3.4.10 Line and Point Spread Functions

Author(s): Nicholas Rowell

A full description of the LSF/PSF calibration pipeline can be found in section 4 of Rowell et al. (2020). In this document we include a brief summary of the main points that can be found therein.

The LSF/PSF calibration pipeline is tasked with solving the parameters of the LSF/PSF models, described in Section 3.3.5, over the whole time range and for all subsets of the data in the SM and AF instruments. The calibration pipeline is split into a series of stages that perform the following operations:

1. 1.

   selection and pre-processing of observations eligible for use in the calibration,

2. 2.

   independent solutions of the LSF/PSF model parameters in 0.5 revolution time steps,

3. 3.

   merger of the independent solutions in a square root information filter, with resets of the solution at major mission events,

4. 4.

   compilation of output products and auto-qualification.

These steps are implemented in IDU in a set of tasks that are executed in sequence. Independent calibrations are made for all distinct combinations of CCD, telescope, gate and window class that are routinely in use, of which there are 1268 in total made up of 1020 2D PSF calibrations and 248 1D LSF calibrations. In practise, and in order to handle some deficiencies in the modelling and noise in the data, independent calibrations of the SM PSF have not been produced and instead were replaced with the calibrations from AF2, which are judged to be similar due to proximity in the focal plane. In addition, the AF window class 2 calibrations were too noisy and were replaced with the corresponding window class 1 calibration; these are coincident in the focal plane and are expected to have identical linear components of the LSF so this is not considered to be a compromise. The LSF/PSF calibration products thus consist of one complete set of 1268 LSF/PSF solutions for every 0.5 revolution time step throughout the Gaia EDR3 period. These are subject to automated qualification algorithms that verify their integrity according to various criteria.

The main conclusions concerning the LSF/PSF calibration for Gaia EDR3 are:

• •

  The LSF and PSF show strong dependences on time, colour and AC position.

• •

  The PSF additionally has a major dependence on AC rate.

• •

  The time variation is significant and is characterised by a gradual deterioration in image sharpness punctuated by step changes at decontaminations and refocusses.

• •

  There is a known problem concerning the PSF calibration for long gates, which is discussed in Section 3.5.4.