BPOLY solutions

NRAO Home > CASA > CASA Cookbook and User Reference Manual

4.4.2.3 BPOLY solutions

For some observations, it may be the case that the SNR per channel is insufficient to obtain a usable per-channel ’B’ solution. In this case it is desirable to solve instead for a best-fit functional form for each antenna using the bandtype=’BPOLY’ solver. The ’BPOLY’ solver naturally enough fits (Chebychev) polynomials to the amplitude and phase of the calibrator visibilities as a function of frequency. Unlike ordinary ’B’, a single common ’BPOLY’ solution will be determined for all spectral windows specified (or implicit) in the selection. As such, it is usually most meaningful to select individual spectral windows for ’BPOLY’ solves, unless groups of adjacent spectral windows are known a priori to share a single continuous bandpass response over their combined frequency range (e.g., PdBI data).

The ’BPOLY’ solver requires a number of unique sub-parameters:

bandtype        =    ’BPOLY’   #   Type of bandpass solution (B or BPOLY)
     degamp     =          3   #   Polynomial degree for BPOLY amplitude solution
     degphase   =          3   #   Polynomial degree for BPOLY phase solution
     visnorm    =      False   #   Normalize data prior to BPOLY solution
     maskcenter =          0   #   Number of channels in BPOLY to avoid in center of band
     maskedge   =          0   #   Percent of channels in BPOLY to avoid at each band edge

The degamp and degphase parameters indicate the polynomial degree desired for the amplitude and phase solutions. The maskcenter parameter is used to indicate the number of channels in the center of the band to avoid passing to the solution (e.g., to avoid Gibbs ringing in central channels for PdBI data). The maskedge drops beginning and end channels. The visnorm parameter turns on normalization before the solution is obtained (rather than after for solnorm).

The combine parameter (§ 4.4.1.5) can be used to combine data across spectral windows, scans, and fields.

Note that bandpass will allow you to use multiple fields, and can determine a single solution for all specified fields using combine=’field’. If you want to use more than one field in the solution it is prudent to use an initial gaincal using proper flux densities for all sources (not just 1Jy) and use this table as an input to bandpass because in general the phase towards two (widely separated) sources will not be sufficiently similar to combine them, and you want the same amplitude scale. If you do not include amplitude in the initial gaincal, you probably want to set visnorm=True also to take out the amplitude normalization change. Note also in the case of multiple fields, that the ’BPOLY’ solution will be labeled with the field ID of the first field used in the ’BPOLY’ solution, so if for example you point plotcal at the name or ID of one of the other fields used in the solution, plotcal does not plot.

For example, to solve for a ’BPOLY’ (5th order in amplitude, 7th order in phase), using data from field 2, with G corrections pre-applied:

bandpass(vis=’data.ms’,          # input data set
         caltable=’cal.BPOLY’,   #
         spw=’0:2~56’,           # Use channels 3-57 (avoid end channels)
         field=’0’,              # Select bandpass calibrater (field 0)
         bandtype=’BPOLY’,       # Select bandpass polynomials
           degamp=5,             #   5th order amp
           degphase=7,           #   7th order phase
         gaintable=’cal.G’,      # Pre-apply gain solutions derived previously
         refant=’14’)            #

[next] [prev] [prev-tail] [front] [up]

More information about CASA may be found at the CASA web page

This code is available under the terms of the GNU General Public Lincense