astro-ph0412171/ms.tex
1: \documentclass[12pt,preprint]{aastex}
2: 
3: \usepackage{verbatim}
4: 
5: \newcommand{\Ha}{H$\alpha$}
6: \newcommand{\ten}{$10^{10}$}
7: \newcommand{\erg}{ergs cm$^{-2}$ s$^{-1}$}
8: \newcommand{\pho}{photons s$^{-1}$ cm$^{-2}$ keV$^{-1}$}
9: \newcommand{\ca}{Ca\,{\sc ii} 8542 \AA}
10: \newcommand{\RHESSI}{$\it RHESSI$}
11: 
12: \renewcommand{\baselinestretch}{1.0}
13: 
14: \shorttitle{CONTINUUM ENHANCEMENT \& HXR EMISSION}
15: \shortauthors{CHEN \& DING}
16: 
17: \begin{document}
18: 
19: \title{
20: ON THE RELATIONSHIP BETWEEN THE CONTINUUM ENHANCEMENT AND HARD X-RAY EMISSION
21: IN A WHITE-LIGHT FLARE}
22: \author{Q. R. Chen and M. D. Ding}
23: \affil {Department of Astronomy, Nanjing University, Nanjing 210093, China}
24: 
25: \begin{abstract}
26: We investigate the relationship between 
27: the continuum enhancement and the hard X-ray (HXR) emission 
28: of a white-light flare on 2002 September 29.
29: By reconstructing the \RHESSI\ HXR images 
30: in the impulsive phase, 
31: we find two bright conjugate footpoints (FPs) 
32: on the two sides of the magnetic neutral line.
33: Using the thick-target model and assuming a low-energy cutoff of 20 keV,
34: the energy fluxes of non-thermal electron beams bombarding FPs A and B 
35: are estimated to be 1.0 $\times$ \ten\ and 0.8 $\times$ \ten\ \erg, 
36: respectively.
37: However, the continuum enhancement at the two FPs 
38: is not simply proportional to the electron beam flux.
39: The continuum emission at FP B is relatively strong 
40: with a maximum enhancement of $\sim 8$\% 
41: and correlates temporally well with the HXR profile; 
42: however, that at FP A is less significant with an enhancement of 
43: only $\sim 4-5$\%,
44: regardless of the relatively strong beam flux.
45: By carefully inspecting the \Ha\ line profiles, 
46: we ascribe such a contrast to different atmospheric conditions at the two FPs.
47: The \Ha\ line profile at FP B exhibits a relatively weak amplitude
48: with a pronounced central reversal, while the profile at FP A is
49: fairly strong without a visible central reversal.
50: This indicates that in the early impulsive phase of the flare, 
51: the local atmosphere at FP A has been appreciably heated 
52: and the coronal pressure is high enough to prevent most high-energy electrons
53: from penetrating into the deeper atmosphere;
54: while at FP B, the atmosphere has not been fully heated,
55: the electron beam can effectively heat the chromosphere and
56: produce the observed continuum enhancement via 
57: the radiative backwarming effect.
58: \end{abstract}
59: 
60: \keywords{
61: line: profiles ---Sun: flares ---Sun: X-rays, gamma rays}
62: 
63: \section{INTRODUCTION}
64: White-light flares (WLFs) are rare energetic events characterized by 
65: a visible continuum enhancement to a few or tens of percent, 
66: which imposes strict constraints on the modeling of solar flares 
67: in terms of energy release and transport processes in the impulsive phase.
68: According to the spectral features, 
69: two types of WLFs have been proposed \citep{mac86}
70: and such a category greatly facilitates our understanding of 
71: the physical conditions and heating mechanisms of WLFs.
72: The spectra of type I WLFs show a Balmer and Paschen jump,
73: strong and broadened hydrogen Balmer lines,
74: and a continuum enhancement that is well correlated with the HXR emission 
75: and microwave bursts \citep{fan95}.
76: However, type II WLFs do not show the above spectral features \citep{din99}.
77: 
78: The continuum enhancement in WLFs is primarily associated with 
79: the impulsive phase \citep{hud92,nei93a}
80: and often persists after the maximum phase 
81: \citep{hud92,mat03}.
82: The close temporal correlation between the continuum enhancement and 
83: the HXR and microwave emission in type I WLFs indicates that 
84: such WLFs are heated by energy deposition of non-thermal electrons 
85: in the chromosphere.
86: This process can be diagnosed using the \Ha\ line, 
87: which appears to be significantly enhanced and Stark broadened
88: \citep{can84,fan93}.
89: However, in most cases, direct collisional heating by the electron beam 
90: in the lower chromosphere and below, where the continuum emission originates,
91: is hardly effective 
92: because only electrons with very high energies can reach there
93: \citep{lin76,nei93b}.
94: Non-LTE computations also show that beam precipitation 
95: cannot produce the continuum enhancement directly 
96: \citep[e.g.,][]{liu01,din03b}.
97: Therefore, the continuum enhancement is supposed to be produced 
98: indirectly via the radiative backwarming effect 
99: \citep{mac89,met90a,met90b,din03b}.
100: This scenario assumes that non-thermal electrons, whose energies are not 
101: necessarily very high, heat the chromosphere first, and then 
102: the enhanced radiation from the upper layers is transported into 
103: deeper layers and causes a heating there.
104: On the other hand, some 
105: authors also attempted to investigate 
106: the spatial coincidence between the continuum enhancement 
107: and the HXR emission.
108: While in some WLFs, such a spatial coincidence holds well 
109: \citep{mat03,met03,xu04};
110: in some other cases, it does not 
111: \citep{syl00,mat02}.
112: 
113: In the last decade, the white-light data from the aspect camera of 
114: {\it YOHKOH}/SXT \citep{tsu91}
115: provide the first chance to study WLFs from space
116: (e.g., Hudson et al. 1992; Matthews et al. 2003 and references therein).
117: The {\it Transition Region and Coronal Explorer} ({\it TRACE})
118: is also capable to observe the continuum emission 
119: in a wide wavelength range covering the visible band \citep[see][]{met03}.
120: However, coincident HXR observation from 
121: {\it YOHKOH}/HXT \citep{kos91}
122: is somewhat limited by the low energy resolution 
123: since the HXT has only 4 broad energy bands (L, M1, M2, and H bands). 
124: The recently launched 
125: {\it Reuven Ramaty High-Energy Solar Spectroscopic Imager} (\RHESSI) 
126: provides unprecedented high resolution imaging spectroscopy 
127: \citep{lin02}.
128: This, together with the ground-based optical spectroscopy,
129: allows us to quantitatively investigate the temporal and spatial 
130: relationship between the continuum enhancement and non-thermal electrons 
131: producing the HXR emission in solar flares.
132: 
133: An M2.6/2B WLF on 2002 September 29 was simultaneously observed by 
134: the imaging spectrograph of 
135: the Solar Tower Telescope of Nanjing University \citep{hua95} and by \RHESSI.
136: A preliminary analysis of observational aspects for this flare has 
137: been presented in a previous paper \citep[hereafter Paper I]{din03a}.
138: A multi-wavelength analysis of this flare was also 
139: carried out by \citet{kul04}. 
140: In this paper, we perform a quantitative analysis of this flare by 
141: deriving the energy flux of 
142: non-thermal electrons and discussing the origin of the continuum enhancement
143: in terms of current WLF models. 
144: 
145: \section{OBSERVATIONS AND DATA ANALYSIS}
146: We first give a brief description of the \Ha\ and HXR emission of this flare, 
147: as presented in Paper I. 
148: This M2.6/2B flare, associated with a filament eruption, 
149: occurred at NOAA 0134 (N12\arcdeg, E21\arcdeg) on 2002 September 29. 
150: It started at 06:32 and peaked at 06:39 UT.
151: As in Paper I,
152: we pay attention to two main \Ha\ kernels, 
153: which are located at different magnetic polarities (see Figure 4).
154: In particular, we select two points (A and B) representative of the two 
155: kernels to check their evolutionary behaviors based on the signatures
156: of the \Ha\ line profile (see \S3.2).
157: Point A, at the center of the first kernel, is already hot 
158: at the start of ground-based observations and cools down gradually.
159: Point B, at the center of the second kernel (also the brightest kernel),
160: is relatively cool at first and is heated rapidly in the impulsive phase.
161: The continuum enhancement (calculated at \Ha+6 \AA) at Point B 
162: rises rapidly and reaches its maximum ($\sim 8$\%) roughly coincident with 
163: the peak of the 25--50 keV HXR emission.
164: It is interesting that the maximum continuum enhancement at Point B
165: is nearly twice that at Point A. 
166: To study the HXR emission, we first 
167: use the CLEAN algorithm \citep[see, e.g.,][]{kru02} to reconstruct 
168: the HXR images.
169: A strong HXR source appears to encompass both kernels
170: in the early impulsive phase,
171: and it then shows a motion across the magnetic neutral line.
172: Compared to data from the 
173: {\em Solar and Heliospheric Observatory}
174: MDI magnetogram, the bright HXR source
175: seems to straddle over
176: the magnetic neutral line at earlier times;
177: therefore, it is thought to contain two 
178: spatially unresolved FP sources;
179: the motion of the HXR source reflects a change of
180: the relative weights of its two components.
181: 
182: In addition, we employ the Maximum Entropy Method
183: (MEM) algorithm provided by 
184: the \RHESSI\ imaging software \citep{hur02}
185: to reconstruct HXR images around the peak of the impulsive phase. 
186: It is worth noting that, the CLEAN algorithm is 
187: a straight forward iterative algorithm involving a convolution of
188: source emission with instrumental Point Spread Function (PSF);
189: thus, it often gives diffuse images with large FWHM 
190: \citep[see, e.g.,][]{asc04}.
191: In comparison, 
192: the MEM algorithm \citep{sat99} generally 
193: yields relatively sharp images. 
194: In this paper, we use both the CLEAN and MEM algorithms
195: for different purposes.
196: Except for the integration time and energy band,
197: the imaging parameters that are explicitly set in this paper
198: are the same as that in Paper I for consistence. In summary, 
199: we use detectors 3 through 8 in image reconstruction
200: (thus with a spatial resolution of $\sim 7$\arcsec),
201: and set the image center at (--290\arcsec, 90\arcsec), 
202: the FOV of $64\arcsec \times 64\arcsec$, 
203: and the pixel size of $2\arcsec \times 2\arcsec$;
204: all the other parameters are taken at their default. 
205: 
206: Figure 1 shows the 15--50 keV HXR image in the impulsive phase 
207: with an integration time from 06:36:00 to 06:36:30 UT.
208: As expected, two conjugate HXR FPs ($black\ contours$), 
209: located at different magnetic polarities,
210: are clearly resolved by the MEM algorithm,
211: in comparison to the elongated CLEAN image ($grey\ scale$). 
212: Taking into account the spatial resolution,
213: the centroids of the two HXR FPs coincide well with 
214: Points A and B ($plus\ signs$)
215: in the two main \Ha\ kernels, respectively.
216: Moreover, a close spatial correspondence between the continuum emission 
217: ($white\ contours$) and the HXR emission at FP B
218: is clearly seen from Figure 1.
219: Note that we draw in the figure two boxes that encompass the two HXR FPs
220: in order to deduce the photon spectra of them.
221: The result revealed in Figure 1 confirms our previous speculation
222: that the HXR source reconstructed with the CLEAN algorithm
223: is in fact two spatially unresolved FP sources (see Paper I). 
224: 
225: We also try other image reconstruction algorithms 
226: and find that the MEM images can be largely reproduced 
227: by the Pixon algorithm \citep{met96} that usually 
228: gives superior noise suppression and photometric accuracy,
229: but is very time-consuming.
230: Thus, the double-footpoint structure in the MEM images should be real,
231: even though the {\it RHESSI} MEM software may not ensure proper
232: photometric convergence especially when 
233: there are too many freedoms \citep{asc04}.
234: Further investigation on this topic is out of the scope of this paper.
235: 
236: We then reconstruct HXR images in 11 logarithmically spaced energy bands 
237: from 10 keV to $\sim 100$ keV, 
238: for imaging spectroscopy in the impulsive phase.
239: Figure 2 shows a number of selected MEM images
240: with pronounced features,
241: together with the CLEAN images for comparison.
242: \citet{asc04} have revealed that the CLEAN algorithm yields
243: a better photometric convergence than the MEM algorithm.
244: Therefore, we further integrate the photon fluxes over the two boxes A and B, 
245: respectively, using the CLEAN images rather than the MEM images. 
246: Figure 3 plots the photon spectra for the two FPs.
247: 
248: We finally reconstruct HXR images in two broad energy bands 
249: (12--25 keV and 25--50 keV) every 3 s with the CLEAN algorithm. 
250: The integration time is $\sim 4$ s.
251: The HXR time profiles at the two FPs are then extracted,
252: which are plotted in Figure 4. 
253: 
254: \section{RESULTS AND DISCUSSIONS}
255: \subsection{NON-THERMAL EMISSION IN THE IMPULSIVE PHASE} \label{bozomath}
256: \RHESSI\ provides for the first time high spatial and spectral resolution
257: imaging spectroscopy for HXR features of solar flares. 
258: It is seen from Figure 2 that in the impulsive phase, 
259: the HXR emission exhibits an evident migration from FP B to FP A 
260: with increasing energies.
261: Below $\sim 15$ keV the emission comes mainly from FP B 
262: while above $\sim 25$ keV FP A is dominant.
263: At intermediate energies the emission from the two FPs 
264: is of comparable magnitude.
265: 
266: We then fit the non-thermal component of the photon spectra 
267: at the two FPs, respectively.
268: In order to avoid possible thermal contamination,
269: the photon spectra are fitted above $\sim 15$ keV.
270: Figure 3 shows that the photon spectra at the two FPs can both be 
271: well fitted with a single power law.
272: FP A has a photon flux of 0.10 \pho\ at 50 keV and a spectral index of 
273: $\gamma_{A}=4.2$, while FP B has a photon flux of 0.04 \pho\ at 50 keV 
274: and a spectral index of $\gamma_{B}=4.7$.
275: Thus, the photon spectrum of FP A is slightly harder than that of FP B. 
276: Considering the uncertainty in defining the areas for flux integration
277: and spectral fitting,
278: such a difference is not significant for the two conjugate FPs,
279: which are bombarded by electron beams whose spectral indices 
280: are generally assumed to be approximately equal. 
281: 
282: Under the assumption that the non-thermal HXR emission at both FPs 
283: is produced via the thick-target bremsstrahlung \citep{bro71}
284: by electrons whose distribution is a single power law 
285: with a spectral index of $\delta=\gamma+1$ 
286: and a low-energy cutoff of 20 keV,
287: we first derive the total power of non-thermal electrons,
288: $P_{20}$ (ergs s$^{-1}$),
289: from the photon spectra presented above 
290: and then deduce the spatial distribution of energy flux, $F_{20}$ (\erg),
291: with the total power partitioned to each pixel
292: whose weight is proportional to the corresponding photon intensity.
293: This is formulated as 
294: \begin{equation}
295: (F_{20})_{ij}=\frac{P_{20}}{A_{ij}} \frac{I_{ij}}{\sum_{ij} {I_{ij}}},
296: \end{equation}
297: where $(F_{20})_{ij}$, $I_{ij}$, and $A_{ij}$ are 
298: the energy flux, photon intensity, and area at pixel ($i,j$), respectively.
299: Finally, we search for the maximum energy fluxes within the two FPs, 
300: which are found to be 1.0 $\times$ \ten\ and 0.8 $\times$ \ten\ \erg\ 
301: at FPs A and B, respectively.
302: We will show in the following that electron beams with such energy fluxes
303: meet well the requirement for producing the continuum enhancement
304: observed in this WLF.
305: 
306: \subsection{RELATIONSHIP BETWEEN THE CONTINUUM ENHANCEMENT
307: AND NON-THERMAL ELECTRONS} 
308: It is seen from Figure 4 that FP B exhibits a significant continuum enhancement 
309: in the impulsive phase that reaches a peak of $\sim 8$\% at around 06:36:35 UT.
310: Moreover, the temporal evolution of the continuum enhancement shows a fairly
311: well correlation with the 25--50 keV HXR emission.
312: This fact indicates that the continuum enhancement is most probably related to 
313: the precipitation of non-thermal electrons into the chromosphere. 
314: In comparison, the continuum enhancement at FP A is less significant
315: while the HXR emission there seems stronger than that at FP B.
316: To get a quantitative view between the continuum emission and non-thermal
317: electrons, we have further derived the energy content of the electron beams
318: at the two FPs (see \S3.1).
319: The results show that in the impulsive phase, the energy flux of 
320: non-thermal electrons precipitating at FP B is slightly less than that at FP A.
321: Therefore, there arises an interesting question:
322: why a stronger electron beam at FP A results in a weaker 
323: continuum enhancement?
324: 
325: To answer the question about the different responses of the continuum emission
326: to the non-thermal electrons at the two FPs,
327: we need to check carefully the \Ha\ spectral signatures that 
328: provide a clue to the atmospheric heating there.
329: Generally speaking, the \Ha\ line emission can be affected by 
330: three different mechanisms: beam precipitation of energetic electrons, 
331: thermal conduction, and enhanced coronal pressure. 
332: In some cases, specific heating mechanisms may be identified unambiguously from
333: the spectral signatures of the \Ha\ line profile (Canfield et al. 1984). 
334: Figure 5 plots the \Ha\ line profiles for the two FPs at 06:36:16 UT. 
335: The figure shows that
336: the \Ha\ line intensity at FP A is much stronger than that at FP B
337: at the start of ground-based observations, 
338: which means that the chromosphere at FP A has already been heated to 
339: a considerable extent before observations.
340: The continuum emission shows a different behavior: 
341: it increases rapidly at the relatively cool FP B in rough coincidence with
342: the HXR emission, while it varies slowly at the relatively hot FP A,
343: as shown in Figure 4.
344: 
345: \subsection{ORIGIN OF THE DIFFERENCE BETWEEN THE TWO FPs}
346: As shown in Figure 5, the \Ha\ profile at FP A 
347: is relatively strong and broad without a visible reversal, 
348: while that at FP B is relatively weak 
349: and shows an appreciable central reversal.
350: According to Canfield et al. (1984), 
351: only a high coronal pressure
352: can produce strong emission profiles without a central reversal,
353: which fits the situation of FP A.
354: Thus, the less significant continuum enhancement at FP A may result from 
355: a high coronal pressure which prevents most energetic electrons accelerated 
356: in the corona from precipitating deep into the chromosphere effectively. 
357: However, the \Ha\ profile at FP B is associated with a relatively 
358: low coronal pressure, 
359: which allows energetic electrons to easily penetrate into the chromosphere. 
360: 
361: We further estimate the coronal column density, $N$,
362: in the loop as follows, 
363: \begin{equation}
364: N=n \frac{L}{2}=\left(\frac{\rm EM}{AL}\right)^{1/2} \frac{L}{2},
365: \end{equation}
366: where ${\rm EM}$, $A$, and $L$ are the emission measure, 
367: the loop footpoint area, and the loop length, respectively,
368: which can be derived from 
369: the {\it GOES} soft X-ray fluxes and \RHESSI\ images.
370: The quantity of $N$ is estimated to be 
371: $\sim 1.0 \times 10^{20}$ cm$^{-2}$ 
372: in the impulsive phase.
373: Since FP A is much denser than FP B,
374: the coronal column density at FP A may be roughly equal to 
375: the value derived above.
376: The corresponding energy $E$, 
377: electrons of energy above which can penetrate to the chromosphere,
378: follows \citep{bro72,ver04} 
379: \begin{equation}
380: E=({3KN})^{1/2}=8.8 N_{19}^{1/2}\ {\rm (keV)}, 
381: \end{equation}
382: where $K=2 \pi e^4 \Lambda$ 
383: (with $e$ the electron charge and $\Lambda$ the Coulomb logarithm) and 
384: $N_{19}$ is the column density measured in $10^{19}$ cm$^{-2}$ .
385: Inserting the quantity $N$ derived above into Eq. (3) yields $E \simeq 27$ keV.
386: The consequence is that only $\sim 30$\% of the beam energy is deposited
387: into the chromosphere at FP A and
388: therefore the backwarming effect is not significant there.
389: 
390: In comparison, we believe that
391: electron heating of the chromosphere followed by the backwarming effect
392: results in the continuum enhancement at FP B.
393: Using the same method as in \citet{din03b},
394: we perform calculations that can predict the continuum enhancement
395: from a model atmosphere that is bombarded by an electron beam.
396: Figure 6 shows the continuum enhancement 
397: at $\lambda=6600$ \AA\ as a function of the beam energy flux.
398: It is seen that an electron beam with an energy flux of 
399: 0.8 $\times$ \ten\ \erg\ can produce a continuum enhancement of $\sim 8$\%.
400: Thus, the energy flux derived for FP B seems enough to meet 
401: the energy requirement of the continuum enhancement. 
402: However, we should  mention that
403: the deduced energy flux suffers a great uncertainty
404: that arises indeed from the uncertainty of the low-energy cutoff
405: of the electron beam.
406: As shown in Figure 2, the nonthermal component of the HXR emission
407: in the two FPs is still visible below 20 keV;
408: therefore, if we select a low-energy cutoff lower than 20 keV,
409: say, 15 keV, the deduced beam energy flux will be 2--3 times
410: that if adopting the usually assumed low-energy cutoff of 20 keV.
411: 
412: According to the atmospheric models computed by \citet{din03b},
413: we obtain the temperature increase in the lower atmosphere in
414: response to the electron beam heating and the backwarming effect.
415: Then, we can estimate the timescale of the backwarming effect as
416: \begin{equation}
417: %\Delta t=\frac{3}{2} (n_{H}+n_{e})k \Delta T / |\Phi_{NT}-\Phi_{T}|,
418: \Delta t=\frac{3}{2} \frac{(n_{\rm H}+n_{e})k \Delta T}{|\Phi_{\rm NT}-\Phi_{T}|},
419: \end{equation}
420: where $\Phi_{\rm NT}$ and $\Phi_{T}$ are the radiative loss rates
421: in the two cases with and without electron beam heating, respectively,
422: the difference of which represents the heating rate due to the
423: backwarming effect. We find that the timescale varies from
424: $\la 1$ s near the temperature minimum region to $\sim 5$ s at
425: the layer of $\tau_{6600}=1$. In deeper layers, however, the timescale
426: becomes much longer and needs $\sim 20$ s, similar to the estimation
427: of \citet{hen90}.
428: As seen from Figure 4,
429: the time delay of the continuum enhancement with respect to
430: the 25--50 keV HXR emission is $\sim 15$ s, 
431: which may be explained partly by the timescale of radiative backwarming
432: and partly by the low temporal resolution of ground-based observations,
433: during which the repetition time for scanning is $\sim 10$ s.
434: 
435: 
436: \section{CONCLUSIONS}
437: In this paper, we discuss the relationship between 
438: the continuum enhancement and the HXR emission of the WLF on 2002 September 29 
439: in terms of current WLF models.
440: The WLF was simultaneously observed by a ground-based imaging spectrograph 
441: and by \RHESSI. The main results are as follows.
442: 
443: 1. Two conjugate FPs are clearly resolved from the \RHESSI\ 
444: HXR images reconstructed with the MEM and Pixon algorithms, 
445: which are located on different sides of the magnetic neutral line. 
446: Around the peak of the impulsive phase, 
447: the energy fluxes of non-thermal electrons
448: bombarding FPs A and B are estimated to be 
449: 1.0 $\times$ \ten\ and 0.8 $\times$ \ten\ \erg, respectively, 
450: in the framework of the thick-target model.
451: 
452: 2. The continuum enhancement differs greatly at the two FPs.
453: At FP B, it increases rapidly in the impulsive phase 
454: reaching a maximum of $\sim 8$\%,
455: and correlates well with the 25--50 keV HXR emission.
456: While at FP A, it is less significant and varies slowly.
457: We show that at FP B, the derived energy flux of non-thermal electrons 
458: (0.8 $\times$ \ten\ \erg) can produce the observed continuum enhancement 
459: ($\sim 8$\%) in terms of WLF models that invoke the radiative backwarming effect. 
460: 
461: 3. The different behaviors of the continuum emission at the two FPs
462: can be explained by different atmospheric conditions,
463: which are revealed by the \Ha\ line profiles. 
464: The \Ha\ spectral signatures indicate that at FP A, 
465: the atmosphere has been heated considerably and 
466: the coronal pressure is high in the early impulsive phase, 
467: which prevents non-thermal electrons effectively penetrating 
468: into the chromosphere;
469: however, at FP B, the preflare heating is relatively low,
470: which allows an electron beam to 
471: easily penetrate into the chromosphere and
472: produce the observed continuum enhancement 
473: via the radiative backwarming effect. 
474: 
475: 
476: \acknowledgments
477: We would like to thank the referee for valuable comments
478: that led to an improvement of the paper.
479: We are very grateful to the \RHESSI\ team 
480: for providing the observational data
481: and well developed analysis softwares.
482: This work was supported by TRAPOYT, NKBRSF under grant G20000784,
483: and NSFC under grants 10025315, 10221001, and 10333040,
484: and FANEDD under grant 200226.
485: 
486: \begin{thebibliography}{}
487: \bibitem[Aschwanden et al.(2004)]{asc04}
488:            Aschwanden, M.J., Metcalf, T.R., Krucker, S., Sato, J.,
489:            Conway, A.J., Hurford, G.J., \& Schmahl, E.J.
490:            2004, \solphys, 219, 149
491: \bibitem[Brown(1971)]{bro71}
492:            Brown, J.C. 1971, \solphys, 18, 489
493: \bibitem[Brown(1972)]{bro72}
494:            Brown, J.C. 1972, \solphys, 26, 441
495: \bibitem[Canfield, Gunkler, \& Ricchiazzi(1984)]{can84}
496:            Canfield, R. C., Gunkler, T. A., \& Ricchiazzi, P. J.
497:            1984, \apj, 282, 296
498: \bibitem[Ding, Fang, \& Yun(1999)]{din99}
499:            Ding, M.D., Fang, C., \& Yun, H.S.
500:            1999, \apj, 512, 454
501: \bibitem[Ding et al.(2003a)]{din03a}
502:            Ding, M.D., Chen, Q.R., Li, J.P., \& Chen, P.F.
503:            2003a, \apj, 598, 683 (Paper I)
504: \bibitem[Ding et al.(2003b)]{din03b}
505:            Ding, M.D., Liu, Y., Yeh, C.-T., \& Li, J. P.
506:            2003b, \aap, 403, 1151
507: \bibitem[Fang \& Ding(1995)]{fan95}
508:            Fang, C., \& Ding, M.D. 1995, \aaps, 110, 99
509: \bibitem[Fang, H\'enoux, \& Gan(1993)]{fan93}
510:            Fang, C., H\'enoux, J.-C., \& Gan, W.Q. 1993, \aap, 274, 917
511: \bibitem[H\'enoux et al.(1990)]{hen90}
512:            H\'enoux, J.-C., Aboudarham, J., Brown, J.C., van den Oord, G. H. J.,
513:            van Driel-Gesztelyi, L., \& Gerlei, O.
514:            1990, \aap, 233, 577
515: \bibitem[Huang et al.(1995)]{hua95}
516:            Huang, Y. R., Fang, C., Ding, M. D., Gao, X. F., Zhu, Z. G.,
517:            Ying, S. Y., Hu, J., \& Xue, Y. Z. 1995, \solphys, 159, 127
518: \bibitem[Hudson et al(1992)]{hud92}
519:            Hudson, H.S., Acton, L.W., Hirayama, T., \& Uchida, Y.
520:            1992, \pasj, 44, L77
521: \bibitem[Hurford et al.(2002)]{hur02}
522:            Hurford, G.J., et al. 2002, \solphys, 210, 61
523: \bibitem[Kosugi et al.(1991)]{kos91}
524:            Kosugi, T., et al. 1991, \solphys, 136, 17
525: \bibitem[Krucker \& Lin(2002)]{kru02}
526:            Krucker, S., \& Lin, R.P. 2002, \solphys, 210, 229
527: \bibitem[Kulinov\'a et al.(2004)]{kul04}
528:            Kulinov\'a, A., Dzif\v{c}\'akov\'a, E., Buj\v{n}\'ak, R., \& Karlick\'y, M.
529:            2004, \solphys, 221, 101
530: \bibitem[Lin \& Hudson(1976)]{lin76}
531:            Lin, R.P., \& Hudson, H.S. 1976, \solphys, 50, 153
532: \bibitem[Lin et al.(2002)]{lin02}
533:            Lin, R.P., et al. 2002, \solphys, 210, 3
534: \bibitem[Liu, Ding, \& Fang(2001)]{liu01}
535:            Liu, Y., Ding, M. D., \& Fang, C. 2001, \apjl, 563, L169
536: \bibitem[Machado et al.(1986)]{mac86}
537:            Machado, M.E., et al. 1986, in The Lower Atmosphere of Solar Flares,
538:            ed. D.F.Neidig (Sunspot: NSO), 483
539: \bibitem[Machado, Emslie, \& Avrett(1989)]{mac89}
540:          Machado, M.E., Emslie, A.G., \& Avrett, E.H. 1989, \solphys, 124, 303
541: \bibitem[Matthews et al.(2002)]{mat02}
542:           Matthews, S.A., van Driel-Gesztelyi, L., Hudson, H.S., \& Nitta, N.V.
543:           2002, in Multi-Wavelength Observations of Coronal Structure and Dynamics,
544:           ed. P.C.H. Martens \& D. P. Cauffman (Amsterdam: Pergamon), 289
545: \bibitem[Matthews et al.(2003)]{mat03}
546:            Matthews, S.A., van Driel-Gesztelyi, L., Hudson, H.S., \& Nitta, N.V.
547:            2003, \aap, 409, 1107
548: \bibitem[Metcalf et al.(2003)]{met03}
549:            Metcalf, T.R., Alexander, D., Hudson, H.S., \& Longcope, D.W.
550:            2003, \apj, 595, 483
551: \bibitem[Metcalf et al.(1990a)]{met90a}
552:            Metcalf, T.R., Canfield, R.C., Avrett, E.H., \& Metcalf, F.T.
553:            1990a, \apj, 350, 463
554: \bibitem[Metcalf, Canfield, \& Saba(1990b)]{met90b}
555:            Metcalf, T.R., Canfield, R.C., \& Saba, J.L.R.
556:            1990b, \apj, 365, 391
557: \bibitem[Metcalf et al.(1996)]{met96}
558:            Metcalf, T.R., Hudson, H.S., Kosugi, T., Puetter, R.C., \& Pi\~na, R.K.
559:            1996, \apj, 466, 585
560: \bibitem[Neidig \& Kane(1993a)]{nei93a}
561:            Neidig, D.F., \& Kane, S.R. 1993a, \solphys, 143, 201
562: \bibitem[Neidig et al.(1993b)]{nei93b}
563:            Neidig, D.F., Kiplinger, A.L., Cohl, H.S., \& Wiborg, P.H.
564:            1993b, \apj, 406, 306
565: \bibitem[Sato, Kosugi, \& Makishima(1999)]{sat99}
566:            Sato, J., Kosugi, T., \& Makishima, K. 1999, \pasj, 51, 127
567: \bibitem[Sylwester \& Sylwester(2000)]{syl00}
568:            Sylwester, B., \& Sylwester, J. 2000, \solphys, 194, 305
569: \bibitem[Tsuneta et al.(1991)]{tsu91}
570:            Tsuneta, S., et al. 1991, \solphys, 136, 37
571: \bibitem[Veronig \& Brown(2004)]{ver04}
572:            Veronig, A.M., \& Brown, J.C. 2004, \apjl, 603, L117
573: \bibitem[Xu et al.(2004)]{xu04}
574:            Xu, Y., Cao, W., Liu, C., Yang, G., Qiu, J., Jing, J., 
575:            Denker, C., \& Wang, H.
576:            2004, \apjl, 607, L131
577: \end{thebibliography}
578: 
579: \clearpage
580: 
581: \begin{figure}
582: \epsscale{.80}
583: \plotone{f1.eps}
584: \caption{
585: \RHESSI\ 15--50 keV HXR image in the impulsive phase with 
586: an integration time from 06:36:00 to 06:36:30 UT. The MEM image 
587: ($black\ contours$, with levels of 10, 20, 40, 60, and 80\%)
588: shows two well resolved HXR FPs across the magnetic neutral line
589: ($white\ dashed\ line$). 
590: The CLEAN image ($grey\ scale$) shows an elongated bright source 
591: covering both magnetic polarities. 
592: The two boxes indicate the areas, covering FPs A and B, respectively,
593: over which the photon fluxes in Figs. 3--4 are integrated.
594: Also shown in the figure is the continuum emission at 06:36:16 UT
595: ($white\ contours$, with levels of 50, 70, and 90\%).
596: \label{fig1}}
597: \end{figure}
598: \clearpage
599: 
600: \begin{figure}
601: \epsscale{1.00}
602: \plotone{f2.eps}
603: \caption{
604: \RHESSI\ HXR images of the flare at different energy bands 
605: from 06:36:00 to 06:36:30 UT, 
606: reconstructed with the CLEAN algorithm ($grey\ scale$).
607: Superposed contours (with levels of 10, 20, 40, 60, and 80\%)
608: are those reconstructed with the MEM algorithm.
609: \label{fig2}}
610: \end{figure}
611: \clearpage
612: 
613: \begin{figure}
614: \epsscale{1.00}
615: \plotone{f3.eps}
616: \caption{Photon spectra for the two FPs, and their power-law fitting.
617: \label{fig3}}
618: \end{figure}
619: \clearpage
620: 
621: \begin{figure}
622: \epsscale{0.70}
623: \plotone{f4.eps}
624: \caption{
625: $Top\ panel$: 
626: Time profiles of the spatially integrated \RHESSI\ HXR emission 
627: in 12--25 and 25--50 keV energy bands. 
628: $Middle\ and\ bottom\ panels$:
629: Net increase of the emission at \Ha\ line center ($solid\ line$) and 
630: at \Ha+6 \AA\ (regarded as the continuum enhancement, $dotted\ line$), 
631: time profiles of HXR emission in 12--25 keV ($grey\ dot$-$dashed\ line$, 
632: scaled by 0.02) and in 25--50 keV ($grey\ solid\ line$) 
633: for FPs A and B, respectively. 
634: The two vertical bars refer to the integration time 
635: for HXR image reconstruction in Figs. 1--2.
636: \label{fig4}}
637: \end{figure}
638: \clearpage
639: 
640: \begin{figure}
641: \epsscale{.80}
642: \plotone{f5.eps}
643: \caption{
644: \Ha\ line profiles at FPs A and B at 06:36:16 UT
645: with the quiet-Sun profile subtracted.
646: The profiles are normalized by the nearby continuum.
647: \label{fig5}}
648: \end{figure} 
649: \clearpage
650: 
651: \begin{figure}
652: \epsscale{.80}
653: \plotone{f6.eps}
654: \caption{
655: Theoretical prediction of the continuum enhancement at $\lambda=6600$
656: \AA\ as a function of the energy flux of the electron beam that
657: bombards the atmosphere. The calculations are similar to those of
658: \citet{din03b}.
659: \label{fig6}}
660: \end{figure}
661: \clearpage
662: 
663: 
664: \end{document}
665: