3 Observations

Photometric observations using Bessel UBRI and Gunn g broad band filters were carried out with FORS1 at the ESO-VLT UT1 during 5 observing runs in visitor mode between August and December 1999. The data were complemented with some additional service-mode observations in the Bessel B and I filters with the same telescope in July and August 2000. Observing conditions were mostly photometric except for the August 1999 run, which was hampered by the presence of clouds and strong winds during some of the nights. In all cases a $2 \times 2$ k TEK CCD in standard resolution mode ( $0\hbox{$.\!\!^{\prime\prime}$ }2$/pixel, FOV $6\farcm8 \times 6\farcm8$), low gain and 4-port readout was used. The Gunn g filter was chosen instead of Bessel V in order to avoid the 5577 Å night sky emission line, thus reducing the background significantly.

From the field-selection images taken with FORS1 it was known that twilight flatfields alone are not sufficient for a data reduction reaching very faint magnitudes. Therefore the images were taken in a jittered mode. A 4 $\times$ 4 grid with a spacing of $8\hbox{$^{\prime\prime}$ }$was adopted in order to maximize the use of the scientific images for flatfielding purposes on the one hand, and to minimize the loss of field-of-view on the other hand. The order of the individual observing positions was such that images with the largest separation were always taken first.

Exposure times for the individual frames were set to 1200 s in U, 515 s in B and g, 240 s in R and 300 s in I. The seeing limit was initially set to $0\hbox{$.\!\!^{\prime\prime}$ }5$ for B and Iand $0\hbox{$.\!\!^{\prime\prime}$ }8$ for the remaining filters. Unfortunately, it became clear after the first observing run that those seeing limits were too strict (mainly due to the La Niña phenomenon at that period (Sarazin & Navarrete 1999; Sarazin 2000), and could not be met within a reasonable amount of telescope time. Therefore the seeing limits were relaxed to 1 $\hbox{$^{\prime\prime}$ }$ for U and g and $0\hbox{$.\!\!^{\prime\prime}$ }8$ for the B filter.

Due to the different seeing goals for each filter and varying seeing conditions during some of the nights, images in 3-5 filters were typically taken during each observing run. This resulted in somewhat longer exposure times on the summed images than initially anticipated (see Sect. 5). Photometric standards from Landolt (1992) were taken at least once during each photometric night.

NIR observations of the FDF in the J and Ks filter bands were acquired using SofI at the ESO NTT during 3 photometric nights in October 1999. Since the field-of-view of SofI with the large field objective is $4\farcm94 \times 4\farcm94$ ( $0\hbox{$.\!\!^{\prime\prime}$ }292$/pixel) only and, thus, significantly smaller than the field-of-view offered by FORS1, the observations were split into 4 subsets to cover the entire FDF.

In order to have as similar observing conditions as possible for all subsets, the observations in both NIR filters were distributed evenly over the three nights. Always at least all four subsets were observed subsequently in one filter for 20 min. Each set of 20 min consisted of 20 exposures of $10\times 6$ s. The positions of the four subsets were chosen so as to cover the entire FDF as observed by FORS with a maximal overlap of the subsets, but to avoid the southernmost 100 pixels of the SofI camera, which show image degradation (see SofI manual). To allow a good sky subtraction, jittered images were taken. We used a random walk jitter pattern within a rectangular box of $22\hbox{$^{\prime\prime}$ }$ border length centered on the central position of each subset. Photometric standard stars from Persson et al. (1998) were observed 3 times during each night to set the zero point.

In the end, the entire FDF was imaged effectively for 100 min in the two NIR filters. Due to the overlap of the individual subsets a narrow region was observed effectively for 200 min and the central region (including the QSO) effectively for 400 min. An overview of the optical and NIR observing runs and the filters used is given in Table 2.