3 Data reduction and calibration

The images were corrected for a constant bias, dark current, and for spatial sensitivity variations using the respective master flats, computed as the median of all available sky flats of the specific run. Afterwards, photometry was performed using the DAOPHOT/ALLSTAR/ALLFRAME software, made available to us by Dr. Stetson (see Stetson [1987], [1994]). A preliminary photometry was carried out in order to construct a short list of stars for each single frame. This list was used to accurately match the different frames. With the correct coordinate transformations among the frames, we obtained a single image, median of all the frames, regardless of the filter. In this way we could eliminate all the cosmic rays and obtain the highest signal/noise image for star finding. We ran the DAOPHOT/FIND routine on the median image and performed PSF fitting photometry in order to obtain the deepest list of stellar objects free from spurious detections. Finally, this list was given as input to ALLFRAME, for the simultaneous profile fitting photometry of all the individual frames. We constructed the model PSF for each image using typically from 60 to 120 stars.

The absolute calibration of the observations to the V-Johnson and I-Cousins systems is based on a set of standard stars from the catalog of Landolt ([1992]). Specifically, the observed standard stars were in the fields: PG0231, SA95 (41, 43, 96, 97, 98, 100, 101, 102, 112, 115), SA98 (556, 557, 563, 580, 581, L1, 614, 618, 626, 627, 634, 642), RUBIN 149, RUBIN 152, PG0918, PG0942, PG1047, PG1323, PG1525, PG1530, PG1633, and Mark A. At least 3 exposures were taken for each standard field, with a total of $\sim$100 standard star measurements per night and per filter.

The reduction and aperture photometry of standard star fields were performed in the same way as for the cluster images. The aperture magnitudes were corrected for atmospheric extinction, assuming A_V=0.14 and A_I=0.08 as extinction coefficients for the V and i filters, respectively.

Figure 2: Calibration equation for each observing night. The upper panels refer to the run of April, while the lower panels refer to theDecember run. In all cases, the V filter curves are on the left side, while the I filter curves on the right

 

 

Table 2: Calibration parameters for each observing night

	Filter V
Date	$a_{\rm m}$	error	Cons.	error
11/Apr.	+0.024	$\pm0.001$	-3.034	$\pm0.002$
12/Apr.	+0.024	$\pm0.002$	-3.025	$\pm0.004$
13/Apr.	+0.024	$\pm0.002$	-3.059	$\pm0.003$
14/Apr.	+0.024	$\pm0.001$	-3.034	$\pm0.002$
15/Apr.	+0.024	$\pm0.003$	-3.057	$\pm0.004$
23/Dec.	+0.022	$\pm0.001$	-2.777	$\pm0.001$
24/Dec.	+0.022	$\pm0.001$	-2.790	$\pm0.001$
	Filter I
Date	$a_{\rm m}$	error	Cons.	error
11/Apr.	-0.012	$\pm0.003$	-4.081	$\pm0.004$
12/Apr.	-0.012	$\pm0.002$	-4.069	$\pm0.003$
13/Apr.	-0.012	$\pm0.003$	-4.086	$\pm0.004$
14/Apr.	-0.012	$\pm0.001$	-4.072	$\pm0.002$
15/Apr.	-0.012	$\pm0.001$	-4.076	$\pm0.004$
23/Dec.	-0.017	$\pm0.002$	-3.805	$\pm0.002$
24/Dec.	-0.017	$\pm0.002$	-3.829	$\pm0.002$

As shown in Fig. 2, a straight line well reproduces the calibration equations. As the seeing and the overall observing conditions were stable during the run, the slopes of the calibration equations for each observing run and for each filter have been computed using the data from all the nights. As it can be seen in Table 2, the standard deviations of the calibration constants for each run and filter is 0.015 mag, corroborating our assumption that all nights were photometric, and that we can assume a constant slope for each filter and run.

Standard stars for which previous problems were reported (PG 1047C, RU149A, RU149G, PG1323A; see Johnson & Bolte [1998]) were excluded, as well as saturated stars, those close to a cosmic ray, etc. After this cleaning, the mean slope was computed, and finally the different night constants were found using this slope to fit the individual data, night by night. The adopted values are presented in Table 2. The typical errors (rms) are also given.

The calibration curves are shown in Fig. 2 for both runs. In this figure, the dotted line represents the best fitting equation, while the continuous line is obtained by best fitting the data imposing the adopted mean slope. The two lines are almost overlapping. The mean number of standard star measures used for computing the curve per night and filter is $\sim75$. Notice the wide color coverage for the standard stars.

The last step on the calibration is the aperture correction. As no available bright and isolated stars exist on the cluster images, we used DAOPHOT to subtract from the image the stars in the neighborhood of the brightest ones, in order to compute the difference between the aperture and the PSF-fitting magnitudes. In view of the stable seeing conditions, we used the same aperture for calculating the aperture photometry of the standard and cluster stars.