2. Observations and reductions
The observations were conducted during two service runs, with the 3.9-m Anglo-Australian Telescope, using the prime focus CCD camera system. The CCD was a 512320 RCA chip, with a pixel size of 30µ (corresponding to 0.492"), readout noise of 37e-, and gain of 4.7e-/ADU.
The cluster NGC 2098 was observed on the night of the 24/25-11-1989, while SL 666 was observed on 2/3-1-1992. Conditions were photometric in both occasions, while the seeing was in the first case and in the second case.
In SL 666, two overlapping regions were observed, A and B, as shown in Fig. 1. For each region, two 300s exposures were obtained through a standard B filter, as well as one 20s exposure. Similarly, two 150s and one 15s exposures were obtained through the R filter.
As in the case of SL 666, two overlapping regions were also observed in NGC 2098 (denoted by A, B, see Fig. 1). Another region well beyond the cluster (F) was also observed for comparison. One short (60sec, for the bright stars in the central regions) and one longer exposure (of 300 sec) were obtained through the B filter for each one of these regions. Correspondingly, 30s and 150s exposures were obtained through the R filter.
Table 1 gives the coordinates of the centres of the various fields observed in and around the two clusters. Additionally, a series of standard stars in E-regions E2,3,4 & 5 (Graham, 1982) and in two `selected areas' SA94 & 96 (Landoldt, 1992) were observed during the two nights, both in B and in R, to provide the necessary transformation of the photometry to the standard Cousins system.
Table 1. Coordinates of the centers of the fields observed
The raw frames were linearised, bias-subtracted and flat-fielded (with dome flats) in the usual way, using standard IRAF 1 procedures. Subsequently, the program DAOPHOT (Stetson, 1987) was used to derive relative photometry of the stars automatically detected by the routine in each frame. The program applies a point spread function (psf) fitting method. After the necessary aperture corrections to the psf-based magnitudes, the resulting `instrumental' magnitudes (binstr, rinstr) were transformed into standard B, R magnitudes. We adopted the mean colour and extinction coefficients for the particular system and site (Bedding & Robertson, 1988), and used the observations of the standard stars to calculate the zero-point corrections and , according to the equations
where secz the airmass. The instrumental zero points and the corresponding errors for SL 666 is for B and for R. For NGC 2098 the values are for B and for R. In the case of SL 666, the two available long-exposure frames were analysed separately and the resulting magnitudes were averaged (by weight of the corresponding errors).
Fig. 2(a,b) shows the distribution of the internal DAOPHOT psf fitting errors as a function of magnitude in SL 666. The same is shown in Fig. 3(a,b) for NGC 2098. The inferior quality of the NGC 2098 data is immediately obvious. Reduced completeness in the detection of stars is expected towards the centres of the clusters. The completeness corrections that have to be applied to the observational luminosity functions and star counts (either overall star counts or star counts in particular ranges of magnitude and colour) were determined as a function of magnitude, from false-star experiments. This method is now common practice and gives quite reliable results (e.g. Stetson & Harris, 1988; Bertelli et al., 1992). Fig. 4 gives the completeness for SL 666 (Fig.4a) and NGC 2098 (Fig. 4b) as a function of B and R magnitude and at different radial distances. It should be noted here, that only data for which the completeness is better than 75% were actually used in the quantitative analysis that follows.
In the case of NGC 2098, the completeness shows a decline at the bright end. Possibly this is a result of the statistical effect due to the completeness experiments.
© European Southern Observatory (ESO) 1998
Online publication: July 20, 1998