It has been known for long time (see MacQueen and Fisher (1983)) that there are two kinds of CMEs in terms of speeds. CMEs belonging to the first category, possibly flare-associated, have high and constant speeds, and those in the second category, possibly associated with filament eruptions, have low speeds but acceleration profiles. This classification has recently been extended to data from SOHO/LASCO, which has significantly higher sensitivity than previous instruments, making it possible to observe diffuse "halo" CMEs. The reference is Sheeley et al., JGR, 1-4, 24739 (1999) (sorry no abstract of paper available at ADS, maybe AGU's policy?). They primarily dealt with halo CMEs using LASCO C2 and C3 data. They found that fast ones, which tend to have ragged structures, decelerate rather quickly (typically 1000 --> 500 km/s in less than an hour), and that slow ones, which tend to have smooth structures, get accelerated to constant speeds sooner than their non-halo counterparts (i.e., those that originate near the limb.)
I would like to know if the correlation between flares and fast CMEs is seen for halo events. In the following tables (separately for slow events and fast events), I replicate the information from the Sheeley et al. paper, but give only typical speeds. Sheeley et al. measured the CME speeds along different radial paths, and for different structures, but we show only one piece of information per event. For slow events, V with suffix "A" stands for the constant speed after the acceleration is over. For fast events, the two speeds with suffices "0" and "1" indicate how the initial high speed decreases (within a half to a few hours).
I inserted the time of the first CME signature in C2 data, after looking at all the
movies. I also looked for the CME signatures in all the available SXT and EIT data,
and entered their locations if I found something. The locations of the signatures
may or may not correspond to active regions. I studied GOES light curves to see
if flares occurred around the CME times. Lastly, I checked the NOAA site for SEP
events and elevated KP indices, in likely association with the CMEs. Some people
may say there should be better data of these kinds, but I list these only for basic
As expected, the events classified as "slow" in the Sheeley et al. paper are hardly correlated with major flares. To me it is just a matter of semantics to call a GOES A-class event a flare. Moreover, lower coronal signatures for these CMEs are mostly found outside active regions. The events labeled as "fast" are a bit problematic, since their correlation with flares or even active regions is not very high. But we have to remember LASCO observations do not tell us whether a CME is moving towards us of away from us. Yes, I think all the ones without signatures at lower corona occurred on the other side of the Sun, or at least way behind the limb.
Here are two examples to illustrate this. First, for the 31 Mar 1998 event, it is clear from the following SXT movie that the backside activity was more spectacular.
About two weeks before this event, we witnessed a rapidly growing active region. This can be seen, for example, in MDI white-light images. This was AR 8179 (at latitude S20), whose central meridian passage was 15 March 1998, and application of the usual differential rotation formula shows that it was about 67 degrees behind the east limb. It is likely that the 31 Mar 1998 CME was associated with a flare from this region, which would have been very intense if it occurred on the visible disk. In less than a week, we welcomed the return of this region as AR 8194, which still produced some major events.
One more example is the 11 May 1998 event, which probably occurred in a similar configuration to the Khan-Hudson events, i,e., involving large-scale trans-equatorial loops. Actially it was only two days after their last event. Because of a larger angle behind the west limb, there was hardly a flare. But the following sequence of SXT images of the west limb (click to enlarge) shows a dynamical evolution, which was probably the CME itself.
Note that we are only rarely fortunate enough to see such direct signatures of
large-scale CMEs, because either the cadence is irregular/relatively low or
the field of view is too small.
There are a few more fast events from the backside. In the new table (click here), they are crossed out. Also, three very fast events were associated with major flares, as they appear blinking in the table. The last event is shown to be not very fast, but if a different path is taken, say in the northeastern direction, the speed may be higher (this represents one difficulty of this kind of analysis on LASCO data). We note all of these were impulsive flares, and as we indicated earlier, intense, impulsive flares, distinct from long-decay events (which we know are well correlated with CMEs), can be associated with fast and extended CMEs. Certain flares are just not an aftermath of a CME, but may actively participate in the acceleration of a CME. Of course, it is still reasonable to think both flares and CMEs may be attributable to a same magnetic instability.
Identification of backside events for the slow CMEs may be less obvious, since one or two events are still hard to understand in terms of solar signatures. The best example is the 6 January 1997 event. Can this possibly be a backside event? One strong reason for this CME being the front side event is the significant magnetic cloud. But in order to understand the solar origin of CMEs, geo-effectiveness can be a distracting factor. In fact, in a paper by Crooker, it is mentioned that "3 of 4 halo backside CMEs also were followed within 5 days by storms." See this plot from the paper.
26 May 2001