AN ANALYSIS OF INTERPLANETARY SOLAR RADIO EMISSIONS ASSOCIATED WITH A CORONAL MASS EJECTION

This favorable configuration of SOHO and STEREO allows us to apply the graduated cylindrical shell (GCS; Thernisien et al. 2009) model to coronagraph images simultaneously recorded by these three spacecraft in order to reconstruct the CME leading edge in three dimensions (3D). We performed this analysis using images taken between 20:00 UT and 23:30 UT on November 29, when the CME was well observed by all three spacecraft (Figure 1). From GCS fitting, we estimate the radial distance of the leading edge r_GCS, leading to a calculated average CME speed of 761 ± 13 km s⁻¹ with a launch time of 19:42 UT. The CME angular half-width λ_GCS was estimated to be 59°. The HEEQ longitude of propagation of the CME from GCS fitting was 128°, which is roughly toward MESSENGER. To calculate the speed further out in the heliosphere, we also applied the self-similar expansion fitting (SSEF) technique to STEREO-A/HI data (Davies et al. 2012, Equation (6)). Using λ_GCS as input, SSEF yielded a CME radial speed of 862 ± 8 km s⁻¹, liftoff at 19:17 UT, and an HEEQ propagation longitude of 135°. These values are in very good agreement with those from the GCS fit. However, note that the GCS and SSEF predicted liftoff times are two hours later than when the CME was first seen by SOHO, as the CME accelerated considerably near the Sun's surface (Gopalswamy et al. 2015). Inspection of the EUVI images shows that the liftoff time is about 16:00 UT and the onset can be characterized as streamer blowout CME.

Zoom In Zoom Out Reset image size

This CME was accompanied by interplanetary type II and type III bursts detected by both STEREO spacecraft on 2013 November 29–30 (Figure 2). Since the CME was launched from the far side of the Sun, ground-based high-frequency radio measurements of the associated radio bursts are absent. Figures 2(a) and (b) display the radio flux density S measured by STEREO-A and STEREO-B, respectively. The type III burst at 22:00 UT on November 29 is followed by the intermittent type II burst, over an interval of seven hours. Within this, we were able to identify three intervals when the signal at both spacecraft was intense enough to enable successful direction-finding analysis: at 22:30 UT (1500 kHz) on November 29 and at 00:30 UT (600 kHz) and 04:00 UT (300 kHz) on the next day.

Zoom In Zoom Out Reset image size

A recent Schmidt & Cairns (2016) data-driven MHD simulation of this event with an analytic quantitative kinetic model for the radio emission concludes that the type II burst intervals of interest here are the fundamental emission. To compare radio measurements with the CME propagation parameters, we converted the radial distances yielded by the GCS fit r_GCS to frequencies f using the density model of Sittler & Guhathakurta (1999) assuming the fundamental emission. We consider the type II burst to be generated at both the CME leading edge (f ∼ r_GCS) and flanks ($f\sim {r}_{{\rm{GCS}}}\mathrm{cos}{\lambda }_{{\rm{GCS}}}$). The type II burst coincides well with the kinematic curve that corresponds to CME flanks (Figure 2(c)). Although type II bursts are generated at interplanetary shock fronts of propagating CMEs, they predominantly appear close to CME flanks (e.g., Gary et al. 1984; Magdalenić et al. 2014). For our wave propagation analysis, we applied the singular value decomposition technique to multicomponent measurements of auto- and cross-correlations of the voltages induced by electric field fluctuations (Santolík et al. 2003; Krupar et al. 2012). Figures 2(c)–(f) present the calculated wave vector directions θ and ϕ in the radial-tangential-normal (RTN) coordinate system (a spacecraft centered coordinate system where the Sun is at (θ = 90°, ϕ = 0°), southward/northward is (θ = 0°, 180°), and east/west is (±ϕ = 90°). An intensity threshold ($S\gt 5\times {10}^{-19}\;{\rm{W}}\;{{\rm{m}}}^{-2}\;{\mathrm{Hz}}^{-1}$) has been applied to suppress the background. The high-frequency portions of the type II and type III bursts (∼>1 MHz) propagate roughly toward STEREO-A, whereas the low-frequency portions propagate east of the spacecraft (Figure 2(e)). For STEREO-B, the type II and type III bursts propagate east of the spacecraft for over the entire interval analyzed (Figure 2(f)).

At this time, the separation angle between the two STEREO spacecraft was 64°, which allows us to accurately locate the sources of the type II and type III bursts by triangulation (Krupar et al. 2014a, 2014b). We note that this favorable separation angle would be also attained for combined spacecraft observations from L₁ and the fifth Sun–Earth Lagrangian point (L₅; Gopalswamy et al. 2011a, 2011b; Vourlidas 2015). Data points around the maximum flux densities of the type II and type III bursts were carefully selected in order to perform the radio triangulation. The triangulated type II and type III radio bursts are clustered along the CME propagation direction that is predicted by the GCS and SSEF fits, with radial distances ranging from 0.12 to 0.31 au (Figure 3). We find that, for this interval, the radio sources are located considerably further from the Sun when compared to the density model of Sittler & Guhathakurta (1999), which predicts radial distances between 0.03 au (1500 kHz) and 0.10 au (300 kHz). This ambiguity is probably caused by radio sources being spatially extended and/or radio propagation effects (Steinberg et al. 1984, 1985; Thejappa et al. 2007; Krupar et al. 2014a). Averaged radio locations of the three analyzed intervals of the type II burst are as follows: 0.23 ± 0.02 au (22:30 UT), 0.21 ± 0.03 au (00:30 UT), and 0.32 ± 0.04 UT (04:00 UT). The source locations derived from the triangulation of the second interval are systematically located closer to the Sun than those of the first one, when it would be expected to be further from the Sun. This discrepancy is probably caused by the superposition of the signal of the second interval of the type II burst with a fast-drifting emission occurring at about the same time. We thus will not include triangulated radio locations of the second interval in our analysis. The calculated average speed of the CME-driven shock based on the triangulated source locations of the first and third interval is 656 $\mathrm{km}\;{{\rm{s}}}^{-1}$, which is lower than those predicted by the GCS and SSEF fits. The third interval was also observed by the Wind/Waves instrument at 332 kHz located at the L₁ point (Bougeret et al. 1995). It yielded a wave vector azimuth ϕ_RTN = −12°, which is a good agreement with the triangulated location by STEREO/Waves, which would be observed at ϕ_RTN = −9° by Wind/Waves.

Zoom In Zoom Out Reset image size

We have examined STEREO-A/SECCHI/COR2 images in order to compare the relative positions of the triangulated source locations of the first interval of the type II burst with coronal structures (Figure 1(c)). The type II burst sources are clustered along positions of two streamers located at the west limb and at the southern CME flank (see the red arrows in Figure 1(c)). This suggests a close relationship between the type II burst and a possible interaction between the CME-driven shock and these streamers. It is consistent with the work of Magdalenić et al. (2014), who reported a similar result for a different CME. We note that another CME is already in progress along the south and appears to overlap with the radio sources. Inspection of the EUVI images shows that the CME is associated with a slow filament eruption starting earlier on the same day and appears to propagate at a plane about 90° away from the source of our CME. Therefore, it is unlikely that it is affecting our observations or interpretation.

Figure 4 shows measurements obtained by MESSENGER and STEREO-A. A fast forward interplanetary shock was observed by MESSENGER, at 0.4 au, at around 14:30 UT on November 30 (Figure 4(a)), identified by a sharp increase in the magnetic field magnitude (from 10 to 55 nT). This shock was likely to have been driven by the CME analyzed in this study. From the CME launch time and the arrival time of the shock at MESSENGER, we estimate the average CME speed between the Sun and MESSENGER to be $\sim 741\;\mathrm{km}\;{{\rm{s}}}^{-1}$. The CME-driven shock was also detected in situ at STEREO-A, at 0.96 au, at around 22:30 UT on December 1 (Figures 4(b)–(d)). Shock arrival corresponds to an abrupt increases in the magnetic field (from 7 to 20 nT), proton density (1 to 12 cm⁻³), and proton bulk velocity (340 to 580 km s⁻¹). From the CME launch time and the time of shock arrival at STEREO-A, we estimated the average CME speed to be 734 km s⁻¹. From the times of shock arrival at MESSENGER and STEREO-A, we estimate a CME speed between the two spacecraft of 729 km s⁻¹. The shock speed calculated by assuming the shock mass flux conservation equation over the shock, measured by STEREO-A, is 602 km s⁻¹. Figure 4(e) presents radio and plasma waves measurements recorded by the low-frequency receiver on board STEREO-A (2.5–160 kHz). We observe large amplitude Langmuir waves and a harmonic emission of a slow drifting type II burst in upstream of the CME-driven shock. We note that it is the first observation of the source region of an interplanetary type II burst by STEREO (Graham & Cairns 2015).

Zoom In Zoom Out Reset image size

Figure 5 shows an overview of the results obtained for the CME and CME-driven shock propagation. We have averaged azimuths ϕ of the type II and type III burst source locations from which we derive their propagation directions. Results for the propagation direction obtained for the type III burst are in a very good agreement with the results of GCS and SSEF fitting (Figure 5(a)). This suggests that in some cases type III bursts could be useful for estimating CME directions although they are related to solar flares instead of CMEs; a more statistical approach is required to confirm whether this is the case generally. Moreover, intense interplanetary type III bursts are more frequently observed than interplanetary type II bursts (Gopalswamy 2011). In contrast, the locations of the type II bursts seem to lie at the CME flanks. Hence, they are consistent with the CME-driven shock rather than the driver itself. We note that the type III burst may not be related to this eruption since it occurred five hours after the CME liftoff. However, it was the first complex type III burst observed since then, and we thus performed the radio triangulation for a comparison with the CME kinematics. Nonetheless, type II bursts occur sometimes without any type III bursts around.

Zoom In Zoom Out Reset image size

Figure 5(b) displays kinematics of the CME leading edge and radio sources. The results from the various techniques are in good agreement with the CME-driven shock arrival times at MESSENGER and STEREO-A. For an analysis of the type II frequency drift, we converted frequencies of the type II burst f to radial distances r using the density model of Sittler & Guhathakurta (1999). As in Figure 2, we consider two possible scenarios: the emission originates at the CME leading edge (r ∼ f) and flanks ($r/\mathrm{cos}{\lambda }_{{\rm{GCS}}}\sim f$). A comparison with white-light and in situ measurements suggests that the latter is valid, and the derived average speed of the CME-driven shock from a frequency drift is then 992 ± 22 km s⁻¹, which is consistent with an assumption that the CME-driven shock propagates faster than the CME itself. For further analysis of the type II burst, we use a relation between the drift rate df/dt of radio emissions measured at the frequency f, and the radial source speed v:

$$\begin{eqnarray}&&\displaystyle \frac{{df}}{{dt}}=\displaystyle \frac{{fv}}{2n}\displaystyle \frac{{dn}}{{dr}},\end{eqnarray} \tag{ 1 }$$

where n is the density and r is the radial distance (Mann et al. 1999). If we assume that below 1 MHz the CME-driven shock is far enough that n ∼ r⁻², we can calculate ξ the angular deviation of the type II burst propagation direction from radial by combining radio measurements with white-light observations as

$$\begin{eqnarray}&&\xi ={\mathrm{cos}}^{-1}\left(\displaystyle \frac{{{fv}}_{{\rm{WL}}}}{\tfrac{{df}}{{dt}}{r}_{{\rm{WL}}}}\right),\end{eqnarray} \tag{ 2 }$$

where r_WL and v_WL are the radial distance and the speed of the CME from white-light observations, respectively. From radio and white-light parameters between the first and the second interval (f = 1.05 MHz, ${v}_{{\rm{WL(GCS)}}}=761$ km s⁻¹, df/dt = 116 Hz s⁻¹, and ${r}_{{\rm{WL(GCS)}}}=16$ R_S), we calculate ξ to be 52°, whereas for parameters between the second and the third interval yield ξ = 62° (f = 450 kHz, ${v}_{{\rm{WL(SSEF)}}}=862$ km s⁻¹, df/dt = 37 Hz s⁻¹, and ${r}_{{\rm{WL(SSEF)}}}\;=32$ R_S). Our results confirm that the type II burst arises from the CME flanks since the calculated values of ξ are comparable to the CME angular half-width λ_GCS.