4 Flat field correction

The division of a CCD frame by a flat field does not necessarily provide a correct calibration of the CCD sensitivity variation. Errors up to 10% are seen on some telescopes depending on the baffling and internal reflections in the optics. Andersen et al. (1995) have developed an algorithm, where a correction is derived from a set of shifted and rotated images. It only works for fields with many stars (distant clusters). Using multiple shifted and rotated CCD frames by itself provides some averaging of the flat field errors, but an additional improvement is gained by the technique we have applied. For a more thorough description of the flat field correction the reader is referred to Bruntt (1999).

In Fig. 1 we show a contour plot of the correction surface in the y filter. Note that an error of $\simeq8\%$ is found from the centre of the CCD to the extreme edges. In the other Strömgren filters the errors are of the same order of magnitude. This clearly shows that serious problems may be present if one observes a star cluster with only one telescope position.

Figure 2: Histograms of the standard deviation of stars before (dashed) and after (solid) the flat field correction in the y filter calculated from (up to) 22 frames

In Fig. 2 the standard deviation of the brightest stars in the cluster NGC 6134 is seen. The standard deviation is calculated from the magnitudes determined from the (up to) 22 y frames.