2 Observations, reduction and analysis

Observations were performed using the ESO facilities in Chile and the Gemini-North telescope in Hawaii. Ten GRB host galaxies, most of them located in the southern hemisphere and selected for having had an optical and/or radio bright afterglow, were imaged at near-infrared wavelengths. Our sample of sources is listed in Table 1 together with a log of the observations.

   
2.1 Near-infrared observations

The NIR data were obtained with the Infrared Spectrometer And Array Camera (ISAAC) on the Very Large Telescope (VLT) at Paranal, the Son OF ISAAC (SOFI) installed on the New Technology Telescope (NTT) at La Silla, and with the Adaptive Optics Hokupa'a / QUIRC instrument on Gemini-North at Mauna Kea. Observations were carried out between March 2000 and September 2001 under photometric conditions. A $K_{\rm s}$filter (2.0-2.3$~\mu$m) was used for the ISAAC and SOFI data, while the observations on Gemini were performed with a K' filter (1.9-2.3$~\mu$m). The focal lens configurations resulted in a respective pixel size of 0 $.\!\!^{\prime\prime}$148, 0 $.\!\!^{\prime\prime}$297 and 0 $.\!\!^{\prime\prime}$02 for the ISAAC, SOFI and Hokupa'a images. The seeing remained rather stable during the observations of one given source, though it varied between 0 $.\!\!^{\prime\prime}$6 and 1 $.\!\!^{\prime\prime}$5 from one night to another. Individual frames were obtained as a co-addition of 12 single exposures of 10 seconds each. The series of acquisition for each object were then carried out in a jitter mode, with a dither of the frames following either a random pattern characterized by typical offsets $\sim$30 $^{\prime \prime }$ on the sky for the ISAAC and SOFI images, or a regular grid with shifts of 5 $^{\prime \prime }$ for the Hokupa'a data. For the ISAAC observations, we reached a total on-source integration time of 1 hour per object.

Data reduction was performed following the standard techniques of NIR image processing. To estimate the thermal background contribution of each frame, a "sky'' map was generated using a median-average of the 9 jittered images directly preceding and following a given acquisition. The corresponding "sky'' was then scaled to the mode of the object frame and subsequently subtracted. This method allowed us to remove in the meantime the contributions of the bias and dark current. Finally, the differential pixel-to-pixel response of the arrays was corrected using flat-field images taken as part of the instrument calibration plans. For the ISAAC and Hokupa'a data, these flat-fields were obtained by observing a blank-field of the sky during twilights, while a white screen within the dome of the NTT was used for the SOFI observations. For the latter, we noticed that the low spatial frequencies of the detector sensitivity were not properly taken into account with the dome images. They were therefore corrected using a low-order polynomial 2D-fit of the array response, a method often refered as the "illumination correction technique''. Photometric calibrations were performed using the NICMOS standard stars from Persson et al. (1998).

Figure 1: Near-infrared images of the gamma-ray burst host galaxies listed in Table 1. Observations were performed with a $K_{\rm s}$ filter, except in the case of the GRB 980329 host for which a K' filter was used. Each frame has a field of view of 45 $^{\prime \prime }$$\times$ 45 $^{\prime \prime }$ and an orientation with the North to the top and the East to the left. From the top to the bottom and the left to the right, they were tentatively ordered with increasing distance of the host from Earth. The GRB 981226 and GRB 001011 host galaxies have been assigned a plausible redshift range as described in Sect. 3.1, while references for the other redshifts are given in Table 2. In the last four images, the targets were not detected.

   

Table 2: Properties of GRB host galaxies.

		Redshift			Photometry
Source	GRB	z	Ref.		K mag	References		R-K colour $^\ddag$	Abs. K mag.$~^\P$
GRB J225559.9+405553	010921	0.45	1		19.05 $\pm$ 0.1	1		2.40 $\pm$ 0.25	-22.50
GRB J145212.5+430106	010222	1.48	2		23.5 $\pm$ 0.0$^\dag$	3		2.20 $\pm$ 0.30	-21.25
GRB J182304.6-505416	001011				21.45 $\pm$ 0.2	this work		3.75 $\pm$ 0.45
GRB J122519.3+200611	000418	1.12	4		21.3 $\pm$ 0.2	this work		2.50 $\pm$ 0.40	-22.65
GRB J015915.5-403933	000210	0.85	5		20.95 $\pm$ 0.2	this work		2.50 $\pm$ 0.30	-22.25
GRB J061331.1-515642	000131	4.5	6		$\geq$22.5	this work
GRB J163353.5+462721	991208	0.71	7		21.7 $\pm$ 0.2	8		2.60 $\pm$ 0.40	-21.00
GRB J223153.1-732429	990712	0.43	9		20.05 $\pm$ 0.1	this work		1.80 $\pm$ 0.30	-21.40
GRB J133807.1-802948	990510	1.62	9		$\geq$22.5	this work
GRB J115450.1-264035	990506	1.31	4		21.45 $\pm$ 0.2	this work		4.05 $\pm$ 0.35	-22.90
GRB J122311.4+064405	990308				$\geq$21.5	this work
GRB J152530.3+444559	990123	1.60	10		21.9 $\pm$ 0.4	8, 11		2.40 $\pm$ 0.80	-23.10
GRB J232937.2-235554	981226				21.1 $\pm$ 0.2	this work		3.40 $\pm$ 0.50
GRB J235906.7+083507	980703	0.97	12		19.6 $\pm$ 0.1	8		2.80 $\pm$ 0.30	-23.95
GRB J070238.0+385044	980329				$\geq$23.0	this work
GRB J115626.4+651200	971214	3.42	13		22.4 $\pm$ 0.2	8		3.20 $\pm$ 0.40	-24.45
GRB J180831.6+591851	970828	0.96	14		21.5 $\pm$ 0.3	14		3.60 $\pm$ 0.60	-22.05
GRB J065349.4+791619	970508	0.83	15		22.7 $\pm$ 0.2	8		2.40 $\pm$ 0.40	-20.45
GRB J050146.7+114654	970228	0.69	16		22.6 $\pm$ 0.3	8, 17		2.00 $\pm$ 0.50	-20.05

Notes:
$\ddagger$ For all sources except the GRB 981226 and GRB 001011 hosts, the R-K colours were estimated using the Rmagnitudes given in Table .1 of Appendix A. The R-band photometry for the host galaxy of GRB 981226 has been derived from Saracco et al. (2001b), Frail et al. (1999) and Holland et al. (2000b), while that of the GRB 001011 host is taken from Gorosabel et al. (2002).
$\P$ Defined as $M_{K}+5 \log_{10}$ h₆₅ assuming a $\Lambda$CDM Universe with $\Omega _m=0.3$ and $\Omega _\lambda =0.7$ (h₆₅ = H₀ (km s^-1 Mpc^-1) / 65).
$\dag$ Estimated from an extrapolation of the afterglow K-band light curve (Frail et al. 2002).

References: (1) Price et al. (2002c); (2) Jha et al. (2001); (3) Frail et al. (2002); (4) Bloom et al. (2002a); (5) Piro et al. (2002); (6) Andersen et al. (2000); (7) Castro-Tirado et al. (2001); (8) Chary et al. (2002); (9) Vreeswijk et al. (2001); (10) Kulkarni et al. (1999); (11) Bloom et al. (1999); (12) Djorgovski et al. (1998); (13) Kulkarni et al. (1998); (14) Djorgovski et al. (2001); (15) Bloom et al. (1998a); (16) Bloom et al. (2001); (17) Fruchter et al. (1999a).

2.2 Photometry

Each galaxy was observed more than 150 days after the date of its hosted GRB event (see Col. 3 of Table 1). Assuming the least favourable case of a bright GRB optical transient (R mag $\sim$ 20 at GRB + 2 days) with a break in the light-curve occuring $\sim$2 days after the burst and a slow decay with time (temporal index $\beta \sim -1.5$), we estimate that all GRB counterparts should have been fainter than R mag $\sim$ 35 at the time of the observations. Taking account of a power law spectrum $F_{\nu} \varpropto \nu^{-1}$ for the modelling of the afterglow emission, we set a lower limit K mag $\approx$ 33. In our data, the flux contribution of any extra light from the fading afterglows should therefore be completely negligible relative to the emission of the host galaxies.

Our final images are presented in Fig. 1. For each observation, the astrometry was performed using foreground stars of the USNO catalog, and the GRB hosts were identified within 1 $^{\prime \prime }$ of the positions of the GRB transients. Among the ten sources of our sample, six host galaxies are clearly detected in our $K_{\rm s}$-band data. Using the task phot within the IRAF package, we measured their total magnitude in an aperture of 5 $^{\prime \prime }$ in diameter centered on the source, with the exception of GRB 990506 host which lies only $\sim$1.8 $^{\prime \prime }$ from another extended object. Since this host galaxy has a very compact morphology at optical wavelengths ( $FWHM \sim 0.14{\hbox{$^{\prime\prime}$ }}$), as revealed by the high resolution HST images (Holland et al. 2000c), we assumed that we get a good estimation of its overall emission within an aperture of $\sim$1.5 $^{\prime \prime }$ in diameter, inside which $\sim$95% of the total flux would be included if the light profile is Gaussian. Given our typical uncertainty on the photometry ($\sim$0.2 mag) and taking account of the prescriptions mentioned in the ISAAC Data Reduction Guide, we found the $K-K_{\rm s}$ colour terms to be negligible in the final conversions to K magnitudes.

The foreground Galactic extinctions in the direction of our targets were derived from the DIRBE/IRAS dust maps of Schlegel et al. (1998) assuming the R_V=3.1 extinction curve of Cardelli et al. (1989). The final dereddened magnitudes of our sources are given in Table 2, together with their redshifts obtained from various papers of the literature. To increase the size of our sample, we also added in our analysis nine other GRB hosts with a determined K-band photometry already published by other authors (see caption of Table 2 for references). Including our results, the number of GRB host galaxies detected in the NIR by October 2002 thus amounts to 15 sources out of the $\sim$35 GRBs which were so far localized with a sub-arcsecond error box (Greiner 2002).