kettenis
commented
4 days ago In the case of VLBI, the INTERFEROMETER_MODEL proposed in MSv3 effectively stores what is effectively the same information four times. And since the PHASE_DELAY is provided in seconds, it becomes identical to GROUP_DELAY (and PHASE_RATE becomes identical to GROUP_RATE). While older correlators such as the VLBA hardware correlator and the Mark4 hardware correlators) may have provided separate phase and group delays and rates, interpretation of those values requires intimate knowledge about the correlator that was used. the MSv3 schema provides no guidance on this.
Providing DISP_DELAY in seconds is somewhat odd as well since the dispersive contribution to the delay is frequency dependent. Once again the MSv3 schema provides little guidance here and omits specifying if a polynomial in time or in frequency should be provided. And if a polynomial in time is intended DISP_DELAY_RATE becomes redundant.
Same for the CLOCK_ERROR and CLOCK_ERROR_RATE. And the MSv3 schema does not specify whether the clock error is or isn't included in the PHASE_DELAY and/or GROUP_DELAY.
We may have to consult the geodetic community to determine what information really is essential here.
To fully support VLBI in MS v4 we would like to include the information that is contained in the following tables :
To demonstrate this we need an MS v2 that contains all these tables so that we can create a conversion example notebook and add unit tests. All of the tables except the Interferometer Model have been implemented. Furthermore, the Interferometer Model as defined in MS v3 does not meet the needs of the VLBI community.