0710.4590/ms.tex
1: \documentclass[12pt,preprint]{aastex}  
2: %\documentclass[12pt,preprint,natbib]{aastex}  
3: %\documentclass[manuscript]{aastex}
4: %\documentclass[preprint2]{aastex}
5: %\documentclass[preprint2,longabstract]{aastex}
6: 
7: %\usepackage{epstopdf}
8: %\usepackage{graphicx}
9: %\usepackage{pdfsync}
10: %\usepackage{times,mathptm} 
11: %\DeclareGraphicsRule{.ps}{eps}{.eps}{`convert #1 `basename #1 .ps`.eps}
12: %\DeclareGraphicsRule{.ps}{eps}{.ps}{}
13: 
14: %\slugcomment{Draft as of \today.}
15: \shorttitle{The Prompt Gamma-Ray Emission of GRB~070125}
16: \shortauthors{Bellm et al.}
17: 
18: 
19: \newcommand{\Ep}{\ensuremath{E_{peak}}}
20: 
21: 
22: \begin{document}
23:  
24: \title{Observations of the Prompt Gamma-Ray Emission of GRB~070125}
25: 
26: 
27: \author{Eric C. Bellm\altaffilmark{1,2}, Kevin Hurley\altaffilmark{1},
28: Valentin Pal'shin\altaffilmark{3}, Kazutaka Yamaoka\altaffilmark{4}, 
29: Mark S. Bandstra\altaffilmark{1}, Steven E. Boggs\altaffilmark{1,5},
30: Soojing Hong\altaffilmark{6},
31: Natsuki Kodaka\altaffilmark{7},
32: A. S. Kozyrev\altaffilmark{8},
33: M. L. Litvak\altaffilmark{8},
34: I. G. Mitrofanov\altaffilmark{8},
35: Yujin E. Nakagawa\altaffilmark{9},
36: Masanori Ohno\altaffilmark{10},
37: Kaori Onda\altaffilmark{7}, 
38: A. B. Sanin\altaffilmark{8},
39: Satoshi Sugita\altaffilmark{4},
40: Makoto Tashiro\altaffilmark{7}, 
41: V. I. Tretyakov\altaffilmark{8},
42: Yuji Urata\altaffilmark{7}, 
43:  \and
44: Claudia Wigger\altaffilmark{11}
45: }
46: \altaffiltext{1}{UC Berkeley Space Sciences Laboratory, 7 Gauss Way,
47: Berkeley, CA 94720-7450, USA}
48: \altaffiltext{2}{ebellm@ssl.berkeley.edu}
49: \altaffiltext{3}{ Ioffe Physico-Technical Institute, 26 Polytekhnicheskaya,
50: St. Petersburg 194021, Russian Federation}
51: \altaffiltext{4}{Department of Physics and Mathematics, 
52: Aoyama Gakuin University, 5-10-1 Fuchinobe, Sagamihara, 
53: Kanagawa 229-8558, Japan}
54: \altaffiltext{5}{Department of Physics, University of California, Berkeley}
55: \altaffiltext{6}{Laboratory of Physics, College of Science and Technology, 
56: Nihon University, 7-24-1 Narashinodai, Funabashi, Chiba 274-8501, Japan}
57: \altaffiltext{7}{Department of Physics, Saitama University,
58: 255 Shimo-Ohkubo, Sakura-ku, Saitama, Saitama 338-8570, Japan}
59: \altaffiltext{8}{Institute for Space Research, Profsojuznaja 84/32,
60: Moscow 117997, Russia}
61: \altaffiltext{9}{Institute of Physical and Chemical Research (RIKEN), 2-1
62: Hirosawa, Wako, Saitama 351-0198, Japan}
63: \altaffiltext{10}{Institute of Space and Astronautical Science, Japan
64: Aerospace Exploration Agency (ISAS/JAXA), 3-1-1 Yoshinodai, Sagamihara, 
65: Kanagawa 229-8510, Japan}
66: \altaffiltext{11}{Paul Scherrer Institute, Villigen PSI, Switzerland}
67: 
68: 
69: \begin{abstract}
70: The long, bright gamma-ray burst GRB~070125 was localized by the 
71: Interplanetary Network. We present light curves of the prompt gamma-ray
72: emission as observed by Konus-WIND, RHESSI, Suzaku-WAM, and
73: \textit{Swift}-BAT.
74: We detail the results of joint spectral fits with Konus and RHESSI data.
75: The burst shows moderate hard-to-soft evolution in its multi-peaked
76: emission over a period of about one minute.  The total burst fluence as
77: observed by Konus is $1.79 \times 10^{-4}$~erg/cm$^2$ (20~keV--10~MeV).  
78: Using the spectroscopic
79: redshift $z=1.548$, we find that the burst is consistent with the
80: ``Amati'' $E_{peak,i}-E_{iso}$ correlation.  Assuming a jet opening angle
81: derived from broadband modeling of the burst afterglow, GRB~070125 is a 
82: significant outlier 
83: to the ``Ghirlanda'' $E_{peak,i}-E_\gamma$ correlation.  Its
84: collimation-corrected energy release $E_\gamma = 2.5 \times 10^{52}$~ergs 
85: is the largest yet observed.
86: 
87: \end{abstract}
88: 
89: \keywords{gamma-rays: bursts} 
90: 
91: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
92: \section{Introduction} 
93: 
94: The prompt gamma-ray emission of gamma-ray bursts (GRBs) is the most
95: extensively studied aspect of these energetic explosions.  Indeed, for
96: twenty-five years after the discovery of GRBs \citep{kleb73}, the prompt
97: emission was the only GRB observable available.  With the
98: first afterglow observations at longer wavelengths \citep{cost97,vanp97},
99: detailed analysis of burst models became possible.  
100: Presently, the \textit{Swift} satellite is detecting $\sim$ 100 bursts
101: per year, most with rapid localization and followup.
102: 
103: The exact mechanism which produces the prompt gamma-ray emission, with its
104: characteristic smoothly broken power-law spectrum, has not been definitively
105: established.  Recent efforts to correlate burst
106: observables with the intrinsic burst energetics have increased the
107: importance of detailed spectral fitting for localized bursts
108: \citep[for a review, see][]{zhan07c}.
109: Some correlations involve the peak spectral energy \Ep,
110: which is often above the $\sim$150 keV cutoff of the \textit{Swift}
111: Burst Alert Telescope (BAT) passband.
112: 
113: Several current observatories are capable of detailed spectral analysis of
114: GRBs over the full range of \Ep.  Konus-W \citep{apte95} 
115: is a double scintillator instrument on the
116: WIND spacecraft.  The Ramaty High Energy Solar Spectroscopic Imager
117: (RHESSI) is a solar observatory which uses
118: nine germanium detectors to image the Sun at X-ray to gamma-ray energies
119: \citep{lin02}.  RHESSI's detectors are unshielded and receive emission
120: from astrophysical sources like GRBs.  The Wide-Band All-Sky Monitor (WAM)
121: \citep{yama05} aboard Suzaku is the large BGO anticoincidence shield for the 
122: Suzaku Hard X-Ray Detector.  AGILE \citep{trav06}
123: and GLAST \citep{ritz07} will give
124: additional coverage at the energy range of \Ep\ and extend
125: spectral coverage for GRBs up to tens of GeV.
126: 
127: In this paper, we present Konus, RHESSI, and Suzaku observations of the
128: bright GRB~070125.  In Section \ref{sec-obs}, we discuss the observations and
129: the localization of the burst by the IPN.  Section \ref{sec-lc} contains
130: the burst light curves, and in Section \ref{sec-fits} we conduct joint
131: spectral fits to the Konus and RHESSI data.
132: 
133: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
134: 
135: \section{Observations} \label{sec-obs} 
136: 
137: GRB~070125 was observed by six spacecraft in the Interplanetary Network
138: (IPN):  RHESSI, Suzaku WAM, and \textit{Swift}-BAT, all in
139: low Earth orbit; the anticoincidence system of the spectrometer aboard
140: the International Gamma-Ray Astrophysics Laboratory (INTEGRAL), at 0.44
141: light-seconds from Earth; Konus-Wind, at 5.4 light-seconds from Earth;
142: and the High Energy Neutron Detector and Gamma Sensor Head aboard Mars
143: Odyssey 2001, at 1130 light-seconds from Earth.  The two other distant
144: missions in the network, Ulysses and MESSENGER (Mercury Surface, Space
145: Environment, Geochemistry, and Ranging), were off.  Since \textit{Swift} was
146: slewing at the time of the burst, it did not immediately localize it.
147: However, the source appeared in a routine image made after the slew was
148: completed, and its 2.5' radius error circle was consistent with the
149: initial IPN localization \citep{igcn070125}.  Even with more than six
150: minutes of elapsed time since the burst onset, the BAT image detections were
151: highly significant at 8.2 sigma \citep{racu07}.
152: 
153: With only one distant spacecraft, the IPN localized the event
154: to a long, narrow error ellipse 
155: whose area ($3\sigma$) is $\sim$ 1200 square arcminutes, centered at 
156: RA(2000) = 07h 51m 17.85s,  Dec(2000) = +31$^\circ$ 06' 12.78".  
157: The chi-squared for this
158: position is 1.57 for 3 degrees of freedom.  Figure \ref{fig-zoom} shows
159: the central region of the error ellipse, with the BAT 90\% confidence
160: error circle and the optical counterpart.  
161: 
162: Initial spectral fits to the prompt emission were reported for RHESSI by
163: \citet{rgcn070125} and for Konus by \citet{kgcn070125}.  The initial RHESSI
164: best fit model was a cutoff power law (equivalent to the Band function below $E_{break}$, see \S \ref{sec-fits}) with $\alpha$
165: = 1.33 $^{+0.11}_{-0.09}$, 
166: \Ep = 980. $\pm$ 300.~keV, and a 30 keV--10~MeV fluence of 
167: $1.5 \times 10^{-4}$~erg/cm$^2$.  The Konus data were best fit by a 
168: Band function with $\alpha$ =
169: -1.10 $^{+0.10}_{-0.09}$, $\beta$ = -2.08 $^{+0.10}_{-0.15}$, 
170: \Ep\ = 367 $^{+65}_{-51}$~keV.  The measured Konus
171: 20 keV--10~MeV fluence was $(1.74 ^{+0.18}_{-0.15}) \times 10^{-4}$ 
172: erg/cm$^2$.  All errors are 90\% C.L.
173: 
174: Pelangeon and Atteia derived a pseudo-redshift for this burst by using
175: the RHESSI parameters \citep{pela07a} and the Konus values
176: \citep{pela07b}.  These were fairly consistent at 1.6 $\pm$ 0.8 and 
177: 1.3 $\pm$ 0.3 respectively.
178: 
179: \citet{cenk07} reported an optical counterpart at RA(2000)=07h~51m~17.75s, 
180: Dec(2000)~=~+31$^\circ$~09'~04.2".  This counterpart was confirmed
181: by \citet{updi07a} in the R band. 
182: 
183: \citet{racu07b} reported detection by the \textit{Swift} XRT.
184: The XRT position was RA(J2000)  = 7h~51m~18.08s, 
185: Dec(J2000) = +31$^\circ$~09'~02.2", 4.7 arcseconds from the optical
186: transient reported by \citet{cenk07}.
187: 
188: Initial afterglow detections in other bands included 
189: \textit{Swift} UVOT in the UV \citep{mars07b}, radio \citep{vand07}, 
190: and IR \citep{bloo07a}.  Milagro \citep{ding07} observations 
191: of the source took place, but no VHE gamma-ray source was detected. 
192: 
193: \citet{fox07} reported a redshift of $z \geq 1.547$ for GRB~070125 from the 
194: identification of the Mg II doublet.  \citet{cenk08} tightened this
195: estimate to $z = 1.5477 \pm 0.0001$.  Independent observations by
196: \citet{proc07}, reported by \citet{updi08},
197: reveal absorption features which are consistent with $z = 1.548$
198: if identified as C IV and Si IV, and the absence of Lyman absorption
199: features requires $z$ to be near this value.
200: 
201: Observations of the decaying afterglow yielded multiple possibilities for a
202: jet break.  The \textit{Swift}-XRT data showed a
203: possible jet break at 1.35 $\pm$ 0.35 days, but were also
204: consistent with no jet break \citep{burr07,racu07}.  Independent optical
205: observations \citep{mira07,garn07} showed a break in the decay at $t \geq
206: 4$ days.  The non-detection by Chandra \citep{cenk07b} was also consistent
207: with a break occuring after 4 days.  \citet{updi08} used a larger optical 
208: dataset to fit a jet break time of $t = 3.73 \pm 0.52$ days, but cautioned
209: that flaring made the best fit break time dependent on the choice of time
210: intervals.  \citet{chan08} found a best fit break time of $t = 3.8$ days 
211: in a joint optical--X-ray fit.  They suggested that the break
212: might be chromatic, as the X-ray data alone did not require a break, and
213: proposed that inverse Compton emission could create a delay between the
214: optical and X-ray breaks.
215: 
216: Extensive observations of the afterglow of GRB~070125 allowed detailed
217: studies of the unusual burst environment.
218: \citet{cenk08} suggested that the low absorbing column densities 
219: inferred from the afterglow spectra indicate that this long burst took
220: place in a low-density galactic halo.  \citet{chan08} performed detailed
221: broadband fitting of the afterglow, and concluded that the immediate
222: environment of the progenitor was likely high density ($n \sim 50$ cm$^{-3}$
223: for a constant density profile).  They also found evidence that the
224: gamma-ray production efficiency for this burst was unusually high
225: ($\eta_\gamma \sim 0.65$).
226: 
227: 
228: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
229: \section{Light Curve} \label{sec-lc}
230: 
231: Figure \ref{fig-lc} shows the Konus, RHESSI, Suzaku-WAM, and
232: \textit{Swift}-BAT light curves
233: corrected for light travel time between the spacecraft.  The Konus trigger
234: time was T$_{0,KW}$ = 07:20:50.853.  Photon travel time
235: from RHESSI to Konus was 5.197 seconds, from Suzaku to Konus was
236: 5.202 seconds, and from \textit{Swift} to Konus was 5.215 seconds.  
237: 
238: The light curves show a qualitatively similar 
239: multi-peaked structure with roughly four major periods of emission.  
240: The RHESSI data in interval A have a slight but significant feature around
241: $T_0$ + 4 seconds whose origin is unclear.  The bump appears in data from
242: all three detectors used in this study.  Examination of hardness ratios
243: suggests that the bump is softer than the rest of the emission in the
244: interval, but insignificantly so ($\sim 1 \sigma$).  The difference is even
245: more negligable when we consider only data above 65~keV.  Accordingly, the
246: bump (if extraneous) should not meaningfully influence the spectral fits 
247: reported in Section \ref{sec-fits}.
248: 
249: T90 for the Konus light curve was 62.2 $\pm$ 0.8 seconds (20--1150~keV), 
250: for RHESSI 63.0 $\pm$ 1.7 seconds (30 keV--2~MeV), and
251: for Suzaku 55 $\pm$ 2 seconds (50 keV--5~MeV).  
252: In the individual Konus bands, the T90s were 62.8 $\pm$ 1.8 seconds 
253: (G1: 20~keV--75~keV), 61.5 $\pm$ 0.9 seconds (G2: 75~keV--300~keV), 
254: and 60.0 $\pm$ 5.6 seconds (G3: 300~keV--1150~keV).  
255: Uncertainties on all T90s are 1-sigma and
256: were obtained by perturbing the light curves with Poisson noise and finding
257: the new T90 values for 1000 trials.
258: \citet{racu07} report a T90 of 60 seconds for the
259: \textit{Swift}-BAT light curve.  Because \textit{Swift} did not trigger
260: on the burst, no BAT event data were stored. The available rate data contain
261: slew artifacts; accordingly, we do not perform further analysis on the BAT
262: data.
263: 
264: Both Konus and RHESSI observed the 64 millisecond peak flux at T-T$_0 =$ 41.472
265: seconds.  Using the spectral fits from Section \ref{sec-fits}, the peak
266: flux (20~keV--10~MeV)
267: observed by Konus was $(1.85 ^{+0.35}_{-0.36}) \times 10^{-5}$ 
268: erg/cm$^2$/s.  RHESSI observed a peak flux of $(2.92 ^{+0.68}_{-0.63}) 
269: \times 10^{-5}$ erg/cm$^2$/s.  
270: While the RHESSI fluences computed in Section
271: \ref{sec-fits} are lower than those measured by Konus, RHESSI recorded a
272: greater proportion of counts in the 64 ms peak interval, implying a 
273: larger peak flux.  These values are moderately sensitive to background
274: subtraction; the errors quoted are purely statistical.
275: 
276: Figure \ref{fig-hardness} shows the fast time evolution of hardness ratios for 
277: Konus and Suzaku.  The burst shows a general softening trend in time,
278: excepting the period of peak flux in interval C, which has comparable
279: hardness to the initial emission in interval A.
280: 
281: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
282: \section{Spectral Analysis} \label{sec-fits}
283: 
284: We performed spectral analysis for the time intervals given in Table
285: \ref{tab-intervals} using the Konus and RHESSI data.   While spectral
286: data are available from Suzaku, the GRB photons passed through the X-Ray
287: Spectrometer (XRS) dewar before reaching the WAM.  This direction is not
288: well-calibrated for the WAM, in part due to uncertain levels of solid Ne in
289: the dewar.  With the detector response poorly understood, it is impossible
290: to determine effectively the spectral parameters.  Accordingly, we
291: omit the Suzaku data in the spectral fits.
292: 
293: Konus 64-channel spectra are available beginning 0.512
294: seconds before the trigger and are integrated over variable timescales.  
295: The detector
296: response, which is a function only of the burst angle relative to the 
297: instrument axis, is
298: generated from Monte Carlo simulations described by \citet{tere98}.
299: 
300: Because of radiation damage to the RHESSI detectors, only three of the
301: nine detectors (rear segments 1, 7, and 8) were usable for this analysis.
302: While the damaged detectors continue to record significant counts, the
303: effect of the radiation damage on the spectral response has proven difficult
304: to model.
305: 
306: To generate the RHESSI spectral response, we simulated monoenergetic
307: photon beams impinging on a detailed mass model in the Monte Carlo suite
308: MGEANT \citep{stur00}.  
309: The response of each detector changes as RHESSI rotates, so we
310: used a beam geometry with photons generated along 60$^\circ$ arcs in rotation
311: angle.  The resulting sector responses were weighted by the burst light
312: curve and added together. Fit results were not appreciably different when
313: using a simple azimuthally averaged response.
314: The beam made an angle of 165$^\circ$ with the RHESSI rotation axis to
315: match the off-axis angle of the GRB (165.2 degrees).  The simulated photons 
316: had initial energies given by 192 logarithmically-spaced bins from 10~keV to 
317: 30~MeV. 
318: 
319: We conducted the spectral fitting in parallel using the spectral fitting
320: packages XSPEC v11\footnote{http://heasarc.gsfc.nasa.gov/docs/xanadu/xspec/}
321: and ISIS v1.4.3 \citep{citeisis}.  The fit
322: parameters obtained from both programs were identical.  Robust fitting
323: required a lower fit bound of 65 keV for RHESSI, slightly higher than the
324: typical 30 keV lower limit.  Because the GRB was arriving from the extreme
325: rear of RHESSI, the photons passed through the back plate of the RHESSI 
326: cryostat and were hence subject to greater attenuation at low energies.
327: The fit ranges were accordingly 20~keV--10~MeV for Konus and 65~keV--10~MeV
328: for RHESSI.   We rebinned the data to a minimum S/N of 2 before performing
329: the spectral fits.  This rebinning did not greatly affect the best fit
330: parameters.
331: Fluence errors were obtained in ISIS by stepping through a grid of
332: fluence values, refitting the free parameters at each grid point, and
333: monitoring the change in chi-squared.  Since it does not assume that the
334: statistic space is quadratic, this method provides more accurate
335: values for the uncertainties than those generated in XSPEC with the
336: \texttt{flux} command.
337: 
338: The data were well-fit in intervals A-C by a Band function \citep{band93}:
339: \[
340: N_E = \left\{
341: \begin{array}{lr}
342:   A (E/E_{piv})^{\alpha} \exp(-E (2+\alpha)/\Ep) & E<E_{break} \\
343:   B	(E/E_{piv})^{\beta}               & E>E_{break}
344: \end{array}
345: \right.
346: \]
347: with $E_{break} \equiv \Ep \frac{(\alpha - \beta)}{(2+\alpha)}$ and 
348: $B \equiv  A (\frac{(\alpha-\beta) \Ep}{(2+\alpha) E_{piv}})^{\alpha-\beta}
349: \exp(\beta-\alpha)$.  For $\beta < -2$ and
350: %$\alpha > -2$, $\Ep \equiv E_0 (2 + \alpha)$ corresponds to the peak of the
351: $\alpha > -2$, \Ep\ corresponds to the peak of the
352: $\nu F_\nu$ spectrum.  The normalization $A$ has units 
353: photons/(cm$^2$~s~keV), and $E_{piv}$ is here taken to be 100~keV. 
354: For joint fits, the Band function parameters $\alpha$,
355: $\beta$, and \Ep\ were tied for both instruments,
356: but the normalizations were allowed to vary independently.
357: For interval D, the best fit model after grouping was a simple
358: power law.
359: %($E_{piv}$ was 1 keV for the cutoff power law fit.
360: We report the best-fit spectral parameters in Table \ref{tab-fitpars}.
361: Figure \ref{fig-spec} shows the spectra in all intervals for the
362: joint fit. 
363: 
364: For single-instrument fits, the Konus data provide superior fit quality
365: and better constraint on the fit parameters, due in part to having about
366: six times more usable counts.  The fit fluence, $\alpha$,
367: and $\beta$ are generally consistent between RHESSI and Konus.  However,
368: the RHESSI data prefer higher \Ep, matching the best fit Konus values 
369: only at the lowest end of rather large error bars.  
370: The Konus fit parameters for the total burst match well the initial values 
371: reported via the GCN \citep{kgcn070125}.  The RHESSI fit \Ep\ typically  is lower here
372: than in the value reported in the GCN \citep{rgcn070125}, but this
373: difference is expected from fitting using the Band function rather than a
374: cutoff power law \citep{band93}.
375: 
376: The spectral parameters for the joint fits are consistent with
377: the Konus-only values.  There are slight improvements
378: in the uncertainties of some of the fit parameters at a cost of an increase
379: in the chi squared.  The RHESSI residuals in the joint fit (Figure
380: \ref{fig-spec}) show a characteristic deviation pattern, indicating that
381: the instruments disagree on the spectral shape.  The Konus data dominate
382: the fit because of their better statistical quality.  The residuals
383: for the RHESSI-only fits do not show any systematic deviation.
384: 
385: For intervals A, and B, the ratio of the RHESSI normalization to the
386: Konus normalization is 0.88.  For interval C, the ratio is 0.95.
387: Characteristic uncertainties for the ratio are 0.04-0.05.  
388: In interval D, the ratio for the power-law fit is
389: 0.84$_{-0.13}^{+0.14}$.
390: Absolute normalizations in photons/(cm$^2$~s~keV) 
391: using $E_{piv} = 100$ keV for the total interval were 
392: (2.50 $^{+0.18}_{-0.15}) \times 10^{-2}$ (Konus) and 
393: (2.25 $^{+0.16}_{-0.14}) \times 10^{-2}$ (RHESSI).
394: 
395: The time-resolved fits show a moderate hard-to-soft evolution.  \Ep\ is
396: largest in the initial broad pulse (539 keV) and then softens to 355 keV in
397: interval B.  The sharp pulse in interval C has a harder spectrum (418 keV).  
398: While the statistically preferred model for the S/N grouped data in interval D 
399: is a simple power
400: law, fitting a cutoff power-law with the Konus data to 2~MeV gives an
401: estimate of \Ep\ at 220~keV.
402: The high-energy spectral index $\beta$ softens monotonically through intervals 
403: A-C.
404: 
405: 
406: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
407: \section{Energetics}
408: 
409: Knowledge of the burst redshift $z=1.548$ makes it possible to draw
410: conclusions about the overall burst energetics.
411: We assume a standard flat cold dark matter cosmology
412: ($\Lambda$CDM), with parameters ($\Omega_\Lambda$, $\Omega_M$, $H_0$) 
413: = (0.761, 0.239, 73~km~s$^{-1}$~Mpc$^{-1}$), 
414: consistent with results from WMAP year 3 \citep{sper07}
415: and large scale structure traced by luminous red galaxies \citep{tegm06}.
416: This particular set of values corresponds to the ``Vanilla model'' of
417: \citet{tegm06}.
418: 
419: Extrapolating to a GRB rest-frame
420: energy band of 1~keV--10~MeV, the isotropic emitted energy for the total
421: burst is $(9.59\pm0.39) \times 10^{53}$~ergs (Konus) and 
422: $(8.67\pm0.38) \times 10^{53}$~ergs (RHESSI) for the joint fit.  
423: Because we allow independent normalizations for the Konus and RHESSI data,
424: we obtain two values of $E_{iso}$ from the joint fit, one for each
425: instrument.
426: 90\% C.L. errors are obtained by exploration of the parameter space as for
427: the fluence; we neglect uncertainty in $z$.  These values, together with
428: the spectral fit of the time-integrated spectrum, are
429: consistent with the ``Amati relation'' correlating $E_{iso}$ with the intrinsic
430: peak energy of the spectrum in the GRB rest frame $E_{peak,i}$
431: \citep{amat02,amat06,ghir08}.  We plot GRB~070125 
432: in the $E_{peak,i} - E_{iso}$ plane in Figure \ref{fig-amati}.
433: 
434: Because the best fit Band function has a hard tail ($\beta \sim -2$), the
435: fluence integral is sensitive to the choice of upper energy bound.
436: If we use the observed energy band 20~keV--10~MeV,
437: corresponding to a GRB frame band of 50~keV--25.5~MeV, 
438: the fluence is 14\% larger than that in the usual bolometric band.
439: For consistency with previous works, we
440: will use the 1~keV--10~MeV band for bolometric estimates.
441: 
442: Converting the 64~ms peak fluxes reported in Section \ref{sec-lc} to
443: bolometric peak luminosities using the best fit Band parameters,
444: we find peak luminosities of (2.59
445: $^{+0.36}_{-0.37}) \times 10^{53}$~ergs/s for Konus and 
446: (4.25 $^{+0.87}_{-0.79}) \times 10^{53}$~ergs/s for RHESSI.
447: 
448: \citet{chan08} performed a broadband fit to afterglow data for GRB~070125.
449: They determined a jet opening angle of $13.2 \pm 0.6$ degrees in their most
450: plausible scenario (a radiative fireball expanding into a constant density 
451: (ISM) medium and emitting via synchrotron and inverse Compton channels).
452: This jet angle was consistent with that inferred from the jet break time
453: $\sim 3.7$ days and an emission radius derived from radio scintillation.  
454: For the collimation-corrected energy
455: $E_\gamma = (1 - \cos \theta) E_{iso}$, we find $E_\gamma = 
456: (2.52 \pm 0.24, 2.27 \pm 0.22) \times 10^{52}$ ergs for (Konus, RHESSI).
457: These values are the largest yet recorded for a burst with measured \Ep\ 
458: (c.f. \citealp{frai06}; \citealp{koce08} also reported lower limits on
459: $E_{\gamma}$ greater than $10^{52}$ ergs for several \textit{Swift} bursts 
460: using the time of the last XRT observation).
461: %\citep[cf.][]{frai06}.
462: We plot GRB~070125 
463: in the $E_{peak,i} - E_{\gamma}$ plane in Figure \ref{fig-ghirlanda} to examine
464: its consistency with the ``Ghirlanda'' $E_{peak,i} - E_\gamma$ correlation 
465: \citep{ghir04,ghir07}. 
466: 
467: In Figures \ref{fig-amati} and \ref{fig-ghirlanda}, we also overplot the best
468: fit correlation lines.  A number of fitting approaches have been considered 
469: in the literature in an effort to account for the apparent
470: extra-statistical spread of the points about the 
471: correlation \citep[for a review, see][]{ghir08}.  We have followed
472: \citet{ghir08} in presenting two least squares fits, one in which the data
473: points are unweighted and a second in which the errors on both axes are
474: considered.  After the fit, we estimate the dispersion of the points
475: perpendicular to the best fit correlation line
476: using the square root of the bias-corrected sample variance.
477: 
478: GRB~070125 is quite consistent with the Amati relation:
479: including it in the fit makes negligible changes in the best-fit correlation
480: slope or the logarithmic dispersion ($0.20$ dex).  However, it is a
481: $5.0\sigma$ outlier to the Ghirlanda correlation fitted without it, using the sample
482: dispersion to estimate $\sigma$.  
483: Including GRB~070125 in an unweighted fit of the bursts in the Ghirlanda
484: sample, the overall dispersion increases to 0.13~dex (from 0.09~dex), and 
485: GRB~070125 remains a $2.8\sigma$ outlier.
486: 
487: The unusual environment of GRB~070125 is responsible for its high value of
488: $E_{\gamma}$.  In particular, the jet opening angle of $13.2 \pm 0.6$
489: degrees derived by \citet{chan08} is larger than all of those
490: presented by \citet{ghir07}.  
491: Retaining the $3.7 \pm 0.5$ day jet break time well-established
492: in the optical \citep{updi08,chan08}, we may derive the jet opening angle
493: assuming adiabatic emission and more conventional parameters \citep{sari99}.
494: Assuming an ISM profile with circumburst density $n = 3$ cm$^{-3}$ and a
495: gamma-ray production efficiency of $\eta_\gamma = 0.2$, the corresponding jet
496: opening angle is $\theta = 5.6 \pm 0.3$ degrees for 
497: Konus.  The resulting collimation-corrected energy would be $E_{\gamma} = (4.6
498: \pm 0.5) \times 10^{51}$~ergs, only $0.8\sigma$ from the best-fit correlation
499: omitting GRB~070125.
500: 
501: \section{Discussion}
502: 
503: While GRB~070125 had a large measured prompt gamma-ray fluence, its spectral
504: properties are unremarkable.  The values of the best-fit spectral parameters
505: are similar to those observed for other bright bursts
506: \citep[e.g.,][]{kane06}, and
507: the spectral evolution observed is similarly common.  The environment
508: of GRB~070125 is unique, however \citep{cenk08, chan08, updi08}, requiring
509: a broad jet opening angle in broadband afterglow models \citep{chan08}.
510: After collimation correction, GRB~070125 has the most energetic prompt
511: emission yet observed and is a significant outlier to the correlation between
512: peak energy and $E_{\gamma}$.
513: 
514: %Granting the existance of these correlations, however,
515: GRB~070125 appears to
516: weaken the claim that the Ghirlanda correlation has low dispersion.  
517: GRB~070125 is not a ``recognizable'' outlier to the Ghirlanda relation
518: in the sense of \citet{ghir07}, as
519: it is highly consistent with the Amati relation.
520: Its jet parameters have been derived from a rich and well-sampled afterglow
521: dataset.  While the circumburst environment of this GRB is unusually dense,
522: this only highlights the assumption of a fairly narrow range of efficiency and
523: density parameters for the majority of GRBs where broadband modeling of the
524: afterglow has not been possible.  The true dispersion of the correlation may in
525: fact be larger.  
526: 
527: %The physical significance of the GRB luminosity correlations has been 
528: %questioned (e.g., \citealp{butl07,butl08}; but see \citealp{ghir08}).  
529: %Selection effects due to
530: %detector thresholds influence the sample, and transforming parameters to the
531: %GRB rest frame can create illusory correlations.
532: The physical significance of GRB spectrum--energy correlations has been
533: questioned \citep[e.g.][]{butl07,butl08}.  In particular, 
534: detector trigger thresholds affect burst detection, and more complex
535: selection effects govern the measurement of peak energies,
536: redshifts, and afterglow breaks.  These effects can influence the sample of
537: GRBs with known redshift, $E_{peak,i}$, and $E_{\gamma}$.  \citet{ghir08}
538: examined the effect of trigger and spectral analysis thresholds in the
539: \Ep--fluence plane, finding that the \textit{Swift}-detected burst sample
540: was truncated by the spectral analysis threshold.  Neither threshold 
541: truncated the pre-\textit{Swift} burst sample.  
542: %and the authors speculate that other selection effects determine the
543: %extent of the pre-\textit{Swift} sample.
544: 
545: We were unable to confirm the source of the systematic shift
546: in \Ep\ and fluence between the two instruments for this burst.
547: Minor radiation damage was becoming noticeable in
548: RHESSI detector 8 near the time of this work, mostly below the 65 keV cut
549: utilized here.  
550: It is also possible that the Monte Carlo simulation
551: of the RHESSI response is less accurate for such extreme off-axis angles,
552: where a greater number of interactions with the cryostat may be expected.  
553: 
554: %\citet{krim06} found for a cutoff power-law
555: %fit to combined \textit{Swift} and Konus data for GRB 050717 a best fit
556: %value of \Ep$=2401 +781/-568$.  A joint \textit{Swift}-RHESSI fit to the
557: %same burst found \Ep$=1950 \pm 350$ keV \citep{bellhead06}.  
558: Our previous work had found excellent agreement in all fit parameters
559: for independent RHESSI and Konus spectral fits for GRB 051103 and 
560: GRB 050717.  For the short GRB
561: 051103, Konus found $\Ep = 1920 \pm 400$~keV and a 20~keV--10~MeV fluence of
562: $4.4 \pm 0.5 \times 10^{-5}$~ergs/cm$^2$ \citep{kgcn051103, fred07}.  A RHESSI
563: fit yielded $\Ep = 1930 \pm 340$~keV and 20~keV--10~MeV fluence of $4.5 \times
564: 10^{-5}$~erg/cm$^2$ \citep{bellhead06}.
565: \citet{krim06} found for a cutoff power-law
566: fit to Konus data for GRB 050717 a best fit
567: value of \Ep $=2101 ^{+1934}_{-830}$~keV.  A RHESSI fit to the same burst found
568: \Ep $=1550 ^{+510}_{-370}$~keV \citep{wigg06}.  
569: Those bursts had RHESSI off-axis angles of 97 and 110 degrees, respectively.
570: 
571: Joint spectral fits to \textit{Swift}-BAT and RHESSI data for 25 bursts
572: co-observed by the two instruments between December 2004 and December 2006
573: indicated that no offset in response normalization was needed for the two
574: instruments \citep{bellsf08}.
575: However, for two of three bursts occurring during or after December 2006, 
576: the RHESSI data showed a significant deficit relative to \textit{Swift}-BAT.  
577: The RHESSI polar angles for all three late bursts were between 90 and 110 
578: degrees.
579: These fits were conducted using only detectors 1 and 7, which do not appear
580: to have radiation damage in background spectra during this interval.  
581: Nonetheless, these results suggest that the observed offset in the RHESSI and 
582: Konus fit parameters found here is more likely a consequence of increased
583: radiation damage in the RHESSI detectors than a geometric effect or a
584: generic offset in the RHESSI simulations.
585: 
586: Future analysis of archival bursts may help identify the source of any
587: systematic effects present here.  It is clear, however, that joint fits
588: between instruments capable of constraining the full range of \Ep\ are
589: valuable in providing the most accurate and precise determination of the fit
590: parameters. 
591: 
592: 
593: \acknowledgements 
594: This work was supported by Swift AO-2 GI grant NNG06GH58G, ``Completing
595: Swift GRB Energy Spectra with Konus and RHESSI'' and by the A0-3 grant 
596: NNX07AE86G.  KH is grateful for IPN support under JPL Contract 1282043, and 
597: NASA grants NNG06GI896, NNX06AI36G, NNG06GE69G, and NAG5-13080.  
598: The Konus-Wind experiment is supported by a Russian Space Agency 
599: contract and RFBR grant 06-02-16070.
600: We thank Bob Lin, David Smith, and Dieter Hartmann for helpful comments.
601: 
602: {\it Facilities:} \facility{RHESSI}, \facility{WIND (Konus)}
603: 
604: 
605: 
606: \bibliography{grb}
607: \bibliographystyle{apj}
608: 
609: \clearpage
610: 
611: \input{tab1} 
612: \input{tab2} %\newpage
613: 
614: %\caption{The IPN $3\sigma$ confidence error ellipse derived from a six 
615: %spacecraft triangulation of GRB~070125.  The asterisk indicates the center of
616: %the ellipse.}
617: 
618: \begin{figure}
619: \plotone{f1.eps}
620: \caption{The central region of the IPN error ellipse, showing
621: the 2.5' radius BAT $3\sigma$ error circle, the optical
622: transient source first reported by \citet{cenk07}, and the
623: center of the ellipse.  The optical source lies 0.048 degrees
624: from the center of the IPN ellipse, on the 87\% confidence contour.}
625:  \label{fig-zoom}
626: \end{figure}
627: 
628: 
629: 
630: \begin{figure} 
631: \plotone{f2.eps}
632: \caption{GRB~070125 light curve for Konus, RHESSI (rear
633: segments 1, 7, and 8 only), Suzaku-WAM, and \textit{Swift}-BAT.  The
634: 	light curves are adjusted for time of flight, with T$_0$ given
635: 		in \S \ref{sec-lc}.  The dashed vertical lines delimit the
636: 		intervals used in the time-resolved spectral fits (\S \ref{sec-fits}).
637: The \textit{Swift} light curve plotted contains all counts observed by
638: \textit{Swift}; in particular, it is not mask-tagged and therefore contains
639: slew artifacts.  
640: }
641: \label{fig-lc}
642: \end{figure}
643: 
644: \begin{figure} 
645: \plotone{f3.eps}
646: \caption{Hardness ratios for GRB~070125.  The Konus energy bands are G1
647: (20--75~keV), G2 (75--300~keV), and G3 (300--1150~keV).
648: The Suzaku hardness ratio plotted here is (520--5000~keV)/(50--240~keV).  
649: %Hardness
650: %ratios of less than zero indicate imperfect background subtraction.  
651: Dashed
652: lines indicate spectral fitting intervals, as in Figure \ref{fig-lc}.
653: Points near $\sim$20 seconds 
654: which are off-scale for the Konus G3/G2 ratio are consistent with
655: zero---there is negligible emission in the G3 band at this time.
656: }
657: \label{fig-hardness}
658: \end{figure}
659: 
660: \begin{figure}
661: \plotone{f4.eps}
662: \caption{Count spectra and residuals for the joint fits.  
663: The Konus data and models are colored black, while the RHESSI data 
664: and models are gray.
665: RHESSI data, model, and errors are divided by 10 
666: in the count spectra plots for clarity.
667: The overplot models differ only in normalization.}
668: \label{fig-spec}
669: \end{figure}
670: 
671: \begin{figure}
672: \plotone{f5.eps}
673: \caption{$E_{peak,i} - E_{iso}$ correlation including GRB~070125.  Values of
674: $E_{peak,i}$ (the intrinsic peak energy in the burst rest frame)
675: and $E_{iso}$ are for the joint Konus-RHESSI fit.  Since the
676: normalization was allowed to vary between the two instruments, we plot
677: separate points for Konus and RHESSI to
678: indicate the corresponding values of $E_{iso}$.  The Konus data point has the 
679: larger value of $E_{iso}$.  Data for other bursts are
680: from Table 1 of \citet{ghir08}, 
681: plotted using the cosmology of this paper ($\Omega_m = 0.239$, $\Omega_\Lambda
682: = 0.761$, $h=0.730$).  
683: The best-fit line for the unweighted data points, omitting GRB~070125, 
684: is overplot with a solid line; the $2\sigma$ scatter about that fit is 
685: indicated with dashed lines.  The dash-dotted line is the best fit when
686: the data points are weighted by their errors on both axes, again 
687: omitting GRB~070125---see text for details.
688: }
689: \label{fig-amati}
690: \end{figure}
691: 
692: \begin{figure}
693: \plotone{f6.eps}
694: \caption{$E_{peak,i} - E_{\gamma}$ correlation including GRB~070125. 
695: Symbols and overplot fit lines are as in Figure \ref{fig-amati}; the Konus data
696: point has the larger value of $E_{\gamma}$.  We also plot the best unweighted
697: fit line including GRB~070125 with a short dotted line.
698: Data for other bursts are from Table 1 of \citet{ghir07}, 
699: assuming an ISM density profile and plotted using the cosmology of this paper. 
700: Bursts with only lower limits on $E_{\gamma}$ were omitted from the fit.
701: }
702: \label{fig-ghirlanda}
703: \end{figure}
704: 
705: \end{document}
706: