Astron. Astrophys. 329, 522-537 (1998)

4. Rejection of poor quality data

Here we present the criteria of selection of usable images, usable measurements and usable stars. The two colours and the four seasons are treated independently.

4.1. Usable images

We eliminate images taken with too bad atmospheric conditions. Fig. 5 presents the distribution of the sky background, absorption, seeing and roundness of the stars, as well as the adopted cuts, which are seen to be rather loose. About 10 % of the images are eliminated.

Fig. 5. Distributions of the sky background, absorption, seeing and roundness of the stars. Dashed lines indicate the adopted cuts.

A first search for variations is performed on light curves (a "variation" is defined in Sect. 5.2; roughly speaking, it is an increase or a decrease of the luminosity). A night is eliminated if too many large variations ( 150) occured during this night. About 15 nights per season (out of 250 nights) are thus removed; the origin seems to be related to electronics and it affects all CCDs.

We then re-do a detection of variations on light curves, excluding the "bad" nights. An image is eliminated if too many main variations ( 20) are detected with a maximum at the time of this image. We eliminate about 5 images per CCD and per season.

4.2. Usable measurements

Columns of a CCD are eliminated if too many main variations are detected on stars located on them. Moreover, we do not take into account a measurement of a star located too close to an edge of a CCD. Generally 90 to 95 % of good CCDs areas are usable (70 % for the worst one). If a star is located on a defective area on an image, this measurement is rejected from the light curve.

We retain a flux measurement only if it is higher than 5 over the local sky background fluctuations ( 10 ADU -Analogic Digital Unit).

4.3. Usable stars

For each detected star on the red reference image, we must associate a star on the blue reference beyond any doubt. Because of blending effects, we are not using any constraint on achromaticity for the detection of a microlensing event; it is thus not needed to have both light curves well sampled: we only ask for at least 50 % good measurements in one colour and 10 % in the other one.

About 75 % of the detected stars satisfy these criteria.

4.4. Summary

Table 2 summarises the number of usable stars and measurements for the four observing seasons.

Table 2. Characteristics of the four years of observation. Column 1 (resp. 2) indicates the number of red (resp. blue) measurements averaged on all CCDs. Column 3 indicates the number of nights and column 4 the number of used stars. The total number of photometric measurements actually used in this analysis, after the cuts described in this section, is about .

© European Southern Observatory (ESO) 1998

Online publication: December 8, 1997
helpdesk@link.springer.de