Non-Potentiality of the Magnetic Field Beneath the ... - Springer Link

NON-POTENTIALITY OF THE MAGNETIC FIELD BENEATH THE ERUPTIVE FILAMENT IN THE BASTILLE EVENT LIRONG TIAN1 , JINXIU WANG1 and DEJIN WU2 1 National Astronomical Observatories, Chinese Academy of Sciences, Beijing, 100012, China (E-mail: [email protected]) 2 Purple Mountain Observatory, Chinese Academy of Sciences, Nanjing, 210008, China

(Received 7 January 2002; accepted 12 June 2002)

Abstract. Based on photospheric vector magnetograms obtained at Huairou Solar Observing Station (HSOS), non-potential characteristics of the magnetic field beneath the filament in active region NOAA 9077 are investigated. We focus on the structure and evolution of the magnetic field from 00:08 UT to 10:25 UT of 14 July before the Bastille event. Particular attention is paid to transverse field strength, shear degree and horizontal gradient of the line-of-sight magnetic field around the filament and filament channel. The following characteristics are found. (1) The magnetic nonpotentiality has an obviously non-uniform distribution. The shear degree of the transverse field (Hagyard et al., 1984) is very large, up to 75◦ in some sites beneath the filament, such as the initial brightening site in TRACE 1600 Å images and the breaking site of the filament. The transverse field and the horizontal gradient of the line-of-sight field are very large in some parts corresponding to the hottest and continuously brightening portions. (2) The mean strength and mean shear angle of the transverse field and mean horizontal gradient of the line-of-sight field have obviously dropped in most parts beneath the filament for two hours before the filament eruption and onset of the X5.7/3B flare. After comparing simultaneous UV and EUV images, Hβ filtergrams and Dopplergrams at solar atmosphere, we suggest that magnetic cancellation is likely to quickly transport the magnetic energy and complexity into the higher atmosphere in these two hours. This leads to magnetic instability in the filament and eventually causes the eruption of filament and onset of the flare.

1. Introduction NOAA 9077 (N18, L = 310) appeared on the visible solar disk from 7 July 2000 until 21 July 2000. This active region was a large group with a magnetic complexity of the βγ δ-magnetic structure, producing four major flares during this period. At the position N22W07, NOAA 9077 produced the fourth and the largest flare (X5.7/3B) in the interval of 10:03–10:43 UT on 14 July associated with a 2600 s.f.u. Tenflare, Type II and IV radio sweeps, and a fast-moving halo-CME. This flare also produced a huge solar proton event with 24000 p.f.u. TRACE had good observations of this event in the lower chromosphere, transition region and coronal atmosphere; a good coverage of vector magnetic field observation was taken at HSOS from 11 to 17 July. This provides us with a chance to examine in detail the evolution of the magnetic field that was associated with the filament eruption, flare occurrence and CME initiation. Solar Physics 209: 375–389, 2002. © 2002 Kluwer Academic Publishers. Printed in the Netherlands.

376

LIRONG TIAN, JINXIU WANG, AND DEJIN WU

Several people have studied this active region and the major event. Zhang et al. (2001) showed magnetic flux cancellation associated with the major event on 14 July; Liu and Zhang (2001) demonstrated the relationship between motion of the spots and the major flare X5.7/3B. Deng et al. (2001) focused their analysis on the daily change of non-potential characteristics in the photospheric magnetic field from 11 to 15 July. Kosovichev and Zharkova (2001) studied magnetic energy release and magnetic transients during the major flare. It is commonly believed that the violent activities in the solar atmosphere, e.g., filament eruption and flare onset, are powered by the magnetic energy stored in the non-potential component of magnetic fields. Coronal mass ejections (CMEs) are observed in close association with the eruption of filaments, which are usually found above magnetic neutral line in complex active regions. Thus, it would be of great importance to investigate magnetic non-potentiality and its evolution under the filament (the vicinity of the magnetic neutral line in the photosphere) before the major events on 14 July. Magnetic shear angle is an important index of the non-potentiality of magnetic field, which has been studied by many people (Hagyard et al., 1984; Sakurai et al., 1992; Lü, Wang, and Wang, 1993; Wang, Ewell, and Zirin, 1994). In these studies, the authors pointed out that the sheared configuration of the active regions is an essential characteristic of strong flare activities. It has been established in the early observations that major flares usually occur near magnetic neutral lines with a strong gradient of the vertical magnetic component and strong shear in transverse component (Zirin and Tanaka, 1973). Studies of the correlation between filament disturbance and corresponding changes of the magnetic environment can help us to understand the possible driving mechanisms of filament eruption and related phenomena, such as flares and coronal mass ejections. More than ten years ago, Leroy (1989) stressed the urgent need for better knowledge of the photospheric magnetic neutral line hosting a filament. Unfortunately, few studies of systematic analysis on the filaments based on vector magnetic field have been made so far. Wang and Li (1998) made the first analysis on the vector magnetic fields beneath filaments in NOAA 6891. In the current study, we will study the evolution of magnetic non-potentiality in the vicinity of the magnetic neutral line in the photosphere on 14 July, and seek how it was associated with the occurrence of the major events. In Section 2 we will describe the data used in this paper. Section 3 shows pre-event activity. Section 4 describes the evolution of vector magnetic fields. In Section 5 we will summarize the non-potential characteristics of the active region and their evolution. The associated filament disturbance will be shown by chromosphere Dopplergrams in Section 6. The conclusions and discussions will be given in Section 7.

NON-POTENTIALITY OF THE MAGNETIC FIELD BENEATH ERUPTIVE FILAMENT

377

2. Data The data used in this paper include the line-of-sight and transverse magnetograms in the photosphere, Hβ 4861 Å filtergrams and Dopplergrams in the chromosphere. They were obtained by the Solar Magnetic Field Telescope, a vector magnetograph system (Ai and Hu, 1986) installed at HSOS of National Astronomical Observatory, with a field of view of about 5.23 arc sec × 3.63 arc sec (512 × 512 pixels of CCD). The photospheric vector magnetic field is measured with the line Fe I 5324 Å. The line-of-sight magnetic field strength (B|| ) and transverse magnetic field strength (B⊥ ) are obtained following the methods introduced by Wang et al. (1996). Here, we use force-free approximation to remove 180◦ ambiguity. The temporal and spatial resolutions of the vector magnetograms are about 5 min and 2 arc sec, respectively. The noise level of the line-of-sight magnetic field is about 10 G. For the transverse field measurements, the noise level is estimated from the standard deviation of the transverse field in the area without magnetic signals in the line-of-sight magnetogram and is found to be about 100 G (see the detailed description by Wang et al., 1996). We also use TRACE and SOHO/EIT data to study the phenomena in the solar corona and the transition region. TRACE has a number of EUV and UV passbands through which it observes the Sun with a spatial resolution of 1.0 arc sec at 10–60 s cadence typically. TRACE generally observes an 8.5 arc min square field of view in broadband visible light, in a UV passband centered on Lα and three wavelength ranges around 1600 Å, or in the coronal 171 Å, 195 Å, and 284 Å passbands.

3. Pre-Event Activities on 14 July In order to investigate the cause of the Bastille event (manifested by a filament eruption, an X5.7/3B flare and a halo CME), we will study magnetic evolution and its association with the pre-event activity shown in different wavelengths. In Table I, we list the pre-event activities on 14 July and their corresponding locations. Disturbance of the filament was obvious in Hβ Dopplergrams and 195 Å images. The pre-event activities can also be seen in the 1600 Å images. Figure 1 is a vector magnetogram marked by AB, CD, DE, FG, and HL along the magnetic neutral line, which represent the hottest position (Schrijver and Title, 2001), the strongest intrusion part of opposite polarity, the continuously brightening position, the obvious magnetic cancellation and the breaking of filament, respectively. We will study the non-potentiality of the magnetic field and their evolution in these parts. The letter ‘P’ denotes a stable sunspot during the day. We make a seeing correction for all magnetograms by assuming this sunspot maintained constant magnetic flux. A detailed description for the seeing correction can be found in Section 4. The letter ‘O’ is the initial point.

378


TABLE I Time sequence of the solar activity events on 14 July. Start time

End time

Phenomena

00:12 UT

01:36 UT

M1 flare, CME dimming SF flare SF flare Disturbance of the filament in Hβ Dopplergrams and 195 Å images Breaking of part of the filament in 195 Å images and Hβ Dopplergrams Breaking of the filament in EIT images X5.7/3B flare, Halo CME

04:30 UT 07:35 UT 07:49 UT

05:20 UT 08:11 UT 09:48 UT

09:48 UT

10:01 UT 10:03 UT

10:43 UT

Location

CD, DE, FG, HML CD, DE, FG, HML AB, CD, FG

DE-FG-HML

HML HML AB, CD, DE, FG, HML

Figure 2 shows the pre-event activities, including the two sub-flares and the disturbance of the filament. The Hβ filtergrams in the chromosphere are in the left panels. TRACE UV 1600 Å images are in the middle panels, which display cool material emission (4.0 × 103 K, Handy et al., 1999) at lower chromosphere. TRACE EUV 195 Å images are in the right panels, which commonly display the emission pattern of transition region and corona. We can find some brightening sites and filament activities in Figure 2. The long, inverted-L-shape filament has existed for a long time. In TRACE 195 Å images (the right panels of Figure 2), we can see two parts in the middle part of the filament (HL in Figure 1), twisting each other (see Figure 2(a) and 195 Å movie). From 07:39 UT, this part began to be disturbed or twisted more tightly. At 09:48 UT, the filament began to break at the position of M in Figure 1. The long-lasting filament began to break from its right side at 10:01 UT seen in TRACE and SOHO EIT 195 Å images. Then the filament erupted from the bottom, which is shown in Figure 3. Similar characteristics can also be seen in the chromospheric Dopplergrams (Figure 7). The UV 1600 Å images from TRACE show a brightening ribbon in the left part of the filament (DE in Figure 1). It began to brighten at 08:31 UT. From the 1600 Å movie, we found that the continuous brightening occurred at this position and extended to the right of the filament before the X5.7/3B flare. In the left panels the corresponding brightening sites in Hβ filtergrams are shown before the major event.


379

Figure 1. A vector magnetogram of NOAA 9077 at 04:14 UT on 14 July. White patches denote positive line-of-sight fields, and black patches denote negative fields. The transverse component is shown by short bars, whose length is proportional to the field strength, an alignment parallel to the transverse field direction. In the left of the magnetograms, letter ‘P’ indicates a sunspot which changed little. Letters O, A, B, C, D, E, F, G, H, M, and L denote some points on the magnetic neutral line. The field of view is about 248 by 173 square arc sec.

4. Vector Magnetic Field of NOAA 9077 on 14 July Figure 4 shows the evolution of the vector magnetic field of NOAA 9077 on 14 July. We make a seeing correction for all magnetograms by assuming the ‘P’ sunspot in Figure 1 had constant magnetic flux on 14 July, so that a correction factor is determined and multiplied to the V , Q, and U images when seeing was bad. On the other hand, inconsistency of the transverse and line-of-sight components limits deduction of the magnetic non-potentiality. We can determine an area where magnetic configuration is approximately potential. The potential transverse components can be calculated theoretically based on the observed line-of-sight component. To make the line-of-sight and transverse measurements consistent, we assume the observed transverse field at the sites of potential configuration to be the same as the transverse field extrapolated from the line-of-sight component. Thus, a correction multiplier for the transverse field calibration is determined (see detailed description by Wang et al., 1996). It is noticed that the magnetogram in the Figure 4(h) has different sensitivity from others, which was obtained at 10:25 UT when the major flares already began. Such big flares probably deform the spectral line profile, therefore, it shows differences from the others. Moreover, it is the last

380


Figure 2. Pre-event activities on 14 July at different wavelengths. Hβ filtergrams in the chromosphere were obtained from HSOS. UV 1600 Å images and EUV 195 Å images are from TRACE.

magnetogram because evening would be coming at HSOS. The seeing correction could not be effective for it. A vector magnetic field movie shows the squeezing motion of flux around the magnetic neutral line in the parts CD and DE. The negative magnetic field squeezed upward into the positive one. The positive field intruded downward into the negative one. Obvious flux decrease is found in two sides of the neutral line at area


381

Figure 3. Filament eruption at 10:10 UT shown in EIT 195 Å image from SOHO.

‘M’ from 06:34 UT to 09:42 UT (the white patches from Figures 4(c–g)). Thus, magnetic flux cancellation is an obvious phenomenon in AR 9077.

5. Non-potentiality of Magnetic Fields beneath the Filament The degree of magnetic shear is a primary non-potential parameter, as emphasized by Hagyard (1997). The shear angle was defined as the angle between the observed transverse magnetic field and the transverse component of potential field, extrapolated from the observed line-of-sight field (Hagyard, Venkatakrishnan, and Smith, 1990). Sheared configuration of the active regions is an essential characteristic for major flare activity. Another quantity is the horizontal gradient of line-of-sight magnetic field, ⊥ Bz =

∂Bz ∂Bz ex + ey , ∂x ∂y

which indicates the degree of squeezing of line-of-sight magnetic fields and the possibility of magnetic reconnection. It is also part of the horizontal current. In this paper, the horizontal gradient of line-of-sight magnetic fields is measured from positive to negative polarity crossing the magnetic neutral line. X0 and Y0 are the coordinates of a point on the neutral line. (Xg0 , Yg0 ) and (Xg1 , Yg1 ) are two points at two sides of the neutral line. The three points are all on a vertical line to

382


Figure 4. Time sequence of vector magnetograms observed at HSOS. The field of view is about 220 by 153 square arc sec.


383

Figure 5. Distribution of non-potentiality of magnetic fields along the neutral line from the beginning point, ‘O’ in Figure 1. (a) The distribution of the transverse fields. (b) The distribution of horizontal gradient of line-of-sight magnetic fields crossing the magnetic neutral line from the positive polarity to the negative one. (c) The distribution of shear angle of transverse fields. The left panels are at 01:01 UT on 14 July; the right panels are at 09:48 UT of the same day. AB, CD, DE, FG, and HL correspond to the parts of the neutral line in Figure 1.

the neutral line at (X0 , Y0 ). S equals to 4 arc sec, half width from (Xg0 , Yg0 ) to (Xg1 , Yg1 ). We calculate the horizontal gradient, Bz , from (Xg0 ,Y g0) to (Xg1 ,Yg1) by 2 Bz = B(X g0 , Yg0 ) − B(Xg1 , Yg1 ). Here, Xg0,1 = X0 ± KS/ (1 + K ), Yg0,1 = Y0 ± S/ (1 + K 2 ) and K = −dx/dy, where dx and dy are the sizes of one pixel in the x- and y-directions.

384


5.1. N ON - POTENTIALITY ALONG THE MAGNETIC NEUTRAL LINE Figure 5 shows the distribution of non-potential characteristics of the magnetic field in the vicinity of the neutral line at 01:01 UT (the left panels) and 09:48 UT (the right panels) on 14 July. X-axis indicates arc length of the neutral line from the beginning point, ‘O’, in the long-axis direction of the filament. Figures 5(a) show the distribution of the transverse field along the neutral line, Figures 5(b) are the distribution of the horizontal gradient of line-of-sight magnetic field crossing the neutral line, Figures 5(c) are the distribution of the shear angle. We find that the magnetic non-potentiality has an obviously non-uniform distribution. The transverse field strength and the horizontal gradient are strongest at AB (the hottest part shown by Schrijver and Title, 2001), and CD (the strongest intrusion part) before the major flare occurred. The strongest transverse field is more than 3000 G. The biggest horizontal gradient is up to 2.0 G km−1 . The largest shear angle occurred at the part of DE, which corresponds to the continuously brightening parts, up to 75◦ . For the breaking site of the filament, HL, though the transverse fields and the horizontal gradient are not very high, the shear angle is large. All these three parameters have some changes along the filament. The detailed evolution will be studied in the next subsection. 5.2. E VOLUTION OF THE MAGNETIC NON - POTENTIALITY BENEATH THE FILAMENT

To understand the detailed evolution of the non-potentiality, we select 5 segments of the filament under which magnetic patches with opposite polarities interact vigorously. For each piece along the magnetic neutral line we measure the mean transverse field, the mean gradient and the mean shear angle. The procedure allows visual selection of pixels of the neutral line in the magnetogram, to measure all the non-potential parameters statistically. To avoid selection effects and to reduce error, for each piece of the magnetic neutral line, we carry out the measurement as follows. For example, for a magnetogram at time t, 5 segments are selected along the neutral line. For each one, for example AB, the selection is done using a cursor and picking the 250 points and then averaged values of transverse fields BT , horizontal gradient Bz , and shear angle ψ are calculated from the 250 points in the AB part. Then the 250 points are again selected – different ones but along AB. This is repeated 10 times. The averaged value of all these measurements is plotted in Figure 6 along with the rms value for the standard deviation. In Figure 6 we exhibit the temporal evolution of the mean transverse fields, the mean horizontal gradient of the line-of-sight magnetic fields and the mean shear angle near the parts of AB, CD, DE, FG, and HL beneath the filament on 14 July. The left panels are the evolution of the mean transverse fields. The middle panels are the evolution of the mean shear angle. The right panels are the evolution of the mean horizontal gradient. In Figure 6 each point denotes a mean value of the corresponding part near the magnetic neutral line. Error bars show the standard


385

Figure 6. Development of the mean strength of transverse fields, the horizontal gradient of the line-of-sight magnetic fields, and the mean shear angle in some key parts (AB, CD, DE, FG, and HL in Figure 1) beneath the filament. The error bars denote the standard deviation.

deviation of the 10 measurements. A total of 17 vector magnetograms on 14 July are used in this investigation. From Figure 6, we found for those parts under the filament that the strongest mean transverse field is up to 2000 G; the steepest mean horizontal gradient is about 1.0 G km−1 ; the biggest mean shear angle is more than 50◦ . On 14 July,

386


we know that some small flares occurred before the X5.7/3B flare. They are listed in Table I, which can also be found in Figure 2. The corresponding change for the three parameters is not obvious after these minor flares. However, we find a general and obvious tendency for each segment. The mean values of the three parameters first increase a bit before the main dropping, then decrease much more at different degrees for different parts of the filament after 08:00 UT. Table II gives the dropping range of the three non-potential parameters after 08:00 UT and corresponding r.m.s. deviation. The large dropping may imply that more free energy was transported from the photosphere to the upper atmosphere. From Table II, we find that a large drop happened at the sections AB, CD, and DE, which correspond to the hottest and continually brightening parts in TRACE images. In the part of HL, the obvious decrease of the parameters happened at about 06:30 UT, which corresponds to the start of the filament disturbance in Figure 2. From the movie of vector magnetograms, we find that more flux cancellation occurred in the parts CD, DE and near point ‘M’ of segment HL. We can predict that the filament above the photosphere would be disturbed because of this obvious change of magnetic structure in the photosphere. TABLE II Drop of magnetic non-potentiality after 08:00 UT and corresponding r.m.s. variations δ in each segment. Parameter

Along AB

Along CD

Along DE

Along FG

Along HL (from 6:30 UT)

BT (G) δ(BT ) Bz (G km−1 ) δ(Bz ) ψ (deg) δ(ψ)

800 ± 100 360 0.40 ± 0.05 0.13 3±1 2.0

1000 ± 100 401 0.35 ± 0.05 0.1 10 ± 1 3.7

800 ± 100 219 0.10 ± 0.01 0.05 8±1 4.2

100 ± 50 108 0.04 ± 0.01 0.03 1±1 1.6

500 ± 50 151 0.06 ± 0.01 0.04 5±1 2.0

6. Disturbance of the Filament in the Chromosphere Figure 6 and Table II exhibit obvious decreasing for the three non-potential parameters approximately from 08:00 UT. Could these changes affect the chromosphere and corona? Is it overwhelming for the major event at 10:03 UT? In order to find the consequence of the significant decrease in magnetic non-potentiality in the photosphere, we will give the evolution of Dopplergrams in the chromosphere in Figure 7.


387

Figure 7. Time sequence of Hβ Dopplergrams in chromosphere. White ribbons indicate red-shifted components (downward movement), and black ribbons indicate blue-shifted components (upward movement). Arrows show the position of the filament disruption and twist of two parts of the filament.

In Figure 7, white ribbons indicate red-shifted components, and black ribbons indicate blue-shifted components. We find that some of the filament material moved downward (white patches), and other filament material moved upward (black patches). This implies that the filament destabilized. At 07:49 UT, the filament seemingly had two parallel parts, shown by two arrows in Figure 7(c). Until 08:20 UT, the two parts form a ‘X’-shape, then more and more join. The TRACE 195 Å images show that the two crossing parts of the filament twisted early from 07:39 UT, afterwards they twisted more and more tightly. Finally, a part of the filament broke off at 09:48 UT. Figure 7(i) shows the start of the filament eruption, when the filament just broke at ‘M’ point, the right side of the inverted-L-shaped filament, shown by an arrow. Later on, some materials were thrown away along the bottom of the inverted-L filament, the DE piece in the neutral line. In Figure 3 and a movie of SOHO EIT 195 Å images, we clearly find a whip-like eruptive motion at the bottom part of the inverted-L filament. Another violent motion of the filament occurred in part DE. From about 08:20 UT, in part DE, the filament showed a more violent motion (white patch in Figures 7(e–h)) when the non-potentiality in the photosphere decreased greatly. At this place, a lot of free energy must have been released to create the major flare. On the whole, the Hβ Dopplergrams observations reveal

388


that the filament had been disturbed violently for more than 2 hours before the eruption and the flare onset, in correspondence to the evolution of non-potentiality in the photosphere.

7. Conclusions and Discussion From the above analysis, we obtain the following results. (1) The magnetic non-potentiality has an obviously non-uniform distribution. The shear degree of the transverse field is very large, up to 75◦ at some sites beneath the filament, such as the initial brightening site in TRACE 1600 Å images and the breaking site of the filament. The transverse field and horizontal gradient of the line-of-sight field are very large in parts corresponding to the hottest and continuously brightening portions. (2) Mean strength and mean shear angle of the transverse field and mean horizontal gradient of the line-of-sight field have obviously dropped in most parts beneath the filament for about two hours before the filament eruption and onset of the X5.7/3B flare. After comparing simultaneous UV and EUV images, Hβ filtergrams and Dopplergrams in the atmosphere of NOAA 9077, we suggest that magnetic cancellation, an indicator of magnetic reconnection, is likely to quickly transport the magnetic energy and magnetic complexity into the higher atmosphere (Martin and Livi, 1992) in two hours. The filament became unstable because the magnetic reconnection affected magnetic structure where the filament was. The instability finally resulted in the filament eruption and the onset of the major flare. From this study, we think that the two hours may be the temporal scale of the instability for the magnetic structure and the filament. The change and instability of the magnetic structure in the two hours is probably critical in the filament eruption and the onset of the major flare.

Acknowledgements This research is supported by NSFC Grant 19973009, 10073013 and NKBRSF G20000784 in China. L. Tian thanks Drs Yang Liu and Jun Zhang for their valuable comments and discussions. The authors are indebted to the anonymous referee for helpful suggestions, and the TRACE and SOHO teams for providing the wonderful data. They are grateful to the HSOS team for good observations of vector magnetic fields.


389

References Ai, G. and Hu, Y.: 1986, Publ. Beijing Astron. Obs. 8, 1. Deng, Y. Y., Wang, J. X., Yan, Y. H., and Zhang, J.: 2001, Solar Phys. 204, 11. Hagyard, M. J.: 1997, in Solar-Terrestrial Prediction V, p. 527. Hagyard, M., Smith, J., Teuber, D., and West, E.: 1984, Solar Phys. 91, 115. Hagyard, M. J., Venkatakrishnan, P., and Smith, J. B., Jr.: 1990, Astrophys. J. Suppl. Ser. 73, 159. Handy, B. N. et al.: 1999, Solar Phys. 187, 229. Kosovichev, A. G. and Zharkova, V. V.: 2001, Astrophys. J. 550, L105. Leroy, J.-L. 1989, in E. R. Priest (ed.), Dynamics and Structure of Quiescent Solar Prominences, Kluwer Academic Publishers, Dordrecht, Holland, p. 77. Liu, Y. and Zhang, H. Q.: 2001, Astron. Astrophys. 372, 1019L. Lü, Y., Wang, J., and Wang, H.: 1993, Solar Phys. 148, 119. Martin, S. F. and Livi, S.: 1992, in Z. Švestka, B. V. Jackson, and M. E. Machado (eds.), Lecture Notes in Physics 399, 33. Sakurai, T., Shibata, K., Ichimoto, K., Tsuneta, S., and Acton, L.: 1992, Publ. Astron. Soc. Japan 44, L123. Schrijver, C. J. and Title, A. M.: 2001, Solar Phys. 200, CD-ROM Wang, H., Ewell M., and Zirin H.: 1994, Astrophys. J. 424, 436. Wang, J. and Li, W.: 1998, in D. Webb, D. Rust, and B. Schmieder (eds.), ‘New Perspectives on Solar Prominences’, IAU Colloquium 167, 98. Wang, J. and Shi, Z.: 1993, Solar Phys. 143, 119. Wang, J., Shi, Z., Wang, H., and Lü, Y.: 1996, Astrophys. J. 456, 861. Zhang, J., Wang, J., Deng, Y., and Wu, D.: 2001, Astrophys. J. 548, L99. Zirin, H. and Tanaka, K.: 1973, Solar Phys. 32, 173.