Received 2000 November 7; accepted 2001 February 5; published 2001 ... High-resolution observations of a large solar flare on 2000 July 14 (âBastille Day ...
The Astrophysical Journal, 550:L105–L108, 2001 March 20 q 2001. The American Astronomical Society. All rights reserved. Printed in U.S.A.
MAGNETIC ENERGY RELEASE AND TRANSIENTS IN THE SOLAR FLARE OF 2000 JULY 14 A. G. Kosovichev W. W. Hansen Experimental Physics Laboratory, Stanford University, 455 via Palou, Stanford, CA 94305-4085
and V. V. Zharkova University of Bradford, Richmond Road, Bradford, West Yorkshire, BD7 1DP, UK Received 2000 November 7; accepted 2001 February 5; published 2001 March 15
ABSTRACT High-resolution observations of a large solar flare on 2000 July 14 (“Bastille Day Flare”) from the Michelson Doppler Imager instrument on the SOHO spacecraft reveal rapid variations of the magnetic field in the lower solar atmosphere during the flare. Some of these variations were irreversible, occurred in the vicinity of magnetic neutral lines, and likely were related to magnetic energy release in the flare. A surprising result is that these variations happened very rapidly on the scale of 10–15 minutes in a large area of ∼50 Mm2 at the beginning of the flare. Other, more localized and impulsive magnetic field variations somewhat similar to “magnetic transients” observed by Zirin and coworkers were accompanied by impulses in continuum intensity and Doppler velocity. These impulses have dynamic characteristics similar to Ellerman’s “bombs” and Severny’s “mustaches” and were probably caused by high-energy particles bombarding the solar surface. Subject headings: Sun: activity — Sun: flares — Sun: magnetic fields — Sun: particle emission — Sun: photosphere — Sun: X-rays, gamma rays — sunspots to minimize effects of possible temporal intensity variations within the line and used to calculate the line shifts in both polarizations using a Fourier tachometer method. Then the difference between these shifts is used to estimate the magnetic field, and the mean shift is used for the velocity estimates. Thus, magnetic field and velocity are obtained from the same eight filtergrams. The continuum intensity is estimated from a filtergram averaging narrowband signals from both sides outside the line. It is taken after the first four line filtergrams. This 27 s observing sequence was repeated every minute. We have used all three observed quantities in this analysis.
1. INTRODUCTION
The mechanism of energy release in solar flares is not understood mainly because of the incomplete knowledge of magnetic field variations in the flares. While some observers reported decrease in the magnetic field strength after the flares (Severny 1964; Zvereva & Severny 1970; Moore et al. 1984; Kosovichev & Zharkova 1999), some others argued that the changes are consistent with a general trend of evolution of active regions (see Sakurai & Hiei 1996 for a review). It has been established in the early observations that strong flares mostly occur near neutral lines of the vertical component of the magnetic field with a strong gradient of the field and where the horizontal component has strong shear, which may experience substantial flare-related variations. However, the results are inconclusive (e.g., Wang et al. 1994), sometimes showing increases of the magnetic shear contrary to what is expected if magnetic energy is released. For a near-limb flare, Cameron & Sammis (1999) observed a weakening of one line-of-sight magnetic polarity and an increase in the other. Surprising strong impulsive variations of magnetic field during solar flares (“magnetic transients”) were detected with a videomagnetograph (Patterson & Zirin 1981; Zirin & Tanaka 1981). However, the interpretation of these observations was unclear, and the observed variations were suggested to be caused by an increase of spectral line emission during the flares (Patterson 1984; Harvey 1985). Here we report on observations of the 2000 July 14 flare carried out by the Michelson Doppler Imager (MDI) instrument on SOHO (Scherrer et al. 1995) in the high-resolution mode (00. 625 per CCD pixel). The MDI data, which include line-ofsight magnetic field, Doppler velocity, and continuum intensity, are obtained from nine filtergrams taken in five different po˚ absorption line in right- and leftsitions in the Ni i 6768 A circularly polarized light every 3 s. The actual frame sequence was R2, R1, r1, r2, c2, l2, l1, L1, L2, where R and B are the red and blue far wings of the spectral line, r and b are the close wings, c is the pseudocontinuum, and 1 and 2 are the two polarization states. The filtergrams are taken in pairs
2. VARIATIONS OF MAGNETIC FIELD DURING THE FLARE
The flare of 2000 July 14 that occurred in Active Region NOAA 9077 started at about 10:10 UT and reached the peak soft X-ray flux (X5.7) at 10:24 UT. The flare was located near the disk center (N227 W077); therefore, on average the MDI measurements are nearly 5 times more sensitive to the vertical component of the magnetic field than to the horizontal components. The first significant variations in the magnetic field are detected in the MDI measurements at 10:12 UT, approximately at the start of the soft X-ray flux recorded by the GOES-10 satellite. During the next 30 minutes magnetic field variations are observed in various areas of the active region. Figure 1a shows the MDI magnetogram taken at the beginning of the flare. The rectangular areas numbered from 1 to 7 indicate the places of the most significant variations. These variations are shown as differences between subsequent magnetograms in panels b–f of the left column. The right column shows the corresponding intensity data. During the impulsive phase, from 10:12 to 10:20 UT, variations occurred in region 1, which contained magnetic fields of opposite line-of-sight polarities within the same penumbra (d-type sunspot), and near two leading unipolar spots in regions 5 and 6. During the maximal phase at 10:20–10:25 UT, magnetic field variations were observed in several other regions, usually reflecting changes in the strength of the existing magnetic fields. Figure 2 shows the rms value of the magnetic field, AB 2 S1/2, in regions 1–7 as a function of time, and the correspondL105
L106
SOLAR FLARE OF 2000 JULY 14
Vol. 550
Fig. 1.—Left column: (a) MDI magnetogram taken at the beginning of the flare at 10:11:30 UT; the gray scale corresponds to the field strength from 21500 G (black) to 1500 G (white); the rectangular boxes indicate regions (1–7) of some of the most prominent variations; (b–f) samples of the 1 minute magnetic field differences; the gray scale corresponds to the 5(30–70) G range; the white and black features show positive and negative variations of 70 G and higher. Right column: corresponding intensity image and differences (the gray scale shows 53% variations).
ing variations of the mean relative intensity and the soft X-ray flux. Evidently, there were two types of magnetic variations: irreversible in regions 1–3 (Figs. 2a–2c) and transient in regions 4–7 (Figs. 2d–2g). The irreversible changes of AB 2 S (or the magnetic energy density estimated from the line-of-sight component) provide evidence of magnetic energy release because AB 2 S in these regions became permanently lower during the impulsive phase. The most significant decrease occurred in region 1 during the impulsive phase, where the rms of the magnetic field decreased by approximately 30 G. This is the mean value for the whole region; in the energy release places it was higher (see § 3). The field variations correlated with intensity increases, which, however, did not exceed 1%. In region 1 the intensity increase started near the middle of the magnetic variation and continued after the magnetic variation ended. The transient variations occurred at the beginning of the impulsive phase in region 5, near the X-ray maximum (region 6) and during the decaying phase (region 7). These variations occurred in unipolar areas and were accompanied by strong (up to 10%) impulsive increases of the mean intensity.
Fig. 2.—Temporal variations of rms magnetic field, AB2S1/2 (thick solid curves), and mean relative background intensity, I0 /I0, max (dotted curves), in each of the seven rectangular regions of Fig. 1, and the whole-Sun soft X-ray flux, FX (shaded profile).
3. DISSIPATION OF MAGNETIC FIELDS
The main site of the magnetic energy decrease was in region 1 (see Figs. 1a, 1b, and 2a). This region includes magnetic fields with opposite line-of-sight polarities. These fields were pushed toward each other before the flare, probably by external flows. This resulted in an increase of the magnetic field gradient and hence stronger electric currents before the flare. Figure 3 shows the locations of the magnetic neutral line before the flare at 9:00:30 UT (thin dashed curve) and 10:11:30 UT (thick dashed curve). This line, which separates the two polarities,
No. 1, 2001
KOSOVICHEV & ZHARKOVA
Fig. 3.—(a) Locations of the neutral magnetic line in region 1 at 9:00:30 UT (thin dashed curve) and 10:11:30 UT (thick dashed curve) and contour lines of the difference between the 30 minute averaged magnetic energy density, AB2/4pS, estimated before (9:42:30–10:11:30 UT) and after (10:23:30– 10:52:30 UT) the flare from the line-of-sight component of the field. The contour levels, white (positive) and black (negative), are (1.25, 1.75, 2.5, 5) # 104 ergs cm23 (from thin to thick). The gray-scale image is the MDI magnetogram at 10:11:30 UT.
was moved southward with a speed of ∼200 m s21, with the larger displacement being in the upper part of region 1. The magnetic field gradient there reached 1.3 G km21 and suddenly began decreasing when the flare started. Most of the magnetic energy release occurred very close to the high-gradient area occupied by the negative polarity. The contour curves calculated from only the line-of-sight component of the field (after removing displacements caused by rotation and by the slow southward drift) do not provide a precise estimate of the magnetic energy release but certainly serve as good indicators of the energy release site. At the same time, above the neutral line in the area of the positive magnetic polarity the magnetic energy changed significantly less and even increased in a small area (black contours in Fig. 3). This demonstrates the complexity of the electrodynamic flare processes near the neutral line, which are not understood. One can imagine a variety of reasons for the apparent asymmetry in the variation of the two polarities. For instance, this might be consistent with the lineof-sight view of shrinkage of magnetic lines connecting the opposite polarities, which may be associated with a reconnection process (Priest & Forbes 2000). The net effect of the observed variations is consistent with magnetic energy dissipation at the beginning of the flare. The integrated magnetic energy density decrease was about 7 # 10 21 ergs cm21. Assuming that the energy release area is determined by the energy density decrease of at least 1.5 # 10 4 ergs cm23, we estimate the characteristic energy release area ∼4.6 # 10 17 cm2. The mean magnetic field strength in this area is about 800 G; it decreased by ∼100 G. The peak variation was about 300 G. If the characteristic height of the magnetic field variation was ∼108 cm, then the energy dissipated in region 1 during the first 10 minutes of the flare was ∼7 # 10 29 ergs. This is comparable to the energy of a typical flare. 4. PHOTOSPHERIC TRANSIENTS
Sudden impulsive variations of the magnetic field were observed during the flare mostly in unipolar areas of various magnetic field strength. These variations resulted in sharp temporary decreases of the magnetic field strength in most cases. After these events lasting 1–10 minutes, the magnetic field strength is returned to the initial values.
L107
Fig. 4.—The line-of-sight magnetic field, B (solid curves), velocity, V (dashed curves), and the relative continuum intensity, I0/I0, max (dotted curves), as functions of time for two magnetic transient events. The gray curves show the corresponding variation for a model of possible rapid line-profile variations.
The first and relatively weak transient event occurred at the very beginning of the explosive phase at 10:13 UT in region 5, approximately 80 Mm from the energy release site in region 1. The highest negative velocity variation (.20.7 km s21) and intensity increase (.20%) occurred at a boundary of the leading sunspot. The variations of V, I0, and B did not completely coincide but were close to each other. In Figure 4a we show the variations of V, I0 /Imax, and B in a small area of 3 # 3 CCD pixels (2 Mm2) as a function of time. It shows a sharp negative velocity signal (Doppler blueshift) lasting only 1 minute followed by a smaller positive signal. This is consistent with an upflow followed by a gradual downflow. The corresponding magnetic field and intensity variations peaked later and lasted longer, ≈4 minutes. The strongest transient associated with the flare was observed in region 6 at ≈10:20 and correlated with the main pulse of the hard X-ray emission. The transient was located at the boundaries of a large sunspot umbra. The MDI magnetic field signal detected a variation of about 1500 G, so in one place the magnetic polarity was even briefly reversed at the peak of the transient. The Doppler velocity variation was negative at the beginning of the transient (DV . 20.6 km s21) and then suddenly changed to positive (DV . 0.4 km s21) when the magnetic variation reached its maximum. In order to determine the effect of possible rapid variations of the line profile, we have carried out computer simulations of the MDI observables. We assumed that the line parameters (continuum intensity, line width, and depth) had the same time dependence given by I0 (t), that the line becomes flatter, and that the line depth did not change by more than 60% (which is twice the variation that we observed during the X2.6 flare of 1996 July 9), and calculated the observing frame sequence given in § 1 using the known properties of the MDI instrument response. Then, we calculated the MDI observables I0 , B, and V and tried to match all three observables with the observations. The time dependence of I0 is basically determined by the input parameters. However, the amplitude of I0 strongly depends also on the line width and depth, thus providing constraints on line
L108
SOLAR FLARE OF 2000 JULY 14
Vol. 550
Fig. 6.—Variations of photospheric intensity measured by MDI in regions A, IA0 (dashed curve), and B, IB0 (dash-dotted curve), shown in Fig. 5, and the total flare intensity, IA0 1 IB0 (solid curve). The lighter gray profile is the soft X-ray flux (1.5–12 keV) from GOES-10; the darker profile (with a gap in the data between 10:13 and 10:20) shows the hard X-ray flux (14–93 keV) from Yohkoh/HXT.
Fig. 5.—Locations of the transient photospheric variations (bright structures) are shown on the maps of the rms intensity variations for two 9 minute intervals: (a) 10:13:30–10:21:30 UT and (b) 10:22:30–10:30:30 UT, superposed over the corresponding MDI intensity images scaled down by a factor of 100.
width and depth variations and, therefore, on two other observables. The results (gray curves in Fig. 4) show that while the rapid changes can cause systematic errors in the MDI measurements they do not produce sufficiently large signals to explain the observed magnetic and velocity variations. In the course of the flare, the transients appeared in various points of a two-ribbon structure (Fig. 5) that seem to be mostly located along the footpoints of a long arcade observed in EUV lines by the Transition Region and Coronal Explorer. During the explosive stage of the flare most of the transients occurred west of the initial energy release site (area A in Fig. 5a), and at a later stage the transients appeared along a new two-ribbon structure east of this site (area B in Fig. 5b). In Figure 6 we compare the intensity variations in these areas with the total soft and hard X-ray fluxes. Evidently, there is a good correspondence between the two sets of photospheric transients and the two main pulses of the hard X-ray flux.
leagues at the Crimean Observatory and later observations by Moore et al. (1984). A surprising result of our investigation is that the irreversible magnetic field variation occurred very quickly during 10 minutes in a large area just at the beginning of the flare, releasing a significant amount of energy, comparable to the total energy of solar flares. We have also observed numerous compact transient events ˚ of continuum brightening in the vicinity of the Ni i 6768 A line, accompanied by variations of magnetic field and velocity. Our simulations of the MDI measurements indicate that it is unlikely that magnetic field and velocity signals are caused by rapid changes in the line emission and profile. The transients were located along a two-ribbon structure formed by the magnetic configuration of the active region. They formed two groups separated both spatially and temporally, which correlated well with the two main impulses of the total hard X-ray flux. It is intriguing that the velocity signal of the transients is similar to chromospheric “mustaches” (or Ellerman “bombs”)—extended spectral line features—studied in detail by Severny (1968) and his colleagues, who also found associated polarization signals that may correspond to impulsive variations of magnetic field. Mustaches can be explained by streams of energetic particles penetrating into the solar chromosphere (Ding, He´noux, & Fang 1998). Therefore, we suggest that our transients observed much closer to the photosphere might be caused by beams of particles accelerated to much higher energies and thus penetrated into much deeper layers than the energetic particles in mustaches. Details of this interpretation have not been developed yet.
5. DISCUSSION
Our results are in agreement with the initial magnetic field measurements in solar flares by Severny (1964) and his col-
We thank Terry Forbes, Jack Harvey, Serge Koutchmy, and Jan Stenflo for useful comments.
REFERENCES Cameron, R., & Sammis, I. 1999, ApJ, 525, L61 Ding, M. D., He´noux, J.-C., & Fang, C. 1998, A&A, 332, 761 Harvey, J. 1985, Abh. Akad. Wiss. Go¨ttingen, 38, 25 Kosovichev, A. G., & Zharkova, V. V. 1999, Sol. Phys., 190, 459 Moore, R. L., Hurford, G. J., Jones, H. P., & Kane, S. R. 1984, ApJ, 276, 379 Patterson, A. 1984, ApJ, 280, 884 Patterson, A., & Zirin, H. 1981, ApJ, 243, L99 Priest, E. R., & Forbes, T. 2000, Magnetic Reconnection (Cambridge: Cambridge Univ. Press)
Sakurai, T., & Hiei, E. 1996, Adv. Space Res., 17, 91 Scherrer, P. H., et al. 1995, Sol. Phys., 162, 129 Severny, A. B. 1964, ARA&A, 2, 363 ———. 1968, in Mass Motions in Solar Flares and Related Phenomena, Ninth ¨ hman (New York: Wiley), 109 Nobel Symp., ed. Y. O Wang, H., Ewell, M. W., Zirin, H., & Ai, G. 1994, ApJ, 424, 436 Zirin, H., & Tanaka, K. 1981, ApJ, 259, 791 Zvereva, A. M., & Severny, A. B. 1970, Izv. Krymskoi Astrofiz. Obs., 41–42, 97
