VLA Monitoring Polarization Calibration Guidelines -------------------------------------------------- S.T. Myers (modified from G. Taylor's guide) Introduction ------------ The VLA polarization monitoring program provides essential calibration information for VLBA observers, especially those interested in polarimetry. The VLA program is also useful for other VLA observers, and for some non-NRAO insruments (e.g., Caltech's Cosmic Background Imager). As such it is essential that the output from the program should be of high quality. If an epoch is in doubt due to weather or instrumental problems, it is better to discard that epoch than to release incorrect results. At the same time, sources change quickly at the high frequencies so the sampling interval of two weeks typically provided may not even be sufficient to track variations, making every epoch precious. Initial Edits ------------- Before running the scripts to calibrate the data it is essential to identify any bad data. This can also be useful to give an indication of the health of the VLA. The AIPS task TVFLG is the way to do this. Be sure to set the y-axis interval to the 3.33 seconds used in the POLCAL runs CPARM(6) 3.33 and for parallel hands set STOKES RRLL (RLLR for cross-hands in a separate run, see below). Before editing you need to be sure to use the SET STOKES in TVFLG to FULL (so all hands are edited when you edit). It is also useful to SWITCH IF FLAG from ONE-IF to ALL-IFS so that if you see something bad in both IFs you can edit it once (if its in one IF or one pol of one IF set this so that only the current IF is edited). The first thing you notice is that if you haven't QUACKed the data, the starts of scans are uneven. You can go back outside of TVFLG and run QUACK to nip off the first 10s of every scan, or you can use the FLAG TIMERANGE (with ALL-IF set) to manually do this in TVFLG --- this is what I do as our scans are short (some are only 30s) and I try to minimize the editing to save integrations. But its a matter of taste, and QUACK can be put in a script if desired. Be sure to cycle through both IFs and both pols (RR and LL). Things to look for are antennas with periods of high rms values in amplitude (indicating unstable gains), and/or intermittent flagging by the online system. Any antenna misbehaving by more than 25% of the time should be flagged completely, at least in the offending IF. When flagging, all polarizations (SET STOKES to FULL) should be flagged for any problems found in RR or LL. Other problems to look for are periods of very low amplitude, either for the entire array or for any given antenna. Very large amplitudes on a single baseline may be caused by failures in the correlator and should also be flagged. Useful things to look at: DISPLAY AMPLITUDE the default, useful to see very high points and variable gains on antennas. but hard to see subtles stuff due to variable source brightness DISPLAY AMPL DIFF these are excellent to see discrepant integrations DISPLAY PHASE DIFF and short term variability. The phase one is good for seeing very noisy antennas (as opposed to badly scaled antennas) as the phase noise is just the noise as a fraction of the source flux. DISPLAY RMS some people like this but I prefer AMPL DIFF Inspection of RL and LR correlations is not generally needed, though it should be considered if the leakage calibration (PCAL) results are suspect. However, I actually check this every time, as it is quick and there are occasional bad correlations not reflected in the RR and LL. Note it will quickly show you antennas that have anomalously high leakage (the will be brighter than the others) Preparation ----------- Currently, there are scripts based on configuration and band, e.g. for C-config X-band use CALCX.001 (the .001 suffix allows AIPS to read it as a runfile no matter what user number you are logged in as: > version 'pwd' > run calcx > version '' Note that you must have started AIPS from the directory the scripts are in. Eventually, we will move to a single script for all configs/bands. Running the Script ------------------ 1. (SETIT) Define calibrators. For each run it is necessary to a number of quantities that are run-specific. These are: abscal - the "absolute" calibrator. This is the calibrator to use for absolute flux and EVPA reference. Generally we use 3C286 (1331+305), or 3C48 (0137+331). 3C138 (0521+166), when present, provides a check on the accuracy of the calibration but is more heavily resolved than the others. polcal - the polariation leakage calibrator. This is a calibrator that is bright and has good parallactic angle coverage such that it can be used to solve for the antenna polarization leakage terms (D-terms). numsrc srclist - the list of all other calibrators (except abscal). Be sure to put the correct numsrc. numcal or nummed phacal or medcal - the other calibrators with medium (1-8 Jy) flux densities. Note that in some vintages of scripts its called phacal rather than medcal. numstrong strongcal - bright (>8 Jy) calibrators that can have shorter solution intervals (important at high frequencies in poor weather) numweak weacal - weak (<1 Jy) calibrators, particularly at K and Q band, which can only have scan-average amp and phase solutions. There are parameters in the script that control the solution intervals, these should default ok but might need to be tweaked if all else fails. Also, there are some configuration and band specific quantities that are set (abscell, absuvr) but for the most part leave those defaulted in the scripts. SETIT takes as input the catalog number (on disk 1) for the multi-source uv file for that band, e.g. >ucat AIPS 1: Catalog on disk 1 AIPS 1: Cat Usid Mapname Class Seq Pt Last access Stat AIPS 1: 1 1553 20040412 .Q BAND. 1 UV 11-AUG-2004 16:39:22 AIPS 1: 2 1553 20040412 .K BAND. 1 UV 11-AUG-2004 16:37:14 AIPS 1: 3 1553 20040412 .X BAND. 1 UV 11-AUG-2004 16:37:14 AIPS 1: 4 1553 20040412 .C BAND. 1 UV 11-AUG-2004 17:00:55 >setit(4) to process the C-band data. 3. CALDATA, gain calibration. Determine the gain and phase for each antenna as a function of time. The output of *.CALIB should be examined to check for smooth variations in the phase and gains. The gains should also be roughly consistent from one antenna to another. Any wildly discrepent (factor 2-3) antennas in gain (either high or low compared to the expected level of 3.0) should be flagged. 4. LISTCAL, examine calibrators. Further check of the calibration derived in step (2). The average flux on the calibrators should be similar to what has been found in previous epochs at the frequency being worked on. Phases should be very close to zero for all calibrators. Any discrepant antenna should be noted, and possibly flagged. 5. CALPOL, determine leakage terms and show cross-polarized results on the abscal. PCAL results should indicate 1-5% leakage for most VLA antennas. Large leakage terms (approaching or above 10%) on any given antenna should be understood and examined closely, and probably that antenna should be flagged and the calibration restarted from the beginning. If many antennas appear to have high leakage then TVFLG should be run on the cross correlations (RL and LR) and these inspected. Alternatively it may be worth recalibrating from the beginning using a different choice of reference antenna. Reasonable polarized flux (10% for 3C286, a bit less for 3C48 -- see previous epochs posted on the web page) should be seen in both IFs, and phases should be consistent about a single value for each IF. 6. Examine LISTR output files. As above. 7. OUTPOL. Maps can be checked to see that all sources appear pointlike in both total intensity and polarized flux. The noise should be fairly uniform, without large stripes or large amounts (more than 1%) of polarized flux off the source for both IFs. 8. The final LISTR outputs should be checked to see that the results are comparable with previous results. Sources may vary by 10s of percent, but should not double in total or linearly polarized flux density. The final values from the two VLA IFs should agree to better than a few percent in flux and a few degrees in polarization angle. If they don't then go back and try to figure out why one IF or another are bad. Check the calibrator polarization - look at the *-3C48.POLC file etc. and the *-ALLCAL.POLC files to see if one antenna is way off. Also, could be a bad refant, go back and try changing it. --- end