Below, we present the more important models applied for the calibration of
the SM and AF CCDs. Some are operated on longer timescales and others on shorter
scales, but they are all operated on shorter timescales than anticipated before
launch. For an overview on the detector performance during the first couple
of years in space, see Crowley et al. (2016).
The CCD cosmetics (see Section 2.3.4) deal with
questions related to individual CCD columns, like saturation level and abnormal
The CCD bias and bias non-uniformity (see
Section 2.3.5) exposes the difficulties involved in
reading CCDs in window mode, where most samples are merely flushed.
As a consequence, the precise timing for reading a particular column within the
short (less than 1 ms) time available for reading a line of pixels, varies
from time to time that column is read, and so does the bias. The model
must therefore take the exact readout details into account.
The astrophysical background model (see
Section 2.3.3) includes in fact several elements. It must
describe the complexities of the two sky areas that overlap in the
focal plane, but the dominating background source is the stray light which
varies strongly with the spin phase of the spacecraft with respect to the Sun,
and produces a complex, intermittent pattern on the focal plane (see Gaia Collaboration et al.2016). In addition,
the background model must also incorporate the charge release trails following
the regular charge injections on each of the AF and BP/RP CCDs.
The PSF/LSF model (see Section 2.3.2) must
encompass many complex effects, but for the second data release, several have
been waived. The optical PSF depends on colour, the field of view, and the
position in the focal plane, but it also changes with time. On top come
effects induced by the scanning law, by the way the CCDs are operated,
and by complex inefficiencies of the charge transfer within the CCDs. A final
complexity is that the chromatic image shifts are included in the PSF model
as shifts of the PSF origin.