2 Observations and basic data reduction

The data set discussed in this work has been obtained with the FOcal Reducer/low dispersion Spectrograph (hereafter FORS1), mounted at the Cassegrain focus of ESO-Antu/Melipal 8.2 m telescopes (Szeifert 2002). The instrument is equipped with a $2048~\times~2048$ pixels (px) TK2048EB4-1 backside thinned CCD and has two remotely exchangeable collimators, which give a projected scale of 0 $.\!\!^{\prime\prime}$2 and 0 $.\!\!^{\prime\prime}$1 per pixel (24 $\mu$m $\times$ 24 $\mu$m). According to the used collimator, the sky area covered by the detector is $6\hbox{$.\mkern-4mu^\prime$ }8\times6\hbox{$.\mkern-4mu^\prime$ }8$ and $3\hbox{$.\mkern-4mu^\prime$ }4\times3\hbox{$.\mkern-4mu^\prime$ }4$, respectively. Most of the observations discussed in this paper were performed with the lower resolution collimator, since the higher resolution is used only to exploit excellent seeing conditions (FWHM $\leq$ 0 $.\!\!^{\prime\prime}$4).

In the current operational scheme, FORS1 is offered roughly in equal fractions between visitor mode (VM) and service mode (SM). While VM data are immediately released to the visiting astronomers, the SM data are processed by the FORS-Pipeline and then undergo a series of quality control (QC) checks before being delivered to the users. In particular, the imaging frames are bias and flat-field corrected and the resulting products are analysed in order to assess the accuracy of the flat-fielding, the image quality and so on. The sky background measurement was experimentally introduced in the QC procedures starting with April 2000. Since then, each single imaging frame obtained during SM runs is used to measure the sky brightness. During the first eighteen months of sky brightness monitoring, more than 4500 frames taken with broad and narrow band filters have been analysed.

As already mentioned, all imaging frames are automatically bias and flat field corrected by the FORS pipeline. This is a fundamental step, since in the case of imaging, the FORS1 detector is readout using four amplifiers which have different gains. The bias and flat field correction remove the four-port structure to within $\sim$1 electron. This has to be compared with the rms read-out noise (RON), which is 5.5 and 6.3 electrons in the high gain and low gain modes respectively. Moreover, due to the large collecting area of the telescope, FORS1 imaging frames become sky background dominated already after less than two minutes. The only significant exception is the U passband, where background domination occurs after more than 10 min (see also Table 1). The dark current of FORS1 detector is $\sim$2.2 $\times$ 10^-3 e^- s^-1 px^-1 (Szeifert 2002) and hence its contribution to the background can be safely neglected.

 

 

Table 1: Typical background count rates expected in FORS1 (SR) images during dark time and at zenith. The last column reports the time required to have a background photon shot noise three times larger than the maximum RON (6.3 ${\rm e}^-$), which corresponds to a 5% contribution of RON to the global noise.

Passband	Count Rate	t3
	(e- px-1 s-1)	(s)
U	0.5	714
B	3.8	94
V	15.8	23
R	26.7	13
I	32.1	11

Since the flat fielding is performed using twilight sky flats, some large scale gradients are randomly introduced by the flat fielding process; maximum peak-to-peak residual deviations are of the order of 6%. Finally, small scale features are very well removed, the only exceptions being some non-linear pixels spread across the detector.

The next step in the process is the estimate of the sky background. Since the science frames produced by FORS1 are, of course, not necessarily taken in empty fields, the background measurement requires a careful treatment. For this purpose we have designed a specific algorithm, which is presented and discussed in Patat (2003). The reader is referred to that paper for a detailed description of the problem and the technique we have adopted to solve it.