E.7 Polarization calibration — RLDLY, PCAL, and RLDIF

You may skip this section unless you have cross-hand polarization data and wish to make use of them. Although there have been major improvements in AIPS polarization routines, they still do not correct parallel hand visibilities for polarization leakage. Thus you need to calibrate polarization only if you wish to make images of target source Q and U Stokes parameters. Polarization calibration is discussed extensively in 4.6; we will discuss changes made because of the wide bandwidths and other aspects of the EVLA.

Frequently, the delay difference between right- and left-hand polarizations must be determined even if FRING was not required for the parallel-hand data. Use POSSM to plot the RL and LR spectra to see of there are significant slopes in phase. If so, use a calibration source with significant polarization, although the EVLA D terms are often large enough to provide a usable signal in the absence of a real polarized signal. Note that 3C286 is significantly polarized and is likely to be the best source to use for this purpose. Then

> TASK RLDLY’ ; INP  C R

to look at the necessary inputs.

> BCHAN c1 ; ECHAN c2  C R

to select channels free of edge effects.

> DOCAL 1 ; GAINUSE 0  C R

to apply the FRING results and all other current calibrations.

> REFANT nr  C R

to select a reference antenna - only baselines to this antenna are used so select carefully. Alternatively, REFANT 0 will loop over all possible (not necessarily good) reference antennas, averaging the result.

> DOIFS j  C R

to set the adverb to the value of APARM(5) used in FRING (E.2. The IFs are done independently (0), all together (= 1), in halves (= 2), or more generally in N groups (= N).

> TIMERANG db , hb , mb , sb , de, he , me , se  C R

 

to specify the beginning day, hour, minute, and second and ending day, hour, minute, and second (wrt REFDATE) of the data to be included. Use an interval not unlike the one you used in FRING.

> INP  C R

to check the inputs.

> GO  C R

to produce a new SN table with a suitable left polarization delay.

Note that RLDLY now always creates an SN table and may be run with multiple calibrator scans. If there is only one calibrator scan and APARM(2)0, it will also copy the CL table which was applied to the input data through GAINUSE to a new CL table applying the correction to the L polarization delay. For all other cases, you must apply the added L polarization delays with CLCAL.

It is probably better to determine a continuum solution for source polarization and antenna D terms before doing the lower signal-to-noise spectral solutions. To find an average solution for each IF:

> DEFAULT PCAL ; INP

to reset all adverbs and choose the task.

> INDI n; GETN m  C R

to select the data set on disk n and catalog number m.

> DOCAL 1 ; DOBAND 3  C R

to apply the delay, complex gain, and bandpass calibration.

> CALSOUR pol_cal1’, ’pol_cal2  C R

to select the polarization calibrator(s) by whatever form of their names appears in your LISTR output. These sources must have I polarization fluxes in the source table.

> ICHANSEL c11,c12, 1,if1,c21,c22, 1,if2,c31,c32, 1,if3,  C R

 

to select the range(s) of channels which are reliable for averaging in each IF. These probably should be the same values that you used in BPASS.

> DOMODEL -1; SPECTRAL -1  C R

to solve for source polarization in a continuum manner.

> PRTLEV 1  C R

to see the answers and uncertainties on an antenna and IF basis.

> CPARM 0,1  C R

to update the source table with the calibrator source Q and U found.

> INP  C R

to review the inputs.

> GO  C R

to find the antenna leakage terms and the source Q and U values on an IF-dependent basis.

PCAL will write the antenna leakage terms in the antenna file and the source Q and U terms in the source table (if CPARM(2) > 0). DOMODEL may be set to true only if the model has Q = U = 0 since PCAL cannot solve for the right minus left phase difference. If SOLINT=0, PCAL will break up a single scan into multiple intervals, attempting to get a solution even without a wide range of parallactic angles.

Having prepared a continuum solution for Q and U, you must also correct it for the difference in phase between R and L polarizations which normally varies considerably between IFs. The task RLDIF will correct the antenna, source, and calibration tables for this difference using observations of a source with known ratio of Q to U. 3C286 is by far the best calibrator for this purpose.

> DEFAULT RLDIF ; INP

to reset all adverbs and choose the task.

> INDI n; GETN m  C R

to select the data set on disk n and catalog number m.

> DOCAL 1 ; DOBAND 3  C R

to apply the delay, complex gain, and bandpass calibration.

> DOPOL 1  C R

to apply the polarization calibration.

> BCHAN c1; ECHAN c2  C R

to average data from channels c1 through c2 only.

> SOURC pol_cal1’, ’pol_cal2  C R

to select the polarization calibrator(s) by whatever form of their names appears in your LISTR output. These sources must have known polarization angles.

> SPECTRAL 0  C R

to do the correction in continuum mode.

> DOAPPLY 1  C R

to apply the solutions to a CL table (making a new modified one) and to the AN and SU tables, updating them in place.

> DOPRINT 0  C R

to omit all the possible printing.

> INP  C R

to review the inputs.

> GO  C R

to determine and apply the corrections.


PIC

Figure E.1: Example spectrum showing D term solutions for one antenna in right and left polarizations covering about 2 GHz at C band


The EVLA polarizers appear to be very stable in time, but to have significant variation with frequency. See Figure E.1. Serious polarimetry with the EVLA will require solving for the antenna polarization leakage as a function of frequency. To compute a spectral solution, assuming you already did the process in the preceding paragraph:

> TGET PCAL  C R

to retrieve the PCAL adverbs.

> SPECTRAL 1  C R

to do the channel-dependent mode.

> DOMODEL 0  C R

to solve for Q and U as a function of frequency. Because PCAL does not solve for a right-left phase difference and that difference is a function of spectral channel, you must solve for a source polarization.

> INTPARM p1,p2,p3  C R

to smooth the data after all calibration has been done while honoring ICHANSEL.

> SPECPARM 0  C R

to determine the calibration source I, Q, and U spectral indices from fluxes in the source table. If you use PMODEL you must provide spectral indices for the model that apply in the frequency range of the data (curvature cannot be specified).

> INP  C R

to review the inputs, the task will take a while to run.

> GO  C R

to run the task writing a PD table of spectral leakages (“D terms”) and, if DOMODEL 0, a CP table of source Q and U spectra.

If the combination of flagging, ICHANSEL, and INTPARM results in no solutions for some channels, the solutions from nearby channels will be interpolated or extrapolated so that all channels get solutions.

After running PCAL in spectral mode, you may examine the resulting PD (polarization D terms) table with POSSM using APARM(8)=6 and BPLOT using INEXT = ’PD’. If a CP table (calibrator polarization) was written, you may also use POSSM with APARM(8) = 7 or 8 and BPLOT with INEXT = ’CP’ to examine the results.

You are almost, but not quite done. The combination of CALIB and BPASS has produced a good calibration for everything except the phase difference between right and left polarizations. This is now a function of spectral channel and needs to be corrected. The task RLDIF has been modified to determine a continuum or spectral right minus left phase difference and to modify the CL or BP table, respectively, to apply a phase change to the left polarization on an IF or channel, respectively, basis. Thus

> DEFAULT RLDIF ; INP

to reset all adverbs and choose the task.

> INDI n; GETN m  C R

to select the data set on disk n and catalog number m.

> DOCAL 1 ; DOBAND 3  C R

to apply the delay, complex gain, and bandpass calibration.

> DOPOL 1  C R

to apply the polarization calibration, spectral if present.

> INTPARM p1,p2,p3  C R

to smooth the data after all calibration has been done.

> BCHAN c1; ECHAN c2  C R

to use solutions from channels c1 through c2 only, extrapolating solutions to channels outside this range.

> SOURC pol_cal1’, ’pol_cal2  C R

to select the polarization calibrator(s) by whatever form of their names appears in your LISTR output. These sources must have known polarization angles.

> POLANGLE p1,p2  C R

to provide the task with the source polarization angle(s) in degrees in source number order. The phase correction will be twice this value minus the observed RL phase. Do not provide values for 3C286, 3C147, 3C48, and 3C138. These are known to RLDIF including rotation measures and other spectral dependence.

> SPECTRAL 1  C R

to do the correction in spectral mode.

> DOAPPLY 1  C R

to apply the solutions to a BP table (making a new modified one) and to the PD and CP tables.

> DOPRINT -1 ; OUTPRI file_name  C R

to write the phase corrections applied to a text file suitable for plotting by PLOTR.

> INP  C R

to review the inputs.

> GO  C R

to determine and apply the corrections.

Use UVPLT, LISTR or POSSM to check that the expected RL and LR phases now appear with all calibrations turned on. Following these steps, you apply the polarization calibration in any task offering DOPOL. A value of 1 will apply the spectral solution if present or the continuum one is there is no PD table. A value of 6 for DOPOL requests the continuum solution despite the presence of a spectral solution. Use POSSM to plot the calibration sources in RL, LR, Q, and U polarization to make sure that all has functioned correctly (these are newly revised tasks).

If you do not have good observations of a polarization angle calibrator, it may be possible to correct the data in some circumstances. See the help files for QUOUT and QUFIX for possible methods.

If you think that the right minus left phase of your reference antenna was not stable over time, but you have produced good model images of Q and U ignoring that, then it may be possible to improve the polarization calibration with the task RLCAL. It does a self-calibration of Q and U to find a time-variable right minus left phase to apply to the data.