Short-term amplitude variability

[jmoldon]

Sometimes, some spw show short-term amplitude modulation that can change in a fraction of minutes. This is very clear in some L-band data of 1331+305. The pipeline currently solves amplitudes using a solint of 180s, and that means that that variability is not corrected.

For example, with 3 min solution interval, this is corrected amp vs time for 1331+305:

And these are 16s solution interval corrected amp vs time for 1331+305:

We have different options to solve this:

• Divide the 'ap' gaincal steps in two: one with a short solint for bright sources (1331+305, 0319+415, maybe also 1407+284), and another with longer solint for the phase calibrators because they are usually not so bright.
• We can follow the same approach as for phases: two tables are computed, one with short interval solint that gives you the best amplitude scale, and a second one per_scan (solint='inf') that is the one actually applied to the targets.
• A mixture of both. Three runs of 'ap' calibration: (1) Short solint (8 or 16sec) for bright sources, (2) medium solution interval (maybe 30 or 60 sec) for phase calibrators [ideally the script should figure out if the source is bright enough to go to shorter solint]. This will be used during calibration so the flux scale is as best as possible. (3) Run on phase calibrators with solint='inf', for solutions that will be applied to the targets.

That will probably mess up things a bit but is probably the most accurate.

[jmoldon]

Fixed in be4fc94f9d1f52d6fd94b3887f82b30962a9c9b0