1: % SAMPLE1.TEX -- AASTeX sample paper with minimal markup.
2: %
3: % V0.1 - sent to all 14/2/2005
4: % V0.2 - SM 16/2/2005
5: % V0.3 - SM 22-23/2
6: % V0.4 - SM 25/2
7: % V1.0 - SM 25/3 revised after referee's comments
8: % V1.1 - SM 30/3
9: %\documentstyle[12pt,aasms4]{article}
10:
11: \documentclass[12pt,preprint]{aastex}
12:
13: \begin{document}
14:
15: %%%%%%%%%%%%%% MY DEFINITIONS
16: \def\pdot {\dot P}
17: \def\Omdot {\dot \Omega}
18: \def\ltsima{$\; \buildrel < \over \sim \;$}
19: \def\lsim{\lower.5ex\hbox{\ltsima}}
20: \def\gtsima{$\; \buildrel > \over \sim \;$}
21: \def\gsim{\lower.5ex\hbox{\gtsima}}
22: \def\msole{~M_{\odot}}
23: \def\mdot {\dot M}
24: \def\sgr {SGR~1806--20}
25: \def\cha {\textit{Chandra~}}
26: \def\xmm {\textit{XMM-Newton~}}
27: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
28:
29:
30: \slugcomment{Submitted to ApJLett on 28/2/2005 - revised 1/4
31: }
32:
33:
34: \title{The first giant flare from \sgr:\\
35: observations with the INTEGRAL
36: SPI Anti-Coincidence Shield}
37: \author{
38: S. Mereghetti$^1$,
39: D. G\"{o}tz\footnote{Istituto di Astrofisica Spaziale e Fisica Cosmica, Sezione di Milano ''G.Occhialini'' - INAF v.Bassini 15, I-20133 Milano, Italy},
40: A. von Kienlin$^2$,
41: A. Rau$^2$,
42: G. Lichti\footnote{Max-Planck-Institut f\"{u}r extraterrestrische Physik, Giessenbachstrasse, Postfach 1312, D-85741 Garching, Germany},
43: G. Weidenspointner$^3$,
44: P. Jean\footnote{Centre d'\'{E}tude Spatiale des Rayonnements, 31028 Toulouse, France }
45: }
46:
47: \begin{abstract}
48:
49: A giant flare from the Soft Gamma-ray Repeater \sgr\ has been
50: discovered with the INTEGRAL gamma-ray observatory on 2004
51: December 27 and detected by many other satellites. This tremendous
52: outburst, the first one observed from this source, was a hundred
53: times more powerful than the two giant flares previously observed
54: from other Soft Gamma-ray Repeaters (SGR). The 50 ms resolution
55: light curve obtained with the Anticoincidence Shield of the
56: INTEGRAL spectrometer SPI, which provides a high effective area
57: above $\gsim$80 keV, shows evidence for emission lasting about one
58: hour after the start of the outburst. This component, which decays
59: in time as $\sim$t$^{-0.85}$, could be the first detection of a
60: hard X-ray afterglow associated to an SGR giant flare. The short
61: (0.2 s) initial pulse was so strong to saturate the detector for
62: $\sim$0.7 s, and its backscattered radiation from the Moon was
63: detected 2.8 s later. The following $\sim$400 s long tail,
64: modulated at the neutron star rotation period of 7.56 s, had a
65: fluence of 2.6$\times10^{-4}$ erg cm$^{-2}$ above 80 keV, which
66: extrapolating to lower energies corresponds to an emitted energy
67: of 1.6$\times10^{44}$~d$^{2}_{15 kpc}$ erg at E$>$3 keV. This is
68: of the same order of that in the pulsating tails of the two giant
69: flares seen from other SGRs, despite the hundredfold
70: larger overall emitted energy of the 2004 December 27 event.
71: \end{abstract}
72:
73: \keywords{stars: individual (SGR 1806--20) -- stars: neutron --
74: X--rays: bursts -- Gamma-rays: bursts}
75:
76: \section{Introduction}
77:
78: Soft Gamma-ray Repeaters (SGRs) are high-energy sources
79: characterized by sporadic periods of activity in which they emit
80: short bursts ($<$1~s) with energy up to $\sim$10$^{41}$ erg. They
81: are believed to be highly magnetized (B$\sim10^{14}$--10$^{15}$ G)
82: neutron stars, or ''magnetars'' (see Woods \& Thompson 2004 for a
83: recent review). Large soft $\gamma$-ray flares, reaching peak
84: luminosities above several 10$^{44}$ erg s$^{-1}$ and lasting a
85: few minutes, have been observed from two SGRs: on 1979 March 5
86: from SGR 0525--66 (Mazets et al. 1979) and on 1998 August 27 from
87: SGR 1900$+$14 (Hurley et al. 1999; Feroci et al. 1999). The
88: detection of only two of such giant flares from different sources
89: in about 30 years implies that these events, involving energy
90: releases of more than 10$^{44}$ ergs, are relatively rare.
91:
92:
93: \sgr\ is currently the most prolific of the four known SGRs. Its
94: level of activity has been increasing in the last few years,
95: during which many bursts have been detected with different
96: satellites, including RXTE (G{\" o}{\u g}{\" u}{\c s} et al. 2001;
97: Ibrahim et al. 2003; Woods et al. 2004) and INTEGRAL (G\"{o}tz et
98: al. 2004). On 2004 October 5 two clusters of strong bursts with a
99: total fluence of $\sim$10$^{-4}$ erg cm$^{-2}$ were emitted
100: within a time span of a few minutes (Mereghetti et al. 2004). It
101: was noted (Golenetskii et al. 2004) that a similar event occurred
102: in SGR 1900$+$14 three months before its 1998 giant flare (Aptekar
103: et al. 2001). The increasing level of activity in \sgr\ was also
104: reflected in the properties of its quiescent X--ray emission.
105: XMM-Newton observations showed that its 2-10 keV luminosity
106: doubled and the spectrum became harder in 2004 (Mereghetti et al.
107: 2005b). A similar trend was present in the persistent 20-150 keV
108: emission discovered with INTEGRAL observations carried out in
109: 2003-2004 (Mereghetti et al. 2005a).
110:
111: The energetic activity of \sgr\ culminated on 2004 December 27,
112: when the INTEGRAL satellite discovered the first giant outburst
113: from this source (Borkowski et al. 2004). More than twenty
114: satellites recorded this exceptional event which started with a
115: hard pulse so intense to saturate most detectors (Hurley et al.
116: 2005; Mazets et al. 2005; Palmer et al. 2005; Terasawa et al.
117: 2005) and to significantly ionize the Earth's upper atmosphere
118: (Campbell et al. 2005). Here we report the properties of this
119: giant outburst derived with the Anti-Coincidence Shield (ACS) of
120: the Spectrometer on board INTEGRAL (SPI, Vedrenne et al. 2003).
121:
122:
123:
124:
125: \section{The Anti-Coincidence Shield of SPI}
126:
127: The ACS consists of 91 bismuth
128: germanate (BGO) scintillator crystals of thickness between 16 and
129: 50 mm and with a total mass of 512 kg. Besides serving as a veto
130: for the SPI germanium spectrometer, the ACS is able to detect
131: gamma-ray bursts from a large fraction of the sky (von Kienlin et
132: al. 2003). The data used here consist of the overall ACS count
133: rate (i.e. the read out from the sum of the 91 crystals) sampled
134: in time intervals of 50 ms. No energy or directional information
135: is available. The ACS is sensitive to photons above a low-energy
136: threshold corresponding to approximately 80 keV. However, due to
137: the different properties of the various BGO blocks, their
138: associated photomultipliers, and the chosen redundancy concept
139: (the signals from two different crystals are fed to the same
140: front-end electronics), this threshold is not sharp and its exact
141: value not well determined.
142:
143: Dead time and saturation effects in the crystals and electronics
144: are negligible ($<$1\%) for total count rates smaller than a few
145: 10$^5$ cts s$^{-1}$. When this condition is verified we can easily
146: convert the observed count rates to an incident energy flux,
147: assuming a spectral shape and knowing the ACS effective area. The
148: latter is a function of the incidence angle, with a maximum for
149: directions nearly orthogonal to the satellite pointing axis and
150: unobstructed by other instruments. At the time of the flare \sgr\
151: was at a zenith angle $\theta$=106$^{\circ}$ from the SPI pointing
152: axis and at an azimuth angle $\phi$=9$^{\circ}$
153: ($\phi$=0$^{\circ}$ corresponds to the satellite Sun-pointing
154: side, i.e. the direction from SPI toward the IBIS instrument). The
155: ACS effective area as a function of energy for this direction has
156: been computed by means of Monte Carlo simulations based on
157: detailed mass modelling (Weidenspointner et al. 2003, and
158: references therein) of SPI and of the surrounding
159: material (satellite structure and other instruments). The
160: effective area increases monotonically with energy; it is
161: $\sim$340 cm$^2$ at 100 keV, $\sim$1150 cm$^2$ at 200 keV, and larger
162: than 3000 cm$^2$ above 1 MeV. Thus the ACS provides the data with
163: the best statistics in the soft $\gamma$-ray range available for this
164: giant flare.
165: For an optically thin thermal bremsstrahlung spectrum with
166: kT$_{br}$=30 keV we obtain a conversion factor of 1 ACS count
167: s$^{-1}$
168: $\sim$4.3$\times10^{-10}$ erg cm$^{-2}$ s$^{-1}$ (80-2000
169: keV).
170:
171:
172:
173: \section{Properties of the giant flare}
174:
175:
176: The total light curve of the flare, obtained by binning at 2.5 s
177: the original ACS data, is shown in Fig.~\ref{total}, while
178: Fig.~\ref{initialfit} displays at full resolution (50 ms) the
179: initial part of the pulsating tail. We have defined t=0 at
180: 21:30:26.55 UT of 2004 December 27, approximately coinciding with
181: the rise time of the flare. The giant flare starts with a short
182: and extremely intense spike\footnote{the Y-axis scale in Fig.~1 is
183: cut at 1.05$\times$10$^{5}$ counts s$^{-1}$ to better show the
184: flare tail}. The measured count rate at the peak,
185: $\sim2\times10^6$ counts s$^{-1}$, is strongly affected by dead
186: time and saturation effects which cannot be easily modelled and
187: prevent reliable flux estimates. A fit with a constant to the
188: count rate in a 2 ks long time interval before the flare
189: yields a count rate of
190: $\sim$8.8$\times$10$^4$ counts s$^{-1}$, which we use in the
191: following analysis as a constant background level and include in
192: all the fits keeping its value fixed.
193:
194:
195: A narrow burst, lasting $\sim$0.2 s, occurs at t=2.8 s (see
196: Fig.~\ref{initialfit}). Since this delay corresponds to the time
197: taken by the wavefront to travel from INTEGRAL to the Moon and
198: back to the satellite, this burst can be explained as
199: backscattered radiation of the initial spike. A similar detection
200: was reported with the Helicon-Coronas-F satellite, which could not
201: observe directly \sgr\ due to Earth occultation (Mazets et al.
202: 2005). Assuming for the backscattered radiation the spectral shape
203: derived by the above authors (dN/dE$\propto$E$^{-0.7}$ exp(--E/800
204: keV) ), we obtain a fluence of $\sim$2$\times10^{-6}$ erg
205: cm$^{-2}$ (E$>$80 keV). The duration and the fluence of the
206: backscattered radiation are consistent with the properties of the
207: initial spike (Mazets et al. 2005; Hurley et al. 2005; Terasawa et
208: al. 2005).
209:
210: The inset of Fig.~\ref{total} shows the light curve of the
211: precursor burst which occurred 143~s before the flare.
212: Triangulation analysis, using the ACS light curve together with
213: data from other satellites, demonstrates that its arrival
214: direction is consistent with the position of \sgr\ (Hurley et al.
215: 2005).
216: The above authors, using the RHESSI satellite, found that the
217: precursor spectrum can be crudely approximated by a thermal
218: bremsstrahlung with kT$_{br}$=15 keV and reported a fluence of
219: 1.8$\times$10$^{-4}$ erg cm$^{-2}$ for E$>$3 keV. Assuming the
220: same spectrum, we obtain a fluence above 80 keV of
221: 4.4$\times$10$^{-6}$ erg cm$^{-2}$, which extrapolated to lower
222: energy overestimates the RHESSI value by a factor ten. Considering
223: the uncertainties in the instruments cross calibration and the
224: extrapolation largely dependent on the poorly determined
225: spectrum, we do not consider this discrepancy as particularly
226: significant. For instance, for kT$_{br}$=30 keV we get F($>$80
227: keV)=4.0$\times$10$^{-6}$ erg cm$^{-2}$ and F($>$3 keV)=10$^{-4}$
228: erg cm$^{-2}$. The phase of the event rise time with respect to
229: the 7.56 s pulsations\footnote{since our data allow to determine
230: the pulsation period with an accuracy of only $\sim$0.1 s, we
231: adopt the more precise value obtained with RXTE (Woods et al.
232: 2005) and define phase $\phi$=0 at t=0 s} is $\phi\sim$0.1, which
233: does not correspond to any of the peaks seen in the pulsating
234: tail.
235:
236:
237: The transition from the initial spike to the pulsating tail is
238: well fitted, in the time interval 1.2--2.2 s, by a power law
239: function F$\propto$t$^{-\delta}$ with $\delta$=2.1$\pm$0.1
240: (Fig.~\ref{initialfit}). The broad bump in the time range
241: $\sim$2--5 s, as well as the following ones after t=10 s, are in
242: phase with the main peak of the 7.56 s pulsations
243: ($\phi$=0.35--0.6) and can therefore be considered as the first
244: appearance of the periodicity. The average profile of the
245: pulsations is shown in Fig.~\ref{fol757-aver} for different time
246: intervals. An evolution of the profile during the flare, with the
247: relative intensity of the peak at $\phi$=0.75 increasing with
248: time, is evident.
249:
250: To characterize the long-term decay profile removing the effect of
251: the pulsations, we binned the data in time intervals of 7.56 s.
252: The resulting light curve (inset of Fig.~\ref{initialfit}) is
253: well fitted in the time interval t$\sim$15--400 s by an
254: exponential function with decay constant $\tau$=138$\pm$5 s. Note
255: that the first bin (t=7.56--16.2 s) does not contain the initial
256: hard spike. Nevertheless it lies significantly above the fitted
257: function. Assuming a thermal bremsstrahlung spectrum
258: with kT$_{br}$=30 keV, similar to the giant flares tails of other
259: SGRs, the fluence in the time interval 1--400 s is
260: $\sim$2.6$\times$10$^{-4}$ erg cm$^{-2}$ for E$>$80 keV.
261: Extrapolating the same spectral shape to lower energy, this
262: corresponds to F($>$3 keV)$\sim$6.4$\times$10$^{-3}$ erg cm$^{-2}$.
263:
264:
265:
266: \section{Afterglow emission}
267:
268: Our data show evidence for a second, separate component in
269: addition to the main flare discussed above. After the end of the
270: pulsating tail, at t$\sim$400 s, the count rate increases
271: again, forming a long bump which peaks around t$\sim$600--800 s
272: and returns to the pre-flare background level at t$\sim$3000-4000
273: s (see Fig.~\ref{longlongtail}).
274:
275: The ACS count rate is dominated by the incident flux of charged
276: particles which varies on different time scales, due to the
277: satellite motion and to the temporal and spatial
278: variations in the flux of cosmic rays and trapped radiation. A
279: long term ACS light curve covering almost one day and including
280: the \sgr\ giant flare is shown in the top panel of
281: Fig.~\ref{psac}. For comparison, the lower panel of
282: Fig.~\ref{psac} shows the light curve
283: obtained in the same time interval
284: with the Plastic Scintillator Anti-Coincidence (PSAC), a
285: thin (5 mm) detector located just below the SPI coded mask
286: aperture, which is sensitive to charged particles and almost
287: completely transparent to hard X--ray photons. It can be noted
288: that both the long term trend and most of the fluctuations on
289: shorter time scales are present in both detectors, as expected for
290: charged particle induced background. On the contrary, the bump at
291: t=400--4000 s is prominent only in the ACS, indicating its hard
292: X--rays nature. Known variable sources, like e.g. Cyg X--1, are
293: not bright enough in the hard X--rays to produce the count rate
294: excess seen in the ACS, and no new transients were reported at the
295: time of this observation. We also checked that the Sun was not
296: particularly active in X--rays, by looking at the publicly
297: available data of the GOES
298: satellites\footnote{http://www.sec.noaa.gov/}. We therefore
299: conclude that the re-brightening after t=400 s and lasting about
300: one hour is associated with \sgr . To our knowledge, this
301: component has not been reported using data from other satellites.
302: This might be due to background variability and source occultation
303: effects in low Earth orbit satellites and to the lack of
304: sufficient sensitivity at hard X--rays in other detectors.
305:
306: After accounting for the background, estimated from a linear fit
307: to the ACS count rate for t$<$0 s and t$>$4000 s, the time profile
308: decay in the interval t=500-4000 s can be reasonably well
309: described as a power law F$\propto$t$^{-\delta}$ with
310: $\delta\sim$0.85. The fluence in the 400-4000 s time interval is
311: approximately equal to that contained in the pulsating tail
312: (t=1--400 s).
313:
314:
315:
316: \section{Discussion}
317:
318: The most striking difference between the 2004 December 27 outburst
319: of \sgr\ and the giant flares previously observed from SGR
320: 0526--66 and SGR 1900$+$14 is the global energetics of the event.
321: For a distance of 15 kpc (Corbel \& Eikenberry 2004), the hard
322: X--ray fluence in the \sgr\ initial spike (Hurley et al. 2005;
323: Terasawa et al. 2005) implies an isotropic-equivalent energy
324: release of several 10$^{46}$~d$^{2}_{15}$ erg.
325: This is at least two orders of magnitude larger than that of the
326: two other giant flares (1.6$\times10^{44}$ erg for SGR 0526--26;
327: $>7\times10^{43}$ erg for SGR 1900$+$14). The much higher involved
328: energy is also reflected in the properties of the transient radio
329: emission, which reached a luminosity a factor 500 (Cameron et al.
330: 2005, Gaensler et al. 2005) larger than that seen in SGR 1900$+$14
331: after the 27 August 1998 event (Frail et al. 1999).
332: On the other hand, the energy output in the \sgr\ pulsating tail
333: (1.6$\times10^{44}$~d$^{2}_{15}$ erg for E$>$3 keV) is similar to
334: that of SGR 1900$+$14 (5$\times10^{43}$ erg) and of SGR 0526--66
335: (4$\times10^{44}$ erg). The giant flare of \sgr\ is thus
336: characterized by a much higher spike-to-tail energy ratio
337: ($\gsim$100) than the two previous events ($\sim$1). According to
338: the magnetar model, essentially all the energy release occurs
339: during the initial $\sim$0.2 s transient phase, when a hot
340: relativistic fireball is launched. A fraction of this energy is
341: trapped by closed field lines in the neutron star magnetosphere,
342: forming an optically thick photon-pair plasma which evaporates
343: giving rise to the radiation observed in the pulsating tail
344: (Thompson \& Duncan 1995). The magnetic field strength limits the
345: amount of energy that can be confined. The fact that this quantity
346: is similar in the three giant flares, despite the much higher
347: total energy release of \sgr , is consistent with a magnetic field
348: of the same order in the three magnetars.
349:
350:
351: The pulse profile and its time evolution observed with the ACS are
352: similar to those seen with RHESSI and Swift (Hurley et al. 2005;
353: Palmer et al. 2005), which are dominated by photons of lower
354: energy. Thanks to the large effective area above 100 keV of the
355: ACS, we can clearly see variations in the pulse profile at high
356: energies. The different evolution of the emission components
357: visible in the pulsating tail (see Fig.~\ref{fol757-aver})
358: suggests separate emission regions. It is interesting to note that
359: the pulsations are present since the very early phase of the
360: flare, as indicated by the broad pulse visible at t=2--5 s
361: (Fig.~\ref{initialfit}). This pulse persists at the same
362: rotational phase in the following cycles, while a secondary peak
363: at $\phi\sim$0.75 gradually emerges.
364:
365: While the above results are similar to those obtained with other
366: satellites, our data provide unique evidence for a second, long
367: lasting component in the hard X--ray light curve. The emission
368: after t$\sim$400 s could originate from the neutron star surface
369: and/or magnetosphere or, alternatively, from the interaction of
370: the relativistic fireball with the surrounding medium. A search
371: for pulsations in the ACS data after t=400 s gave a
372: negative result, favoring the second interpretation. The afterglow
373: produced by the \sgr\ giant flare has been observed in the radio
374: band (Gaensler et al. 2005; Cameron et al. 2005), yielding
375: estimates of the minimum energy in magnetic field and relativistic
376: particles, E, of several 10$^{43}$ erg (see also Nakar, Piran \&
377: Sari 2005). The hard X--ray fluence in the 400-3000 s time
378: interval is consistent with this value. With simple gamma-ray
379: burst afterglow models based on synchrotron emission we can
380: estimate the bulk Lorentz factor $\Gamma$ from the time t$_{0}$ of
381: the afterglow onset: $\Gamma\sim$15 (E / 5 10$^{43}$~erg)$^{1/8}$
382: (n / 0.1 cm$^{-3}$)$^{-1/8}$ (t$_{0}$ / 100 s)$^{-3/8}$, where n
383: is the ambient density (see, e.g. Zhang \& Meszaros 2003, and
384: references therein). This is smaller than the typical values for
385: gamma-ray bursts, but consistent, considering the large involved
386: uncertainties, with the mildly relativistic outflow inferred from
387: the modelling of the radio data (Granot et al. 2005).
388: Alternatively, the observed re-brightening could be due to an
389: Inverse Compton component, implying a high density environment
390: (n$\gsim$10 cm$^{-3}$) and a high electron radiation efficiency.
391:
392:
393: \acknowledgments
394:
395: We thank K. Hurley, G. Ghisellini, J. Borkowski and M. Feroci for
396: useful comments. This work has been partially funded by the
397: Italian Space Agency. The SPI-ACS is supported by the German
398: "Ministerium fur Bildung und Forschung" through the DLR grant
399: 50.OG.9503.0.
400:
401: \begin{references}
402:
403: \reference{} Aptekar R.L., Frederiks D.D., Golenetskii S.V. et al. 2001, ApJSS 137, 227
404: \reference{} Borkowski J., G\"{o}tz D., Mereghetti S. et al. 2004, GCN Circ. n. 2920
405: \reference{} Cameron P.B. et al. 2005, submitted to Nature, astro-ph/0502428
406: \reference{} Campbell P. et al. 2005, GCN Circ. n. 2932
407: \reference{} Corbel S. \& Eikenberry S.S. 2004, A\&A 419, 191
408: \reference{} Feroci M. et al. 1999, ApJ 515, L9
409: \reference{} Frail D.A., Kulkarni, S.R. \& Bloom, J.S., 1999, Nature, 398, 127
410: \reference{}Gaensler, B.M., et al., 2005, submitted to Nature, astro-ph/0502393
411: \reference{}G{\" o}{\u g}{\" u}{\c s}, E., Kouveliotou, C., Woods, P.M., et al. 2001, ApJ, 558, 228
412: \reference {} Golenetskii S.V., Aptekar R., Mazets E. et al. 2004, GCN Circular n. 2769
413: \reference{}G\"{o}tz, D., Mereghetti S., Mirabel F.I. \& Hurley K. 2004, A\&A 417, L45
414: \reference{} Granot J., Ramirez-Ruiz E., Taylor G.B. et al. 2005, Submitted to ApJ, astro-ph/0503251
415: \reference{} Hurley K., Cline T., Mazets E. et al. 1999, Nature 397, 41
416: \reference{} Hurley K. et al. 2005, submitted to Nature, astro-ph/0502329
417: \reference{} Ibrahim A.I., Swank J.H. \& Parke W. 2003, ApJ 584, L171
418: \reference{} Mazets E.P. et al. 1979, Nature 282, 587
419: \reference{} Mazets E.P. et al. 2005, astro-ph/0502541
420: \reference{} Mereghetti S., G\"{o}tz D., Borkowski J. et al. 2004, GCN Circular n. 2763
421: \reference{} Mereghetti S., G\"{o}tz D., Mirabel I.F \& Hurley K. 2005a, A\&A, in press, astro-ph/0411695
422: \reference{} Mereghetti S., Tiengo A., Esposito P. et al. 2005b, submitted to ApJ, astro-ph/0502417
423: \reference{} Nakar E., Piran T. \& Sari R. 2005, astro-ph/0502052
424: \reference{} Palmer D. et al. 2005, submitted to Nature, astro-ph/0503030
425: \reference{} Terasawa, T. et al., 2005, submitted to Nature, astro-ph/0502315
426: \reference{}Thompson, C., \& Duncan, R.C. 1995, MNRAS, 275, 255
427: \reference{} von Kienlin A., Beckmann V., Rau A., et al. 2003, A\&A, 411, L299
428: \reference{} Vedrenne G., Roques J.-P., Sch\"{o}nfelder V., et al. 2003, A\&A 411, L63
429: \reference{} Weidenspointner G., Kiener J., Gros M. et al. 2003, A\&A 411, L113
430: \reference{} Woods P.M. \& Thompson C. 2004, astro-ph/0406133
431: \reference{} Woods P.M., Kouveliotou C., G{\" o}{\u g}{\" u}{\c s} E., et al. 2004, The Astronomer's Telegram, 313
432: \reference{} Woods P.M., Finger M., Patel S. et al. 2005, GCN Circular n. 2950
433: \reference{} Zhang B. \& M\'{e}sz\'{a}ros P. 2003, astro-ph/0311321
434:
435:
436: \end{references}
437:
438:
439: \clearpage
440: \figcaption{ACS light curve of the whole giant flare binned
441: at 2.5 s. The peak of the flare, reaching an observed count rate
442: $\gsim2\times10^{6}$ counts s$^{-1}$, is not shown. The inset
443: shows the light curve of the precursor burst at full resolution
444: (50 ms). \label{total}}
445: \epsscale{.7}
446: \plotone{f1new.eps}
447:
448: \clearpage
449: \figcaption{Declining part of the initial spike and initial part of the pulsating tail.
450: The line is a fit with a power law plus a constant (fixed at the pre-flare level) to the count rate in the
451: time interval 1.2--2.2 s. The narrow burst at t=2.8 s is due to
452: radiation from the initial spike backscattered by the Moon. The
453: inset shows the light curve of the tail binned at the spin period
454: (7.56 s) and fitted with an exponential function.
455: \label{initialfit}}
456: \epsscale{.7}
457: \plotone{f2new.eps}
458:
459: \clearpage
460: \figcaption{Averaged pulse profile obtained by folding the data in three time
461: intervals at the spin period of 7.56 s. \label{fol757-aver} }
462: \epsscale{.7}
463: \plotone{f3new.eps}
464:
465: \clearpage
466: \figcaption{SPI ACS light curve of \sgr\ binned at 50 s. The count rate settles back
467: to the pre-flare level after t$\sim$3000--4000 s.
468: \label{longlongtail}}
469: \epsscale{.7}
470: \plotone{f4new.eps}
471:
472:
473: \clearpage
474: \figcaption{ Comparison between the
475: ACS (top) and PSAC (bottom) light curves. The long lasting
476: emission after the giant flare is not visible in the PSAC, while
477: particle induced variations are visible in both detectors (the
478: narrow pulse at t$\sim-10^4$ s in the ACS is a gamma-ray burst,
479: unrelated to \sgr ).
480: \label{psac}}
481: \epsscale{.7}
482: \plotone{f5new.eps}
483:
484:
485:
486: \end{document}
487: