Fitting the Beams to the Parallel-hands Gain Amplitudes
The user has the option to solve for the un-normalised gain amplitudes (G-terms)
across the entire frequency band (not recommended for wide-bands) or in a number
of frequency chunks for the parallel-hands polarisations. In the latter case,
the top and bottom part of the chunks are discarded (so using a frequency chunk
of two is not permissible) and the beams are then fitted to the time-frequency
averaged G-terms in each chunk for each parallel-hand polarisation. These fitted
offsets for the parallel-hands polarisations are then averaged. This option can
be requested with the --fit_to_sep_pol command line argument. A concern
associated with this approach is the possibility of low SNR compared to fitting
to the Stokes I amplitude. It is therefore possible that some dishes could have
invalid fits in one or both parallel-hand polarisations because of low SNR. This
option should therefore only be used if the telescope is well understood,
possibly via repeated pointing calibrations.
The un-normalised G-terms for each scan for each parallel-hand polarisation, over number of dishes with one curve per frequency chunk in the 950 - 1350 MHz band with RFI mask applied for the sample 9-point scan with the MeerKAT array is shown below.
The Gaussian fits in cross-elevation and elevation in each frequency chunk for the parallel-hands are shown below for dish m000 from the sample 9-point scan with the MeerKAT array.
The weighted-average of the fitted parameters for all dishes from fitting to the parallel-hands polarisations gain amplitudes from the sample 9-point scan with the MeerKAT array
