The basic CCD reduction was carried out using IRAF's CCDRED package
(Tody 1986).
The raw images were trimmed and overscan subtracted. No
additional bias subtraction was found necessary. Multiple dome
flat-fields (sky flat-fields in the case of the fields)
were averaged, and the data flat-fielded using the resultant images.
Further details of this process can be found in Massey
(1992).
Since the CCD pixel size (0.87 arcsec) was large compared to typical seeing conditions encountered at Las Campanas we measured the full-width at half-maximum (FWHM) of a number of moderately exposed stellar images on each frame and found the average value to be close to 2 pixels. Thus we judged the stellar images to be undersampled for determining and applying the point-spread function (PSF) method of photometry needed for the crowded cluster fields.
To overcome this we first used the aperture photometry routines within the DAOPHOT (Stetson 1987; Stetson et al. 1990) package in IRAF for measuring the Landolt standards. An aperture of radius 16 pixels or 14 arcsec was chosen to match Landolt's original aperture as closely as possible. These measurements were used to transform measurements from the instrumental system to standard magnitudes (Johnson-Cousins). The equations used were of the form:
Because of crowding, particularly a problem for the fainter stars, it was necessary to use PSF photometry on the cluster frames. A best fit PSF was determined and subtracted for each object frame independently. The residuals from the PSF subtraction were found to be mainly in the core of the profile and these were measured using aperture photometry with a radius equal to the FWHM of the PSF. This latter quantity was typically 10% of the PSF flux itself, independent of magnitude. A PSF correction was accordingly determined for each cluster star individually and applied to the PSF-derived magnitude followed by an "aperture correction'' as described above. The accuracy of this method was tested using isolated stars on the object frames and was found to produce errors which were less than those from other sources (see Table 2 (click here) below). These instrumental magnitudes were then transformed to the standard system using the transformation equations described above.
The three sets of V, R, I photometry, resulting from the three
individual deep CCD exposures in each colour, were then combined in order
to eliminate spurious detections resulting from the undersampling (e.g.
cosmic rays). Only
stars which were common to at least two of the three sets were used. The
photometry was then averaged. 77% of the stars thus selected had a
range in measured V magnitude less than the averaged internal photometric
error for that star. The resulting deep-frame photometry was then merged
with that determined using the short-exposure frames.
The positions of each star was determined using
the STARLINK package ASTROM and star positions from the UKST/COSMOS
database, and as such are accurate to arcsec.