The proton event in question was observed around 00 UT on 5 June 2011 at the first Lagrangian point (L1) and geosynchronous orbit (GEO). Before discussing the event itself, we describe how we isolated it from the ongoing SEP events. In Fig. 1, we show time profiles of proton flux at similar energies from detectors at five locations, including the twin spacecraft of Solar Terrestrial Relations Observatory (STEREO; Kaiser et al. 2008), i.e., STEREO-A (STA) and STEREO-B (STB), which were located at 95 degrees west and 93 degrees east of the Sun–Earth line. During 4–6 June 2011, the proton flux was much more elevated at STA than at L1 (or GEO) and STB due to two SEP events that were attributable to fast CMEs at 06:48 UT and 22:05 UT on 4 June. These SEP events have been already discussed by Lario et al. (2013) and Lawrence et al. (2014) with a particular focus on energetic electrons and neutrons, respectively, but we focus on energetic protons in this paper.
When measured by Large Angle and Spectrometric Coronagraph Experiment (LASCO) onboard SOHO, the two CMEs were as fast as 1440 km/s and 2524 km/s (projected on the plane of the sky), respectively, and associated with flares at N15W140 and N17W148. From the STA location, these flares were disk events. The 195 Å EUV data obtained by EUV Imager (EUVI) onboard STA indicate that these flares would have been X1 and X8 flares if they had been observed by GOES (Nitta et al. 2013). The estimated CME speeds and flare magnitudes indicate that the second event was more energetic, which also reflected in the proton flux at STA. At STB, from where the flare region was far behind the east limb, only the second SEP event was seen, starting late on 5 June (light blue curve in Fig. 1). Around the Sun–Earth line, both two SEP events were seen, although at much lower levels than at STA (red, green and blue curves). In particular, the Energetic and Relativistic Nuclei and Electron (ERNE) instrument onboard SOHO detected the two SEP events more clearly than Electron, Proton, Alpha Detector (EPEAD) onboard GOES because of its lower background level.
The spiky proton event we study was observed during the decay of the first SEP event and before the onset of the second SEP event at Earth around 03:30 UT on 5 June (see the profile of SOHO/ERNE). Note that it stands out most prominently in the GOES data. At L1, it is considerably weaker in the Wind data and hardly noticeable in SOHO/ERNE data although SOHO/ERNE has much lower background level. The Solar Isotope Spectrometer (SIS) onboard ACE also recorded the event in the 10–30 MeV and 30–80 MeV channels, but the data are available only as quick-look and browse data, so we do not include them in this study. We can rule out the possibility that the second SEP event mentioned above may have contributed to our event, because we expect the onset time (with respect to the CME on the Sun) and rise time to the peak to be correlated—there is no reason to expect a sharp increase of the particle flux long after the CME occurred. In this case, the proton rise time was only of the order of 10 min, whereas the onset time was delayed by more than two hours after the CME. Then, an eruption from a well-connected longitude may have produced a prompt event with short rise time like our event. However, there was no flare or CME around the time of the spiky proton event. Therefore, we expect this event to be produced in situ rather than escaping from an IP shock close to the Sun.
Now, we compare the temporal variations of energetic protons and solar wind parameters to see if our proton event was associated with an IP shock. Figure 2 plots proton fluxes at several energies and solar wind plasma and magnetic field from 15:00 UT on 4 June to 07:00 UT on 5 June. The top four panels (a)–(d) show the proton fluxes from GOES in four energy ranges from 2.5 to 30.6 MeV. The proton fluxes from SOHO in the 26–32 MeV range and Wind in the 19–28 MeV channel are plotted in panels (e) and (f). Panels (g)–(k) shows the solar wind parameters obtained by Wind.
We immediately spot the arrival of a shock wave around 20:06 UT on 4 June at the Wind satellite as indicated by the vertical line labeled 3. Yang et al. (2019), in a statistical study of electron acceleration at IP shocks detected by Wind, described this shock in detail, showing the enhancement of 0.1–4 MeV protons as well as 0.2–40 keV electrons across the shock front. This IP shock had quasi-perpendicular geometry, and the electron energy spectra were interpreted with shock drift acceleration (Yang et al. 2019). The shock was likely driven by the halo CME at 07:40 UT on 2 June 2011 that resulted from two successive eruptions at S19E25 involving NOAA active regions 11,226 and 11,227 (Zhang et al. 2015; Palmerio et al. 2017). The speed of the CME projected on the plane of the sky was 976 km/s as measured in SOHO/LASCO data. During its early evolution, this CME did not produce any SEP events at L1, STA or STB (Kihara et al. 2020).
As the IP shock propagated to the magnetosphere, the 6.5 MeV protons increased at GEO around 20:40 UT (Fig. 2b, line 1). This was coincident with the sudden storm commencement. The flux at this energy stayed high until 01:30 UT on 5 June, passing a second jump around 00:10 UT (line 2) which was more prominent at higher energies (Fig. 2c). However, soon after this second jump (approximately 1 h), the flux decreased rapidly. At higher energies (in particular ~ 30.6 MeV), the proton flux started to increase even before 00:10 UT. It is to be noted that GOES/EPEAD measures energetic proton flux in its eastward and westward directions and, in general, they can differ due to the geomagnetic cutoffs. However, in this event, the fluxes were comparable and showed very similar time profiles due likely to the enhanced solar wind dynamic pressure and subsequent inner motion of the geomagnetic cutoffs (Rodriguez et al. 2010). Thus, the average of the two fluxes is shown in Fig. 2.
A similar, short-duration enhancement was also seen at L1. The proton flux in the > 20 MeV range started to increase ~ 40 min earlier as marked by line 4 (Fig. 2e, f), and this ~ 40 min difference is comparable to our estimate of the time it took for the solar wind to propagate from L1 to Earth. Note that SOHO/ERNE detected other enhancements even before the shock arrival (owing to its lower background level) but they can be attributed to a separate SEP event, as described above.
The short-duration enhancement of energetic proton flux at L1 occurred in the compressed sheath region bounded by the IP shock that arrived about four hours earlier and an ICME flux rope that started around 02:00 UT on June 5 (line 6). The leading edge of this ICME appears to be a reverse-shock-like structure with sharp decreases of the density and magnetic field, correlated with a sharp increase of the flow speed [see Tsurutani et al. (2018) for another example of a “reverse wave” at the leading edge of an ICME]. Thus, the density was relatively high in the sheath region. Also, the solar wind speed increased monotonically across the sheath, indicating that this sheath region was, in fact, being compressed from behind and evolving.
During the time period of the energetic proton enhancement, the magnetic field magnitude of the solar wind dropped substantially and fluctuated rapidly (between lines 4 and 5 in Fig. 2). Because the temporal drop was correlated with the proton enhancement, the drop (which we hereafter refer to as magnetic cavity) will be examined in more detail later. Here, we note that there was a solar wind velocity jump at the peak time of the proton flux, i.e., 00:09 (at Wind) on 5 June, from |V|~ 480 to ~ 530 km/s. The density remained relatively steady across this jump, indicating that this velocity jump was not causing a substantial plasma compression.
To better understand the plasma environment associated with the short-duration, energetic proton enhancement, we studied the orientation of the observed structures using data from Wind, ACE, and ARTEMIS. Figure 3 shows the positions of the relevant spacecraft around the time of the passage of the velocity-jump structure. While GOES was in the magnetosphere, ARTEMIS was in the pristine solar wind and measured basically the same structure. Thus, we used data from Wind, ACE, and ARTEMIS and applied the two-dimensional timing method assuming that the structure was planar and the direction of motion was parallel to its normal direction. The derived normal direction was n ~ (− 0.89, − 0.46) and the propagation speed was V ~ 439 km/s. This result is also shown by the gray line and arrow in Fig. 3. Similar results were obtained when we used the timing of detection of the shock structure (at 6/4 20:06 at Wind in Fig. 2; n ~ (− 0.79, − 0.61), V ~ 432 km/s) and the reverse-shock-like structure (at 6/5 01:58 at Wind in Fig. 2; n ~ (− 0.87, − 0.49), V ~ 307 km/s). The estimated orientation is reasonable considering the fact that this IP shock has been attributed to solar eruption events at S19E25. While we interpreted earlier that the ICME sheath region was being compressed from behind and evolving, the substantially smaller speed of the reverse-shock-like structure further suggests that the ICME sheath region was, in fact, expanding and hence the reverse-shock-like structure was propagating sunward in the ICME-sheath rest frame.
Based on the 6-h difference between the forward shock and reverse-shock-like structure (i.e., between lines 3 and 6), the thickness of the compressed region (or ICME sheath) is estimated to be at least ~ 1040 RE, (= 307 km/s times 6 h) which is much larger than the gyroradius of ~ 30 MeV protons with the magnetic field magnitude of ~ 15 nT, i.e., ~ 8 RE. Also, the estimated propagation speeds are roughly equal to the instantaneous solar wind speed, indicating that the time profiles in Fig. 2 are spatial variations rather than temporal variations. In fact, the 40-min difference mentioned above (the difference between lines 1 and 3) as well as the peak time difference (the difference between lines 2 and 4) can be roughly explained by the travel distance between Wind and GOES.
Let us now examine more details of the magnetic cavity identified above. Figure 4 shows zoomed-in views of the magnetic cavity observed by Wind (brown curves), ARTEMIS (blue curves), and ACE (black curves), demonstrating a few notable features of the magnetic fluctuations in the cavity. Here, the ACE data are shifted in time to account for the solar wind propagation between spacecraft. It is evident that all spacecraft observed various fluctuations in the cavity. For example, at 00:12 UT, Wind and ACE detected a current-sheet-like structure, i.e., a sharp drop in the magnetic field magnitude (Fig. 4a) and a large rotation of the magnetic field direction (Fig. 4c). The same magnetic shear was also observed by ARTEMIS at 00:44 UT (Fig. 4e, f). However, discrepancies can also be found between ACE and Wind (Panels a, b, and c). For example, in the Wind data, there was a magnetic field rotation during 23:35–23:55 on 4 June but not in the ACE data. While the latitudinal angle (θB) varied from ~ 80° to ~ − 60°, the longitudinal angle (ϕB) varied from ~ 180° to ~ − 30°, although the two angles remained steady in the middle 23:40–23:50. There were also much shorter-scale (a few second) variations throughout the cavity in the Wind data (but much less frequently in the ACE and ARTEMIS data). Many of these shorter-scale variations are the so-called magnetic holes in which magnetic field magnitude decreases down to a few nT (e.g., Turner et al. 1977; Winterhalter et al. 1994; Haynes et al. 2015; Roytershteyn et al. 2015; Volwerk et al. 2020; Madanian et al. 2020). Magnetic holes are typically characterized as localized (a few second duration) depressions in the magnetic field magnitude with little (< 10°) rotation. It has been reported that magnetic holes are often observed in a high-beta condition associated with stream-interaction regions.
Because of the presence of the various magnetic fluctuations during the proton flux enhancement, we further examined the frequency spectrum and hence the turbulence properties. Figure 5a shows the magnetic field spectra obtained just before (pink; 4 June 2011 22:30:00–23:15:00 UT) and during (light blue; from 4 June 2011 23:30:30 to 5 June 2011 00:30:00 UT) the observation of the magnetic cavity, demonstrating the presence of developed turbulence. Here, the power spectral densities are normalized by the background magnetic field magnitude squared. It is evident that the spectra show a power law and that the turbulence was enhanced in the cavity. In the frequency range near and below the ion cyclotron frequency (the arrows), the power law indices before (red) and during (blue) the cavity were ~ 1.7 and ~ 1.8, respectively, comparable to the typical value in the solar wind.
Figure 5b shows the background-subtracted, proton energy spectrum obtained by SOHO/ERNE, demonstrating that the spiky flux enhancement above 10 MeV (i.e., inside the cavity, colored blue) exhibited a power-law spectrum that is harder than that of before the spike (i.e., outside the cavity, colored red). Also shown for reference is the background spectrum (dashed curve) which contained a bump in the highest energy range (> 50 MeV) due to an artificial noise. The spectrum in the < 10 MeV range stayed almost the same across the cavity, suggesting that the acceleration mechanism in the cavity was different from the mechanism that produced < 10 MeV protons. Another caveat is that the SOHO/ENRE was also detecting a decay phase of an unrelated SEP event, as described earlier and shown in Fig. 1. Thus, to minimize the effect of the SEP, we used a narrow time period to show the pre-spike spectrum, i.e., 22:00–23:00 UT of 4 June 2011, as opposed to the entire time period of ICME sheath region before the spike.