STEREO observations of flares and their associations with CMEs
From RHESSI Wiki
|Stereoscopic Flares and CMEs|
|1st Author:||Nariaki Nitta|
|Published:||9 November 2009|
|Next Nugget:||Dips and Waves|
|Previous Nugget:||Imaging through visibility interpolation: uv-smooth|
Recent RHESSI Science Nuggets (e.g., 12 August 2009, 31 August 2009) discussed interesting observations that may give a clue to the long-standing problem of the relation between CMEs and flares. The observations given in these Nuggets primarily come from the time when the Sun was still active. Wishing for another solar maximum to come again in our lifetime, this Nugget takes up a similar subject on the basis of STEREO observations of flares during 2007-2009. In retrospect, December 2006 was the last active period, having produced four X-class flares. We had 11 M-class and 95 C-class flares between January 2007 and October 2009. They have allowed us to reconfirm that flares, if not X-class, are not highly correlated with CMEs, and to find that one of the CME-related signatures, namely large-scale coronal dimming, is a good indicator of a flare-associated CME. A "dimming" event is simply the sudden disappearance, or depletion, of a part of the corona. A summary of this study is given here with all the associated movies and plots. The summary table, which is occasionally updated, includes all the M- and C-class flares between March 2007 and October 2009 as well as many isolated B- and A-class flares or even non flares.
CME formation close to the Sun
We were made aware that a spectacular CME can be launched without appreciable signatures in the corona by a striking example. But if the CME is somehow associated with a flare, we can expect the first CME manifestation to be seen close to the Sun, say within 0.5 solar radius above the solar surface. The C1 telescope of SOHO LASCO delivered such observations, but it stopped functioning in June 1998. Now each STEREO spacecraft carries a coronagraph (COR1) whose starting height is 1.3 solar radius from the center. By combining data from COR1 and EUVI (which is an instrument like SOHO EIT), we can easily confirm the occurrence of a CME in association with the given flare. The increasing separation of the STEREO spacecraft reduces the dependence of CME visibility due to Thomson scattering. Even if one CME is largely out of the plane of the sky as viewed by one coronagraph, it can be a limb CME as viewed by another coronagraph. We used the FESTIVAL software as included in SolarSoft. The movies made for the individual flares look like the figure below. The reader may try running the movies for the events on 31 December 2007 and 26 April 2008. These are spectacular events, but most flares are less eventful than this typical example (a C flare located near the southwest limb as viewed from STEREO B).
Decoupling of flares from CMEs -- dependence of ARs
By comparing these EUVI-COR1 movies with COR2 movies available with, e.g., the NRL SECCHI movie maker, it is found that only a small number of M- and C-class flares were associated with big CMEs. Examples are: 31 December 2007 00 UT, 2 January 2008 09 UT, 25 March 2008 18 UT and possibly 5 April 2008 15 UT (which occurred behind the west limb, so the flare emission was severely attenuated). Including less energetic CMEs, only 1 or 2 M-class flares (out of 11) and 6-8 C-class flares (out of 86 from March 2007 onward) were unambiguously associated with a CME. One can argue that experts can identify more CMEs, but are those CMEs that the experts label "Very Poor" (see, for example, the CDAW LASCO CME catalog) really proper CMEs? It would be interesting to compare with events selected by automated CME detection schemes such as CACTus. We note that the poor association of these M- and C-class flares with CMEs is largely affected by the fact that they occurred in a small number of active regions that happened to be CME-poor. They are AR 10960 (June 2007), AR 10963 (July 2007), AR 10978 (December 2007), AR 11024 (July 2009) and AR 11029 (October 2009). These regions had relatively large (200-600 millionth) sunspot areas, and high flux near sheared polarity inversion (as indicated by MDI data), which may indeed be used to forecast major flares. It is interesting to note that two major CMEs associated with C-class flares occurred in AR 10980, which had the sunspot area of only 10 millionths. This region was CME poor in the previous rotation. See Figures 2 and 3.
Apart from AR 10956 (May 2007), the CME-productive regions or those that produced at least one decent CME did not have particularly large spot areas or sheared polarity inversion lines. Most flares in these regions were below C-class. Note that the 26 April 2008 CME associated with a B-class flare occurred in a spotless region. Moreover, there were a number of large CMEs of front-side origin that had nothing to do with flares. Therefore, one needs to be careful when using the terms flares and CMEs interchangeably. They may indeed be synonymous for extremely big events, but not for smaller ones we more commonly observe.
Dimming and waves
We have analyzed EUVI data to associate the presence or absence of a CME with a coronal dimming or a large-scale EUV wave. In order to observe these phenomena, we should use two kinds of difference images, that is a running difference (successive images of a sequence differenced) and a base difference (the first or some pre-event reference image is subtracted). The EUV waves were first identified in running difference images, but base difference images capture long-lasting dimmings better. There are technical issues regarding how to correct for solar differential rotation, so there is a limit to the span of a useful base-difference image. There are several publications that show both sets. We routinely make movies of both types of difference images to identify the two phenomena with respect to the X-ray light curves (see Figure 4). A good example may be the event on 13 February 2009. It was observed in quadrature by the two STEREO spacecraft, which were ~90 degrees apart. Try the movies from EUVI-B and EUVI-A. But do not forget to check out a boring event, which is more representative of all the flares in this solar minimum.
The impression from repeated viewing of the movies is usually consistent with more quantitative analysis as to the magnitude of the dimming that may be correlated with the energy of the associated CME. Even though the CME may start small and become larger-scale as it propagates, its associated dimming is expected to be large. Therefore we study the light curves in summed pixels or macro pixels. Here we summed 40x40 EUVI full resoluton pixels (about 1.6 arcseconds). An example is shown in the "confusogram" here. If there are more than 10 macropixels which start to dim around the flare onset and go down below 90% of the pre-event value, at least in two wavelength channels, then a substantial CME will be seen. This serves as a rule of thumb for flare-associated CMEs, and we know that many CMEs, not associated with flares, do not leave as clear dimming signatures as the above confusogram plot. We also observed a number of events with clear wave-like propagations. But they may not necessarily signal major CMEs. Indeed, some of them are associated with only minor (diffuse, narrow, slow) CMEs if dimming is not extended or deep.
The main purpose of this nugget is to remind the reader that flares and CMEs represent essentially different things, taking advantage of simple magnetic field configurations during solar minimum. A number of major CMEs of front-side origin occur without appreciable flares. Conversely, there are only a handful of M-class and C-class flares that are associated with real CMEs propagating beyond 10 solar radii. In these special events, the flare may be integrated into the CME processes. But a majority of flares are just local phenomena. During solar maximum, many things occur in combination, often blurring the true relationship between CMEs and flares. For prediction business, large sunspot areas and strong shear (as found in longitudinal magnetograms) may not necessarily signal CMEs, although they may be linked to major flares. Coronal dimming in EUV images could partly be due to a temperature effect, but if it is large and deep enough to be clearly identified, it is a good indicator of flare-associated CMEs. We are aware of the cases where a CME (typically originating from areas outside active regions) does not leave any low coronal signatures including dimming. Some homework for the RHESSI workers is to determine whether it is possible to distinguish flares associated with major CMEs from those not associated, on the basis of hard X-ray images.