Heresy you say. Of course I must calibrate. I love to calibrate. Slowly but surely, we are moving in the direction of producing fully calibrated data from the ATCA. We are not there yet, but we are part of the way. This section gives some hints on how useful the on-line calibration is, but does really assume you have more knowledge than has so far been presented in this manual.
When you began your observation, you did some set up chores which involved a program called CACAL (formerly you would have used DELCOR). These chores were really to make a basic on-line calibration of antenna delay and complex gain. These are the very same gains that the usual offline secondary calibration process wants to redetermine. Because CACAL knows the spectrum of 1934-638 (if you set up with it) your data should already be on a flux density scale that is essentially correct (if 1934-638 was at a very low elevation it might be a bit wrong). That is, the antenna complex gain amplitudes should be pretty much right.
The usual idea behind secondary calibration is that the atmospheric
phase changes on a scale of size of roughly a few degrees on the sky (at
cm wavelengths), so that the phase you determine from the primary
calibrator (either on line or off line) is probably not going to
transfer unless you happen to be observing very close to it. In
addition, the atmosphere is dynamic and the atmospheric phase changes
with time; sometimes very rapidly and sometimes not so rapidly. The
secondary calibrator is used to determine the phase close to your
program source and to track it with time. The main component of the
change of gain phase with time is the atmosphere. The receivers are
very phase (and amplitude) stable and are unlikely to drift significantly
during your observation. We generally also use the secondary
calibration to redetermine the gain amplitudes with time because the
corrections made on line to account for attenuation effects with
elevation (
correction) are usually not perfect.
Now although the phase may wander with time, it very unlikely that the differential phase between the XX and YY feeds will change. Since the basic on-line calibration results in XX and YY being in phase, they will remain so, even if their phase changes with time. This means that in AIPS, rather than examining the XX and YY data separately, you can combine it straight away and examine Stokes I. For example, when (if necessary) you are editing your data with TVFLG, say (see § 7), you could request Stokes I and pass through the data once, rather than requesting XX and then YY and passing through the data twice.
Now all this means that if the phase stability of your observation was good, it is possible that the basic on-line calibration that you did will be useful at some level. Unfortunately, we do not yet produce an on-line band-pass calibration, so if you are doing spectral-line work, then you need to work your way though the off-line calibration procedure. If you plan to do an optimum calibration in MIRIAD, you should do just that. But let us say you have decided to remain within AIPS. Let us also say that you do not need a band-pass calibration and you have averaged all of your channels together (see § 6 for details). You could then make an image of Stokes I (no other Stokes parameters are available in AIPS for ATCA data) directly from the on-line calibrated data. Then, if your source was strong enough, you could rely upon self-calibration (see § 17) to improve the complex gains.