Low-Coronal and Coronagraphic Images: Their Complementary Roles in

Understanding Geo-Effective Eruptions

N. V. Nitta, Lockheed Martin Solar and Astrophysics Lab.

It has been established that coronal mass ejections (CMEs) are the primary driver of severe space weather. By definition, CMEs are observed by white-light coronagraphs. However, a question arises as to whether modern coronagraphs (e.g., SOHO/LASCO) are sensitive enough to capture all the Earth-directed and potentially geo-effective CMEs. In the recent list of interplanetary coronal mass ejections (ICMEs) during 1996-2002 by Cane and Richardson (2003), 30-50% of ICMEs have no associated halo CMEs. Even though no features in X-ray and EUV images are known to be 100% correlated with CMEs (Hudson and Cliver 2001), these images are useful for understanding the origin of geo-effective CMEs.

CME Source Regions

• The source region of a fast CME, as observed by LASCO, can usually be located unambiguously in EUV/X-ray images taken within ~1 hour before the first detection of the CME.

• If we do not see in EUV/X-ray images any signatures (e.g., waves, dimming, post-eruption arcades, etc.) that often accompany CMEs, we may conclude that the fast CME comes from backside.

• But how well is the source region identified, if the CME is slow and diffuse, and its first detection by LASCO C2 comes hours later than a CME-associated phenomenon?

Fast CMEs

Perhaps nobody dares to question the link between the CME and the flare/filament eruption on 14 July 2000. Note that low coronal images combined with magnetic field data/extrapolation, not coronagraph images, help us understand its origin and its possible geo-effectiveness.

We already know from the major SEP event (starting as early as 10:50 UT) that the eruption drove an IP shock. LASCO data are useful for predicting the shock arrival at Earth (combined with DH type II burst data) – but not in this event because of too much contamination with SEPs.

Figure 1: Bastille Day 2000 event observed by LASCO, EIT and TRACE

Slow CMEsFigure 2: LASCO height-time plot superposed with the GOES X-ray light curve. Extrapolation of the height-time relation (noisy!) puts the CME departure time at 03:05 UT from the active region at S17 W19.

But this fit ignores the data point at 04:40 – no CME was detected above 2Rsun (red). In fact, EUV/X-ray images suggest no CME-like activity until the flare onset at 04:31 UT, so the true height-time relation may be something like the blue line.

If we blindly use the LASCO height-time relation without reference to the data coverage or to low coronal images, we often conclude the CME onset time to be in the middle of nowhere. This has a danger of unduly separating CMEs from what we know in the corona.

Use of Height-Time Relation

Figure 3: Tracing the ejection from the low corona. (a) Height-time plot superposed with the GOES X-ray light curve. Linear fit to the LASCO C2/C3 height-time relation puts the CME lift-off time to be ~2:00 UT, well before the flare onset. Inclusion of the C1 data makes a better link between the CME and the flare X-ray ejection. However, plotting the ejections in X and Y directions separately (in b and c) indicates that things are not so simple.

To relate the phenomena in EUV/X-ray images with CMEs, it is important to observe 1.1-2 Rsun. According to the above figure, however, (1) CMEs may be highly non-radial in the beginning (then the “height” may not be a good representation), or (2) CMEs observed by C2/C3 may be physically separate from the low coronal counterparts.

Earth-directed CMEs – Halo CMEs

Figure 5: As predicted, a clear magnetic cloud started on 1997 January 10 and triggered a reasonably severe storm. Hence the importance of halo CMEs established to predict storms.

Figure 4: Partial halo CME on 1997 January 6. This occurred during a CDAW, where it was predicted that the ejected material would be observed near Earth 3-4 days later and that it might cause a geomagnetic storm.

Figure 6 (Movie): But the solar signatures in Yohkoh/SXT images around the estimated CME liftoff time are minimal (no EIT data). There was a bigger eruption on the previous day, but no halo CME was detected.

Movie not included in CD. Get it from the workshop ftp site or from here.

How can we be sure that a halo CME is Earth-directed, when we do not see characteristic patterns for CMEs in the low corona?

Alternative Scenario

Figure 7:Correlation between the maximum ICME speed and the shock transit speed, which was used by Webb et al. (1998) to argue that the eruption on January 6 is a better candidate for the magnetic cloud on January 10 than that on January 5. But the scatter is already large, and even the case for January 6 makes an outlier.

If we do not start from the halo CME – ICME association as given, the following scenario may be a possibility.

1. The magnetic cloud on January 10 originated from the eruption (with LDE and filament disappearance) on January 5.

2. The January 6 halo CME was a backside event, not related with the disappearance of a filament, as noted by Webb et al., in the NW part of the region (note that the halo CME was seen mostly over the S limb).

3. Neither filament eruption on January 5 nor January 6 caused a CME intense enough to be observed by LASCO.

But SXT may have failed to detect key signatures from the eruption on January 6, which may have indeed been responsible for the halo CME. These signatures may be found only with simultaneous images in a broad temperature range.

Or can coronagraph images tell us convincingly whether the CME was front-sided?

Sensitivity of Coronagraphs

Figure 8: On 1997 May 10, MURI region AR 8038 produced an eruption less global than the one on May 12. LASCO did not observe a CME on May 10.

When geo-effectiveness is studied in terms of halo CMEs, the sensitivity of the instrument to motions off the plane of the sky is seldom addressed, as if the halo CME were a necessary condition. But in the ICME list of Cane and Richardson (2003), many ICMEs are not associated with LASCO CMEs.

Weird Geo-effective Events

Figure 9: Some geomagnetic storms are not associated with fast CMEs like the Bastille Day 2000 event. Zhang et al. (2003) argued that they are attributable to faint slow partial halo CMEs from the over the east limb.

But during the periods of interest, multiple eruptive events occurred closer to the disk center but without CME association. For example, the cusp-like appearance in SXT images often corresponds to a post-eruption arcade. Why are they not seriously considered?

This may again be related to the issue of the sensitivity of coronagraphs.

We should identify the coronal manifestation of ICMEs, at least magnetic clouds which may have a more organized origin than other kinds of ICMEs.

CMEs and Non-eruptive Flares X-ray/EUV images occasionally reveal ejections. But some of them do not move long distances and are not associated with CMEs.

Figure 10: SXT images of X-class flares without CMEs. The first two events show slow (<80 km/s) ejections as indicated by thin arrows, but they do not move beyond the overlying loop structures as indicated by thick arrows. The third event does not even show ejections, but slowly (<10 km/s) rising loops.

Figure 11: Wind/WAVES dynamic spectrum for 1997 November 27, overplotted with the GOES X-ray light curve. Note that the X-class flare at 12:10 UT was very quiet in these wavelengths. However, there was a metric type II burst.

EUV Ejections and CMEs

Figure 12 (Movie) : TRACE example of “unsuccessful ejection”, not leading to a CME. The motion decelerated and halted (Ji et al. 2003).

Movie not included in CD. Get it from the workshop ftp site or from here.

Figure (Movie) 13: Similar example. This one seems to be a “hesitant (?)” ejection. A slow CME seems to be associated. The white feature in the encircled area, rather than the filament itself, may be the CME signature.

Movie not included in CD. Get it from the workshop ftp site or from here.

Filament eruptions not associated CMEs are characteristically decelerated. There is a variety to this class of ejections. The filament eruption in the right example is not spectacular, but it may be responsible for the slow CME observed an hour later.

Dimming in Multiwavelength ImagesFigure 14: Coronal dimming in a sigmoidal region as observed by SXT, SXI and EIT. Images dominated by hot plasma (SXT and SXI B_THN_A and P_THICK) show dimming only in Area 3. Area 2 shows decrease in flux but it is due to a brightening around 13:00 UT in most images but EIT and SXI OPEN. Area 1 shows dimming only in EIT and SXI OPEN. The quicker recovery in SXI OPEN images seems to be due to flare brightening. Note the different behaviors of SXI OPEN and P_THN_A even though their temperature responses are close.

We need a broad temperature coverage to tell mass depletion (due to the CME) from temperature effects.

It appears that dimming in EIT images is a reliable indicator for mass depletion due to CMEs. But it is important to supplement it with data that sample different temperatures so that we can distinguish real mass loss from heating/cooling.

We need part II of

which should focus on disk events, calibrating both coronagraphic and low coronal data.

Conclusion