¹University of Hertfordshire, U.K.
²University of Exeter, U.K.

Semesters 2005A and 2005B were dramatic ones for gamma-ray burst science. This was thanks largely to the Swift satellite, which saw first light in December 2004, and normal science operations from 1 April 2005. Swift out-performs all previous GRB observatories in terms of depth, accuracy and speed of positional localisations (Gehrels et al. 2004, ApJ, 611, 1005). For about 10 GRBs per month, Swift's onboard Burst Alert Telescope (BAT) detects and provides arcminute position for bursts positions which can often be refined to within a few arcseconds accuracy by the X-Ray Telescope (XRT) and distributed to observers on the ground within minutes of the burst occurring. In the case of bright afterglows, the UV-optical Telescope (UVOT) also provides simultaneous optical photometry.

Still, a great deal of the promise of Swift is only realised if there are ground-based telescopes able to make rapid observations of these positions, to monitor the fading afterglows and measure refined, sub-arcsecond positions for the bursts. Ultimately much science relies on obtaining redshifts and characterising the host galaxies of the GRBs.

The eSTAR project, being led by the University of Exeter and Liverpool JMU, provides this link between Swift and ground based telescopes by making use of the emerging field of intelligent agent technology to provide real-time, autonomous decision making in software. Initially funded to build a distributed intelligent robotic telescope network, the project has now also been deployed at UKIRT making it the largest telescope in the world with a fully automatic system for responding to GRBs.

Since all aspects of an observation programme at UKIRT are either software readable or software controllable, this allows the eSTAR software to fully specify an MSB on the basis of the information automatically distributed by Swift. Therefore when the system receives an alert it can place an MSB directly into the queue as a high priority ToO, sounding an audible alarm in the control room alerting the observer to initiate rapid response observations. The data obtained is automatically reduced in real-time by the ORAC-DR system, and returned to the eSTAR software. This allows UKIRT observers to get on target within minutes of the burst occurring, and, potentially, allows automatic evaluation and further follow-up of the initial first-look data.

Figure 1: The spectral energy distribution of the afterglow of GRB 050814, obtained at UKIRT and NOT, showed it to be at z=5.3 (Jakobsson et al. 2006 A&A 447 897).

The first successful observations by eSTAR using UKIRT were of GRB 050716 where the system responded to the Swift XRT position alert and queued observations at UKIRT within 48 seconds of the initial alert being received from Swift. Unfortunately operational difficulties (since resolved) meant that these observations were not carried out until 56 minutes after the initial alert. Ultimately, it should be possible to reduce this "time to target" down to under 5 minutes on some occasions.

As a dedicated IR facility, UKIRT has a very important role to play in GRB follow-up, a good example being detecting and providing photometric redshifts for high-z bursts. Its well-developed ORAC-DR pipeline ensures data is rapidly available for assessment: something of considerable value in GRB observing. All this has allowed us to follow a substantial fraction of the observable bursts through 2005, providing valuable data that has already appeared in many papers. Since there have been so many developments, and so many bursts to which UKIRT has contributed, we can only describe a couple of highlights here.

Very High Redshift GRBs

Because of their great luminosity, GRBs would be detectable, if they exist, out to redshifts z~20. While luminous examples of galaxies and quasars are thought to become increasingly rare at the highest redshifts, GRBs - because they have stellar progenitors - may well be just as bright at early cosmic times as they are at lower redshifts. Indeed, the unenriched environments prevailing in the early universe may even favour GRBs, since various strands of evidence suggest that their luminosities are enhanced at low metallicities (Fruchter et al. 2006, Nature, submitted). Furthermore, cosmic time-dilation works to our benefit in the sense that observations at a given observer time correspond to earlier (and hence brighter) times in the rest-frame of the GRB.

Prior to the launch of Swift, the median GRB redshift was only around 1, and the highest, z=4.5. Relative to previous missions, Swift is more sensitive at both softer photon energies, and longer burst durations. In both respects, redshift works in our favour: so in theory Swift should detect more high redshift bursts than previous missions. The first burst inferred to be at z>5 (GRB 050814) occurred in August, and thanks to observations with UKIRT, combined with data from the Nordic Optical Telescope, a photometric redshift of z~5.3 was measured (Figure 1). This burst was followed only a few weeks later by GRB 050904. Again, UKIRT played a key role in identifying the high redshift nature of this burst (Figure 2), for which a photometric redshift of z>6 was inferred (Haislip et al. 2006 Nature, 440, 181 ), which was later confirmed spectroscopically by observers at Subaru to be z=6.3 (Kawai et al. 2006 Nature, 440, 184).

Figure 2: UKIRT image of GRB 050904 at z=6.3. Note that again the afterglow is bright and easily detected. This burst retains the title of the most distant spectroscopically confirmed GRB at the time of writing.

In January 2006, during a period of particularly high burst activity, the afterglow of GRB 060116 was discovered at UKIRT (Kocevski et al. 2006 GCN 4540). Followup observations suggested a photometric redshift of z=6.7 (Grazian et al. 2006 GCN 4545), although rather complex foreground reddening in this case makes the conclusion uncertain (Tanvir et al. 2006 GCN 4602).

In fact, the mean redshift of Swift-detected bursts is nearly z=3. As further GRBs continue to be found at high redshift, they will open up a new window for the exploration of the very early universe, providing a cosmic beacon for absorption line studies of the intergalactic medium, and pinpointing their host galaxies. Indeed, the afterglows can be used to obtaining redshifts and frequently abundances etc. for their hosts, which are often far too faint for direct spectroscopic determination.

Short Duration Bursts

The CGRO/BATSE catalogue of GRBs clearly revealed two distinct classes of event: (1) those with long durations (typically 2-200 sec) and softer gamma-ray spectra, whose afterglows have been studied since 1997, and (2) those with short durations (less than 2 sec) and harder gamma-ray spectra, which had remained largely mysterious until the launch of Swift. The study of short-duration bursts took off in May 2005 when the first Swift X-ray afterglow was found for a short burst. Although no optical afterglow was seen, despite efforts at a number of large telescopes including UKIRT, the X-ray position was good enough to identify the likely host galaxy as a cD elliptical at redshift z=0.23. Thus it appears that, unlike long-duration bursts, short bursts can be produced in older stellar populations. This is a prediction of the favoured scenario for producing short-duration bursts; namely from the merger of two neutron stars (NS-NS). The short-bursts also appear to be less luminous events than the long-duration events.

Subsequently, several other short bursts have been tracked down and generally the NS-NS picture seems in good shape. However, in December 2005 another short burst (GRB 051221) was identified by Swift that has given us pause for thought. In this case the gamma-ray flux was particularly strong, and yet the burst was determined to be at a redshift of z=0.55. Thus the total absolute gamma-ray luminosity for this short burst is much closer to that typical of long-duration bursts (thought to be produced in extreme core-collapse events; so-called Collapsars). Observations made at UKIRT, beginning a couple of hours after the burst, showed a bright IR afterglow that, it seems, faded surprisingly slowly in the infrared. It also took place in a very faint, blue host galaxy, another characteristic of long-duration bursts. The final word on the nature of this burst has not yet been had, but it has become clear that in individual cases, deciding whether a burst should be in the long or short duration class can be a tricky business.