Terrestrial frame motion correction responsibilities

There are a number of well known effects that are apparent in the position of celestial objects due to the motion of the ITRF with respect to the celestial frames of reference. Not limited to the precession of the spin axis, the nutation of the axis, polar motion (the motion of the crust relative to the spin axis). There are also effects that are due to the path light takes from the object to the observer (aberration, tropospheric delay, ionospheric shifts etc)

The telescope model needs to parameterise all of these effects, and perhaps others into a delay model, usually described as a polynomial with some validity, for some period of time. This polynomial is evaluated by the correlator elements so that a delay can be applied to the signals from each telescope in order to form a correlation in the desired direction.

Once the visibilities are measured in that direction, they are passed on the SDP receive workflow and are written into measurement sets. The measurement sets are then provided to various workflows for the purposes of imaging and calibration.

The visibility is essentially the amplitude and phase of a correlation at a particular spatial frequency - the spatial frequency is encapsulated in the position in the cube of U, V and W at which the measurement is made, W is equivalent to the "delay" in the correlator delay polynomial.

As the CBF does not send the UVW associated with the delay model for each baseline, something in the ingest process has to determine the UVW of each visibility and add it to the measurement set. If the UVW are measured in metres only one measurement is required for all frequencies for a given baseline.

The precision with which this UVW is determined and the frame of reference in which it is made makes a substantial contribution to the quality and fidelity of an image. If the visibility is placed in the UVW cube at a point which does not correspond to the place at which it was measured then the image quality will suffer.

Ideally <all> the effects that are in the delay model should be applied to the calculation of the UVW which are in the measurement set. But it is my understanding that is not routinely done by observatories. Also there are some interesting features of precession that need to be accounted for in imaging. For example it is insufficient to simply precess and transform the J2000 source position to an apparent position as is done for the delay model, , then calculate the UVW via the standard XYZtoUVW transformation - as although precession will be correct for the phase centre, the rest of the image will not be properly corrected and consequently all sources away from the image centre will be rotated away from their catalogue position due to precession. It should also be noted that all effects on the apparent source position that are applied to the delay model - but not applied to the UVW for imaging - will result in slight rotations of the image and some loss of image quality.

This issue is a request for clarification as to the following issues:

1) Confirmation of how the delay polynomial will be determined and how it will be provided to the CBF.

2) How many of the effects of the Earth's motion and phase centre direction that are applied to the delay model are required to be included in the UVW calculation.

3) Which (sub)element is responsible for the UVW calculation for the purposes of inclusion in a measurement set.