2. Instrumentation and observations
The first ISO observations were scheduled on 5 April, with the CAM (Cesarsky et al. 1996) and PHT (Lemke et al. 1996) instruments. Filters CAM LW10 (8-15 µm) and PHT C 160 (134-218 µm) were used respectively. The second set of observations was performed on 13 April. In order to cover the full 3 radius GRB error box and the 50 radius BeppoSAX error box of the GRB970402, the observing strategies listed in Table 1 were adopted.
Table 1. Journal of the ISO observations presented in this paper.
The dominant sources of noise for the CAM long-wavelength (LW) detector for faint source work are cosmic ray glitches and flat-fielding imperfections. Both problems are alleviated by ensuring that each point on the sky is visited by as many array pixels as possible within the available time, while still ensuring adequate dwell time per position for minimal detector stabilisation to occur, and to support deglitching. Splitting the observation into two separate rasters improves glitch rejection.
Once the images were sky-substracted, flat-fielded and the glitches were removed, photometry was acquired. A 3 pixel diameter aperture (18 ) was used.
In regions of high diffuse illumination or high source density, CAM's sensitivity may be degraded from the expectations due to source confusion "noise". Note that the diameter of the central spot of the ISO PSF is, in arcseconds, 0.84 times the wavelength in µm, or 8 .4 at 10 µm. Such confusion effects limited the achieved sensitivity for these measurements on the crowded field of GRB 970402, relative to what could be achieved on a relatively empty field (5- 100 Jy for the strategy used).
The data was reduced using ISOCAM Interactive Analysis Software (CIA) (Ott et al. 1997) and post-processed to remove residual glitches and unstable background using a wavelet transform-based program developed at CEA, Saclay (Starck et al. 1995). Processing steps involve deglitching, dark field subtraction, flat-fielding and responsive transient correction. Increased confidence in the photometry and glitch rejection was achieved by comparing in detail the different rasters recorded on each of the two days used for ISO observations.
Careful inspection of the noise statistics at different parts of the raster images reveals a 5- detection limit of about 140 µJy for point sources inside the GRB error circle, with the sensitivity falling to about 300 µJy at 5- towards the edges of the raster coverage, outside the GRB error box. In Fig. 1 a profile of the 5- sensitivity limit as a function of position across the raster is presented for a coaddition of the two 5 April 6 /pixel field of view (pfov) rasters. This is the relevant measure of precision for this paper which compares results from the raster pairs taken on different days. A somewhat better sensitivity limit will be achieved when these ISO observations are all coadded for comparison with future follow-up observations on this field. Also, more sophisticated photometry routines could be applied to reduce scatter in the measurements and eventually further constrain the upper limit from CAM.
The PHT observations were taken in the C 160 filter which is centred at 174 µm with / = 0.51. This filter is used in combination with a 2x2 detector array with 92 square pixels (Lemke et al. 1996). At 175 µm the point-spread function has a FWHM of 80 . An area of 8 x8 was mapped using a 9 x 9 grid raster with 46 steps in both directions. The resulting noise level in the central 6 x6 is about 70 mJy/pixel. However, at 174 µm the sensitivity is limited by the cirrus confusion noise which has an amplitude of more than 70 mJy for spatial frequencies of order 90 at the galactic latitude (b = -9°) of our target. The resulting 5- point source PHT sensitivity is 350 mJy.
© European Southern Observatory (ESO) 1998
Online publication: January 8, 1998