3 Observations and reduction

The observations were performed in two runs (August and September 1995), using the Nordic Optical Telescope (NOT) at La Palma, equipped with the ARcetri Near-Infrared CAmera (ARNICA, Lisi et al. 1994) and they were part of the 1995 International Time Programme for the telescopes on the Canary Islands on optical/IR follow-up studies of WENSS sources.

ARNICA relies on a 256$\times$256 HgCdTe NICMOS array and was used with a scale of 0.55 arcsec/pixel and a field of view of 2.35 arcmin.

Tables 1, 2, 3 list the observed sources for the USS, GPS, and FS samples respectively. In the first columns of the Tables we give some information from other authors on the optical identifications of the three samples.

All sources were observed in K-band. Some of USS and GPS sources, in particular those which were not clearly detected on the K-band images after the quick-look reduction at the telescope, were also observed in the J-band to increase the chances of an infrared detection. We used the method developed by Tyson (1986) and collaborators for deep CCD imaging, in order to ensure background-limited images and avoid the saturation of the brightest sources. To this aim, for each object, the total exposure time was broken up into a number of shorter exposures, background-limited, between which the telescope was moved in a 9 position raster. For calibration purposes we observed standard stars, typically four per night, from the ARNICA list (Hunt et al. 1998). The seeing was good throughout our runs, between 1.0 and 0.5 arcsec and conditions were mostly photometric.

The reduction of the images employed the IRAF procedures available in the ARNICA data reduction package. In order to eliminate the spatial variations in the detector response, arising from camera vignetting and variation in the quantum efficiency of the detector, we applied the flat field correction, that is the division into each frame of a sky flat frame. As a rule, for each of the 9 positions of the telescope, the data reduction procedures compute a weighted median of the images of the other 8 positions (or a subset of them) to form the sky frame for flat-fielding. However, in the case of K-band images, it was necessary to do the background subtraction from each frame before flat-fielding by means of normalized differential sky flats (Hunt et al. 1994) in order to eliminate the contribution of the thermal emission from the telescope that becomes significant for this band. Finally, the frames were combined together to produce the final source frame, using a new algorithm, invented by Hook & Fruchter (1997) in order to combine multiple stacks of dithered, undersampled image frames.

 

Table 1: Sample of USS sources

 

Table 2: Sample of GPS sources

^*G = galaxy; FG = faint galaxy; Q = quasar.

^**These apertures were used for comparison with optical photometry (Snellen, PhD thesis).

 

Table 3: Sample of FS sources