HIGH-RESOLUTION OBSERVATIONS OF MULTIWAVELENGTH EMISSIONS DURING TWO X-CLASS WHITE-LIGHT FLARES

Transcription

1 The Astrophysical Journal, 641: , 2006 April 20 # The American Astronomical Society. All rights reserved. Printed in U.S.A. HIGH-RESOLUTION OBSERVATIONS OF MULTIWAVELENGTH EMISSIONS DURING TWO X-CLASS WHITE-LIGHT FLARES Yan Xu, 1,2 Wenda Cao, 1,3 Chang Liu, 1 Guo Yang, 1 Ju Jing, 1 Carsten Denker, 1,3 A. Gordon Emslie, 2 and Haimin Wang 1,3 Received 2005 March 30; accepted 2005 December 16 ABSTRACT We observed two X-class white-light flares (WLFs) on 2003 October 29 (20:40 UT) and November 2 (17:16 UT) using the Dunn Solar Telescope (DST) and its High-Order Adaptive Optics (HOAO) system in several wavelengths. The spatial resolution was close to the diffraction limit of DST s 76 cm aperture, and the cadence was as high as 2 s. This is the first time that WLFs have been observed in the near-infrared (NIR) wavelength region. We present a detailed study in this paper comparing photospheric continuum observations during the two events with corresponding line-of-sight magnetograms from the Solar and Heliospheric Observatory (SOHO) Michelson Doppler Imager (MDI) and hard X-ray (HXR) data from the Ramaty High-Energy Solar Spectroscopic Imager (RHESSI ). We also discuss several models that provide possible mechanisms to explain these continuum enhancements, especially in the NIR. Subject headinggs: Sun: activity Sun: flares Sun: photosphere 1. INTRODUCTION Solar flares with emissions in the visible continuum or integrated light are identified as white-light flares ( WLFs; Neidig 1989); they may be associated with the most energetic particles accelerated during flares (Neidig et al. 1993b). The flare energy is generally supposed to be released in, for example, current sheets in a region of magnetic reconnection; the accelerated particles subsequently propagate downward along field lines and precipitate into the lower atmosphere. It is not yet clear, however, where and how the released energy is converted into whitelight radiation. A substantial amount of work has been undertaken trying to understand the following two fundamental (and intertwined) problems: (1) Where does the visible and near-infrared (NIR) emission originate? (2) What is the energy source? The objective of this study is to compare observations with the theoretical models discussed above and to provide motivation for future research into the origin of WLFs. In x 2 we briefly review the various mechanisms that have been proposed to account for enhanced continuum emission in WLFs. In x 3we present the characteristics of recent observations of NIR emission, concurrent with RHESSI hard X-ray (HXR) and SOHO MDI observations; the results and implications of these observations are presented in x 4. In x 5 we discuss the results obtained. 2. REVIEW OF WHITE-LIGHT FLARE EMISSION MECHANISMS 2.1. Direct Heating by Nonthermal Electrons and Protons The initial theory describing WLFs was based on observations that white-light flare emission was always associated with HXR emission. Therefore, the same nonthermal electrons responsible for the HXR bremsstrahlung (or possibly protons accelerated concomitantly with these electrons) were assumed to be the energy carrier (Hudson 1972; Rust & Hegwer 1975; Lee et al. 1996). Under this hypothesis, the white-light emission is caused 1 Center for Solar-Terrestrial Research, New Jersey Institute of Technology, 323 Martin Luther King Boulevard, Newark, NJ 07102; yx2@njit.edu. 2 Department of Physics, Oklahoma State University, Stillwater, OK Big Bear Solar Observatory, North Shore Lane, Big Bear City, CA by nonthermal electrons and protons (Najita & Orrall 1970) precipitating down to the chromosphere and photosphere and depositing most of their energy in the lower atmosphere, near the 5000 ¼ 1 level. These high-energy particles are thermalized by collisions and can create locally enhanced ionization, which enhances the bound-free and free-free continuum emission (Najita & Orrall 1970; Hudson 1972; Ding 2003; Ding et al. 2003). However, since the photospheric column density exceeds some cm 2, only electrons with an extremely high initial energy can contribute to the heating, rendering this mechanism decidedly inefficient. The relation between the column density ( per square centimeter) of a certain layer and the initial energy of an electron required to penetrate to such a layer has been given by Emslie (1978); from these results we find that only electrons with injected energies around a few MeV, or protons with an injected energy higher than 30 MeV, can penetrate to the 5000 ¼ 1 level. For electron and/or proton spectra with typical steepness, the total energy content in such high-energy particles may not be sufficient to excite the 5000 ¼ 1 layers to the required degree Chromospheric Back-warming Many observations have shown that a flare-associated temperature enhancement can occur in the lower atmosphere around the minimum-temperature region (MTR) and photosphere (Machado et al. 1978; Cook 1979; Mauas et al. 1990; Metcalf et al. 1990a; Liu et al. 2001). However, given the energetic difficulties associated with the hypothesis that such layers are directly heated by high-energy particles ( Metcalf et al. 1990b; Ding et al. 2003), a model invoking chromospheric heating through radiative backwarming was proposed. In this model, electrons with a moderate energy of 20 kev stop in the transition region and the upper chromosphere, where they produce continuum extreme-ultraviolet (EUV) emission through ionization and recombination. Some of this emission is reradiated downward to lower layers, where it induces secondary ionization ( photon ionization) at a level where there is significant opacity to such EUV radiation. As a result of this secondary heating, continuum emission is generated from deeper layers (Hudson 1972; Aboudarham & Hénoux 1986; Metcalf et al. 1990b).

2 TWO X-CLASS WHITE-LIGHT FLARES H Emission The upper photosphere and below is heated by H continuum emission. According to Aboudarham & Hénoux (1987), the nonthermal hydrogen ionization and excitation reach their maximum around the MTR. Hence, if the electron number density is strongly increased as a result of the flare, this leads to an increase of the H population and in turn to an increase of the absorption of continuum radiation of both photospheric and chromospheric origins (Machado et al. 1989; Mauas et al. 1990; Metcalf et al. 1990b, 2003; Ding et al. 1994, 2003). The continuum emission is then mainly due to H emission in the upper photosphere and is unrelated to a Balmer or Paschen jump Summary The brief overview above shows that WLF emission could be produced not only by the direct effect of precipitating electron beams but also by secondary processes, such as back-warming and enhanced absorption of photospheric or chromospheric radiation. Multiple sources of continuum emissions may be intermixed and contribute to the observed emission during a WLF. Aboudarham & Hénoux (1986, 1987) combined the back-warming model and H emission model. They showed that all layers from the chromosphere to the photosphere contribute to the whitelight emission process. In particular, they concluded that even the MTR heating could be fully attributed to nonthermal electrons, with the photosphere subsequently heated via radiative back-warming. Neidig et al. (1993a) compared the observed center-to-limb variation of 86 WLFs. They concluded that the observed center-to-limb variation of white-light emissions is most likely compatible with a source in the middle photosphere or with a source in the middle photosphere combined with one at higher altitude. Recently, Ding (2003) proposed a model of nonthermal electron beam heating plus H emission to explain the continuum enhancement near Ca ii k8542, which is formed mostly in the photosphere. For a review of other mechanisms, such as heating by dissipation of electric currents, nonthermal proton beams, soft X-ray radiation, UV radiation, and dissipation of Alfvén waves, see Metcalf et al. (1990b). Most interestingly in the context of the current work, Neidig et al. (1993a) proposed that flare kernels consist of a bright inner core, corresponding to the direct heating in the chromosphere, and a weaker outer region, corresponding to the back-warming emission. In the present work we explore this idea further through observations of NIR white-light emission with unprecedented spatial and temporal resolution. 3. OBSERVATIONS AND DATA REDUCTION Xu et al. (2004) reported the first observations of flare continuum emission in the NIR at 1.56 m. In the undisturbed atmosphere, 1.56 m is considered to be the opacity minimum; i.e., it corresponds to the deepest layer in the solar atmosphere, some 50 km lower than the 5000 ¼ 1 level (Wang et al. 1998). Therefore, the NIR observations provide additional information for theoretical modeling. In this paper, we present a comprehensive study of two X-class WLFs (that in the previous paper [Xu et al. 2004] plus a new one) observed in the visible continuum at 520 nm, the NIR at 1.56 m, and the G band around nm. The data sets have both high cadence and high spatial resolution. The seeing condition in the NIR is very stable; an autocorrelation analysis for granulation areas of all the image sequences showed that the rms contrast variation in the NIR is about 6.5% and in the other two wavelengths are over 10%. This NIR image quality was much more stable than those in the visible continuum and G band, rendering NIR observations an excellent resource for studying flare morphology and dynamics. NOAA AR produced four X-class flares in just 8 days as it moved across the solar disk. We obtained comprehensive observations for the X10 flare on 2003 October 29 (first event) and the X8 flare on 2003 November 2 (second event). Our results are based on observations in the visible continuum at 520 nm with a 52 nm bandpass, the NIR continuum at 1560 nm with a 5 nm bandpass, and the G band at nm with a 0.5 nm bandpass. There are a few Fraunhofer absorption lines within the NIR bandpass, but these contribute less than 3% of the overall radiation. Therefore, the NIR continuum contamination is negligible. The observations were carried out with the DST s HOAO system (Rimmele 2000; Rimmele et al. 2003) at the National Solar Observatory (NSO) at Sacramento Peak. Two newly developed photometric imaging systems for the NIR and visible wavelength regions were provided by the Big Bear Solar Observatory (BBSO) and temporarily moved to NSO at Sacramento Peak. Detailed information about the instruments and setups can be found in Denker et al. (2005) and Xu et al. (2004). The field of view (FOV) is ; for the NIR, ; for the green continuum, and ; for the G band. With the 1024 ; 1024 pixel detector, the image scales were 0B089, 0B079, and 0B074 pixel 1, respectively. In order to cover more of the flaring area, we enlarged the FOVof the NIR observations on 2003 November 2 to 122B2 ; 122B2 resulting in an image scale of 0B119 pixel 1. The FOV was not changed for the other two wavelengths. The cadence was 1 minute for the 520 nm continuum and for the NIR continuum of the first event. All other data were observed with a cadence of 2 s. All images are dark and flat-field corrected and registered with respect to MDI white-light full-disk images. Figure 1 shows the sample images of the NIR, visible continuum, and G band during the flares. RHESSI HXR data and MDI magnetograms during the flare periods were also used for comparison. 4. RESULTS 4.1. Flare Morphology The 2003 October 29 (X10) and 2003 November 2 (X8) flares were typical two-ribbon flares. Preliminary results of the first event in the NIR were presented by Xu et al. (2004). The contrast enhancement reached its maximum at 20:42 UT in the October 29 event, which was located around S17,W10 in the highly active region NOAA AR The November 2 event occurred near S17,W63 in the same active region and reached its maximum at 17:17 UT. We define the flare contrast enhancement as C ¼ I f I b I 0 ; ð1þ where I f is the intensity of the flaring area, I b is the intensity in the same area right before the flare, and I 0 is the undisturbed quiet-sun background. A summary of the basic NIR and visible continuum properties of the two events is given in Table 1. The maximum NIR continuum contrast enhancements were about 25% and 66% for the two events, respectively. The visible contrasts were higher than those in the NIR, i.e., 45% and 76%, respectively. The G-band results are also included in Table 1. Although the CCD camera was saturated, the lower limits of G-band emission during the flares are 75% and 230%.

3 1212 XU ET AL. Vol. 641 TABLE 1 Basic Measurements of Two White-Light Flares Measurement October 29 November 2 Disk location (deg)... S17 W10 S17 W63 GOES a X-ray class... X10 X8 Start time ( UT)... 20:39 17:15 Peak time ( UT)... 20:42 17:17 NIR contrast (%) Visible contrast (%) G-band contrast (%) A NIR total (10 19 cm 2 ) Vavg NIR (km s 1 ) E avg (V cm 1 ) a Geostationary Operational Environmental Satellite. Fig. 1. Short-exposure images of NOAA AR obtained with frame selection and the HOAO system at 20:42 UT on 2003 October 29 (left) andat 17:17 UT on 2003 November 2 (right). Top, NIR emission; middle, visible emission; bottom, G-band emission. Figure 1 shows the selected images of the NIR, visible continuum, and G band during the flares. Similar G-band observations have been carried out by Hudson et al. (1992) with a bandpass of 3 nm. Although the origin of the G-band radiation is still not clear (Rutten et al. 2001; Steiner et al. 2001), it does show a strong ribbon-like emission similar to that of the continuum. For both events, the NIR and visible flare kernels are very similar, while the G-band images also exhibit kernels in the same locations but across much larger flaring areas. The NIR images are much less blurred than the visible images. In order to show the flare ribbons more clearly, contrast-enhanced difference images (Xu et al. 2004) in all three wavelengths are shown in Figure 2. The flare kernels have two distinct features, a bright inner core and a surrounding faint halo. Thresholds of 10% and 20% of the contrast enhancement were used to distinguish between core and halo emission in the two events. Neidig et al. (1993a) observed similar structures with distinct temporal behaviors in the 1989 March 7 WLF. In the November 2 event, the size of the core structures relative to the halo are larger than in the October 29 event the ratios between core area and halo area (core-halo ratio) is about 4% (October 29) and around 25% (November 2) White-Light NIR versus Hard X-Ray Light Curves Because the NIR data are less subject to seeing variations and have a higher temporal resolution, we chose to compare the NIR light curves with the HXR light curves in different energy channels (Fig. 3). Two kinds of NIR light curves are plotted. The red crosses show the average light curve (ALC), i.e., the average flux across the whole FOV, with a contrast threshold of 1% imposed. To highlight the time history of the brightest inner core in the image, the black plus signs also show the maximum light curve (MLC), defined as the average contrast of the brightest (90th percentile) pixels; for these points, a contrast threshold of 7% was used to exclude bright patches unrelated to the flare. The observational cadence was 1 minute for the first event and 2 s for the second event. It should be noted that the seeing conditions were less stable during the November 2 event, especially during the period from 17:15:21 to 17:16:51 UT, as plotted in Figure 3 (bottom). The effect of mediocre seeing leads to inferred intensities that are much lower than their real value. However, this should not affect the distinct peak between 17:16 and 17:17 UT, when the seeing was improving. For both events, the ALC and MLC curves are broadly similar to the HXR light curves at all energies from 50 kev upward; the ALC light curve has a time dependence similar to the hard X-rays, with the MLC light curve peaking somewhat earlier (about 2 minutes earlier than the ALC for the October 29 event and about 20 s earlier for the November 2 event). (Note that the cadence was 1 minute for the first event and 2 s for the second event, so both of these time differences are significant.) AccordingtoHénoux et al. (1990), Liu et al. (2001), and Neidig et al. (1993a), the finite time required for heating of the photosphere and the MTR via chromospheric recombination leads to a time lag of somewhat less than 1 minute between the hard X-rays and the visible continuum, in particular the continuum near Ca ii k8542. For the October 29 event, the hard X-rays peak around 20:43 UT and the NIR ALC reaches its peak about 30 s later. However, in view of the 1 minute observational cadence for this event, this delay is only marginally significant. On the other hand, the cadence for the November 2 event is of the order of 2 s, so the observed delay of about 20 s between the hard X-rays and the NIR ALC is much more significant. For the October 29 event, the NIR MLC light curve, corresponding to the inner core of the flare, shows a peak at 20:42 UT, a time when the HXR light curves also show a relatively small, but very impulsive, spike. In addition, this HXR spike is more obvious in the kev range. This correlation between high-energy hard X-rays and inner core white-light emission supports the hypothesis of direct nonthermal heating of the inner core by very energetic electrons Hard X-Ray Spectra We used the SPEX software package to fit the X-ray photon spectra observed by RHESSI near the flare peak time (Fig. 4).

4 No. 2, 2006 TWO X-CLASS WHITE-LIGHT FLARES 1213 Fig. 2. Background-subtracted difference images in the NIR (top), visible (middle), and G band (bottom) for the 2003 October 29 event (left column; scaled to 0.35) and 2003 November 2 event ( right column; scaled to 1.0). These spectral fitting results are preliminary, since the pulse pileup (Datlowe 1975; Hurford et al. 2002) correction, caused by multiple low-energy photons arriving within the resolving time and counted as one single higher energy photon, is likely not very accurate for these extremely large events; however, such considerations should not significantly affect the conclusions below. The spectral fitting was conducted in the energy range of and kev for the two flares, respectively. For both events, data of one orbit after the flare (the duration of a RHESSI orbit is about 100 minutes) are used for background subtraction. The photon spectrum was fitted with a thermal component plus a nonthermal broken power law. For the October 29 event, the photon spectrum power-law indices for the low-energy

5 1214 XU ET AL. Vol. 641 Fig. 3. Light curves of the contrasts for the NIR and HXR fluxes for the October 29 (top) and November 2 (bottom) events. The red crosses show the contrast for the core (MLC) NIR emission and the black plus signs the contrasts for the average (ALC) NIR emission, which includes the fainter halo components. The lowest plot in the lower panel shows the seeing variation in the November 2 event. For the November 2 event, the uncertainties for the NIR ALC and MLC are 0.02 and 0.031, respectively. and high-energy portions of the photon spectrum are 3.6 and 4.2, with a spectral break at 127 kev. For the November 2 event, the corresponding values are 2.8 and 3.6, with a spectral break at 109 kev. The relatively harder HXR spectrum in the November 2 event implies more accelerated high-energy electrons at 50 kev and above, making the likelihood of direct excitation of the photosphere and MTR region more plausible in that event Cooling Time According to Najita & Orrall (1970), the lower solar atmosphere relaxes primarily by radiation after being heated by a flare. A certain relaxation time or cooling time is associated with each height in the atmosphere. With our high-cadence and high-resolution data, it is possible to measure the cooling time. We plot the average intensity of small areas with 3 ; 5 pixels covering the brightest kernel as a function of time for the flares in Figure 5. Plots A1 A4 are NIR light curves in the first event of four different areas, which have a maximum contrast enhancement. C2 C4 are for the G band in the same event, and the last four plots (D1 D4) are for the NIR light curves in the second event. All the light curves show two kinds of relaxation profiles, a rapid plus a gradual cooling; similar results were obtained by Hudson et al. (1992). Since the size of the selected flare areas are much smaller than the width of the flare ribbons, this cooling profile is unaffected by the halo structures surrounding the flare cores, and the cooling time inferred for the fast component (about 20 s for the G band) represents the relaxation profile of the core structure. In the October 29 event, the measured NIR cooling time is about 2 minutes; since this is close to the image cadence, the real value could be substantially less than this. Due to the much higher observational cadence for the November 2 event, the inferred cooling of about 30 s is much more meaningful. Fig. 4. RHESSI spectra at the flare peak of the October 29 event (top) and the November 2 event (bottom), each fitted with a combination of an isothermal component (dotted line) and a broken power-law nonthermal component (dashed line). For the October 29 event, the power-law indices are 3.6 and 4.2, with a break energy of 127 kev. For the November 2 event, the spectrum is significantly harder, with power-law indices of 2.8 and 3.6 and a break energy of 109 kev. The triangles represent the background. In the photosphere, the predicted cooling time is on the order of a few seconds (Mein 1966), while it increases to the order of 100 s in the chromosphere. Thus, with the hypothesis that the core emission is coming from a rather deep layer, i.e., the photosphere, the observed 30 s cooling time is consistent with the predicted value. 5. DISCUSSION In this paper, multiwavelength observations of two X-class WLFs are presented. This is the first time that NIR continuum contrasts have been measured, providing new and important clues for understanding flare energetics. According to standard solar atmospheric models, the opacity minimum is about 1.6 m. Therefore, the NIR continuum should originate from the deepest layers in the quiet atmosphere, about 50 km deeper than the visible continuum (Vernazza et al. 1976; Wang et al. 1998). (Although the opacity during the flare must change due to the increasing electron population, causing the NIR emission to originate somewhat higher in the photosphere or chromosphere [Ohki & Hudson 1975], it is beyond the scope of this paper to derive the opacity change during these two flares.) The flare energy budget cannot be balanced if only direct heating of the emitting region (Xu et al. 2004) is involved; some other heating mechanisms, such as chromospheric back-warming and enhanced absorption of H emission, are required. Chromospheric back-warming and H emission models have been used to explain relatively weak flares (Aboudarham & Hénoux 1986,

6 No. 2, 2006 TWO X-CLASS WHITE-LIGHT FLARES 1215 Fig. 5. NIR and G-band light curves of selected regions during flare peaks. A1 A4 show the NIR cooling patterns during the October 29 event, C1 C4 are the corresponding G-band curves during the same event, and D1 D4 are the corresponding NIR light curves for the November 2 event. 1987, 1989; Ding 2003; Ding et al. 2003; Machado et al. 1989; Metcalf et al. 1990b). The findings and results of our observations are briefly summarized as follows: 1. Significant intensity enhancements appeared in the visible and NIR continua and G band during the impulsive phase of the flares. The maximum intensity enhancements were 45% of the white light and 25% of the NIR continuum during the first event and 76% of the white light and 66% of the NIR continuum for the second event. 2. The core NIR emission and the impulsive HXR emission up to 500 kev are well correlated, with the average NIR emission peaking some tens of seconds after the core emission. This supports the hypothesis that the core emission is predominantly associated with direct particle heating, with the halo emission caused by other, indirect heating mechanisms. 3. The cooling light curve for the flare photospheric emission can be characterized by two steps: a rapid temperature drop, related to the cooling of the bright cores, and a relatively slower decay, related to the halo structures. The timescale is on the order of less than 30 s and a few minutes for these two steps, respectively. These timescales are consistent with the hypothesis of radiative cooling of the two energized regions. Since ionization and recombination occur very quickly, we cannot distinguish the heating effects of different mechanisms by studying the morphology of flares alone. However, according to Hénoux et al. (1990), Liu et al. (2001), and Neidig et al. (1993a), indirect heating should result in a delay in the white-light

7 1216 XU ET AL. emission compared to the HXR flux, caused by the timescales for chromospheric recombination to occur. Our observations show that while the light curve for the core NIR emission peaks at approximately the same time as the hard X-rays, the light curve of the total NIR emission reaches its peak somewhat later. This result is consistent with the presence of two components of heating, direct heating by energetic particles in the flare cores and indirect heating in the halos; this suggestion was also made by Neidig et al. (1993a). The hypothesis of direct heating for the core areas is also supported by the observational evidence that the core areas are much smaller than the halo areas and by the fact that there is a stronger correlation between HXR emission and NIR core emission in the November 2 event, which has a harder HXR spectrum (and hence more high-energy electrons) than in the October 29 event, which has a somewhat softer HXR spectrum. The flares studied were typical two-ribbon flares; the NIR observations showed ribbon separation speeds of about 28 km s 1 in the October 29 event and 24 km s 1 in the November 2 event. We can calculate the electric field in the reconnecting current sheet assuming a standard reconnection model, using the equation (Forbes & Lin 2000) E c ¼ V k B n ; ð2þ where B n is the normal component of the magnetic field (measured by the SOHO MDI) and V k is the speed of the ribbon separation, derived from NIR observations. During the second event, the magnetogram had some temporary saturation effects as the ribbons swept through sunspot umbra. Therefore, the actual magnetic field should be stronger than its measured value, and the derived electric fields listed in Table 1 are therefore lower limits. In Table 1, A NIR total is the average total area with flare emissions, Vavg NIR is the average speed of the ribbon separation after correcting for geometric foreshortening, and E avg is the inferred average electric field in the reconnection region; we obtained values of 23 and 22 V cm 1 in each event. These electric field values are the largest that have been reported using this method so far and, for plausible values of the density and temperature in the reconnecting region, correspond to several orders of magnitude above the Dreicer field E D 10 8 n(cm 3 )/T(K). Such electric fields should therefore accelerate all the electrons within the current sheet in a very short period (see, e.g., Litvinenko 1996), adding further support for the hypothesis that the core region of NIR emission is energized by direct bombardment of the photosphere by nonthermal particles. In summary, extraordinarily stable, high-resolution, highcadence images in the near-infrared have allowed us to identify two main regions of white-light emission. The core emission is characterized by synchronism with the HXR light curves and a rapid cooling time and is more prominent in events with harder HXR spectra. These features strongly suggest direct bombardment by high-energy accelerated particles as the energy source. The other component of emission is a halo component, characterized by a delay relative to the HXR peak and a longer cooling time. This component is interpreted as being energized by secondary processes, such as radiative back-warming and enhanced H absorption. Obtaining the excellent data would not have been possible without the help of the dedicated DST observing staff. The National Solar Observatory is a division of the National Optical Astronomy Observatories, which is operated by the Association of Universities for Research in Astronomy, Inc., under cooperative agreement with the National Science Foundation. This work is supported by the NSF under grants ATM and ATM , by NASA under grants NAG and NNGO-4GG21G, and by the Air Force under grant F and by NSFC Aboudarham, J., & Hénoux, J. C. 1986, A&A, 156, , A&A, 174, , Sol. Phys., 121, 19 Cook, J. W. 1979, ApJ, 234, 378 Datlowe, D. W. 1975, Space Sci. Instrum., 1, 389 Denker, C., Mascarinas, D., Xu, Y., Cao, W., Yang, G., Wang, H., Goode, P. R., & Rimmele, T. 2005, Sol. Phys., 227, 217 Ding, M. D. 2003, J. Korean Astron. Soc., 36, S49 Ding, M. D., Fang, C., Gan, W. D., & Okamoto, T. 1994, ApJ, 429, 890 Ding, M. D., Liu, Y., Yeh, C. T., & Li, P. J. 2003, A&A, 403, 1151 Emslie, A. G. 1978, ApJ, 224, 241 Forbes, T. G., & Lin, J. 2000, J. Atmos. Sol.-Terr. Phys., 62, 1499 Hénoux, J.-C., Aboudarham, J., Brown, J. C., van den Oord, G. H. J., & van Driel-Gesztelyi, L. 1990, A&A, 233, 577 Hudson, H. S. 1972, Sol. Phys., 24, 414 Hudson, H. S., Acton, L. W., Hirayama, T., & Uchida, Y. 1992, PASJ, 44, L77 Hurford, G. J., et al. 2002, Sol. Phys., 210, 61 Lee, S. W., Lee, J., Yun, H. S., Fang, C., & Hu, J. 1996, ApJ, 470, L65 Litvinenko, Y. 1996, ApJ, 462, 997 Liu, Y., Ding, M. D., & Fang, C. 2001, ApJ, 563, L169 Machado, M. E., Emslie, A. G., & Avrett, E. H. 1989, Sol. Phys., 124, 303 Machado, M. E., Emslie, A. G., & Brown, J. C. 1978, Sol. Phys., 58, 363 Mauas, P. J. D., Machado, M. E., & Avrett, E. H. 1990, ApJ, 360, 715 Mein, P. 1966, Ann. d Astrophys., 29, 153 REFERENCES Metcalf, T. R., Alexander, D., Hundon, H. S., & Longcope, D. 2003, ApJ, 595, 483 Metcalf, T. R., Canfield, R. C., Avrett, E. H., & Metcalf, F. T. 1990a, ApJ, 350, 463 Metcalf, T. R., Canfield, R. C., & Saba, J. L. R. 1990b, ApJ, 365, 391 Najita, K., & Orrall, F. Q. 1970, Sol. Phys., 15, 176 Neidig, D. F. 1989, Sol. Phys., 121, 261 Neidig, D. F., Kiplinger, A. L., Cohl, H. S., & Wiborg, P. H. 1993a, ApJ, 406, 306 Neidig, D. F., Wiborg, P. H., & Gilliam, L. B. 1993b, Sol. Phys., 144, 169 Ohki, K., & Hudson, H. S. 1975, Sol. Phys., 43, 405 Rimmele, T. 2000, Proc. SPIE, 4007, 218 Rimmele, T. R., et al. 2003, Proc. SPIE, 4839, 635 Rust, D. M., & Hegwer, F. 1975, Sol. Phys., 40, 141 Rutten, R. J., Kiselman, D., Rouppe van der Voort, L., & Plez, B. 2001, in ASP Conf. Ser. 236, Advanced Solar Polarimetry Theory, Observation, and Instrumentation, ed. M. Sigwarth (San Fracisco: ASP), 445 Steiner, O., Hauschildt, P. H., & Bruls, J. 2001, A&A, 372, L13 Vernazza, J. E., Avrett, E. H., & Loeser, R. 1976, ApJS, 30, 1 Wang, H., Spirock, T. J., Goode, P. R., Lee, C., Zirin, H., & Kosonocky, W. 1998, ApJ, 495, 957 Xu, Y., Cao, W., Liu, C., Yang, G., Qiu, J., Jing, J., Denker, C., & Wang, H. 2004, ApJ, 607, L131