As noted above, mfcal
solves for antenna gains and bandpass function.
It can also solve for the delay terms, but this is rarely useful. However
it does assume an unpolarised calibrator and that the polarisation leakage
terms are zero. If you have measured only the XX and YY correlations,
you have to make these assumptions. However if you have measured
all four polarisation products, then you can correct
for these assumptions by running gpcal
Determining Gains and Bandpass Functions - MFCAL
We will now discuss some of the inputs to mfcal.
- vis: The input dataset - the dataset containing the primary
calibrator in this case.
- line: The standard linetype parameter, used to select and
average channels. See the discussion of this in Section 5.4.
Assuming the 128 MHz / 33 channel continuum system, you
will want to discard the first and last few channels, calibrating only
the central 25 channels of so. Otherwise you
will not normally want to perform any averaging.
- edge: This is an alternative way of indicating the channels
to discard. When working with single IF datasets, it is probably
easier to use the line parameter.
- interval: This gives one or two numbers,
in minutes, being the maximum solution interval length, and
maximum solution gap, in minutes. The latter can usually be allowed to
default. The default maximum interval length is
During stable observing
conditions (good phase stability) you should set the maximum solution length
to no more than the calibrator scan time.
Set it to a small value if phase stability is poor (e.g. 1 to 0.1 minutes).
If you set it to a smaller value, however, the solution process becomes
more sensitive to bad data - make sure you flag it well. Additionally you
might run into program memory limitations, where it aborts due to ``buffer
overflows'' or the like.
- select: Normal data-selection.
- refant: The reference antenna. Set this to the antenna
which had the cleanest XY phase.
The default is 3 (which may not be very appropriate).
- flux: This gives the flux density and spectral index of the calibrator.
- options: There are a few extra options, which are not of
great use or interest.
- Note that old versions of mfcal
to solve for a delay by default. This is no longer the case. To solve
for delays (which is strongly discouraged when calibrating a single IF band),
you now have to explicitly use the delay option.
- The interpolate
option may be useful if you cannot determine the bandpass for some
completely flagged channels but still need an estimate of the bandpass
function. With this option, the bandpass is spline
fit (in real and imaginary) and evaluated for the flagged channels.
Be careful if a substantial fraction of channels are bad - the algorithm
is only implemented if less than 50% of the channels are flagged bad.
- If you are calibrated data that are to be combined
or compared with ATCA data reduced before August 1994, you will generally
want to use the `oldflux' option to select the pre-August 1994 ATCA
flux scale. See Section 12.6 for more information.