Abstract

GRB 080810 was one of the first bursts to trigger both Swift and the Fermi Gamma-ray Space Telescope. It was subsequently monitored over the X-ray and UV/optical bands by Swift, in the optical by Robotic Optical Transient Search Experiment (ROTSE) and a host of other telescopes, and was detected in the radio by the Very Large Array. The redshift of z= 3.355 ± 0.005 was determined by Keck/High Resolution Echelle Spectrometer (HIRES) and confirmed by RTT150 and NOT. The prompt gamma/X-ray emission, detected over 0.3–103 keV, systematically softens over time, with Epeak moving from ∼600 keV at the start to ∼40 keV around 100 s after the trigger; alternatively, this spectral evolution could be identified with the blackbody temperature of a quasi-thermal model shifting from ∼60 to ∼3 keV over the same time interval. The first optical detection was made at 38 s, but the smooth, featureless profile of the full optical coverage implies that this is originated from the afterglow component, not from the pulsed/flaring prompt emission.

Broad-band optical and X-ray coverage of the afterglow at the start of the final X-ray decay (∼8 ks) reveals a spectral break between the optical and X-ray bands in the range of 1015–2 × 1016 Hz. The decay profiles of the X-ray and optical bands show that this break initially migrates blueward to this frequency and then subsequently drifts redward to below the optical band by ∼3 × 105 s. GRB 080810 was very energetic, with an isotropic energy output for the prompt component of 3 × 1053 and 1.6 × 1052 erg for the afterglow; there is no evidence for a jet break in the afterglow up to 6 d following the burst.

1 INTRODUCTION

Gamma-ray bursts (GRBs) emit large amounts of energy across the full range of the electromagnetic spectrum, so obtaining panchromatic data allows a more thorough investigation of the processes involved. Even with the rapid slewing capability of Swift (Gehrels et al. 2004), few bursts have good, simultaneous multiband observations of the prompt emission. On rare occasions, a trigger on a precursor has allowed the X-ray Telescope and UV/Optical Telescope (XRT and UVOT; Burrows et al. 2005a; Roming et al. 2005) to be on target for (most of) the main event (e.g. GRB 060124 –Romano et al. 2006; GRB 061121 –Page et al. 2007), whereas, in about 10 per cent of Swift-detected GRBs, the duration of the burst has been such that the narrow-field instruments have been able to observe the tail-end of the prompt emission (e.g. GRB 070616 –Starling et al. 2008; GRB 060607A –Ziaeepour et al. 2008). Thus, whenever prompt emission is detected by more than just gamma-ray instruments, the GRB becomes potentially more interesting and informative, allowing models to be tested more rigorously.

In the case of GRB 080810, bright, highly variable emission detected by the Burst Alert Telescope (BAT; Barthelmy et al. 2005) continued for more than 100 s, with the XRT and UVOT on target and collecting data from ∼80 s (Page et al. 2008a); a bright source was detected in both the X-ray and optical bands. The gamma-ray emission was also observed by the Fermi1 GRB Monitor (Meegan et al. 2008), making GRB 080810 one of the first GRBs to trigger both it and the BAT; Konus–Wind also detected this burst (Sakamoto et al. 2008b). Many telescopes reported the detection of the optical afterglow (see Page et al. 2008b for a summary) while Prochaska et al. (2008) announced a tentative redshift of z= 3.35 using the Keck/HIRES spectrograph, which was then confirmed by RTT150 (Burenin et al. 2008) and Nordic Optical Telescope (NOT; de Ugarte Postigo et al. 2008a). The Very Large Array also detected the radio afterglow at 8.46 GHz (Chandra & Frail 2008) 3–4 d after the burst.

Section 2 presents the observations and preliminary results, covering the gamma-ray (Section 2.1), X-ray (Section 2.2) and optical (Section 2.3) bands. Section 3 discusses the redshift determination (Section 3.1), using a thermal interpretation to provide an alternative spectral fit (Section 3.2) and the multiwavelength, broad-band modelling (Section 3.3). The conclusions are given in Section 4. Throughout this paper, we follow the convention of Fν,t∝ν−βt−α (photon spectral index Γ=β+ 1), where Fν,t is the flux density, ν is the observed frequency and t is the time since the onset of the burst. Errors are given at 90 per cent confidence unless otherwise stated.

2 OBSERVATIONS AND ANALYSES

Swift and Fermi both triggered on GRB 080810 at 13:10:12 ut on 2008 August 10 (this time is used as T0 throughout the paper), with the Swift-XRT and UVOT detecting the afterglow as soon as they were on target. The best Swift position is that determined from the UVOT refined analysis (Holland & Page 2008): forumla (J2000; estimated uncertainty of 0.6 arcsec), consistent with the Robotic Optical Transient Search Experiment III (ROTSE-III; Rykoff 2008) and NOT (de Ugarte Postigo et al. 2008a) localizations.

2.1 Gamma-rays

GRB 080810 was clearly detected by the Swift–BAT (Sakamoto et al. 2008a) over all energy bands (Fig. 1), although the emission above about 100 keV is weaker than at the lower energies. The T90 (15–150 keV) is 108 ± 5 s (estimated error including systematics); the fluence over this time is 4.2 × 10−6 erg cm−2. The slow rise of the emission, over which there are multiple, overlapping peaks, started about 20 s (observer's frame) before the trigger.

Figure 1

The Swift–BAT (top four panels) and Fermi–GBM (bottom five panels) light curves, over their standard energy bands. Note that the ordinate scale for the 100–150 keV BAT curve (fourth panel) and the 300–1000 keV NaI curve (eighth panel) are different from the lower energy bands because the emission was much weaker. The BGO light curve is shown down to 100 keV, but spectral analysis is performed only for data >200 keV. The Swift light curves are in units of count s−1 (fully illuminated detector)−1, while the Fermi curves are count s−1.

Figure 1

The Swift–BAT (top four panels) and Fermi–GBM (bottom five panels) light curves, over their standard energy bands. Note that the ordinate scale for the 100–150 keV BAT curve (fourth panel) and the 300–1000 keV NaI curve (eighth panel) are different from the lower energy bands because the emission was much weaker. The BGO light curve is shown down to 100 keV, but spectral analysis is performed only for data >200 keV. The Swift light curves are in units of count s−1 (fully illuminated detector)−1, while the Fermi curves are count s−1.

Konus–Wind also detected GRB 080810, but observed the burst in waiting mode (Sakamoto et al. 2008b), meaning only 3-channel spectra were available, covering 20 keV to 1 MeV.

The Fermi Gamma-Ray Burst Monitor (GBM) (von Kienlin et al. 2004; Meegan et al. 2009) triggered on the burst as well (Meegan et al. 2008), identifying the same pulses as did the BAT (Fig. 1). Unfortunately, the burst was outside the field of view of the Large Area Telescope (LAT), which is sensitive to higher energy photons than GBM (0.02–300 GeV, compared to 8 keV–40 MeV).

The Fermi–GBM NaI detectors provide similar T90 estimates to that measured by the BAT, while the higher energy Bismuth Germanate scintillators (BGO) durations are shorter; values are given in Table 1. This is a consequence of the hard-to-soft evolution (see, e.g. Table 2), combined with the different sensitivities of the instruments.

Table 1

T90 measurements over a range of energy bands. The longer duration at lower energies shows that the emission softened over time.

Instrument Band (keV) T90 (s) 
Average 
Swift–BAT 15–150 108 ± 4 
Fermi–NaI 10–1000 113 ± 2 
Fermi–BGO 200–20 000 73 ± 7 
Energy-sliced 
Swift–BAT 15–100 105 ± 4 
Swift–BAT 100–150 55 ± 9 
Fermi–NaI 10–50 107 ± 1 
Fermi–NaI 50–100 81 ± 1 
Fermi–NaI 100–300 73 ± 1 
Instrument Band (keV) T90 (s) 
Average 
Swift–BAT 15–150 108 ± 4 
Fermi–NaI 10–1000 113 ± 2 
Fermi–BGO 200–20 000 73 ± 7 
Energy-sliced 
Swift–BAT 15–100 105 ± 4 
Swift–BAT 100–150 55 ± 9 
Fermi–NaI 10–50 107 ± 1 
Fermi–NaI 50–100 81 ± 1 
Fermi–NaI 100–300 73 ± 1 
Table 2

Power law (PL), cut-off power law (CutPL) and quasi-thermal (PL+Therm) fits to time-sliced spectra from GBM and BAT; the Fermi NaI n7, n8, nb and BGO b1 detectors were fitted simultaneously each time. The Epeak for the BAT single power-law fits are estimated from the relation given by Sakamoto et al. (2009); see text for details.

Time (s since trigger) Detectors Model Γ Epeak (keV) BB kT (keV) χ2/d.o.f. 
0–10 BAT PL 0.92 ± 0.13 – – 66/56 
0–10 GBM PL 1.48 ± 0.04 – – 582/488 
0–10 GBM CutPL 1.06+0.13−0.17 807+1113−470 – 509/485 
0–10 GBM+BAT PL 1.43 ± 0.03 – – 694/544 
0–10 GBM+BAT CutPL 0.95+0.12−0.13 602+537−252 – 580/543 
0–10 GBM+BAT PL+Therm 1.63 ± 0.11 – 62 ± 9 597/542 
10–20 BAT PL 1.24 ± 0.10 159+738−92 (est.) – 30/56 
10–20 GBM PL 1.56 ± 0.04 – – 568/486 
10–20 GBM CutPL 1.07+0.15−0.18 346+378−160 – 505/485 
10–20 GBM+BAT PL 1.53 ± 0.03 – – 622/544 
10–20 GBM+BAT CutPL 1.09+0.11−0.12 353+275−139 – 536/543 
10–20 GBM+BAT PL+Therm 1.69+0.10−0.09 – 46+8−8 553/542 
20–27 BAT PL 1.23 ± 0.11 162+757−99 (est.) – 39/56 
20–27 GBM PL 1.53 ± 0.04 – – 538/486 
20–27 GBM CutPL 1.11+0.13−0.14 434+425−190 – 476/485 
20–27 GBM+BAT PL 1.50 ± 0.03 – – 596/544 
20–27 GBM+BAT CutPL 1.12+0.10−0.11 452+356−176 – 516/543 
20–27 GBM+BAT PL+Therm 1.64+0.09−0.08 – 53+11−9 538/542 
40–53 BAT PL 1.60 ± 0.10 82+160−32 (est.) – 60/56 
40–53 GBM PL 1.69 ± 0.06 – – 620/486 
40–53 GBM CutPL 1.27+0.23−0.24 188+464−114 – 602/485 
40–53 GBM+BAT PL 1.67 ± 0.05 – – 682/544 
40–53 GBM+BAT CutPL 1.41+0.13−0.16 230+538−175 – 664/543 
40–53 GBM+BAT PL+Therm 1.78+0.15−0.11 – 28+15−9 668/542 
100–106 BAT PL 1.71 ± 0.19 69+97−52 (est.) – 55/56 
100–106 GBM PL 2.14+0.28−0.23 – – 600/486 
100–106 GBM CutPL 0.99+0.83−1.47 49+472−49 – 593/485 
100–106 GBM+BAT+XRT PL 1.46 ± 0.02 – – 904/585 
100–106 GBM+BAT+XRT CutPL 1.05+0.07−0.08 39+12−9 – 725/584 
100–106 GBM+BAT+XRT PL+Therm 1.46 ± 0.03 – 2.6+0.4−0.7 797/583 
Time (s since trigger) Detectors Model Γ Epeak (keV) BB kT (keV) χ2/d.o.f. 
0–10 BAT PL 0.92 ± 0.13 – – 66/56 
0–10 GBM PL 1.48 ± 0.04 – – 582/488 
0–10 GBM CutPL 1.06+0.13−0.17 807+1113−470 – 509/485 
0–10 GBM+BAT PL 1.43 ± 0.03 – – 694/544 
0–10 GBM+BAT CutPL 0.95+0.12−0.13 602+537−252 – 580/543 
0–10 GBM+BAT PL+Therm 1.63 ± 0.11 – 62 ± 9 597/542 
10–20 BAT PL 1.24 ± 0.10 159+738−92 (est.) – 30/56 
10–20 GBM PL 1.56 ± 0.04 – – 568/486 
10–20 GBM CutPL 1.07+0.15−0.18 346+378−160 – 505/485 
10–20 GBM+BAT PL 1.53 ± 0.03 – – 622/544 
10–20 GBM+BAT CutPL 1.09+0.11−0.12 353+275−139 – 536/543 
10–20 GBM+BAT PL+Therm 1.69+0.10−0.09 – 46+8−8 553/542 
20–27 BAT PL 1.23 ± 0.11 162+757−99 (est.) – 39/56 
20–27 GBM PL 1.53 ± 0.04 – – 538/486 
20–27 GBM CutPL 1.11+0.13−0.14 434+425−190 – 476/485 
20–27 GBM+BAT PL 1.50 ± 0.03 – – 596/544 
20–27 GBM+BAT CutPL 1.12+0.10−0.11 452+356−176 – 516/543 
20–27 GBM+BAT PL+Therm 1.64+0.09−0.08 – 53+11−9 538/542 
40–53 BAT PL 1.60 ± 0.10 82+160−32 (est.) – 60/56 
40–53 GBM PL 1.69 ± 0.06 – – 620/486 
40–53 GBM CutPL 1.27+0.23−0.24 188+464−114 – 602/485 
40–53 GBM+BAT PL 1.67 ± 0.05 – – 682/544 
40–53 GBM+BAT CutPL 1.41+0.13−0.16 230+538−175 – 664/543 
40–53 GBM+BAT PL+Therm 1.78+0.15−0.11 – 28+15−9 668/542 
100–106 BAT PL 1.71 ± 0.19 69+97−52 (est.) – 55/56 
100–106 GBM PL 2.14+0.28−0.23 – – 600/486 
100–106 GBM CutPL 0.99+0.83−1.47 49+472−49 – 593/485 
100–106 GBM+BAT+XRT PL 1.46 ± 0.02 – – 904/585 
100–106 GBM+BAT+XRT CutPL 1.05+0.07−0.08 39+12−9 – 725/584 
100–106 GBM+BAT+XRT PL+Therm 1.46 ± 0.03 – 2.6+0.4−0.7 797/583 

Time-sliced spectra from both the BAT and Fermi (NaI and BGO) detectors, covering 0–10, 10–20, 20–27, 40–53 and 100–106 s after the trigger, were fitted with single and cut-off power laws and the results given in Table 2; the thermal fits are discussed in Section 3.2. The useful energy ranges for the BAT, NaI and BGO spectral fitting are 15–150, 8–1000 and 200–40 000 keV, respectively. The spectra and models were extensively tested in both xspec (Arnaud 1996) and the Fermi software package rmfit (Mallozzi, Preece & Briggs 2005); these methods provided consistent results and so the numbers given in this paper are those from xspec. Because BAT spectra are created already background-subtracted and have non-Poissonian errors, Cash/Castor statistics cannot be used;2 hence all results were obtained using χ2 statistics.

Using the F-test, the Band function (Band et al. 1993) is not a statistical improvement over the simpler cut-off power law, with β, the higher energy index, unconstrained in each case. For the fits presented here, the normalizations of the GBM detectors were tied together at a value of 1.23 relative to the normalization of the BAT, which was itself fixed at unity. This constant of normalization for the GBM was determined by simultaneously fitting all five intervals of data, but allowing the other fit parameters to vary between the intervals. The cross-calibration between Swift–BAT and Fermi–GBM is discussed in Section 2.1.1.

Sakamoto et al. (2009) found a correlation between the photon index from a simple power-law fit to a BAT GRB spectrum and Epeak, thus allowing an estimate of the peak energy from the limited BAT energy bandpass. The correlation for a source 15° off-axis (GRB 080810 was approximately 20° off-axis, so this is the closest of the relationships), log(Epeak) = 3.184–0.793Γ (where 1.3 ≤Γ≤ 2.3), was used to produce the estimated Epeak values given in Table 2 (marked as ‘est.’); BAT slewed during the interval 12–64 s after the trigger, so Epeak was estimated for the last (100–106 s) spectrum using the on-axis approximation. The spectrum extracted for 0–10 s after the trigger has too hard a photon index to allow the use of this approximation, while 10–20 s is just consistent with the range. The 1σ spread of the relation has been included in the error estimation. These estimated peak energies are consistent with those found from jointly fitting the Fermi and Swift data, although the error bars on the measurements are very large.

The numbers show that the peak energy moves to lower values over time; this is demonstrated graphically in Fig. 2. The single power-law fits also show that Γ increased (softened) until at least 53 s. The spectrum from 100–106 s covers a flare in the XRT emission, which explains the harder (flatter) photon index (see Fig. 6).

Figure 2

Epeak, measured from joint FermiSwift fits, moves to lower energies with time.

Figure 2

Epeak, measured from joint FermiSwift fits, moves to lower energies with time.

Figure 6

From top to bottom: the joint BAT–XRT flux light curve, in erg cm−2 s−1 over 0.3–10 keV (black histogram – BAT; grey crosses – XRT), during the first orbit; the white and v-band UVOT light curve (the data before the vertical line are white, normalized to align with the v band; afterwards –v); unfiltered ROTSE-III light curves; the BAT 15–150 keV light curve; the BAT hardness ratio over (50–150 keV)/(15–50 keV); time-sliced spectral fits to the BAT data; the XRT light curve; the XRT hardness ratio over (1.5–10 keV)/(0.3–1.5 keV); time-sliced spectral fits to the XRT data.

Figure 6

From top to bottom: the joint BAT–XRT flux light curve, in erg cm−2 s−1 over 0.3–10 keV (black histogram – BAT; grey crosses – XRT), during the first orbit; the white and v-band UVOT light curve (the data before the vertical line are white, normalized to align with the v band; afterwards –v); unfiltered ROTSE-III light curves; the BAT 15–150 keV light curve; the BAT hardness ratio over (50–150 keV)/(15–50 keV); time-sliced spectral fits to the BAT data; the XRT light curve; the XRT hardness ratio over (1.5–10 keV)/(0.3–1.5 keV); time-sliced spectral fits to the XRT data.

Extracting simultaneous BAT and GBM spectra over T0− 4–T0+ 26 s (i.e. the brightest interval; Fig. 3), a cut-off power-law model is significantly better than a single power law, with Γ= 1.05+0.07−0.08 and Epeak estimated to be 569+290−181 keV, corresponding to an isotropic energy release of Eiso∼ 3 × 1053 erg (1 keV–10 MeV in the rest frame; z= 3.355 from Prochaska et al. 2008).

Figure 3

GBM and BAT spectra covering −4 to 26 s over the trigger. The residuals are plotted in terms of sigma, with error bars of 1σ.

Figure 3

GBM and BAT spectra covering −4 to 26 s over the trigger. The residuals are plotted in terms of sigma, with error bars of 1σ.

Fig. 4 plots the Swift–XRT and BAT and Fermi–NaI spectra between 100–106 s. There is only a tenuous detection in the BGO at this time, so that spectrum has not been included in the plot for clarity; it was, however, used in the fit to help constrain Epeak.

Figure 4

Fermi–GBM, BAT and XRT spectra covering 100–106 s after the trigger. The residuals are plotted in terms of sigma, with error bars of 1σ.

Figure 4

Fermi–GBM, BAT and XRT spectra covering 100–106 s after the trigger. The residuals are plotted in terms of sigma, with error bars of 1σ.

Note that, in both of the above plots, the BAT data appear lower down the ordinate axis simply because of the way the normalizations are defined. Intercalibration between Swift and Fermi is discussed in Section 2.1.1.

2.1.1 Intercalibration

The intercalibration of the Swift–BAT and the Fermi–GBM is a work in progress and preliminary simulations were discussed by Stamatikos, Sakamoto & Band (2008); further results will be presented in Stamatikos et al. (in preparation). As mentioned in the previous section, the normalizations of the GBM detectors were all tied together for the current paper, finding a mean value of 1.23 ± 0.06 compared to a BAT value of unity. Allowing the normalizations to vary between the GBM detectors, while again simultaneously fitting all the data sets with a cut-off power law, produced the following relative constants: NaI n7 = 1.24+0.09−0.08, n8 = 1.18 ± 0.08, nb= 1.32+0.10−0.09 and BGO b1 = 1.88+0.31−0.27, where, as before, the BAT constant was fixed to unity. Thus, the GBM detectors agree quite well, with a typical discrepancy of ≲20 per cent.

2.2 X-rays

Swift–XRT identified and centroided on an uncatalogued X-ray source in a 2.5 s Image Mode frame, 76 s after the BAT trigger. The source was bright enough such that the XRT stayed in Windowed Timing (WT) mode throughout the first orbit, which ended about 460 s after the trigger. The WT data showed two large flares, with smaller peaks superimposed (Fig. 5), with an underlying decay of α= 1.05+0.17−0.14.

Figure 5

0.3–10 keV X-ray light curve. The decay has been fitted with a doubly broken power law and overlying flare/Gaussian components (described in the text). The lower panel shows the ratio between the data and the model.

Figure 5

0.3–10 keV X-ray light curve. The decay has been fitted with a doubly broken power law and overlying flare/Gaussian components (described in the text). The lower panel shows the ratio between the data and the model.

By the time of the second orbit, the source had faded sufficiently that data could be collected in Photon Counting (PC) mode. The ‘canonical’ X-ray light curve identified from Swift bursts (Nousek et al. 2006; Zhang et al. 2006) follows a steep-shallow-normal pattern (possibly with a later jet-break) and, while not all bursts show all these components (Evans et al. 2009), in this case the first PC orbit could be identified as the end of the shallow plateau phase. Swift is in a low-Earth orbit, so can only observe a given target for a maximum of about 1.5–2 ks per 96 min revolution, leading to orbital gaps in the data. Because of these intervals of no data, the start time of the plateau in the GRB 080810 X-ray light curve is very uncertain. Nevertheless, a flattening in the decay at this time was a statistical improvement. The best-fitting parameters obtained by fitting the complete light curve with a doubly broken power law (plus flare components, as shown in Fig. 5) are given in Table 3; because of the limited data, the plateau phase was poorly constrained. There was a deviation from the late-time power-law decay around ∼150 ks, when the X-ray emission briefly rebrightened.

Table 3

The doubly broken power-law model fitted to the decay continuum; the ‘plateau’ phase is poorly constrained. A number of flare components are also required to give a good fit. The three main early flares peak at 107, 210 and 281 s; a later rebrightening occurs around 150 ks. See text for more details.

Parameter Value 
α1 1.05+0.17−0.14 
Tb,1 2679+1677−1335
α2 −0.03+0.57−0.58 
Tb,2 5909+969−518
α3 1.76+0.10−0.08 
Parameter Value 
α1 1.05+0.17−0.14 
Tb,1 2679+1677−1335
α2 −0.03+0.57−0.58 
Tb,2 5909+969−518
α3 1.76+0.10−0.08 

The early flares (almost certainly due to the continuation of the prompt emission component) are well fitted with Fast Rise Exponential Decay-like (FRED) profiles, while the deviation around 105 s can be equally well-modelled by either a FRED-like profile or a Gaussian. However, it should be noted that the data are much better sampled at early times, so the shape of the later flare is more uncertain. A model comprising a power-law decay with six super-imposed, sometimes over-lapping flares (at ∼T0+ 91, 102, 107, 118, 210 and 281 s; the strongest peaks are those at 102 and 210 s) provides a good fit to the WT data, with each flare component being statistically significant. There are also additional deviations, but adding further flares to the model does not improve the fit.

Including these six burst components at early times, and a Gaussian to model the later-time rebrightening, we obtain χ2/d.o.f. = 134/130 (χ2ν= 1.03).

The spectrum of the plateau orbit (∼T0+ 4–5 ks) can be modelled with a power law of Γ= 1.95 ± 0.07 absorbed by the Galactic column of NH= 3.28 × 1020 cm−2. After the temporal break, the photon index is Γ= 2.00 ± 0.09, so there is no sign of spectral evolution at this point. A spectrum extracted just for the late-time ‘bump’ in the light curve is also consistent with these values, with Γ= 1.99+0.28−0.24. The lack of measured intrinsic absorption is not unusual for bursts at medium to high redshift (see, e.g. Grupe et al. 2007) and is also consistent with the optical data (see Section 3.1).

2.3 Optical

Swift–UVOT detected a bright optical afterglow in the v, b and white filters (Holland & Page 2008); the non-detection in u and the UV filters (∼4 ks after the trigger) is consistent with the redshift of 3.355.

ROTSE-III (Akerlof et al. 2003), at the Siding Spring Observatory in Australia, imaged GRB 080810 35 s after the burst (Rykoff 2008), detecting a counterpart which brightened for about 30 s, before fading first with a shallow decay, then with a steeper slope. These unfiltered magnitudes have been normalized to the R band for subsequent analysis; the method is described by Rykoff et al. (2009).

A target-of-opportunity program was triggered on the Keck telescopes to obtain HIRES spectroscopy (Vogt et al. 1994) of the afterglow of GRB 080810, with the observations beginning around 37 min after the trigger (Prochaska et al. 2008); these data showed the redshift of the burst to be z∼ 3.35 (see Section 3.1).

The NOT observations (de Ugarte Postigo et al. 2008a) began 10.59 h after the burst, providing confirmation of the redshift. This was followed by observations with the 1.5-m Observatorio de Sierra Nevada (OSN; de Ugarte Postigo, Aceituno & Castro-Tirado 2008b) telescope and the 1.54-m Danish telescope at La Silla, Chile (Thöne, de Ugarte Postigo & Liebig 2008).

Data from the Faulkes Telescopes North & South, the Liverpool Telescope (LT; Guidorzi, Steele & Tanvir 2008; Guidorzi, Bersier & Tanvir 2008), the Isaac Newton Telescope (INT/WFC) and the NOT (de Ugarte Postigo et al. 2008a) were reduced in standard fashion, including bias subtraction and flat-fielding; i′-band observations were additionally defringed. Seeing-matched aperture photometry was performed using the Image Reduction and Analysis Facility (iraf).3 Absolute calibration was performed using the Sloan Digital Sky Survey (SDSS; Adelman-McCarthy et al. 2008), with conversion to Johnson–Cousins.4

All the optical data included in this paper are listed in Table 4, including data from KANATA, obtained from the Gamma-Ray Bursts Coordinates Network (GCN; Ikejiri et al. 2008). These magnitudes have not been corrected for Galactic extinction (although the fluxes used later have been).

Table 4

Optical data obtained for GRB 080810; errors are at the 1σ level. No correction for Galactic extinction has been made. FTS/N = Faulkes Telescope South/North; NOT = Nordic Optical Telescope; OSN = Observatorio de Sierra Nevada; DK = Danish telescope at La Silla; LT = Liverpool Telescope; INT/WFC = Isaac Newton Telescope/Wide Field Camera. KANATA data were taken from Ikejiri et al. (2008). The UVOT u, b and v filters are close to the standard UBV system. (Poole et al. 2008). Errors are at the 1σ level.

Filter Mag. Time (s since burst) Exp. time (s) Telescope 
I 19.36 ± 0.09 47863 300 1.5m-OSN 
I 19.34 ± 0.08 50548 300 1.5m-OSN 
I 19.42 ± 0.13 52581 300 1.5m-OSN 
I 19.51 ± 0.08 53558 300 1.5m-OSN 
i 19.91 ± 0.03 52186 1800 LT 
i 20.67 ± 0.06 124762 1800 LT 
i 21.11 ± 0.06 174122 1500 FTN 
i 21.08 ± 0.05 177534 1800 FTN 
i 21.54 ± 0.06 224122 1800 LT 
i 21.60 ± 0.07 242905 1800 FTN 
i 21.66 ± 0.07 261896 1800 FTN 
i 21.90 ± 0.09 311386 900 INT/WFC 
i 22.63 ± 0.10 401674 1200 INT/WFC 
R 15.95 ± 0.12 3320 300 FTS 
R 16.30 ± 0.12 3697 180 FTS 
R 16.27 ± 0.12 3886 180 FTS 
R 16.46 ± 0.12 4075 180 FTS 
R 16.48 ± 0.12 4263 180 FTS 
R 16.66 ± 0.12 4453 180 FTS 
R 16.39 ± 0.12 4712 300 FTS 
R 16.97 ± 0.12 5689 300 FTS 
R 16.84 ± 0.12 6624 300 FTS 
R 16.97 ± 0.12 7422 300 FTS 
R 21.55 ± 0.09 244918 1800 FTN 
R 21.56 ± 0.06 265404 1800 FTN 
R 19.31 ± 0.07 38117 120 NOT 
R 19.66 ± 0.07 43110 300 1.5m-OSN 
R 19.78 ± 0.04 47548 300 1.5m-OSN 
R 19.81 ± 0.06 48212 300 1.5m-OSN 
R 19.75 ± 0.04 48868 300 1.5m-OSN 
R 19.78 ± 0.06 50016 300 1.5m-OSN 
R 19.81 ± 0.06 50215 300 1.5m-OSN 
R 19.76 ± 0.05 50884 300 1.5m-OSN 
R 19.83 ± 0.05 51559 300 1.5m-OSN 
R 19.85 ± 0.06 52255 300 1.5m-OSN 
R 19.89 ± 0.06 52916 300 1.5m-OSN 
R 19.83 ± 0.08 53879 300 1.5m-OSN 
R 19.65 ± 0.06 54530 300 1.5m-OSN 
R 20.76 ± 0.09 140260 900 1.5m-OSN 
R 22.44 ± 0.31 419360 3000 1.5m-DK 
R 22.95 ± 0.38 497928 5400 1.5m-DK 
R 19.02 ± 0.05 37420 1800 1.5m-DK 
r 20.18 ± 0.01 55987 1800 LT 
r 20.96 ± 0.02 142301 1800 LT 
r 21.72 ± 0.04 216691 2100 LT 
r 22.10 ± 0.09 307411 780 INT/WFC 
r 22.59 ± 0.06 395798 5700 INT/WFC 
Unfilt 13.70 ± 0.03 38 ROTSE-III 
Unfilt 13.00 ± 0.03 52 ROTSE-III 
Unfilt 12.79 ± 0.03 67 ROTSE-III 
Unfilt 12.78 ± 0.04 81 ROTSE-III 
Unfilt 12.81 ± 0.03 95 ROTSE-III 
Unfilt 12.63 ± 0.12 124 ROTSE-III 
Unfilt 12.84 ± 0.11 138 ROTSE-III 
Unfilt 13.06 ± 0.19 152 ROTSE-III 
Unfilt 12.94 ± 0.03 173 20 ROTSE-III 
Unfilt 12.87 ± 0.19 261 20 ROTSE-III 
Unfilt 13.65 ± 0.10 386 20 ROTSE-III 
Unfilt 13.76 ± 0.03 444 20 ROTSE-III 
Unfilt 13.91 ± 0.03 493 60 ROTSE-III 
Unfilt 14.06 ± 0.04 562 60 ROTSE-III 
Unfilt 14.14 ± 0.16 631 60 ROTSE-III 
Unfilt 14.36 ± 0.07 700 60 ROTSE-III 
Unfilt 14.37 ± 0.07 770 60 ROTSE-III 
Unfilt 14.49 ± 0.06 839 60 ROTSE-III 
Unfilt 14.66 ± 0.13 908 60 ROTSE-III 
Unfilt 14.77 ± 0.14 977 60 ROTSE-III 
Unfilt 14.70 ± 0.07 1046 60 ROTSE-III 
Unfilt 14.86 ± 0.04 1115 60 ROTSE-III 
Unfilt 15.16 ± 0.03 1504 682 ROTSE-III 
Unfilt 15.75 ± 0.05 2196 684 ROTSE-III 
Unfilt 15.79 ± 0.1 2823 552 ROTSE-III 
Unfilt 16.45 ± 0.05 4438 682 ROTSE-III 
Unfilt 16.83 ± 0.05 5129 682 ROTSE-III 
Unfilt 16.94 ± 0.04 5821 682 ROTSE-III 
Unfilt 17.10 ± 0.06 6512 682 ROTSE-III 
Unfilt 17.29 ± 0.04 7203 682 ROTSE-III 
Unfilt 17.42 ± 0.06 7898 690 ROTSE-III 
Unfilt 17.39 ± 0.15 9298 690 ROTSE-III 
V 20.27 ± 0.07 49190 300 1.5m-OSN 
V 20.33 ± 0.08 51228 300 1.5m-OSN 
V 20.35 ± 0.09 53239 300 1.5m-OSN 
V 20.44 ± 0.11 54203 300 1.5m-OSN 
V 13.7 ± 0.1a 390 33 KANATA 
v 13.73 ± 0.08 196 10 UVOT 
v 13.81 ± 0.08 206 10 UVOT 
v 13.80 ± 0.08 216 10 UVOT 
v 13.74 ± 0.08 226 10 UVOT 
v 13.89 ± 0.08 236 10 UVOT 
v 13.83 ± 0.08 246 10 UVOT 
v 13.95 ± 0.08 256 10 UVOT 
v 13.74 ± 0.08 266 10 UVOT 
v 13.91 ± 0.08 276 10 UVOT 
v 14.04 ± 0.08 286 10 UVOT 
v 13.98 ± 0.08 296 10 UVOT 
v 13.93 ± 0.08 306 10 UVOT 
v 14.14 ± 0.08 316 10 UVOT 
v 14.00 ± 0.08 326 10 UVOT 
v 14.27 ± 0.09 336 10 UVOT 
v 14.14 ± 0.08 346 10 UVOT 
v 14.09 ± 0.08 356 10 UVOT 
v 14.12 ± 0.08 366 10 UVOT 
v 14.21 ± 0.09 376 10 UVOT 
v 14.28 ± 0.09 386 10 UVOT 
v 14.27 ± 0.09 396 10 UVOT 
v 14.16 ± 0.08 406 10 UVOT 
v 14.49 ± 0.09 416 10 UVOT 
v 14.39 ± 0.09 426 10 UVOT 
v 14.32 ± 0.09 436 10 UVOT 
v 14.46 ± 0.09 446 10 UVOT 
v 14.50 ± 0.10 456 10 UVOT 
v 17.53 ± 0.10 5148 197 UVOT 
v 21.10 ± 0.49 63077 1079 UVOT 
v 20.79 ± 0.44 103938 777 UVOT 
v 22.05 ± 0.49 303764 5964 UVOT 
B 21.14 ± 0.12 49887 300 1.5m-OSN 
B 21.23 ± 0.13 51901 300 1.5m-OSN 
b 17.89 ± 0.08 4532.3 197 UVOT 
b 20.28 ± 0.16 27235 569 UVOT 
b 21.45 ± 0.32 56302 867 UVOT 
b 21.83 ± 0.47 79387 778 UVOT 
b 21.60 ± 0.38 102520.4 831 UVOT 
b 22.55 ± 0.52 164312 2162 UVOT 
white 14.87 ± 0.10 90 10 UVOT 
white 14.78 ± 0.11 100 10 UVOT 
white 14.68 ± 0.12 110 10 UVOT 
white 14.71 ± 0.12 120 10 UVOT 
white 14.89 ± 0.10 130 10 UVOT 
white 14.83 ± 0.11 140 10 UVOT 
white 14.90 ± 0.10 150 10 UVOT 
white 14.93 ± 0.10 160 10 UVOT 
white 14.93 ± 0.10 170 10 UVOT 
white 14.93 ± 0.10 180 10 UVOT 
Filter Mag. Time (s since burst) Exp. time (s) Telescope 
I 19.36 ± 0.09 47863 300 1.5m-OSN 
I 19.34 ± 0.08 50548 300 1.5m-OSN 
I 19.42 ± 0.13 52581 300 1.5m-OSN 
I 19.51 ± 0.08 53558 300 1.5m-OSN 
i 19.91 ± 0.03 52186 1800 LT 
i 20.67 ± 0.06 124762 1800 LT 
i 21.11 ± 0.06 174122 1500 FTN 
i 21.08 ± 0.05 177534 1800 FTN 
i 21.54 ± 0.06 224122 1800 LT 
i 21.60 ± 0.07 242905 1800 FTN 
i 21.66 ± 0.07 261896 1800 FTN 
i 21.90 ± 0.09 311386 900 INT/WFC 
i 22.63 ± 0.10 401674 1200 INT/WFC 
R 15.95 ± 0.12 3320 300 FTS 
R 16.30 ± 0.12 3697 180 FTS 
R 16.27 ± 0.12 3886 180 FTS 
R 16.46 ± 0.12 4075 180 FTS 
R 16.48 ± 0.12 4263 180 FTS 
R 16.66 ± 0.12 4453 180 FTS 
R 16.39 ± 0.12 4712 300 FTS 
R 16.97 ± 0.12 5689 300 FTS 
R 16.84 ± 0.12 6624 300 FTS 
R 16.97 ± 0.12 7422 300 FTS 
R 21.55 ± 0.09 244918 1800 FTN 
R 21.56 ± 0.06 265404 1800 FTN 
R 19.31 ± 0.07 38117 120 NOT 
R 19.66 ± 0.07 43110 300 1.5m-OSN 
R 19.78 ± 0.04 47548 300 1.5m-OSN 
R 19.81 ± 0.06 48212 300 1.5m-OSN 
R 19.75 ± 0.04 48868 300 1.5m-OSN 
R 19.78 ± 0.06 50016 300 1.5m-OSN 
R 19.81 ± 0.06 50215 300 1.5m-OSN 
R 19.76 ± 0.05 50884 300 1.5m-OSN 
R 19.83 ± 0.05 51559 300 1.5m-OSN 
R 19.85 ± 0.06 52255 300 1.5m-OSN 
R 19.89 ± 0.06 52916 300 1.5m-OSN 
R 19.83 ± 0.08 53879 300 1.5m-OSN 
R 19.65 ± 0.06 54530 300 1.5m-OSN 
R 20.76 ± 0.09 140260 900 1.5m-OSN 
R 22.44 ± 0.31 419360 3000 1.5m-DK 
R 22.95 ± 0.38 497928 5400 1.5m-DK 
R 19.02 ± 0.05 37420 1800 1.5m-DK 
r 20.18 ± 0.01 55987 1800 LT 
r 20.96 ± 0.02 142301 1800 LT 
r 21.72 ± 0.04 216691 2100 LT 
r 22.10 ± 0.09 307411 780 INT/WFC 
r 22.59 ± 0.06 395798 5700 INT/WFC 
Unfilt 13.70 ± 0.03 38 ROTSE-III 
Unfilt 13.00 ± 0.03 52 ROTSE-III 
Unfilt 12.79 ± 0.03 67 ROTSE-III 
Unfilt 12.78 ± 0.04 81 ROTSE-III 
Unfilt 12.81 ± 0.03 95 ROTSE-III 
Unfilt 12.63 ± 0.12 124 ROTSE-III 
Unfilt 12.84 ± 0.11 138 ROTSE-III 
Unfilt 13.06 ± 0.19 152 ROTSE-III 
Unfilt 12.94 ± 0.03 173 20 ROTSE-III 
Unfilt 12.87 ± 0.19 261 20 ROTSE-III 
Unfilt 13.65 ± 0.10 386 20 ROTSE-III 
Unfilt 13.76 ± 0.03 444 20 ROTSE-III 
Unfilt 13.91 ± 0.03 493 60 ROTSE-III 
Unfilt 14.06 ± 0.04 562 60 ROTSE-III 
Unfilt 14.14 ± 0.16 631 60 ROTSE-III 
Unfilt 14.36 ± 0.07 700 60 ROTSE-III 
Unfilt 14.37 ± 0.07 770 60 ROTSE-III 
Unfilt 14.49 ± 0.06 839 60 ROTSE-III 
Unfilt 14.66 ± 0.13 908 60 ROTSE-III 
Unfilt 14.77 ± 0.14 977 60 ROTSE-III 
Unfilt 14.70 ± 0.07 1046 60 ROTSE-III 
Unfilt 14.86 ± 0.04 1115 60 ROTSE-III 
Unfilt 15.16 ± 0.03 1504 682 ROTSE-III 
Unfilt 15.75 ± 0.05 2196 684 ROTSE-III 
Unfilt 15.79 ± 0.1 2823 552 ROTSE-III 
Unfilt 16.45 ± 0.05 4438 682 ROTSE-III 
Unfilt 16.83 ± 0.05 5129 682 ROTSE-III 
Unfilt 16.94 ± 0.04 5821 682 ROTSE-III 
Unfilt 17.10 ± 0.06 6512 682 ROTSE-III 
Unfilt 17.29 ± 0.04 7203 682 ROTSE-III 
Unfilt 17.42 ± 0.06 7898 690 ROTSE-III 
Unfilt 17.39 ± 0.15 9298 690 ROTSE-III 
V 20.27 ± 0.07 49190 300 1.5m-OSN 
V 20.33 ± 0.08 51228 300 1.5m-OSN 
V 20.35 ± 0.09 53239 300 1.5m-OSN 
V 20.44 ± 0.11 54203 300 1.5m-OSN 
V 13.7 ± 0.1a 390 33 KANATA 
v 13.73 ± 0.08 196 10 UVOT 
v 13.81 ± 0.08 206 10 UVOT 
v 13.80 ± 0.08 216 10 UVOT 
v 13.74 ± 0.08 226 10 UVOT 
v 13.89 ± 0.08 236 10 UVOT 
v 13.83 ± 0.08 246 10 UVOT 
v 13.95 ± 0.08 256 10 UVOT 
v 13.74 ± 0.08 266 10 UVOT 
v 13.91 ± 0.08 276 10 UVOT 
v 14.04 ± 0.08 286 10 UVOT 
v 13.98 ± 0.08 296 10 UVOT 
v 13.93 ± 0.08 306 10 UVOT 
v 14.14 ± 0.08 316 10 UVOT 
v 14.00 ± 0.08 326 10 UVOT 
v 14.27 ± 0.09 336 10 UVOT 
v 14.14 ± 0.08 346 10 UVOT 
v 14.09 ± 0.08 356 10 UVOT 
v 14.12 ± 0.08 366 10 UVOT 
v 14.21 ± 0.09 376 10 UVOT 
v 14.28 ± 0.09 386 10 UVOT 
v 14.27 ± 0.09 396 10 UVOT 
v 14.16 ± 0.08 406 10 UVOT 
v 14.49 ± 0.09 416 10 UVOT 
v 14.39 ± 0.09 426 10 UVOT 
v 14.32 ± 0.09 436 10 UVOT 
v 14.46 ± 0.09 446 10 UVOT 
v 14.50 ± 0.10 456 10 UVOT 
v 17.53 ± 0.10 5148 197 UVOT 
v 21.10 ± 0.49 63077 1079 UVOT 
v 20.79 ± 0.44 103938 777 UVOT 
v 22.05 ± 0.49 303764 5964 UVOT 
B 21.14 ± 0.12 49887 300 1.5m-OSN 
B 21.23 ± 0.13 51901 300 1.5m-OSN 
b 17.89 ± 0.08 4532.3 197 UVOT 
b 20.28 ± 0.16 27235 569 UVOT 
b 21.45 ± 0.32 56302 867 UVOT 
b 21.83 ± 0.47 79387 778 UVOT 
b 21.60 ± 0.38 102520.4 831 UVOT 
b 22.55 ± 0.52 164312 2162 UVOT 
white 14.87 ± 0.10 90 10 UVOT 
white 14.78 ± 0.11 100 10 UVOT 
white 14.68 ± 0.12 110 10 UVOT 
white 14.71 ± 0.12 120 10 UVOT 
white 14.89 ± 0.10 130 10 UVOT 
white 14.83 ± 0.11 140 10 UVOT 
white 14.90 ± 0.10 150 10 UVOT 
white 14.93 ± 0.10 160 10 UVOT 
white 14.93 ± 0.10 170 10 UVOT 
white 14.93 ± 0.10 180 10 UVOT 

The white and v-band UVOT data are plotted in the second panel of Fig. 6, with the ROTSE-III data in the third. The UVOT white data have been normalized to align with the v band at 180–190 s (shown by the vertical line in the plot). There is an indication that the white magnitude increases around the same time as the first flare seen by the XRT, but this is only significant at the 1.5σ level. After this time, the UVOT data appear to follow a smooth decline, with no indication of a brightening at the time of the second X-ray flare (210 s after the trigger).

The ROTSE-III data show a clear ∼1 mag brightening between T0+ 40 s and T0+ 60 s. Although there is apparent structure in the light curve between ∼120–260 s, the limiting magnitudes were shallow at these times (caused by cloud cover), meaning these variations are not very significant.

The lack of obvious flaring in the optical coincident with the X-ray flares implies that this emission is strongly dominated by the afterglow and has very little (if any) contribution from the prompt emission.

3 DISCUSSION

In this section, we first present details about how the redshift was determined from the Keck data, then an alternative model for the prompt data and finally discuss the panchromatic observations and what can be learnt from them.

3.1 Redshift determination

Fig. 7 presents a Keck velocity plot for a series of transitions corresponding to a δv≈ 1300 km s−1 interval around z≈ 3.355. The top panel shows a pair of broad transitions (centred at v≈ 0 and +700 km s−1) which mark the reddest Lyα features detected in the spectrum. Blueward of these data is a series of Lyα features which correspond to the intergalactic medium (IGM) at z < 3.35. The termination of the IGM establishes a rough redshift (conservatively, a lower limit) for the GRB host galaxy, that is zGRB≈ 3.355.

Figure 7

Keck/HIRES velocity plots for a subset of the transitions detected at zzGRB. In the figure v= 0 km s−1 corresponds to an arbitrary z= 3.35104 and orange-dashed lines indicate blends with coincident features. Within the velocity interval shown, we detect two strong (broad) Lyα lines centred at v= 0 km s−1 and v≈+700 km s−1 and corresponding metal-line absorption, including very strong C iv absorption. These Lyα lines are the reddest observed in the afterglow and we conclude that one (or both) are associated with the GRB host galaxy. Both complexes also show significant resonance, low-ion absorption from Si ii and C ii ions, but only the gas at z≈ 3.35 shows significant absorption from fine-structure levels of these ions.

Figure 7

Keck/HIRES velocity plots for a subset of the transitions detected at zzGRB. In the figure v= 0 km s−1 corresponds to an arbitrary z= 3.35104 and orange-dashed lines indicate blends with coincident features. Within the velocity interval shown, we detect two strong (broad) Lyα lines centred at v= 0 km s−1 and v≈+700 km s−1 and corresponding metal-line absorption, including very strong C iv absorption. These Lyα lines are the reddest observed in the afterglow and we conclude that one (or both) are associated with the GRB host galaxy. Both complexes also show significant resonance, low-ion absorption from Si ii and C ii ions, but only the gas at z≈ 3.35 shows significant absorption from fine-structure levels of these ions.

Fig. 7 reveals that each of these Lyα features also show corresponding metal-line transitions of Si ii, C ii (both around 0, 50 and 760 km s−1) and C iv (from −80 to +80 km s−1 and +580 to almost +800 km s−1). Also detected is absorption from O i, Si iv, Al ii and a number of other ions (not shown in the plot). However, it is only the gas at v≈ 0 km s−1 (i.e. z≈ 3.351) that shows significant absorption from fine-structure levels of the Si ii, C ii and O i ions/atoms. Absorption from these fine-structure levels (e.g. Si ii* 1264) is a strong signature of gas near the GRB afterglow because the afterglow itself is the excitation mechanism (Prochaska, Chen & Bloom 2006). Therefore, we are inclined to associate the GRB with the redshift of this material. Under this hypothesis, the gas at v≈+700 km s−1 might be considered a neighbouring galaxy with a very large, positive peculiar velocity.

An alternative hypothesis is that the gas at v≈+700 km s−1 (i.e. z≈ 3.36) marks the ambient interstellar medium of the host galaxy and that the material at v≈ 0 km s−1 is due to a fast outflow of material near the GRB afterglow. This hypothesis is challenged by the fast wind speed required and the absence of fine-structure absorption in the Si ii and C ii gas at v≈+770 km s−1. An emission-line measurement of the host galaxy would be required to resolve these two possibilities and so we infer zGRB= 3.355 ± 0.005.

Independent of the correct interpretation of the two systems, we can draw a few conclusions about the gas associated with GRB 080810. First, the total hi column is low relative to the mean value observed in GRB afterglow spectra (Jakobsson et al. 2006). For both Lyα lines, we set a conservative upper limit of NH i < 1019.5 cm−2 based on the absence of strong damping wings. We also place a lower limit to the column density of the z= 3.36 absorber of NH i > 1017.7 cm−2 based on Lyman limit absorption at λ < 3976Å. Both systems show relatively weak low-ion absorption, consistent with the low NH i values. In contrast, each shows very strong high-ion absorption, a characterstic of gas associated with GRB host galaxies (Prochaska et al. 2007). In agreement with this optical measurement, no absorption above the Galactic value was required when fitting the X-ray spectrum (Section 2.2).

We note in passing that GRBs 020813 (Barth et al. 2003) and 021004 (Fynbo et al. 2005) both showed similarly unusual absorption systems.

3.2 Thermal emission

Ryde (2004) and Ryde & Pe'er (2009) discuss the possibility that thermal emission may be ubiquitous in GRBs, fitting samples of spectra with a power-law-plus-blackbody model, and Lazzati, Morsony & Begelman (2009) present simulations of photospheric emission. Such thermal components may explain the hard early-time spectra which are inconsistent with synchrotron emission (Preece et al. 1998; Ghisellini, Celotti & Lazzati 2000; Savchenko & Neronov 2009). A similar quasi-thermal model was applied to the GRB 080810 gamma-ray spectra and the results are included in Table 2. In most of the spectra presented by Ryde & Pe'er (2009), a distinct break is seen in the temporal evolution of the blackbody temperature; this is not apparent in our data, though this may be related to the fact that our first spectrum spans T0+ 0–10 s, and most of the breaks seen occur within 10 s of the burst. Following Pe'er et al. (2007), we have estimated the photospheric radius (the radius beyond which the outflow becomes optically thin) to be ∼2 × 1011Y0 cm, where Y0 is the ratio between the fireball energy and the energy emitted in gamma-rays (1 ≲Y0 < 3–5; see Pe'er et al. 2007). We also determine that the coasting Lorentz factor is ∼570 Y1/40. These calculations assume that the blackbody temperature is the maximum measured (62 keV), since we do not see a break in the behaviour of the temperature. The ratio of the blackbody flux to the total flux over the gamma-ray band (10–104 keV) is 0.2 for this spectrum. With the exception of the spectrum after 100 s, where the blackbody is much weaker and cooler, this ratio remains between 0.1 and 0.2 for the spectra considered here. We do point out, however, that the fits systematically have a worse χ2 than the cut-off power-law models.

3.3 Multiwavelength data

3.3.1 Broad-band modelling

The top panel of Fig. 6 shows the joint BAT–XRT flux light curve. The data were converted into 0.3–10 keV fluxes using power-law fits to time-sliced spectra; where there were simultaneous BAT and XRT detections, the spectra were fitted jointly. The figure also shows how the BAT and XRT hardness ratios and spectra evolve over time. A general softening trend can be seen across both bands, though the data harden during the flares, as is typical (e.g. Golenetskii et al. 1983; Ford et al. 1995; Borgonovo & Ryde 2001; Goad et al. 2007; Page et al. 2007). This softening with time likely explains why there was no significant gamma-ray emission detected at the time of the second large flare seen by the XRT, just after T0+ 200 s.

The flare around 100 s was detected over both gamma-ray (BAT and GBM) and X-ray bands, though; as mentioned in Section 2.3, there is only a slight hint of an increase in the UVOT white magnitude at this same time. Extracting and fitting simultaneous BAT and XRT spectra demonstrates that there is curvature between the gamma-ray and X-ray bands; this is also shown by the differences in the independent power-law fits shown in Fig. 6. Broken power-law models are preferred for spectra from both the rising and falling portions of the flare. The break energy is not well-constrained in either case, but is located towards the top end of the XRT bandpass (between ∼3–10 keV).

Instead of using a series of broken power laws to fit GRB light curves, an alternative method was proposed by O'Brien et al. (2006) and Willingale et al. (2007): the light curve is parametrized by one or two components each comprising an early exponential rise which then rolls over into a power-law decay. The first of these components accounts for the gamma-ray and early X-ray emission (the ‘prompt’ phase), while the second (which is not always seen – see fig. 2 of Willingale et al. 2007) forms what we typically see as the afterglow at later times. These components allow an intrinsically curved light curve to be fitted without relying on a series of abrupt breaks in a power-law model. Excluding the times where the flares are seen, this exponential-to-power-law model can easily account for the BAT–XRT light curve of GRB 080810. The second, ‘afterglow’ component begins to dominate in the interval 500–3500 s, between the end of the first and beginning of the second orbit of data. The optical data can also be modelled with the exponential-to-power-law combination. Using this model, the afterglow Eiso was calculated to be 1.6 × 1052 erg, approximately a 10th of the prompt Eiso (Section 2.1). As for the prompt measurement, this is also high, compared with typical values of 1051 for afterglows (Willingale et al. 2007).

Fig. 8 brings together all the photometric data considered in this paper, from the gamma-ray regime to radio wavelengths. The gamma-ray and X-ray light curves have been converted to flux densities by assuming simple power-law fits (changing over time where there is evidence for spectral evolution). Note that all light curves except the X-ray and radio points have been vertically offset from their actual positions, to avoid overlap and make the plot clearer.

Figure 8

Multiwavelength flux density light curves, showing the gamma-ray, X-ray, optical and radio data collated for GRB 080810. Four fiducial times are marked by dotted vertical lines and curved lines indicate model fits; both sets of lines are explained in the text. With the exception of the X-ray and radio data, the light curves have been vertically offset to make the plot clearer; these offsets are given in the figure.

Figure 8

Multiwavelength flux density light curves, showing the gamma-ray, X-ray, optical and radio data collated for GRB 080810. Four fiducial times are marked by dotted vertical lines and curved lines indicate model fits; both sets of lines are explained in the text. With the exception of the X-ray and radio data, the light curves have been vertically offset to make the plot clearer; these offsets are given in the figure.

The early high-energy photon index evolves from harder to softer, as indicated by the fact that the Fermi (8–1000 keV data from the summed NaI detectors, normalized to 100 keV) and BAT data extend till ∼100 s, while the softer, 1 keV XRT data show apparent prompt emission out to at least 400 s; at this time Swift entered the South Atlantic Anomaly and so stopped collecting data for the remainder of the orbit.

The exponential-to-power-law model fits (O'Brien et al. 2006; Willingale et al. 2007) to the BAT, XRT and R-band data are shown as solid lines in the figure, with the dotted sections showing the continuation of these models beyond the extent of the data. The model jointly fitted to the BAT–XRT data (with the BAT data extrapolated to 1 keV) is also shown as the higher dotted blue line; the 20 keV magenta line is the same model (though with a different normalization) as the joint BAT–XRT fit. Each optical band is initially consistent with the same slope (αopt= 1.22 ± 0.09 when the power-law decay takes over), with slightly different normalizations between the filters. Note that the second of the radio points is only a marginal detection, at the 2.4σ level.

The second component fitted to the X-ray and optical profiles also contains an exponential rise. Such an increase is usually assumed because the second component is expected to turn on some time after the prompt trigger; however, in this burst, the early R-band points actually show an initial increase, peaking at ∼100 s after the trigger, and provide us with a method of fitting the turn-on of the afterglow. With only three or four points between the initial ROTSE detection and the peak of the optical emission, however, quantifying the speed of the rise would be overinterpreting the data.

We do not have a detailed theoretical model for the onset of the afterglow with which to compare the data. It is possible that part of the rise seen could be caused by spectral evolution (that is, the passage of the peak of the spectrum through the bandpass), although it is reasonable to assume that we are mainly seeing an increase in peak flux as the external (afterglow) shock begins and strengthens, ∼30 s after the burst.

The vertical dotted lines mark four times at which spectral energy distributions (SEDs) have been created (Fig. 9): Tp, at 50 s after the trigger, is during the prompt phase, where Fermi, BAT and ROTSE-III data are available; Tpa is the time at which the prompt and afterglow model fits are equal in X-ray flux (∼600 s after the trigger); Ta indicates the time at which the power-law decay starts in the X-ray band (∼8 ks); Tlate is the point around which we have X-ray, optical and radio measurements (shortly before ∼300 ks). The optical flux values in Fig. 9 have been corrected for the Lyman forest absorption (Curran et al. 2008b) and the X-ray data corrected for the hydrogen column. The error bars on the optical points are smaller than the markers plotted.

Figure 9

SEDs of GRB 080810 at the times of the vertical lines in Fig. 8. The stars indicate the approximate (cooling) break frequencies for the third and fourth SEDs. See text for details.

Figure 9

SEDs of GRB 080810 at the times of the vertical lines in Fig. 8. The stars indicate the approximate (cooling) break frequencies for the third and fourth SEDs. See text for details.

The first SED, at Tp, is a distinctly different shape from the later, afterglow-dominated plots. The prompt spectrum is shown as a cut-off power law (see Table 2) with the extension below the X-ray band shown by the curved dotted line, indicating that we assume this component peaks somewhere between the optical and X-ray bands since, as can be seen in Fig. 8, the optical emission at this time is totally dominated by the afterglow component. The power-law dotted line from the optical band at this time is parallel to the optical–X-ray slope at Tpa– i.e. it represents the afterglow component. There is no evidence for a prompt optical flash (caused by reverse shock emission), since this should be brighter than an extrapolation of the (forward shock) X-ray spectrum. The inferred break between the X-ray and optical bands will be due to synchrotron self-absorption, νa. Shen & Zhang (2009) discuss how to constrain the location of the emission region from the self-absorption frequency in a prompt optical-to-gamma-ray SED, assuming the optical and gamma-ray emission are both from the same synchrotron continuum. Their equations give an approximate range of 5 < R (1014Γ3/4300B1/44 cm) <16 (for νa > νopt) for the radial distance of the prompt emission from the centre of the GRB explosion. Here, Γ is the bulk Lorentz factor (Γ= 300 ×Γ300) and B is the magnetic field strength (B=B4× 104 G). This distance is of the same order as that estimated by Kumar et al. (2007).

Yost et al. (2007) discuss a number of GRBs where optical observations were obtained during the burst, but prompt optical emission was not detected, as we believe is the case for GRB 080810. They compare the optical-to-gamma-ray spectral indices (βopt-γ) with the gamma-ray spectral slopes (βγ) and find that most of their values imply that there is a spectral rollover between the optical and higher frequencies. The numbers for GRB 080810 at time Tpopt-γ∼ 0.53 and βγ∼ 0.7) are entirely consistent with their measurements. It seems likely that these events are part of the same population as GRBs with prompt optical detections, but are at the faint end of the distribution.

There is no significant evolution of the X-ray spectral index over the entire period Tpa to Tlate, with βX= 1.00 ± 0.09. The optical spectral index also remains constant until at least Ta, with βopt= 0.51 ± 0.22 (from a comparison of the R, v and b measurements at Ta). Thus, we find Δβ across the break is ∼0.5 as expected for a cooling break, although the uncertainties on the β values are large.

The SED at Tpa has been created using only the apparent afterglow flux for the X-ray data – the lower of the blue lines in Fig. 8. At this time, the spectral break must be at or below the optical band because the optical flux clearly lies on the direct extrapolation of the X-ray spectrum. Between Tpa and Ta, the X-ray temporal decay is slow, while the optical has already passed into the power-law phase; that is, the X-ray decay starts later and is then more rapid than in the optical [αX= 1.81 ± 0.20 compared to αopt= 1.22 ± 0.09; we note that, using this method of parametrizing the light curve, αX is slightly steeper than – though consistent with – the value obtained from the series of broken power laws (see Table 3)]. Even including the contribution from the residual prompt component (i.e. using the flux from the joint BAT-XRT model – the upper dotted blue line in Fig. 8), the X-ray data still decay more gradually than the optical over this time frame.

As an aside, we note that Granot, Ramirez-Ruiz & Perna (2005) find that a structured jet could lead to a steeper decay in the optical than in the X-ray – the opposite of the measurements for GRB 080810. Using the standard fireball closure relationships (e.g. Zhang & Mészáros 2004), the optical data at this time are consistent with a wind medium undergoing slow cooling, with νm < ν < νc. In this case, α= (3β+ 1)/2 =[(3 × 0.51) + 1]/2 = 1.27, compared to the measured value of αopt∼ 1.22.

The most straightforward way to explain this difference in the X-ray and optical behaviour is to have the cooling break moving towards higher frequencies (i.e. moving from the optical towards the X-ray). If the blastwave is indeed moving through a wind medium, then the cooling break is expected to increase with time as t1/2. This would, however, necessitate a transition from a wind to a homogeneous medium for the break then to move down to lower frequencies (see later), which would lead to a change in the optical decay which is not clearly seen. An alternative explanation for the movement of the cooling break to higher energies is energy injection. Using the equations for the luminosity index, q, given by Zhang et al. (2006), we find that this part of the decay is indeed consistent with energy injection (that is, q < 0). The approximate frequency of this cooling break is marked by a star in Fig. 9; it cannot be determined accurately because of the lack of constraint on the optical index, but lies within the range 1015–2 × 1016 Hz.

BVRI data from OSN at 50 ks after the trigger (the closest measurements were taken and then extrapolated to this exact time by using the slope of the light curve) provide a spectral slope of βopt= 0.98 ± 0.38, consistent with the X-ray data; this confirms that the cooling break has passed through the optical frequencies by Tlate and the data are consistent with the relevant closure relation [α= (3β− 1)/2 for ν > νc] within the uncertainties. As the cooling break moves through the optical band, we would expect to see a change in slope of the light curve; however, we do not have sufficient coverage or statistical precision at this late time to confirm or deny such a break. Furthermore, such a change in slope is likely to be spread over a decade in time, since the spectral break is a smooth, rather than abrupt, transition. With no radio index and the lack of measurements between the optical and radio regimes, it is not possible to say where the peak of the spectrum lies. The star again marks an approximate position for the break, although we note that there will be multiple breaks in this part of the spectrum, corresponding to this cooling break, the peak frequency and the synchrotron self-absorption frequency. The radio data are consistent with the extrapolation of a synchrotron spectrum from the optical and X-ray bands, though.

The optical spectral measurements are quite poorly constrained, both at early and late times and, within the errors, βopt could be consistent with a constant value from Tlate to Ta. We also note that the X-ray data during the power-law decay (βX= 1.00 ± 0.09 and αX= 1.81 ± 0.20) are not consistent with being above νc (specifically, the temporal slope is too steep and is, in fact, steeper than is typical for a pre-jet-break slope; see, e.g. Evans et al. 2009), indicating the decay may not be governed by the standard fireball model (e.g. Rees & Mészáros 1992; Mészáros & Rees 1994, 1997; Sari, Piran & Narayan 1998) with a single synchrotron emission component. This lack of agreement with the standard fireball model appears to be the case for a growing number of afterglows, with many bursts for which we have good, multiwavelength data presenting challenges which still need to be accounted for (e.g. Willingale et al. 2007; GRB 061007 –Schady et al. 2007; Mundell et al. 2007; GRB 061121 –Page et al. 2007).

There is no evidence for a jet break (Rhoads 1999) in the data out to at least 6 d (which is within the range of Swift jet break times discussed by Racusin et al. 2009). This places a limit on the jet opening angle of θj > 4° (Sari, Piran & Halpern 1999; Frail et al. 2001) and, hence, a lower limit on the total gamma-ray energy, Eγ > 8 × 1050 erg. The beaming fraction, Eγ/Eiso, is therefore >0.0027, which is of the same order as the those found for the samples in Frail et al. (2001) and Bloom, Frail & Kulkarni (2003).

The so-called canonical XRT light curve (Nousek et al. 2006; Zhang et al. 2006) typically starts with a steep decay, with α∼ 3, which is generally attributed to off-axis emission, often referred to as the ‘curvature effect’ (Fenimore, Madras & Nayakshin 1996; Kumar & Panaitescu 2000; Dermer 2004). However, the early X-ray light curve of GRB 080810 shows a significantly slower decline, with α∼ 1. In fact, the beginning of the GRB 080810 light curve is very similar to that of GRB 060607A (Molinari et al. 2007; Ziaeepour et al. 2008), both in X-rays and the optical: X-ray flares, superimposed on a relatively slow decline, while the optical emission rises smoothly, independently of the variability seen by the XRT and BAT. Both bursts also show low NH i. This slower-than-expected decay could be (at least partly) caused by repeated flaring on top of the underlying continuum. Alternatively, there could be input from the X-ray afterglow; the optical afterglow has clearly risen by this time (with little evidence of prompt emission; Fig. 8), but the X-ray light curve cannot be fully deconvolved into the prompt and afterglow components.

No plateau phase is seen in the optical data – the emission rises and then fades steadily away – which could be explained by the lack of a strong optical component during the prompt phase: from the exponential-to-power-law model, the plateau phase often seen in the X-ray may be a consequence of the bright prompt emission dominating at early times as the afterglow emission rises (see, e.g. Page et al. 2009 for an example of this). The final optical decay starts around 300 s, while the X-ray plateau extends to ∼8 ks before the onset of the power-law decay. These findings (the optical data showing no plateau and beginning the power-law decay phase earlier) were also noted for GRB 061121 (Page et al. 2007), another burst for which good, multiband data were obtained.

3.3.2 Onset of the afterglow

Measurement of the peak time of the afterglow component in the optical band gives us the opportunity to calculate the initial Lorentz factor, Γ0 (Molinari et al. 2007). Estimating the peak to occur at ∼100 s (from the ROTSE-III data), and taking Eiso to be 3 × 1053 erg, Γ0 is found to be ∼640(η0.2n0)−1/8 (for a constant density medium), at the high end of the range of estimates from other bursts (Molinari et al. 2007; Page et al. 2009; Rykoff et al. 2009). Here, η is the radiative efficiency and n is the particle density of the surrounding medium in cm−3. However, there is some evidence that the data are better suited to a wind environment, rather than constant density, model. In this case, Γ0∼ 235(η0.2)−1/4, taking A= 3 × 1035 cm−1, where the number density, n(r) =Ar−2 (Chevalier & Li 2000; Molinari et al. 2007).

Zhang, Kobayashi & Mészáros (2003) and Jin & Fan (2007) discuss the possible appearance of the reverse shock in optical light curves. These papers consider three types of reverse shock emission: Type I light curves show both forward- and reverse-shock emission peaks; Type II shows a single peak, corresponding to the reverse shock, with a later flattening caused by the forward shock as it starts to dominate the emission; Type III light curves show no sign of the reverse shock, with just a single rise and fall of the optical emission. The light curve of GRB 080810 appears to be a member of this third class, with a single power-law decay seen from around 100 s until several hundred kiloseconds after the burst trigger. The synchrotron frequency, νm, at the time the reverse shock crosses the outflow is therefore very small, well below the optical band.

Panaitescu & Vestrand (2008) found an anticorrelation between the peak optical flux and the time of the peak for fast-rising optical afterglows. K-correcting our values to the fiducial redshift of z= 2 used by Panaitescu & Vestrand gives a peak time of ∼85 s and a corresponding peak flux of ∼94 mJy, meaning that the optical afterglow of GRB 080810 is consistent with their findings. This correlation is explained by a structured outflow being seen off-axis.

Although the rise seen in the optical light curve is naturally explained as the onset of the afterglow, there are other possible mechanisms – for example, off-axis or structured outflows or two-component jets (e.g. Granot et al. 2002). These other explanations are discussed for a sample of Swift–UVOT bursts by Oates et al. (2009). They find that the onset of the forward shock or an off-axis viewing angle could best explain the rises seen in the light curves.

3.3.3 Flares

Prior to Swift, X-ray observations tended to start hours or days after the burst trigger, with only a few light curves showing flaring activity (Piro et al. 1998, 2005). However, flares are now regularly detected by the XRT (see Chincarini et al. 2007; Falcone et al. 2007 for survey papers), with about 50 per cent of Swift bursts showing them, typically in the first thousand seconds or so (Burrows et al. 2007; Chincarini et al. 2007). Some light curves, however, do show rebrightenings at later times (e.g. GRB 050502B –Falcone et al. 2006; GRB 050724 –Campana et al. 2006; GRB 070710A –Covino et al. 2008; GRB 070311 –Guidorzi et al. 2007; see also Kocevski, Butler & Bloom 2007). Confirming the interpretation established by Burrows et al. (2005b) and Zhang et al. (2006), Curran et al. (2008a) investigated late-time flares, finding that their properties are consistent with those of early ones, implying that the central engine may sometimes be active for up to 100 ks or that it can be restarted. In the case of the deviation from the power-law decay at about 150 ks in the light curve of GRB 080810 presented here, the Δt/t and ΔF/F values are within the range of the values plotted in Fig. 3 of Curran et al. for their sample of late-time flares. Because of the uncertainties on the values, this particular flare could be consistent with internal shocks (Δt/t < 1), refreshed shocks or patchy shells (see Ioka et al. 2005. Both the early and late flares seen here lie within the distribution of Δt/t found by Chincarini et al. (2007).

4 SUMMARY AND CONCLUSIONS

Plentiful broad-band data were collected for the bright GRB 080810 which triggered both Swift and Fermi. The redshift was found to be 3.355 ± 0.005, and the burst was energetic, with an isotropic energy of 3 × 1053 erg in the prompt emission component and 1.6 × 1052 erg in the afterglow. There is no evidence for a jet break up to 6 d after the burst occurred.

The prompt component (detected from 0.3–103 keV) is seen to evolve from hard to soft, with Epeak decreasing from ∼600 to ∼40 keV over 110 s. Despite being detected at the same time as the gamma-ray emission, the optical data appear to be strongly dominated by the afterglow, with little (if any) prompt component. SEDs created at fiducial times during the observations show the movement of the cooling break with time. By Ta (8 ks after the burst), the break frequency has moved from the optical band towards the X-rays, lying in the range 1015 to 2 × 1016 Hz; at later times (Tlate∼ 300 ks), this cooling break has evolved to <3 × 1014 Hz. Although the optical data conform to the standard fireball interpretation, the decay of the X-ray afterglow (after about 104 s) is too rapid to be consistent with the model.

Well-sampled bursts such as GRB 080810 enable us to investigate more thoroughly the myriad of models which exist for GRBs, with the ultimate goal of a complete and consistent description of GRB emission from early to late times.

1
Formerly known as Gamma-ray Large Area Space Telescope (GLAST).
3
iraf is distributed by the National Optical Astronomy Observatories, which are operated by the Association of Universities for Research in Astronomy, Inc., under cooperative agreement with the National Science Foundation.

The authors gratefully acknowledge support for this work at the University of Leicester by STFC, in Italy by funding from ASI and at PSU by NASA contract NAS5-00136. ER thanks the NOVA-3 network for support. JXP is partially supported by NASA/Swift grant NNX07AE94G. ROTSE-III has been supported by NASA grant NNG-04WC41G and the Australian Research Council and ESR would like to thank the TABASGO Foundation. AJvdH was supported by an appointment to the NASA Postdoctoral Program at the MSFC, administered by Oak Ridge Associated Universities through a contract with NASA. FY was supported by NASA Swift Guest Investigator grants NNG-06GI90G and NNX-07AF02G. The DARK cosmology centre is funded by the DNRF. This paper is partly based on observations made with the Nordic Optical Telescope, operated on the island of La Palma jointly by Denmark, Finland, Iceland, Norway and Sweden, in the Spanish Observatorio del Roque de los Muchachos of the Instituto de Astrofisica de Canarias. This work is partly based on observations with the INT, operated on the island of La Palma by the Isaac Newton Group in the Spanish Observatorio del Roque de los Muchachos of the Instituto de Astrofisica de Canarias. We also thank P. Chandra and D. Frail for help with the radio data, G. Marcey and D. Fischer for scheduling the ToO during their Keck observing time, Peter Jonker for performing the INT/WFC observations and C. Thöne for working on the data from the NOT and Danish telescope. We extend our thanks to the whole of the Fermi-GBM team for their work on this new mission. Finally, we thank the anonymous referee for their detailed comments, which improved the paper.

REFERENCES

Adelman-McCarthy
J. K.
et al.,
2008
,
ApJS
 ,
175
,
297
Akerlof
C.
et al.,
2003
,
PASP
 ,
115
,
132
Arnaud
K. A.
,
1996
, in
Jacoby
G.
,
Barnes
J.
, eds,
ASP Conf. Ser.
  Vol.
101
, Astronomical Data Analysis Software and Systems V.
Astron. Soc. Pac.
, San Francisco, p.
17
Band
D.
et al.,
1993
,
ApJ
 ,
413
,
281
Barth
A. J.
et al.,
2003
,
ApJ
 ,
584
,
L47
Barthelmy
S. D.
et al.,
2005
,
Space Sci. Rev.
 ,
120
,
143
Bloom
J. S.
Frail
D. A.
Kulkarni
S. R.
,
2003
,
ApJ
 ,
594
,
674
Borgonovo
L.
Ryde
F.
,
2001
,
ApJ
 ,
548
,
770
Burenin
R.
et al.,
2008
,
GCN Circ.
 ,
8088
Burrows
D. N.
et al.,
2005
,
Space Sci. Rev.
 ,
120
,
165
Burrows
D. N.
et al.,
2005
,
Sci
 ,
309
,
1833
Burrows
D. N.
et al.,
2007
,
Phil. Trans. R. Soc. A
 ,
365
,
1213
Campana
S.
et al.,
2006
,
A&A
 ,
454
,
113
Chandra
P.
Frail
D.
,
2008
,
GCN Circ.
 ,
8103
Chevalier
R. A.
Li
Z.-Y.
,
2000
,
ApJ
 ,
536
,
195
Chincarini
G.
et al.,
2007
,
ApJ
 ,
671
,
1903
Covino
S.
et al.,
2008
,
MNRAS
 ,
388
,
347
Curran
P. A.
Starling
R. L. C.
O'Brien
P. T.
Godet
O.
Van Der Horst
A. J.
Wijers
R. A. M. J.
,
2008
,
A&A
 ,
487
,
533
Curran
P. A.
Wijers
R. A. M. J.
Heemskerk
M. H. M.
Starling
R. L. C.
Wiersema
K.
Van Der Horst
A. J.
,
2008
,
A&A
 ,
490
,
1047
Dermer
C.
,
2004
,
ApJ
 ,
614
,
284
De Ugarte Postigo
A.
Thöne
C. C.
Hjorth
J.
Jakobsson
P.
Banhidi
Z.
Grundahl
F.
Arentoft
T.
,
2008
,
GCN Circ.
 ,
8089
De Ugarte Postigo
A.
Aceituno
F.
Castro-Tirado
A. J.
,
2008
,
GCN Circ.
 ,
8090
Evans
P. A.
et al.,
2009
,
MNRAS
 ,
397
,
1177
Falcone
A. D.
et al.,
2006
,
ApJ
 ,
641
,
1010
Falcone
A. D.
et al.,
2007
,
ApJ
 ,
671
,
1912
Fenimore
E. E.
Madras
C. D.
Nayakshin
S.
,
1996
,
ApJ
 ,
473
,
998
Ford
L. A.
et al.,
1995
,
ApJ
 ,
439
,
307
Frail
D. A.
et al.,
2001
,
ApJ
 ,
562
,
L55
Fynbo
J. P. U.
et al.,
2005
,
ApJ
 ,
633
,
317
Gehrels
N.
et al.,
2004
,
ApJ
 ,
611
,
1005
Ghisellini
G.
Celotti
A.
Lazzati
D.
,
2000
,
MNRAS
 ,
313
,
L1
Goad
M. R.
et al.,
2007
,
A&A
 ,
468
,
103
Golenetskii
S. V.
Mazets
E. P.
Aptekar
R. L.
Ilinskii
V. N.
,
1983
,
Nat
 ,
306
,
451
Granot
J.
Panaitescu
A.
Kumar
P.
Woosley
S. E.
,
2002
,
ApJ
 ,
570
,
L61
Granot
J.
Ramirez-Ruiz
E.
Perna
R.
,
2005
,
ApJ
 ,
630
,
1003
Grupe
D.
Nousek
J. A.
Vanden Berk
D. E.
Roming
P. W. A.
Burrows
D. N.
Godet
O.
Osborne
J.
Gehrels
N.
,
2007
,
AJ
 ,
133
,
2216
Guidorzi
C.
Steele
I.
Tanvir
N.
,
2008
,
GCN Circ.
 ,
8093
Guidorzi
C.
Bersier
D.
Tanvir
N.
,
2008
,
GCN Circ.
 ,
8099
Guidorzi
C.
et al.,
2007
,
A&A
 ,
474
,
793
Holland
S. T.
Page
K. L.
,
2008
,
GCN Circ.
 ,
8095
Ikejiri
Y.
Uemura
M.
Ohsugi
T.
Kawabata
K.
Arai
A.
Yamanaka
M.
Sakimoto
K.
,
2008
,
GCN Circ.
 ,
8081
Ioka
K.
Kobayashi
S.
Zhang
B.
,
2005
,
ApJ
 ,
631
,
429
Jakobsson
P.
et al.,
2006
,
A&A
 ,
460
,
L13
Jin
Z. P.
Fan
Y. Z.
,
2007
,
MNRAS
 ,
378
,
1043
Kocevski
D.
Butler
N.
Bloom
J. S.
,
2007
,
ApJ
 ,
667
,
1024
Kumar
P.
Panaitescu
A.
,
2000
,
ApJ
 ,
541
,
L51
Kumar
P.
et al.,
2007
,
MNRAS
 ,
376
,
L57
Lazzati
D.
Morsony
B. J.
Begelman
M. C.
,
2009
,
ApJ
 , submitted (arXiv:0904.2779v1)
Mallozzi
R. S.
Preece
R. D.
Briggs
M. S.
,
2005
,
RMFIT: A Lightcurve and Spectral Analysis Tool
 .
Univ. Alabama
, Huntsville
Meegan
C. A.
et al.,
2008
,
GCN Circ.
 ,
8100
Meegan
C. A.
et al.,
2009
,
ApJ
 ,
702
,
791
Mészáros
P.
Rees
M. J.
,
1994
,
MNRAS
 ,
269
,
L41
Mészáros
P.
Rees
M. J.
,
1997
,
ApJ
 ,
476
,
232
Molinari
E.
et al.,
2007
,
A&A
 ,
469
,
L13
Mundell
C. G.
et al.,
2007
,
ApJ
 ,
660
,
489
Nousek
J. A.
et al.,
2006
,
ApJ
 ,
642
,
389
Oates
S. R.
et al.,
2009
,
MNRAS
 ,
395
,
490
O'Brien
P. T.
et al.,
2006
,
ApJ
 ,
647
,
1213
Page
K. L.
et al.,
2007
,
ApJ
 ,
663
,
1125
Page
K. L.
et al.,
2008
,
GCN Circ.
 ,
8080
Page
K. L.
Holland
S. T.
Palmer
D. M.
Barthelmy
S. D.
Burrows
D. N.
Roming
P. W. A.
Gehrels
N.
,
2008
, GCN Report 157
Page
K. L.
et al.,
2009
,
MNRAS
 ,
395
,
328
Panaitescu
A.
Vestrand
W. T.
,
2008
,
MNRAS
 ,
387
,
497
Pe'er
A.
Ryde
F.
Wijers
R. A. M. J.
Mészáros
P.
Rees
M. J.
,
2007
,
ApJ
 ,
664
,
L1
Piro
L.
et al.,
1998
,
A&A
 ,
331
,
L41
Piro
L.
et al.,
2005
,
ApJ
 ,
623
,
314
Poole
T. S.
et al.,
2008
,
MNRAS
 ,
383
,
627
Preece
R. D.
Briggs
M. S.
Mallozzi
R. S.
Pendleton
G. N.
Paciesas
W. S.
Band
D. L.
,
1998
,
ApJ
 ,
506
,
L23
Prochaska
J. X.
Chen
H.-W.
Bloom
J. S.
,
2006
,
ApJ
 ,
648
,
95
Prochaska
J. X.
Chen
H.-W.
Dessauges-Zavadsky
M.
Bloom
J. S.
,
2007
,
ApJ
 ,
666
,
267
Prochaska
J. X.
Perley
D.
Howard
A.
Chen
H.-W.
Marcy
G.
Fischer
D.
Wilburn
C.
,
2008
,
GCN Circ.
 ,
8083
Racusin
J. L.
et al.,
2009
,
ApJ
 ,
698
,
43
Rees
M. J.
Mészáros
P.
,
1992
,
MNRAS
 ,
258
,
41
Rhoads
J. E.
,
1999
,
ApJ
 ,
525
,
737
Romano
P.
et al.,
2006
,
A&A
 ,
456
,
917
Roming
P. W. A.
et al.,
2005
,
Space Sci. Rev.
 ,
120
,
95
Ryde
F.
,
2004
,
ApJ
 ,
614
,
827
Ryde
F.
Pe'er
A.
,
2009
,
ApJ
 ,
702
,
1211
Rykoff
E. S.
,
2008
,
GCN Circ.
 ,
8084
Rykoff
E. S.
et al.,
2009
,
ApJ
 ,
702
,
489
Sakamoto
T.
et al.,
2008
,
GCN Circ.
 ,
8082
Sakamoto
T.
et al.,
2008
,
GCN Circ.
 ,
8101
Sakamoto
T.
et al.,
2009
,
ApJ
 ,
693
,
922
Sari
R.
Piran
T.
Narayan
R.
,
1998
,
ApJ
 ,
4997
,
L17
Sari
R.
Piran
T.
Halpern
J. P.
,
1999
,
ApJ
 ,
519
,
L17
Savchenko
V.
Neronov
A.
,
2009
,
MNRAS
 ,
396
,
935
Schady
P.
et al.,
2007
,
MNRAS
 ,
380
,
1041
Shen
R.-F.
Zhang
B.
,
2009
,
MNRAS
 , in press
Stamatikos
M.
Sakamoto
T.
Band
D. L.
,
2008
, in
Huang
Y.-F.
,
Dai
Z.-G.
Zhang
B.
, eds, AIP Conf. Proc. 1065,
Nanjing Gamma-Ray Burst Conference
 .
Am. Inst. Phys.
, New York, p.
59
Starling
R. L. C.
et al.,
2008
,
MNRAS
 ,
384
,
504
Thöne
C. C.
De Ugarte Postigo
A.
Liebig
C.
,
2008
,
GCN Circ.
 ,
8106
Vogt
S. S.
et al.,
1994
,
SPIE
 ,
2198
,
362
Von Kienlin
A.
et al.,
2004
,
SPIE
 ,
5488
,
763
Willingale
R.
et al.,
2007
,
ApJ
 ,
662
,
1093
Yost
S. A.
et al.,
2007
,
ApJ
 ,
669
,
1107
Zhang
B.
Mészáros
P.
,
2004
,
Int. J. Mod. Phys. A
 ,
19
,
2385
Zhang
B.
Kobayashi
S.
Mészáros
P.
,
2003
,
ApJ
 ,
595
,
950
Zhang
B.
Fan
Y. Z.
Dyks
J.
Kobayashi
S.
Mészáros P.Burrows
D. N.
Nousek
J. A.
Gehrels
N.
,
2006
,
ApJ
 ,
642
,
354
Ziaeepour
H.
et al.,
2008
,
MNRAS
 ,
385
,
453

Author notes

*
This paper is dedicated to the memory of Professor Martin Turner, who sadly passed away during its writing. Martin was an influential figure in X-ray Astronomy and an excellent PhD supervisor. He will be greatly missed.
NASA postdoctoral program fellow.