HST/STIS Images of GRB 980326

Images of the field about GRB 980326, taken approximately 1000 days after outburst. On the left is the drizzled image shown at native resolution. The field of view is 13". On the right is the same image convolved with the point-spread function (PSF). This procedure brings out a faint feature just visible in the left-hand image, highlighted in both images by a red circle. The centroid of the object lies only one pixel (0."025) from the astrometric position of the GRB. This offset is less than the width of the PSF (3.5 pixels) and is within the astrometric uncertainty. The diagonal streaks are diffraction spikes from nearby bright stars.

We observed the field of GRB 980326 with HST/STIS on 31 December 2000, just over 1000 days after the burst. The total exposure time was 7080 seconds in the 50CCD aperture. The images were reduced by the standard HST pipeline, and drizzled to a combined image with a pixel scale of 0."0254.

We projected the location of the early afterglow from a KeckII/LRIS R band image, taken by Filippenko, Leonard and Riess on March 27, 1998 (see GCN 33). We estimate the 1-sigma error radius to be less than 1.5 drizzled HST pixels, which includes both the scatter in the transformation (0.9 pixels, using 6 reference stars) and the measurement error of the transient in the Keck image (0.8 pixels).

Within one pixel of the estimated position, there was evidence of a small source. We therefore convolved the image with a gaussian with a FWHM equal to that of the PSF (3.5 drizzled pixels). This procedure uses the PSF as a matched filter to enhance the the signal-to-noise ratio of unresolved sources. In this convolved image an object is found, again offset by one pixel from the expected position with a significance of ~4.5 sigma. We estimate a magnitude (in the broad STIS 50ccd filter) of V = 29.25 +/- 0.25.

Bloom et al., (1999, Nature, 401, 453) have suggested that the unusual light curve of this GRB might be explained by an underlying supernova, and on the basis of this estimated a redshift of about z~1. A supernova at this redshift (which in this model peaks around a magnitude of 25) would be expected to be significantly fainter than V=29 by now (as would the rapidly decaying GRB afterglow). Therefore, we may be detecting a faint host galaxy. Deep HST images have in all other cases detected a host underlying the GRB, when good astrometry was available for the GRB. If we are observing a host galaxy, then the above magnitude could underestimate its brightness by as much as a few tenths of a magnitude, due to extended emission missed in this measurement. A galaxy at a z~1 with this observed magnitude would lie a rather remarkable 7 magnitudes below the L*, the knee of the of the galaxy luminosity function, at that redshift.

Given the apparent point-source nature of the detected object, and the lack of color information, we cannot exclude the possibility (which we consider remote) that we are observing a light echo, from either the GRB or the supernova.

This result was reported as GCN...