprehensive report #475) and found that the event had 0.1, 0.017, and 4 × 10⁻⁴ particles per (cm².s.sr.MeV)⁻¹ in the 13.7–25.2 MeV, 20–40 MeV, and 40–80 MeV channels, respectively. By fitting an approximate spectrum to these values and assuming that there were no significant contribution from particles with energies >100 MeV, we estimated a >10 MeV flux of 14.5 pfu for the 1981 December 5 FE-SEP event. Thus the FE-SEP events in the present study are one to two orders of magnitude larger than the 1981 December 5 event.

The spectral index γ at the peak SEP intensity in the 10–100 MeV energy range was generally high (4.15–4.69) matching that of the 1981 December 5 event (Kahler et al. 1986) and in the upper range of the two-point spectral indices of Kahler (2001). We compared the spectra of FE-SEP events with those of two subsets of SEP events: (i) large SEP events associated with weaker soft X-ray flares (C-class or smaller), and (ii) large SEP events with low proton intensity (<30 pfu in the >10 MeV GOES channel). We now examine these two subsets in more detail.

3.3.1. SEP Events Associated with C-class Flares

The large SEP events associated with C-class flares are shown in Table 4 (excluding the three FE-SEP events associated with C-class flares). The soft X-ray importance was in the range C3.7–C9.7. The SEP intensities (Ip) were in the range 14–55 pfu, smaller than those of FE-SEP events. The spectral index γ was obtained by fitting a curve of the form dN/dE = kE^−γ to the ERNE measurements in the 10–100 MeV range as in Figure 2. The sky-plane speeds (V_CME) and widths (W) of the CMEs were obtained from the SOHO/LASCO CME catalog. The eruption locations were all in the western hemisphere similar to the FE-SEP events. The association of a metric type II burst (y = Yes, n = No) burst and/or a DH type II burst (Y = Yes, N = No) is indicated in the last column of Table 4. All events had DH type II bursts and only two (2000 April 4 and 2010 August 14) had both metric and DH type II bursts (yY events). The metric type II burst during the 2000 April 4 event was extremely weak. The fundamental component was reported to be at frequencies below 80 MHz. Our examination of the dynamic spectrum reported in Kahler et al. (2007) indicated that the starting frequency was around 40 MHz. The metric type II burst during the 2010 August 14 event was also brief but had clear fundamental harmonic structure in the e-CALLISTO dynamic spectrum obtained at the Gauribidanur Radio Observatory near Bangalore in India (http://soleil.i4ds.ch/solarradio/qkl/2010/08/14/GAURI_20100814_094458_59.fit.gz.png). The fundamental component started at 60 MHz, indicating larger CME height at type II onset. The empirical formula used in Section 3 suggests that the shock heights were at 1.72 and 1.54 Rs, respectively for the 2000 April 4 and 2010 August 14 events. The DH type II bursts were intense and extended to frequencies below 1 MHz, except for the 2010 August 14 event. In this event, the DH type II burst was extremely narrow banded and ended at ∼3 MHz. Since most of the events in Table 4 had only DH type II bursts, we infer that the large CME height is an important characteristic of these events, a property shared by the FE-SEP events. The main difference between the events in Table 4 and the FE-SEP events is the higher >10 MeV intensities of the latter.

Table 4.  List of Large SEP Events Associated with C-class Flares

S. No.	Date	Ip	γ	VCME	Wa	X-ray	Location	Type IIb
		(pfu)		km s−1		Imp
1	2000 Apr 04	55	4.76	1188	H	C9.7	N25W55	yY
2	2000 Oct 25	15	3.82	770	H	C4.0	N10W66	nY
3	2004 Apr 11	35	4.01	1645	314	C9.6	S14W49	nY
4	2010 Aug 14	14	3.38	1205	H	C4.4	N17W52	yY
5	2012 Sep 28	28	3.25	1035	H	C3.7	N06W34	nY

^aCME width in degrees; H denotes that the CME is a halo, so its true width is unknown. ^bThe lowercase letters (y, n) denote whether or not a metric type II burst was associated with the SEP event (y = yes; n = no); the upper case letters provide similar information for DH type II bursts.

Download table as:  ASCIITypeset image

Figure 9 shows the SEP spectra of the five events in Table 4. The spectral index ranged from 3.25 to 4.76, overlapping with that of the FE-SEP events. Note that two events had spectral index >4 and one had spectral index close to 4 (2000 October 25). The spectral indices of the last two events in Table 4 were well below those of the FE-SEP events.

Zoom In Zoom Out Reset image size

The 2000 April 4 event had the largest γ, even larger than the largest value (4.69) for the FE-SEP events. Puzzled by this, we examined the source region of the 2000 April 4 event in more detail. We found that the eruption was ∼10° to the east of NOAA AR 8933. Therefore, this event should have been classified as a purely FE event. This is further demonstrated in Figure 10 using H-alpha images and magnetograms from the BBSO combined with SOHO/EIT images. A large north-south filament was located to the east of NOAA AR 8933 as shown the H-alpha image taken at 16:34 UT on 2000 April 3 (Figure 10(a)). The filament centroid was at N14W55, whereas the active region was located at N16W66. The BBSO videomagnetogram taken around the same time as the H-alpha image shows that the filament was located in a weak field region (Figure 10(b)). The negative polarity of the weak field region appears to be an extension of the AR polarity. A post eruption arcade observed by SOHO/EIT at 195 Å on April 4 at 17:36 UT (Figure 10(c)) was also clearly to the east of the active region. Finally, the H-alpha image taken after eruption shows that the two-ribbon structure had formed clearly to the east of the active region. Thus, the 2004 April event is indeed an FE event, similar to the other FE-SEP events considered in this paper.

Zoom In Zoom Out Reset image size

The source region of the 2004 April 11 event also was somewhat similar to the 2000 April 4 event: there was a circular filament mostly outside AR 0588. This event was studied in great detail to show the relation between type III bursts, type II bursts, and SEP events (Gopalswamy & Mäkelä 2010). The associated flare had some impulsive phase emission as indicated by the microwave emission at frequencies up to 15,400 MHz (intensity ∼220 sfu with peak value of 920 sfu at 2695 MHz). This event was listed as a disappearing solar filament associated with AR 0588. Thus the 2004 April 11 event is also likely to be an FE event, although it is not as clear as the 2000 April 4 event.

The 2000 October 25 event was also an FE event, but the filament was located between two large active regions 9201 and 9199 in the western hemisphere. Yohkoh soft X-ray images show a large arcade formed between these two ARs. The spectral index was also relatively high (3.82). The remaining two events in Table 4 were from active regions and the spectral indices were smaller.

3.3.2. Low-intensity SEP Events

In the FE-SEP event (1981 December 5) reported by Kahler et al. (1986), the interplanetary type II burst was barely discernible, in contrast to the events in Table 1, all of which show intense type II bursts in the IP medium that could be tracked for long distances from the Sun. The 1 AU shocks were also readily identified. One important outcome of this study is that the lack of a metric type II burst is not necessarily a significant characteristic. However, shock formations at heliocentric distances close to or larger than ∼2 Rs are indicated in the FE-SEP events. The Alfvén speed profiles typically peak around 3 Rs, so the shock formation happens at a distance where the Alfvén speed is still increasing with height. The large CME speeds measured from white-light imagery and type II radio observations make them sufficiently super-Alfvenic to accelerate particles. The continued acceleration of CMEs in the FE-SEP events make them more super-Alfvenic when they reach the declining portion of the Alfvén speed profile as indicated by the intense and long-lasting type II bursts in the IP medium. Mäkelä et al. (2015) reported that the shock formation height was >3 Rs in a few SEP events with metric type II bursts. One has to examine the property of the ambient corona and the CME acceleration profile in detail for these cases, including consideration given to the possible type II formation at the shock flanks.