Scattered light and illumination correction
The VST dome and telescope are very open designs and the telescope is therefore particularly prone to scattered light entering the optical train from non-astronomical sources - as is clear from the beautiful reflections visible in these pictures.
Our master flatfields are made from twilight flats which include both scattered light from non astronomical sources as well as scattered light generated from mulitple internal reflections within the optics. We use the master flats to: correct for pixel-to-pixel sensitivity variations within detectors; place all the detectors on a common internal gain system; and correct for any vignetting of the telescope optics.
Scattering within the optical train and vignetting tend to produce radially symmetric scattered light patterns however, non astronomical scattered light can add quite subtle variations. Unfortunately this latter source of scattered light is present in different patterns everytime new twighlight flatfield frames are taken.
It is usually impossible to tell directly from simple examination of a flatfield frame if scattered light is going to be a serious issue, although as in the i-band flatfield image for VST below, it is easy to spot the presence of multiple internal reflections from the optics.
Why is this a problem ? The flatfielding process is predicated on uniform illumination from, in this case, the twilight sky. Although the sky may well be uniform to better than 1% the extra scattered light most certainly won't be. This causes an equivalent zero-point variation of the photometric calibration over the field.
Application of the flatfields yield reasonably flat dark sky science frames since the astronomical scattered light is the same in dark sky and twilight skies. However, the non-astro scattered light is generally different in both cases and may be visible as subtle gradations in background over the field.
The simplest way to measure the impact of this on the photometry is to acquire images of dense well-sampled accurately calibrated photometric fields and examine the magnitude residuals as a function of position on the array. Unfortunately such regions do not generally exist over such fields, 1 sq deg in this case.
In lieu of this, and in order to make an accurate estimate of the scale of the problem, we have taken ATLAS i-band and z-band observations processed using the same master flats for the period 16th August - 30th September and stacked photometric residuals by using 2MASS as a photometric catalogue. The i-band and z-band colour equations required for this are available in the VISTA technical section. The ATLAS survey is ideal for this as there are several hundred independent pointings in both bands available over this period.
Examples of the required illumination corrections (i.e. magnitude corrections) derived in this way are shown below together with the overall variation as a function of radius, where the latter is an attempt to demonstrate scattered light (and vignetting) in the optics (i.e. the radially symmetric component) from external scattered light effects.