The astrophysical background incident on the focal plane of Gaia is dominated by
stray light (Gaia Collaboration et al. 2016), in part because the compact and folded
design (Safa et al. 2004)
of the various optical instruments leaves little opportunity for
stray light path baffling. The lower rows (1–4)
of the focal plane have a diffuse background signal that is dominated by
sunlight scattered around the DSA while in the higher rows (5, 6, and 7) it
is the diffuse optical background from the Zodiacal Light and Milky Way that dominates.
Hence, the background consists of a high amplitude, rapidly changing
photo-electric component that repeats on the satellite spin period. Furthermore,
this component evolves slowly in both amplitude, with the L2 elliptical orbital
solar distance, and phase, as the scanning attitude changes with respect to
the ecliptic and Galactic planes. Superposed on this
are transient spikes in background due to very bright stars and bright Solar
system objects transiting across or near the focal plane. Moreover, there are
two non-photo-electric components to the background signal. These are a
charge release signal associated with electronic charge injections
which are used for on-board radiation damage
mitigation (Prod’homme et al. 2011) and the dark current signal.
The photo-electric background signal
varies routinely over three orders of magnitude depending on instrument and
spin phase with values as low as and up to electrons per
pixel per second.
Regarding the non-photo-electric signals, the charge release signal currently
varies between 1 and 10 electrons per pixel per second in the first TDI line
immediately after the last injection line but rapidly diminishes to 1% of
this level after the following TDI lines, while the
dark signal is all but negligible (see Section 2.3.4).
The parametric model for the astrophysical background consists of the
outer product of two one-dimensional spline functions (van Leeuwen 2007a) which
defines a flexible, two-dimensional surface model known colloquially as a
‘bispline’.
In the along-scan direction, the spline function is quadratic with knots evenly spaced at intervals that
adapt to the available data density. In regions of high density that generally
exhibit higher background fluctuations on smaller spatial scales, and in which
there is a higher density of data to constrain the model, knots are more
closely spaced. Typically, the along-scan knot interval is in the range 1
to 10 arcminutes.
In the across-scan direction, the spline function is linear, allowing for
sudden discontinuities in the gradient of the stray light pattern. Knot
positions are placed at fixed, unevenly spaced positions to best follow
rapid changes in the stray light pattern in the across-scan direction.
The charge release
model consists of a simple, empirical look-up table (LUT) as a function of
TDI line index following the last charge injection line with a power-law
scaling as a function of that injection level present at the same column
position on the CCD. This enables a single charge release LUT calibration per
device when applied in conjunction with the across-scan profile of the
charge injection for the same device which is also characterised via a LUT.
Compared to Gaia DR1, which employed background following from the daily
processing chain (IDT), in Gaia DR2 the astrophysical background spline models are improved by being refit in
the cyclic processing (IDU). In particular, additional data beyond the
along–scan limits of each independent processing interval are employed to better
constrain the bispline surface edge along that dimension. This leads to
more accurate background following near the boundaries between each
interval’s independently fitted model and consistent behaviour across those
boundaries.