INTERPLANETARY SHOCKS LACKING TYPE II RADIO BURSTS

Figure 2 shows the distribution of shock and EJ speeds measured at 1 AU. Both EJ and shocks are generally faster than the slow solar wind. The average speed of the EJ is smaller than that of the shocks in all the three sets (RQ, RL, and all shocks). The RQ shocks and their EJ have below-average speeds, while the RL shocks and their EJ have above-average speeds. The average speed of RQ shocks is ∼34% smaller than that of the RL shocks, with a similar difference for EJ (∼28%). Thus there is a clear indication that the RQ shocks are generally weaker than the RL shocks. The median speeds confirm the same tendency that RQ shocks and their EJ are generally weak compared to the RL populations.

Zoom In Zoom Out Reset image size

In order to further quantify the strength of the shocks, we determined the Alfvénic Mach number (M_A) for each shock from in situ data. The Mach numbers were obtained from one of the following three methods: (1) from the compilation of shock properties made available online by J. C. Kasper (http://www.cfa.harvard.edu/shocks/) we extracted the Mach numbers for 151 shocks (Mach numbers are average values computed using eight different methods); (2) for 64 events with good solar wind data but not listed in the above Web site, we used the Shock and Discontinuities Analysis Tool (SDAT; SDAT uses an extension of the Vinas & Scudder 1986 analysis method based on the Rankine–Hugoniot conservation equations); and (3) for 15 events with low data quality we used the following approximate method to estimate M_A. Assuming that the shock normal and speed are along the radial direction, the shock speed (V_sh) and M_A can be obtained from the mass conservation relation

$$\begin{equation} (V_d - V_{{\rm sh}})N_d = (V_u - V_{{\rm sh}})N_u, \end{equation} \tag{ 1 }$$

where the subscripts d and u to the solar wind speed (V) and density (N) indicate the downstream and upstream values, respectively. Inputting V_d and V_u into Equation (1), we get V_sh. From the upstream magnetic field (B_u) and N_u, we obtain the upstream Alfvén speed V_A and hence M_A = (V_sh − V_u)/V_A. We randomly selected eight events in J. C. Kasper list and computed M_A using the SDAT and approximate methods. M_A differs by 0.55 between methods (1) and (2) and by 0.73 between (1) and (3). Similarly, the estimated shock speed differs by 38 km s⁻¹ between (1) and (2), and 82 km s⁻¹ between (1) and (3).

Figure 3 shows the distribution of M_A for all the shocks and the RQ and RL subsets. The average M_A for all the shocks is ∼3.2. For RL shocks, M_A = 3.4, which is ∼31% higher than that of the RQ shocks (M_A = 2.6). The Mach number difference reflects the CME speed difference for RQ and RL shocks and confirms that the RQ shocks are weaker. Since the Mach number distributions are not symmetric, we have also given the median values. Although the median M_A values of the RL and RQ shocks are above and below the median M_A for all shocks, we see that the difference is not very high. The sheath interval (standoff distance) for the RQ and RL shocks are 10.4 hr and 11.3 hr, respectively. We shall discuss the reason for this in the next section.

Zoom In Zoom Out Reset image size

Figure 4 shows the speed and width distributions of CMEs associated with all, RQ, and RL shocks. The speeds and widths are in the sky plane and no attempt was made to correct for projection effects. The average speed (535 km s⁻¹) of CMEs associated with RQ shocks is only slightly greater than that of the general population of CMEs (457 km s⁻¹). The average CME speed of RL shocks is a factor of ∼ 2 higher than that of the RQ shocks (1237 km s⁻¹ versus 535 km s⁻¹). Note that the difference between the two populations is more pronounced at the Sun than at 1 AU (see Figures 2 and 3). This is expected because of the momentum exchange between CMEs and the solar wind during the IP passage (see Section 4.3). Among the RL shocks, 11 (or ∼8%) were associated with purely metric type II bursts and 16 (or ∼11%) were associated with purely kilometric type II bursts. The average speeds CMEs associated with purely metric and purely km type II bursts are 740 km s⁻¹ and 729 km s⁻¹, respectively. These are significantly below the average CME speed of all RL shocks (1237 km s⁻¹), but well above that of the RQ shocks (535 km s⁻¹). This means the RQ CMEs were not able to drive shocks that are strong enough to accelerate electrons even to produce purely m or purely km type II bursts. The CME speeds of the purely metric and kilometric type II bursts are consistent with the hierarchical relationship between CME kinetic energy and the wavelength range over which the type II bursts occur (Gopalswamy et al. 2005). In the CME speed distribution of RL shocks, we have shown the average speed (1521 km s⁻¹) of CMEs faster than 900 km s⁻¹, which is consistent with other studies that considered fast and wide CMEs irrespective of IP shock association (Gopalswamy et al. 2008c).

Zoom In Zoom Out Reset image size

Table 3 summarizes the CME and shock speeds for all, RQ, and RL shocks divided according to the associated IP drivers (MC, EJ, or DL). We find that the MC-associated shocks have the lowest CME speed (891 km s⁻¹), the EJ-associated shocks have intermediate CME speed (941 km s⁻¹), and the DL shocks have the highest CME speed (1308 km s⁻¹; see Table 3). This pattern applies not only to all the shocks, but also to the RQ and RL subsets, as can be seen in Table 3. The differences in CME speeds can be attributed to the fact that the MC, EJ, and DL shocks have different solar source distributions: the MC-associated CMEs mostly originate close to the disk center, the EJ-associated CMEs originate at intermediate central meridian distances, and the CMEs responsible for the DL shocks originate close to the limb. Accordingly, the CMEs are subject to projection effects of varying extents: large (MC), intermediate (EJ), and minimal (DL). CMEs associated with the DL shocks are also expected to be very energetic because they need to produce a shock signal at Earth in spite of their origin near the solar limb (see Gopalswamy et al. 2007, 2009e). The shock speeds at 1 AU have a behavior opposite to that of the CME speeds: the MC-associated shocks are measured at their noses, so they are of the highest speed (625 km s⁻¹); the EJ-associated shocks are measured between their nose and flanks, so they are of intermediate speed (549 km s⁻¹); the DL shocks appear slow (average speed ∼ 483 km s⁻¹) because they are actually the flanks of shocks propagating orthogonal to the Sun–Earth line. This pattern is also seen for RL and RQ shocks, with a single exception that the MC- and EJ-associated RQ shocks have roughly the same speed.

Apart from the speed, the CME width is also an indicator of the CME energy. Halo CMEs are known to be faster on the average (Gopalswamy et al. 2007), although we do not know their true width. However, we know that there is a good correlation between CME speed and width when CMEs erupting near the limb are considered (Gopalswamy et al. 2009e). Therefore, we expect the halo CMEs to be wide and hence more energetic. Furthermore, CME width and mass are correlated (Gopalswamy et al. 2005), so faster and wider CMEs generally have higher kinetic energy. Thus, the fraction of halo and partial halo CMEs in a population is a good indicator of how energetic the population is. The width distributions in Figure 4 show that the fraction of full-halo CMEs is the largest for RL shocks (∼72%) and the smallest for the RQ shocks (40%), differing by a factor of ∼2. If we combine full (width = 360°) and partial halos (width ≥120°, but <360°), we see that the fraction of such wide CMEs is still higher for the RL shocks, but to a smaller extent (97% versus 72%). Only a small fraction of shocks are associated with non-halo CMEs, but even these CMEs are very wide: the average CME width is 102° for RL shocks versus 78° for RQ shocks. The speed and width distributions thus indicate that the CMEs driving the RQ shocks are definitely more energetic than the average CMEs, but less energetic than those driving RL shocks.

One of the striking differences in the kinematic properties of CMEs associated with the RQ and RL shocks is the sign of the acceleration within the coronagraphic field of view. Figure 5 shows that the average CME acceleration is close to zero negative (mean ∼ −0.1 m s⁻²; median ∼ 0.2 m s⁻²) for all the shocks taken together. However, the average CME acceleration is positive for RQ shocks (mean ∼ +6.8 m s⁻² and median ∼ +3.5 m s⁻²), while it is negative for RL shocks (mean ∼ −3.6 m s⁻² and median ∼ −2.4 m s⁻²). This result provides an important clue for understanding their radio-quietness: the CMEs continue to accelerate and attain super-Alfvénic speeds only at a distance where the CME speed is high enough and the Alfvén speed is low enough to form a shock. Even there, they seem to be only marginally super-Alfvénic so the shocks are not strong enough to accelerate electrons. Where such CMEs might become super-Alfvénic is illustrated in Figure 6.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The CME acceleration measured for the two RQ events in Figure 6 is close to the average CME acceleration of RQ shocks. The variation of the Alfvén speed, the CME speed, and the solar wind speed is shown as a function of the heliocentric distance. The Alfvén speed profile was obtained using models of magnetic field and plasma density as in Gopalswamy et al. (2001a). The 2003 August 14 CME at 20:06 UT originated from S10E02 and had an average speed of ∼378 km s⁻¹ within the LASCO field of view. A second-order fit to the height–time measurements yields a constant acceleration of 4.4 m s⁻². From the second-order fit, we have plotted the speed as a function of distance in Figure 6(a). In the inner corona, the CME had a speed of only about 300 km s⁻¹, which is below the local Alfvén speed. No shock is expected until about 8 Rs because of the higher Alfvén speed in the ambient medium. The CME attained a speed of ∼550 km s⁻¹ when it reached the edge of the LASCO field of view. Beyond 8 Rs, the CME maybe able to drive a shock provided the CME speed exceeds the sum of the Alfvén speed and the solar wind speed.

The 2006 August 16 CME at 07:31 UT originated from the southwest quadrant (S01W19) and had an average speed of ∼ 563 km s⁻¹ within the LASCO field of view. A second-order fit to the height–time measurements gives a constant acceleration of ∼5.1 m s⁻². When plotted as a function of heliocentric distance (see Figure 6(b)), the CME speed was found to be briefly above the Alfvén speed in the inner corona, turning sub-Alfvénic due to higher Alfvén speed until about 5.5 Rs, and then becomes super-Alfvénic. Once the CME speed exceeds the sum of the Alfvén speed and the solar wind speed, a shock forms. Clearly the CME behavior (and its shock driving ability) is similar to the previous event beyond ∼5.5 Rs. One would expect a metric type II burst in the 2006 August 16 event when the CME was super-Alfvénic briefly in the inner corona, but there was none. The Alfvénic Mach number (∼1.3) was probably too small for the shock to be able to accelerate particles in sufficient numbers. It must be pointed out that the Alfvén speed profiles shown in Figure 6 were derived from models of density and magnetic field variation in the corona. The actual density and magnetic field on the days of the CMEs may be quite different. Both the CMEs in Figure 6 also originated close to the disk center, so their true speeds may be higher. In this sense, the above discussion should be considered qualitative.

As noted before, 16 RL shocks had type II bursts only in the km wavelength domain. CMEs associated with 13 of these 16 shocks (or 81%) had positive acceleration, similar to those of the RQ shocks. These CMEs also must have attained super-Alfvénic speeds only far from the Sun, but their shocks are slightly stronger, so they produce type II emission. Interestingly, the subset of RL shocks that had radio emission only in the metric domain shows an opposite CME behavior: CMEs associated with nine out of the 11 RL shocks (or 82%) had deceleration within the coronagraphic field of view suggesting that the shocks were efficient in electron acceleration only near the Sun (within the first 2–3 Rs) and remained subcritical thereafter.

The different acceleration profiles of RL and RQ CMEs obtained near the Sun, and the expected interaction with the solar wind can explain the reduced contrast between the RL and RQ shock properties at 1 AU (see Figures 2 and 3). The evolution of CMEs during their IP passage is such that fast CMEs tend to decelerate to the speed of the surrounding solar wind by the time they get to 1 AU (notable exceptions are the fastest events that have speeds well above the solar wind speed). On the other hand, slow CMEs speed up to attain the solar wind speed. The two cases crudely correspond to the CMEs of RL and RQ shocks, respectively, consistent with the acceleration distributions in Figure 5.

It was possible to identify the soft X-ray flares and their sizes (peak X-ray flux expressed as flare class) in the source regions of 48 RQ and 137 RL shocks. In several cases, the source locations were identified using the associated filament eruption, but the X-ray intensity was not above the background level, so no flare information is listed for these events. The flare-size distributions are shown in Figure 7 for all shocks and the RQ and RL subsets. The median size of flares associated with RQ shocks (C3.4) is smaller by an order of magnitude compared to that in RL shocks (M4.7). The flares associated with RQ shocks are predominantly of C-class, while M- and X-class flares dominate in RL shocks (see Figure 7). The difference in flare sizes is consistent with the difference in speeds and widths of RQ and RL CMEs (there is a reasonable correlation between X-ray flare size and CME kinetic energy; see Gopalswamy et al. 2009e). We also examined the flare sizes for a subset of events in which the central meridian distance (CMD) of the flares ≤ 85° to avoid the possibility that some limb flares may be partially occulted and hence we may not know their true sizes. There were 47 such RQ shocks and 130 RL shocks whose median flare sizes are C3.4 and M4.9, respectively. These are not too different from the case when no CMD restriction is used.

Zoom In Zoom Out Reset image size

The longitude and latitude distributions of the solar sources of the RQ and RL shocks are compared in Figure 8. The numbers of RQ and RL shock sources decline toward limbs. The longitudinal distribution is mainly due to the selection effect because for a given CME speed, the shock from a disk-center CME has a better chance to be detected at 1 AU. For a disk-center eruption, the measurement is likely to be made at the nose of the shock compared to an eruption at a larger CMD. In the latter case, the measurement will be made at the shock flanks where the shock strength is typically lower (see discussion in Section 4.1). The latitude distributions have a sharp cutoff around ±30° latitude for both subsets. Furthermore, the latitude distribution is bimodal, with peaks in the northern and southern active region belts. This suggests that the shock-driving CMEs originate in the active region belt. Active regions have a higher magnetic field, thus resulting in more energetic CMEs.

Zoom In Zoom Out Reset image size

It must be pointed out that the longitude distribution of CME sources associated with RL shocks is quite different from the solar sources of RL CMEs reported in Gopalswamy et al. (2008a). The center-to-limb variation of the number of fast and wide RL CMEs showed an increase, which is opposite to what is seen in Figure 8. This can be explained by the fact that many limb and even behind-the-limb CMEs may produce type II bursts without producing a shock signal at Earth (Gopalswamy et al. 2008a). On the other hand, the RQ and RL shocks have their solar sources clustered near the disk center (there are only six behind-the-limb events).

A different representation of the source distribution can be seen in Figure 9, where the heliographic coordinates of the eruption regions are plotted for RQ and RL shocks separately and together. There are some clear differences between the source distributions. (1) There are almost no CME sources at the east limb for RQ shocks, whereas there are several RL sources at both limbs. The RQ sources are also more concentrated near the central meridian (with a slight eastern bias) compared to those of RL shocks, which show more uniform distribution in longitude (with a slight western bias). The limb sources of RL shocks are consistent with the fact that the associated CMEs are more energetic to produce a shock signature near Earth. (2) The CME sources are generally within the ±30° latitude range, with only a small number of exceptions (the CME sources of five RQ shocks and six RL shocks outside; see also Figure 8).

Zoom In Zoom Out Reset image size

The variation of source latitudes over the solar cycle shown in Figure 10 explains the reason for the higher latitude sources in Figure 9. The higher-latitude sources occur mostly during the rise phase of the solar cycle, similar to the solar sources of CMEs that produce a signature near Earth in the form of MCs or non-cloud EJ (Gopalswamy et al. 2008b, 2009a). This behavior has been attributed to the control by the global dipolar field, which has maximum strength during the solar minimum phase. In the beginning of a solar cycle, active regions emerge at higher latitudes, but the prominences and CMEs from these regions tend to move toward the equator under the influence of the global dipolar field (Gopalswamy et al. 2000a; Gopalswamy & Thompson 2000; Filippov et al. 2001; Plunkett et al. 2001; Gopalswamy et al. 2003b; Cremades et al. 2006). During the rest of the solar cycle, the sources continue to follow the sunspot butterfly diagram.

Zoom In Zoom Out Reset image size

Figure 10 shows that both RQ and RL shocks have a peak rate of occurrence during the maximum phase of the solar cycle (see also Table 2). This simply reflects the fact that more energetic eruptions occur during the maximum phase. The active regions in which such eruptions occur are also closer to the equator. It has been shown that the average speed of CMEs increases by a factor of 2 between the minimum and maximum phases (see, e.g., Howard et al. 1985; Gopalswamy 2004b, 2006a). The number of shocks in the northern and southern hemispheres also peaks at different times. The latitude–time plot in Figure 10 for RL shocks shows two clusters: one toward the end of year 2000 in the north and the other in the year 2002 in the south. High-latitude activities around these times are known to be associated with the solar polarity reversal (Gopalswamy et al. 2003a), but it is not clear why such a high level of activity exists at low latitudes. One possibility is the presence of super active regions during these periods. On the other hand, the peaks in the RQ shock numbers are not very strong and seem to be due to statistical noise. Although the occurrence rates of RQ and RL shocks have a peak in the maximum phase, they have different behavior during the rise and declining phases: there are more RQ shocks during rise phase, contrary to the higher rate of RL shocks during declining phase. The number of RQ shocks drops sharply in the declining phase. This can be appreciated better when we plot the ratio of the number of RL shocks to that of the RQ shocks as a function of time (see Figure 11). The ratio is ≤1 during the rise phase, increases to ∼2 during the maximum phase and peaks at ∼6 in the declining phase. There is a notable dip at the beginning of the maximum phase (year 1999), when there is a general dearth of energetic phenomena (see, e.g., Gopalswamy et al. 2004). The ratio starts decreasing in the late declining phase, but still remains above 1 toward the end of the study period. There are two possible explanations for the behavior of the RL to RQ shock number ratio. The first explanation is concerned with the solar cycle variation of Alfvén speed. During the rise phase, the ambient medium typically has a lower density so the Alfvénic speed is expected to be higher, making it difficult to form strong shocks, thus rendering many shocks RQ. The 1 AU observation that there is no change in the Alfvén speed between solar minima and maxima (Mullan & Smith 2006) poses problem to this explanation. However, we do not know if the solar cycle variation of the Alfvén speed observed at 1 AU applies for the entire inner heliosphere where type II bursts are observed. The second explanation is the influence of super active regions, which are prolific producers of energetic CMEs and flares. During the declining phase of cycle 23, there were many super active regions that produced many energetic eruptions (Gopalswamy et al. 2006). Shocks from these eruptions might have skewed their number in favor of RL shocks. It is not clear if such a behavior will be present in all cycles.

Zoom In Zoom Out Reset image size

In order to make a direct comparison with the past works on IP shocks, Figure 12 shows the transit speed distributions of the shocks in our list. The average transit speed of all the shocks is ∼775 km s⁻¹, with the RQ and RL shocks having the values 629 km s⁻¹ and 851 km s⁻¹, respectively. The median transit speeds have similar relationships among the three populations. Note that these values are very similar to the average Sun–HELIOS transit speed (745 km s⁻¹) for a smaller number of shocks reported by Sheeley et al. (1985) (see also Watari & Detman 1998 and references therein). The transit speeds are always greater than the in situ shock speeds; they are also less than the corresponding CME speeds near the Sun, except for the RQ shocks. The RQ shocks thus point to a different kinematics as discussed below.

Zoom In Zoom Out Reset image size

One of the distinct kinematic properties of the RQ shocks is the positive acceleration shown by most of the associated CMEs compared to the negative acceleration of the CMEs associated with RL shocks. This means that the shocks form generally at large distances from the Sun (≥10 Rs) where the driving CMEs become super-Alfvénic. CMEs associated with purely kilometric type II bursts show a similar kinematic behavior (Gopalswamy et al. 2005; Gopalswamy 2006b). The km type II bursts typically start at frequencies below 1 MHz when the CMEs are typically at heliocentric distances >10 Rs. Thus, the RQ shocks seem to be similar to those producing km type II bursts, except that the former are even weaker. One can see that the shocks are ordered according to the CME kinetic energy. The RQ shocks are the weakest, followed by shocks producing the km type II bursts, and then the shocks producing radio emission at higher frequencies. There is also further hierarchy between CME kinetic energy and the frequency range of type II bursts reported elsewhere (Gopalswamy et al. 2005). It is worth pointing out that the average speeds of CMEs associated with metric type II bursts (740 km s⁻¹) and km type II bursts (729 km s⁻¹) are similar, but the evolution of the underlying shocks is quite different: in the case of purely metric type II bursts the shock forms very close to the Sun and quickly becomes subcritical; in the case of km type II bursts, the shock forms far away from the Sun, similar to most of the RQ shocks. When Sheeley et al. (1985) reported the connection between CMEs and IP shocks, the metric type II bursts were thought to be "neither necessary nor sufficient for the occurrence of IP shocks." Here we see that purely metric type II bursts are indicative of shocks producing radio emission only near the Sun and remaining subcritical for the rest of the Sun–Earth distance. The IP shocks associated with purely metric type II bursts correspond to the lowest CME kinetic energy among all the RL shocks (see also Gopalswamy et al. 2008b). When there is a metric type II with no associated shock at Earth, either the shock completely dissipates or it simply misses the spacecraft making in situ observations.

RQ CMEs have been studied before (Sheeley et al. 1984; Gopalswamy et al. 2001b; Lara et al. 2003; Shanmugaraju et al. 2003; Gopalswamy et al. 2008c, 2008b). Sheeley et al. (1984) reported CMEs lacking metric type II bursts. Gopalswamy et al. (2001b) reported that ∼60% of CMEs with speeds exceeding 900 km s⁻¹ were not associated with type II bursts in DH wavelengths, but the average CME width was relatively small. Lara et al. (2003) considered CMEs associated with metric and DH type II bursts and confirmed that the speed and width of CMEs are important in deciding their type II association. Shanmugaraju et al. (2003) confirmed the results of Gopalswamy et al. (2001b) and Lara et al. (2003), using a sample of CMEs associated with flares of importance >C1.0. In addition, they point out the importance of flare duration, size, and rise time in the CME-type II association and contend that CMEs need not be the drivers of shocks producing type II bursts. Assuming that all type II bursts are due to CME-driven shocks, Gopalswamy et al. (2008b) concluded that slow CMEs with type II busts and fast CMEs without type II bursts can be explained by the variation in the Alfvén speed of the ambient medium. Thus, RQ CMEs either did not drive shocks, or the shocks were not able to accelerate electrons in sufficient numbers. In the case of RQ shocks, the underlying CMEs did drive shocks, so the inability to accelerate electrons seems to be the case. As noted before, many of the RQ shocks formed only at large distances (>10 Rs) from the Sun, so no shocks are expected in the corona.

The low Mach number of the RQ shocks observed at 1 AU is consistent with the low-energy drivers associated with them. From numerical simulations, Burgess (2006) showed that supercritical shocks are required for accelerating electrons in the energy range relevant to the production of type II radio bursts. We suggest that the RQ shocks are subcritical, i.e., their Mach number is typically below the first critical Mach number (see Edmiston & Kennel 1984; Mann et al. 2003): the critical Mach number ranges between 1 and 2 for quasi-parallel shocks and between 2.3 and 2.7 for quasi-perpendicular shocks for the coronal plasma. The actual value depends on the plasma beta of the ambient medium through which the shock propagates (Gary 2001). For example, the solar wind at 1 AU has beta ∼1 (Mullan & Smith 2006), so the critical Mach number can be anywhere between 1 and 2.3 (Mann et al. 2003), the higher values being for quasi-perpendicular shocks. Recall that the average Mach number for RL shocks is 3.4, making them supercritical. On the other hand, the average Mach number for RQ shocks (2.6) is similar to the critical Mach number for quasi-perpendicular case, given the uncertainty in the estimation of Mach numbers (see Section 3.1). Furthermore, the Mach numbers are estimated only at 1 AU in a high-beta situation. The Mach number in the near-Sun IP medium and the corona can be very different because both the CME speed and ambient Alfvén speed change with heliocentric distance. The density and magnetic field models of the corona indicate a peak Alfvén speed of ∼600 km s⁻¹ at ∼3 Rs from the Sun (see Figure 6), which suggests an Alfvén Mach number of ∼6 for the fastest detected CMEs. For CMEs producing DH type II bursts, the average speed is ∼1000 km s⁻¹ indicating an Alfvénic Mach number of ∼1.7 near the Sun. The corona contains dense streamers and tenuous regions, so the Alfvén speed and Mach number can vary by a factor of 4 near the Sun (Gopalswamy et al., 2008b). Both the Alfvén speed and the CME speed change (mostly decrease) with heliocentric distance, so does the Mach number. The existence of a type II burst thus depends on how these two speeds vary relative to each other between the Sun and Earth. The reduced contrast in Alfvénic Mach number between RQ and RL shocks can be attributed to the CME evolution in the inner heliosphere. It is well known that fast CMEs tend to decelerate to the speed of the surrounding solar wind by the time they get to 1 AU. On the other hand, slow CMEs speed up to attain the solar wind speed. The two cases crudely correspond to the driving CMEs of RL and RQ shocks, respectively (the acceleration distributions are consistent with this).

It was previously shown that RQ CMEs were not associated with large SEP events, implying lack of ion acceleration. We expect something similar in the case of RQ shocks. Since the RQ shocks are likely to be subcritical, it is of interest to know whether ESP events are also absent. Furthermore, RQ shocks seem to be quasi-perpendicular, which can also be verified by determining the shock normal angle at 1 AU. A systematic investigation energetic particle data at the time of RQ shock arrival at 1 AU seem to support these conclusions, as reported in an accompanying paper (Mäkelä et al. 2009). Such an analysis might explain the large fraction of IP shocks without ESP events reported by Ho et al. (2008). The suggestion that most of the RQ shocks may be quasi-perpendicular can be confirmed from the time profiles of ESP events, which substantially differ depending on the shock normal angle. In particular, quasi-perpendicular shocks are supposed to produce spike events (Sarris & van Allen 1974).

At the microscopic level, one expects a substantial difference between low and high Mach number shocks such as the mechanism of electron acceleration. In low Mach number shocks, the acceleration occurs via the classical fast Fermi acceleration, which increases the energy of the electrons only by a factor of 2–3 and that too for nearly perpendicular shocks. High Mach number shocks acquire a dynamic rippled character and the energization can occur over a wider range of shock normal angles and to higher energy levels (see Burgess 2006 for details). As noted before, we have very little information on the shock geometry except at the location of the observing spacecraft (at Sun–Earth L1 point). When the spacecraft passes through the shock, it might encounter superthermal particles (ESP events), which may provide information different from what the type II bursts provide. Type II bursts require electrons escaping from the shock front, while ESP events correspond to what is present at the shock front. Thus, we expect some slightly different information on the shock strength from the ESP events and radio bursts. High Mach number shocks (supercritical) have also the possibility of accelerating electrons via the Bunemann instability of the reflected ion beam at the shock front (Cargill & Papadopoulos 1988). Low Mach number (subcritical) shocks do not have ion reflection, so they cannot energize electrons by this mechanism. At 1 AU, the difference between the Mach numbers of RQ and RL shocks exists, but is not substantial. There are indications from CME observations that the difference may be substantial near the Sun. The electron acceleration efficiency depends on additional factors also: the availability of seed particles and enhanced turbulence in the ambient medium into which the shock propagates.

In a study aimed at identifying the IP drivers of shocks, Oh et al. (2007) reported a small fraction of shocks in which they could not identify the IP driver. They discussed one of the "unidentified" driver events (2000 February 11 at 02:33 UT) in detail and suggested that it is a "blast-wave-type IP shock." Our investigation revealed that this shock is indeed associated with an IP EJ, which is ∼4 hr in duration (see Table 1 and Figure 13). The EJ is clearly identifiable from the temperature signature as shown in Figure 13. There was also a second shock on this day associated with another EJ. The solar sources of the two CMEs are in the NE and SW quadrants, respectively on 2000 February 8 at 09:30 UT and on the next day at 19:54 UT as shown in Figure 13. The two CMEs were also associated with IP type II bursts, and hence were driving shocks near the Sun. Therefore, we conclude that the blast-wave scenario is incorrect in this event because the IP shock has a driver identified both in the IP medium and near the Sun.

Zoom In Zoom Out Reset image size

In all, Oh et al. (2007) reported that 10 IP shocks had "unidentified" drivers. Five of these shocks are not in our list, probably because they are either CIR shocks or slow shocks. Of the remaining five, two are in our list as DL shocks (2007 October 23 and 2001 March 27 shocks for which we know the CME driver at the Sun); one was associated with an MC (2000 February 11 shock) and two were associated with EJ (1998 May 3 and 2001 September 13 shocks). Note that we were able to identify the driving CMEs near the Sun for all the 42 IP shocks with no discernible IP drivers. The reason for the lack of IP drivers is simply the large CMD of the eruption regions, such that we just observe the shock flanks at Earth. The drivers head roughly orthogonal to the Sun–Earth line, so they are not intercepted by the observing spacecraft (see Gopalswamy et al. 2001a, 2009c). Thus, none of the DL shocks qualifies to be a blast wave.

During our study period overlapping with that of Oh et al. (2007), we identified 14 DL shocks. Four of these did not figure in the Oh et al. (2007) list: (1996 August 16 and November 11, 2000 July 26, and 2001 April 7). As we noted above, only two were designated as "unidentified" by Oh et al. They identified six of the shocks with MCs, one with a high-speed stream and one with an EJ. Thus, there is clearly disagreement on the IP drivers of at least eight shocks. Our DL shock on 2001 January 31 was designated as an HSS shock by Oh et al. (2007), which we think is incorrect. The speed never increased more than 450 km s⁻¹ for the next two days. If anything, there was a tiny EJ (1–2 hr long) around 18 UT on 2001 February 1, which is more than 34 hr after the shock. Even if this EJ is associated with the shock, it is consistent with the DL shock classification (the driver is barely seen). The solar source was a CME on 2001 January 28 at 15:54 UT from S04W59, which suggests that just the eastern flank of the shock should have arrived at Earth. In addition to these differences, we note a predominance of MCs as drivers in Oh et al. (2007) list. This may be because they seem to have combined both MC and cloud-like events as MCs. We designated the cloud-like events as EJ according to the list maintained by R. P. Lepping (http://lepmfi.gsfc.nasa.gov/mfi/MCL1.html).