4.3 Beginning the calibration

Handling EVLA data is quite different from the situation with the historic VLA. The principal difference is the very wide bandwidth of the EVLA which makes every data set have a large number of spectral channels in every spectral window. This requires corrections for delay errors and bandpass shape which were not needed with historic VLA continuum data. It also means that almost all observations are contaminated by radio frequency interference (RFI) at a level requiring that some portion of the data be marked as bad (“flagged”) and omitted from later use. Writing a general list of calibration steps is rather difficult because the steps needed depend on the problems found in the data set. Low frequency data (bands less than 8 GHz) are likely to be heavily contaminated with RFI, while the highest frequency bands have very little. But the high frequencies suffer greatly from rapidly varying “instrumental” (i.e., atmospheric) phase variation. Nonetheless, we will make a list of the general steps that need to be taken following those already described.

1. Correct the data for antenna locations with VLANT (§4.3.1)
2. Correct 3-bit data (only) for post-detector gain with TYAPL (§4.3.2)
3. Look at some of your calibrator data with POSSM using the TV display (§4.3.3)
4. If there is significant narrow-band RFI causing spectral ringing, copy the data applying a Hanning smooth with SPLAT (§4.3.4)
5. If there is significant RFI on your calibrator sources, flag the bad channels and times using UVFLG, FTFLG, RFLAG, or some other flagging task. (§4.3.5)
6. If the phases show significant slope with frequency, run FRING to find delay corrections and apply them with CLCAL (§4.3.6)
7. Run SETJY with OPTYPE ’CALC’ on the primary flux calibrators and set CALCODEs for other calibrators if needed. (§4.3.7)
8. Run BPASS to find corrections for bandpass shape and examine them with BPEDT, perhaps flagging some calibrator data and re-running BPASS. (§4.3.8)
9. Run VLACALIB separately for each calibration source (§4.3.9)
10. Run GETJY to determine the calibration sources’ flux from that of the primary flux calibrator (§4.3.10)
11. Use EDITA to examine the resulting SN table for bad values and to flag calibrator data as needed. (§4.3.11)
12. If significant editing was done, run SETJY with OPTYPE ’REJY’ on the secondary calibrators; then re-run VLACALIB and GETJY (§4.3.9, §4.3.10)
13. Run CLCAL to apply the SN table to the CL table (§4.3.12)
14. Examine your data to see if there remains significant RFI. Use POSSM or SPFLG on a limited set of baselines or even FTFLG although this last will encourage you to flag too much data.
15. If there is significant RFI, use multiple passes of RFLAG to address the problem. (§4.3.13)
16. If you have deleted a bunch of calibrator data, delete the calibration tables except for CL tables 1 and 2. (§4.3.14)
17. For 8-bit data (only), examine the SY table with EDITA or SNPLT. Correct it as needed with TYSMO and then apply it to the data with TYAPL or SYSOL for Solar data. (§4.3.15)
18. Now return to the FRING step and repeat.
19. If you wish to calibrate polarization, run RLDLY to correct right - left delay, run PCAL to find the antenna D terms, and then RLDIF to correct the right - left phase difference. Note that this requires a calibration source of known linear polarization. (§4.3.16)
20. Now examine your target source data with POSSM and edit them with RFLAG
21. Calibrate the weights with REWAY if TYAPL has not already done this. Shift spectral-line data to constant velocity with CVEL. (§4.3.17)
22. Try some quick images with IMAGR to check the calibration (§4.3.18)
23. Back up your calibrated data with FITTP or FITAB (§4.3.19)
24. Apply the calibration and editing to the target sources, writing single-source files for imaging and self-calibration with SPLIT (§4.3.20)

4.3.1 Baseline corrections

Sometimes, e.g., during a VLA array re-configuration, your observations may have been made when one or more of the antennas had their positions poorly determined. The positional error is usually less than a centimeter at the VLA, but even this may affect your data significantly. The most important effect is a slow and erroneous phase wind which is a function of source position and time. Since this error is a function of source position, it cannot be removed exactly using observations of a nearby calibrator, although the error will be small if the target source is close to the calibrator. In many observations, the target sources and calibrators are sufficiently close to allow this phase error to be ignored. Self-calibration will remove this error completely if you have enough signal-to-noise to determine the correction during each integration.

The maximum phase error introduced into the calibrated visibility data by incorrect antenna coordinates Δϕ_B, in radians, by a baseline error of ΔB meters is given by

Note, however, that the error due to the phase-wind is not the only error introduced by incorrect antenna positions. A further, but much smaller effect, will be incorrect gridding of the data due to the erroneous calculation of the baseline spatial frequency components u, v and w. This effect is important only for full primary beam observations in which the antenna position error is of the order of a meter. It is highly unlikely that such a condition will occur. Note too, that this error cannot be corrected by the use of self-calibration. However, after correcting the antenna position with CLCOR, you may run UVFIX to compute corrected values of u,v, and w. The maximum phase error in degrees, Δϕ_G, caused by incorrect gridding of the u,v,w data is

If baseline errors are significant they need to be removed from your data before calibration. It is important to do this to CL table 1, right after running BDF2AIPS. For the EVLA, use the task VLANT. This task determines and applies the antenna position corrections found by the VLA operations staff after your observation was complete. To run VLANT:

For arrays other than the VLA, use CLCOR to enter the antenna position corrections (in meters) in a new CL table and the old AN table. This must be done for each affected antenna in turn. CLCOR puts the corrections into the AN table as well as the CL table, so it is wise to save the AN table before running CLCOR by running TASAV.

The program will need to be run as many times as there are antennas for which positional corrections must be made. Set GAINUSE and GAINVER both to 2 after the first correction. Otherwise, with the above adverbs, CLCOR will make multiple CL table versions each with only one correction in them. Note that subsequent calibration must be applied to CL table 2 to create higher versions of the calibration table. This new CL table (version 2) will replace version 1 in all of the subsequent sections on calibration. Thus, in subsequent executions of CALIB, you must apply these corrections by specifying DOCALIB TRUE ; GAINUSE 0 (for highest version). Note too that CLCOR and VLANT change the antenna file for the changed antenna location(s). Therefore, it is wise to save the AN table before running CLCOR or VLANT by running TASAV.

Note that NRAO’s data analysts use the  task LOCIT to determine the antenna position corrections; see Appendix V.. These are available to the general user, but a data set designed to determine antenna corrections is normally required. Such data sets consist of about 100 observations of a wide range of phase calibrators taken as rapidly as possible.

4.3.2 Correcting EVLA 3-bit data

With the historic VLA, the system temperature was measured in real time and a correction for it applied to the data as they were archived. With the EVLA this is not the case. Instead the SY table records the total power when the switched noise tubes are on and the total power when they are off. These values may be used to correct data taken with 8-bit correlation (§4.3.15) to values which would be fully calibrated in Janskys if the noise tube levels and antenna efficiencies were accurately known. For 3-bit data, frequently used to observe very wide bandwidths at high frequency, the the function to correct the visibilities from these measurements is not well known. As a consequence, for such data, we can only correct for any post-detector gain changes. These are also recorded in the SY table, so use

Reminder — do this only for 3-bit correlator data. Your 8-bit data will be discussed later.

4.3.3 Check your data with POSSM

The next step is to take a look at your data. Do not be obsessive about this. Choose your main calibration source (that you plan to use for FRING, BPASS, and absolute gain calibration), choose a few baselines (say all baselines to one interior antenna), and use the  TV display. At this point you want to look for antennas that appear out of the ordinary (i.e., dead) and for frequencies damaged by RFI. Use

4.3.4 Remove spectral ringing with SPLAT

If there is very narrow-band RFI, the POSSM plots will show ringing - a sin(x)∕x pattern as a function of spectral channel. If this occurs in your data, and it will not occur in some bands, then

4.3.5 General considerations in flagging

At this point, it is essential only to flag the most egregiously bad data samples. These are, with the EVLA, usually either dead antennas or spectral regions affected more or less uniformly by RFI. We will however use this section for a more general discussion of flagging, most of which will apply more directly to later stages in the reduction. Probably FTFLG may be used carefully at this stage to flag only those channels seen to be generally bad in the POSSM plots.

The philosophy of editing and the choice of methods are matters of personal taste and the advice given below should, therefore, be taken with a few grains of salt. When interferometers consisted of only a couple of movable antennas, there was very little data and it was sparsely sampled. At that time, careful editing to delete all suspect samples, but to preserve all samples which can be calibrated, was probably justified. But modern instruments produce a flood of data, with the substantial redundancy that allows for self-calibration on strong sources. Devoting the same care today to editing is therefore very expensive in your time, while the loss of data needlessly flagged is rarely significant. A couple of guidelines you might consider are:

• Don’t flag on the basis of phase. At least with the VLA, most phase fluctuations are due to the atmosphere rather than the instrument. Calibration can deal with these up to a point, and self-calibration (if you have enough signal) can refine the phases to levels that you would never reach by flagging. The exceptions are (1) IF phase jumps which still happen on rare occasions, and (2) RF interference which sometimes is seen as an excursion in phase rather than amplitude.
• Don’t flag on minor amplitude errors, especially if they are not common. Except for very high dynamic range imaging, these will not be a problem, and in those cases, self-calibration always repairs or sufficiently represses the problem.
• Don’t flag if CALIB reports few closure errors and the SN tables viewed with EDITA, SNPLT, and LISTR and the calibrator data viewed with the matrix format of LISTR show only a few problems.

There are three general methods of editing in . The “old-fashioned” route uses LISTR to print listings of the data on the printer or the user’s terminal. The user scans these listings with his eyes and, upon finding a bad point, enters a specific flag command for the data set using UVFLG. While this may sound clumsy, it is in fact quite simple and by far the faster method when there are only a few problems. In a highly corrupted data set, it can use a lot of paper and may force you to run LISTR multiple times to pin down the exact problems. The “hands-off” route uses tasks which attempt to determine which data are bad using only modest guidance from the user. The most general of these are RFLAG and FLAGR mentioned below. The third and “modern” route uses interactive (“TV”-based) tasks to display the data in a variety of ways and to allow you to delete sections of bad data simply by pointing at them with the TV cursor. These tasks are TVFLG (§O.1.6) for all baselines and times (but only shows one IF, one Stokes, and one spectral channel at a time), SPFLG (§10.2.2) for all spectral channels, IFs, and times (but only shows one baseline and one Stokes at a time), FTFLG for all spectral channels, IFs, and times (but shows all baselines combined, one Stokes at a time), EDITA (§4.3.11) for editing based on SY (SysPower), TY (T_ant), SN or CL table values, EDITR (§5.5.2) for all times (but only shows a single antenna (1–11 baselines) and one channel average at a time) and WIPER for all types of data (but with the time, polarization, and sometimes antenna of the points not available while editing). TVFLG is the one used for continuum and channel-0 data from the VLA, while FTFLG is only used to check for channel-dependent interference. SPFLG is very useful for spectral-line editing in smaller arrays, such as the Australia Telescope and the VLBA, but is tedious for arrays like the VLA. Nonetheless, it may be necessary for the modern EVLA. (The redundancy in the spectral domain on calibrator sources helps the eyes to locate bad data.) EDITR is more useful for small arrays such as those common in VLBI experiments. EDITA has been found to be remarkably effective using VLA system temperature tables. All four tasks have the advantage of being very specific in displaying the bad data. Multiple executions should not be required. However, they may require you to look at each IF, Stokes, channel (or baseline) separately (unless you make certain broad assumptions); EDITA and EDITR do allow you to look at all polarizations and/or IFs at once if you want. They all require you to develop special skills since they offer so many options and operations with the TV cursor (mouse these days). A couple of general statements can be made

• For data corrupted by RFI, try FTFLG but be careful. It displays frequency versus time for the selected sources and polarization, but averages all baselines together. A few bad baselines can make it look like all are bad and cause you to flag too much. Flagging channels seen to be bad in all POSSM plots is a good thing and checking channels that appear bad in FTFLG more carefully elsewhere is also good.
• The ultimate tool for EVLA flagging is SPFLG but that shows a frequency-time display for each baseline individually. That me be reasonable for the ATCA or VLBA, but 351 baselines in the VLA is a lot of work. Nonetheless, SPFLG may help you identify channels you may wish to flag more generally.
• For highly corrupted data (say with considerable RF interference, significant cross-talk between antennas, or erratic antennas) TVFLG may be useful. It gives an overall view of the data which is far superior to that given by LISTR. RFI and similar problems are more troublesome at lower frequencies, so TVFLG is probably preferred for L, P, and “4” bands.
• Most VLA data at higher frequencies are of good quality and the flexibility of TVFLG is not needed. In such cases, LISTR with OPCODE = ’MATX’ can find scans with erroneous points efficiently.
• The displays given by TVFLG and, to a lesser extent, LISTR in its MATX mode are less useful when there are only a few baselines. Thus, for arrays smaller than the VLA, users may wish to use SPFLG on spectral-line data sets and EDITR on continuum data sets.
• A reasonable strategy to use is to run LISTR first. If there are only a few questionable points, use LISTR and UVFLG, otherwise switch to an interactive task, such as EDITA followed by TVFLG.
• Task FLAGR is a somewhat experimental task to measure the rms in the data on either a baseline or an antenna basis and then delete seriously discrepant points and times when many antennas/correlators are questionable. It also clips amplitudes and weights which are outside specified normal ranges. Task FINDR reports the rmses and excessive values to assist in running FLAGR.
• Task RFLAG flags data on the belief that RFI is highly variable in time and/or frequency. It can do plots showing the time and/or frequency statistics as a function of spectral channel and recommends flagging levels. These, or user-chosen values, may be applied to the data to produce large flag tables; see §4.3.13. RFLAG is best used after data have some initial calibration to remove delay and bandpass issues and equalize the flux scale.
• Task CLIP makes entries in a flag table, applying calibration and then testing amplitudes for reasonableness on a source-by-source basis. In 31DEC23, it even has an option to flag on the difference of parallel-hand correlations. Use that option only after you have good calibration. CLIP can be very useful for large data sets, but does not show you the bad data to evaluate yourself. ACLIP does the same operation on auto-correlation data.
• Task FGSPW performs a matrix scalar average over all spectral channels on a per-IF (aka “spectral window”) per polarization per baseline per SOLINT or scan basis. It then flags those IFs with excessive amplitude. This operation is designed to catch those windows where amplitudes have overflowed the hardware due to RFI within the spectral window and may catch other problems as well.
• Task WIPER makes entries in a flag table for all data samples wiped from a UVPLT-like display of any uv data set parameter versus any other parameter. The source, Stokes, IF, time, etc. of the points are not known during the interactive editing phase, but some baseline information is displayed. It can plot and edit any choice of STOKES in one execution and can flag/unflag by baseline.
• Task DEFLG makes entries in a flag table whenever the phases are too variable as measured by too low a ratio of vector-averaged to scalar-averaged amplitudes. This may be useful when applied to the calibrator source in phase-referencing observations and for other data at the highest and lowest frequencies which are affected by atmospheric and ionospheric phase variability.
• Task SNFLG makes entries in a flag table whenever the phase solutions in an SN or CL table change excessively between samples on a baseline basis. It can also flag data if the amplitude solutions differ from their mean excessively. It can instead flag data when the amplitude and/or phase solutions or solution weights are outside user-specified ranges.
• Task REWAY examines spectral rms to try to determine better weights for the data. It is better to use it after delay, bandpass, and complex gain issues have been addressed at some level. Run it with no flagging of the output for bad values of the spectral rms. Then plot the weights with VPLOT or ANBPL to look for weights that are seriously abnormal (high or low). Those data may need to be flagged. High weights mean that the data are of abnormally low amplitude, whilst low weights mean that the data are very noisy. REWAY uses robust methods to find the rms and so a few channels of RFI may not cause very low weights, but lots of RFI or receiver failures will make the weights abnormally low. REWAY now displays statistical information to help you assess what weights are “high” and “low”.
• Task WETHR makes entries in a flag table whenever various weather parameters exceed specified limits. WETHR also plots the weather (WX) table contents.
• Task VPFLG flags all correlators in a sample whenever one is flagged. Observations of sources with circular polarization (Stokes V) require this operation to correct the flagging done on-line (which flags only known bad correlators).
• Task FGPLT plots the times of selected flag-table entries to provide you information on what these powerful tasks have done.
• Task FGPRT prints a matrix of the number of flags applying to each baseline in a user specified range of time, spectral channel, IF, source, etc.

Many of the above tasks produce large flag tables. The task REFLG attempts to compress such tables, sometimes quite remarkably and sometimes incorrectly. Task FGDIF may be used to confirm that the flag tables before and after flag the same things. FGCNT may also help confirm this.

4.3.6 Correcting delay errors with FRING

We have had difficulty setting all of the delays in the EVLA to values which are sufficiently accurate. If the delay is not set correctly, the interferometer phase will vary linearly with frequency, potentially wrapping through several turns of phase within a single spectral window (“IF band”). Your POSSM displays probably show such an effect. Delay errors are a problem familiar to VLBI users and  has a well-tested method to correct the problem. Using your LISTR output, select a time range of about one minute toward the end of a scan on a strong point-source calibrator, usually your bandpass calibrator. Then

The different IFs in current EVLA data sets may come from different basebands and therefore have different residual delays. The option APARM(5) = 3 to force the first N_if∕2 IFs to have one delay solution while the second half of the IFs has another is strongly recommended, but only when the first half all come from one of the “AC” or “BD” basebands (hardware IFs) and the second half come from the other. The 3-bit data path of the EVLA actually has four hardware IFs, so APARM(5) = 4 produces four delay solutions, dividing the IFs in quarters. The Widar correlator may now separate the hardware basebands into unequal numbers of spectral windows. APARM(5) = -1 will cause FRING to use BPARM to specify the IFs used in each delay solution. Note that, at low frequencies, the phases may also be affected by dispersion (phase differences proportional to wavelength). FRING now offers APARM(10) to enable solving for a single delay plus dispersion from the fitted single-IF delays. If you have a large number of narrow spectral channels, CHINC may be used to reduce the memory and time requirements of FRING. This SN table will need to be applied to the main CL table created by INDXR or OBIT.

Be sure to apply this (or higher) CL table with DOCALIB 1 in all later steps.

4.3.7 Primary flux density calibrators

Careful measurements made with the D array of the VLA have shown that the Baars et al.¹ formulæ for “standard” calibration sources are in error slightly, based on the assumption that the Baars’ expression for 3C295 is correct. Revised values of the coefficients have been derived by Rick Perley and Brian Butler. Task SETJY has these formulae built into it, giving you the option (OPTYPE ’CALC’) of letting it calculate the fluxes for primary calibrator sources 3C48, 3C123, 3C138, 3C147, 3C196, 3C286, 3C295, and 1934-638. The default setting of APARM(2) = 0) will calculate the flux densities based on Perley-Butler 2017 values which cover the range 50 MHz to 50 GHz for the primary calibrators:3C48, 3C138, 3C147, 3C286, and 3C295 plus 3C123 and 3C196, all with a number of synonyms. Other sources which will be computed, but which may not be good calibrator sources, are J0444-2809, PictorA, 3C144 (Taurus A, Crab Nebula), 3C218 (Hydra A), 3C274 (Virgo A), 3C348 (Hercules A), 3C353, 3C380, 3C405 (Cygnus A), 3C444, and 3C461 (Cassiopeia A). multiple names are also allowed for these sources; see EXPLAIN SETJY. Most of the sources in the extra list have limits on the frequency range over which the function is valid; SETJY will tell you if the frequencies are out of range. Higher values of APARM(2) select older systems of coefficients if you need them to match previous data reductions. SETJY will recognize both the 3C and IAU designations (B1950 and J2000) for the standard sources. See EXPLAIN SETJY for details of all models. You may also insert your own favorite values for these sources instead (OPTYPE = ’ ’) and you will have to insert values for any other gain calibrators you intend to use (or use GETJY, see §4.3.10). Adverbs SPECINDX and SPECURVE allow you to enter spectral index information to help set the calibrator fluxes.

Unfortunately, since all the primary flux calibrators are resolved by the VLA in most configurations and at most frequencies, they cannot be used directly to determine the amplitude calibration of the antennas without a detailed model of the source structure, see Figure 4.1 as an example. Beginning in April 2004, model images for the calibrators at some frequencies are included with . Models of 3C286, 3C48, 3C138, and 3C147 are available for all 6 traditional bands of the VLA except 90 cm and even for S band of the EVLA. Type CALDIR  C R to see a list of the currently available calibrator models. Sources which are small enough to be substantially unresolved by the VLA have variable flux densities which must be determined in each observing session. A common method used to determine the flux densities of the secondary calibrators from the primary calibrator(s) is to compare the amplitudes of the gain solutions from the procedure described below.

Use SETJY to enter/calculate the flux density of each primary flux density calibrator. The ultimate reference for the VLA is 3C295, but 3C286 (1328+307), which is slightly resolved in most configurations at most frequencies, is the most useful primary calibrator. CALIB has an option that will allow you to make use of Clean component models for calibrator sources. You are strongly encouraged to use the existing models. If you follow past practice at the VLA, you may have to restrict the uv range over which you compute antenna gain solutions for 3C286, and may therefore insert a “phony” flux density appropriate only for that uv range at this point. In both cases, the following step should be done. CALIB will scale the total flux of the model to match the total flux of the source recorded by SETJY in the source table. This corrects for the model being taken at a somewhat different frequency than your observations and for the model containing most, but not all, of the total flux. An example of the inputs for SETJY, where you let it calculate the flux, would be:

Or you can set the flux manually as shown below:

Note that, although SOURCES can accept a source list, ZEROSP has room for only one set of I, Q, U, V flux densities. To set the flux densities for several different sources or IFs, you must therefore rerun SETJY for each source and each IF, changing the SOURCES, BIF, EIF, and ZEROSP inputs each time. Alternatively, set ZEROSP to the flux at 1 GHz and enter SPECINDX and SPECURVE adverbs to describe the dependence with frequency.

CALIB will use the V polarization flux in the source table if one has been entered. The RR polarization will be calibrated to I+V and the LL to I-V. While this has little practical use with circular polarizations because V is almost always negligible, it can be used for linearly polarized data from the WSRT. That telescope has equatorially mounted dishes, so the VV polarization is I-Q and the HH is I+Q independent of parallactic angle. For WSRT data, you should relabel the polarizations to RR/LL and enter I, 0, 0, -Q for ZEROSP, since Q is not negligible in standard calibrators.

In 31DEC22, more extensive calibration of linear polarization became available. CALIB will use the source table Q and U fluxes and the parllactic angles to compute proper gains for the V and H polarizations of point sources. A new procedure VHCALIB and task VHCAL have been written to apply full Clean-component polarization models to determine the parallel-hand gains of linear systems such a Meerkat.

If the LISTR scan listing shows calibration sources with a blank CALCODE, you should also use SETJY to correct this. Set OPTYPE = ’ ’, set the desired SOURCE name, and set CODETYPE to the desired CALCODE. Expert users sometimes prefer TABPUT for this function; run PRTAB to see what values must be set in PIXXY.

4.3.8 Calibrating the bandpass shape with BPASS

Bandpass calibration is for the modern VLA required rather than merely recommended. Having chosen those channels which may be reliably used to normalize the bandpass functions,

Do not use spectral smoothing at this point unless you want to use the same smoothing forever after. Apply the flag table. A model for the calibrator may be used; see §4.3.9.1.

Note: if your have a strong bandpass calibrator, observed frequently, you may use BPASS instead of CALIB to calibrate the gains. You must, if course, turn off the bandpass normaization entirely to use this option.

BPASS now contains the adverbs SPECINDX and SPECURVE through which the spectral index and its curvature (to higher order than is known for any source) may be entered. For the standard amplitude calibrators 3C286, 3C48, 3C147, and 3C138, these parameters are known and will be provided for you by BPASS. For other sources, you may provide these parameters, but BPASS will fit the fluxes in the SU table for a spectral index (including curvature optionally) if you do not. Note that, if no spectral index correction is applied, the spectral index of the calibration source will be frozen into the target source. Bandwidths on the EVLA are wide enough that this is a serious problem. If you do not know the spectral index of your calibration source, BPASS itself or the new task SOUSP may be used to determine the spectral indices from the SU table. Of course, that means that GETJY must already have been run. Since BPASS must usually be run before CALIB and hence GETJY, this suggests that one may have to iterate this whole process at least once. SOUSP now offers the option of correcting one or more SN tables after it adjusts the source fluxes for the spectral index it determined. This may reduce the need for further iterations.

Note that the bandpass parameters shown above assume that the phases are essentially constant through each scan of the bandpass calibrator. This may not be true, particularly at higher frequencies. In this case, you have two choices. One is to set BPASSPRM(5) to 0 which will determine the vector average of the channels selected by ICHANSEL at every integration and divide that into the data of that integration. This will remove all continuum phase fluctuations, but runs a risk of introducing a bias in the amplitudes since they do not have Gaussian statistics. BPASSPRM(5) = -1 now applies a phase-only correction on a record-by-record basis. A better procedure, which is rather more complicated, is as follows. Use SPLIT to separate the bandpass calibrator scans into a separate single-source file applying any flags and delay calibration and the like. Then run CALIB on this data set with a short SOLINT to determine and apply a phase-only self-calibration. On the uv data set written out by CALIB, run BPASS using the parameters described in the previous paragraph. Finally, use TACOP to copy the BP table back to the initial data set.

The spectral quality of the final images has been found to be determined in part by the quality of the bandpass solutions. In particular, for reasons which are not yet known, the bandpasses are not exactly antenna dependent especially in the edge channels. This “closure error” may be measured in individual and statistical ways by BPASS and reported to you. To check on this problem for your data set, set

It is probably a good idea to set MINAMPER and MINPHSER fairly high (i.e., 20 and 12) to make a big deal only about major excursions, but to set BPASSPRM(6) and (7) fairly low (i.e., 0.5 and 0.5) to view the spectrum of closure errors (which will look a lot like the spectrum of noise on your final Clean images). There is even a task called BPERR which will summarize and plot the error reports generated by BPASS and written to text files by PRTMSG.

The bandpass solutions are calculated at each bandpass calibrator scan. As a consequence, they are likely to be unevenly spaced in time and may even have times (due to on-line or later editing) at which there are solutions for some IFs and polarizations but not all. When the latter happens, program source data will be lost unless the missing solutions are filled in. The task BPSMO may be used for this purpose or to create a new BP table at regular time intervals using one of a number of time-smoothing functions. Set APARM(4) = -1 for the “repair” mode or set APARM(4) to the desired BP interval.

After the bandpasses have been generated, you can examine them using tasks BPLOT and POSSM. You can obtain an average from all antennas with

To view each antenna individually, using the TV to save paper

POSSM shows each screen for 30 seconds before going ahead. You can cause it wait indefinitely by hitting button A, speed it up by hitting TV buttons B or C, or tell it to quit by hitting button D. If DOTV = -1, then POSSM makes multiple plot extension files, which can be sent to the printer (individually or collectively) by LWPLA. You might want to use a larger value of NPLOTS to reduce the number of pieces of paper.

BPLOT is used to create one or more plots (on the TV or in plot files) of the selected bandpass table. The plots will be a set of profiles separated on the vertical axis by an increment in time or antenna number (depending on the sort selected). More than one plot for more than one antenna or more than one time may be generated. Multiple IFs and polarizations will be plotted along the horizontal axis if they are present in the BP table and selected by the adverbs. Thus, BPLOT is useful for plotting the change in bandpass shape as a function either of time or of antenna.

Task BPEDT offers a TV-interactive graphical way to check your BP table. It displays the amplitude and phase of one antenna to be edited and up to 10 more antennas for comparison. The task is like EDITA except that the horizontal axis is spectral channel rather than time. You may walk through the entire bandpass table selecting the editable antenna and time in sequence. If some of the solutions show residual effects from RFI, then you may generate flags in a flag table to delete certain spectral channels for certain antennas and times. If you do this, of course, you should re-run BPASS applying the new flags. In 31DEC21, BPEDT offers the option of flagging the BP values instead and even interpolating over the flagged values. Task BPFLG may be used with large BP tables after you have determined appropriate clip levels. Task BPEPL may be used to make plots similar to those of BPEDT without the interactivity. It will be faster if you have a lot of times in the BP table. If you do have lots of times in your BP table, BPPLT is another way to view a portion of the table as a function of time.

The BP tables are applied to the data by setting the adverb DOBAND > 0 and selecting the relevant BP table with the adverb BPVER. There are three modes of bandpass application. The first (DOBAND 1) will average all bandpasses for each antenna within the time range requested, generating a global solution for each antenna. The second mode (DOBAND 2) will use the antenna bandpasses nearest in time to the data point being calibrated. The third mode (DOBAND 3) interpolates in time between the antenna bandpasses and generates the correction from the interpolated data. This mode has been found to be required for VLA data. If BPSMO was used to make a fairly finely sampled BP table, then DOBAND 2 may be used. Modes DOBAND 4 and DOBAND 5 are the same as modes 2 and 3, respectively, except that data weights are ignored.

It is often not possible to observe a strong bandpass calibrator many times during a run. In this case, one can run BPASS on the single scan on the strong calibrator and then remove the main bandpass shape with DOBAND 1 in task SPLAT. Corrections to this basic bandpass shape as a function of time may then be determined with adequate signal-to-noise using task CPASS. This task can be used to fit the residual bandpass with a small number of parameters (<< the number of spectral channels) at each calibrator scan. The results may then be applied with DOBAND 2. Check the output of CPASS carefully — it is capable of making bandpass shapes with large ripples that are not present in the data. CPASS also knows about the flux coefficients of all standard sources and can fit the spectral index of other sources from the flux values in the SU table.

4.3.9 Calibrating the complex antenna gains with VLACALIB

4.3.9.1 Using calibrator models

It is now considered standard practice to use flux calibrator models and you are strongly encouraged to do so. As mentioned above, all the primary flux calibrators are resolved at most frequencies and configurations. Figure 4.1 shows the visibilities and image of the commonly used calibrator 3C48 at X-band, it is obvious this source is far from being point like. Since April 2004, source models have been shipped with  as FITS files. Currently, models for 3C48, 3C286 and 3C138 are available for all bands except 90 cm and 3C147 at all bands except, X, C and P. Rick Perley has many models on two different web sites, one for 2014 and one for 2019 observations. These may be reached in 31DEC20 by setting DATAIN appropriately. Additional models are in the works, so you should always check to see what is available:

You can load a model with the procedure VLACALIB (see below) or you can load the model in manually using CALRD:

Set DATAIN = ’2019’ to select models from Perley’s 2019 website. Soon the task MOD3D will enable muti-facet models at P and 4 band to be distibuted with a “3D” model attached to a single facet.

4.3.9.2 Calibrating complex gains

The primary calibration task in  is CALIB. Most of the complexity of CALIB can be hidden using the procedure VLACALIB. Before attempting to use this procedure, you must first load it by typing:

Type HELP VLAPROCS for a full list of the procedures available in VLAPROCS.

The procedure VLACALIB automatically downloads and uses calibrator models, if one is available and the inputs are set correctly. After you have loaded VLACALIB you may invoke the calibrator model usage by setting:

This procedure will load the calibrator model (using CALRD) and use it when it runs CALIB, then print any messages from CALIB about closure errors on the line printer, and finally run LISTR to print the amplitudes and phases of the derived solutions. Plots of these values may be obtained using task SNPLT.

Then you may select the model image with GET2N for use directly in CALIB. Example inputs for CALIB are:

CALIB will use the clean components table attached to the model to find antenna gain solutions. It will sum the clean components within a certain radius of the center of the map (so that confusing sources that are part of the model do not influence the gain) and scale them to the flux in the SU table. Therefore, you must still run SETJY before running CALIB.

After running CALIB check the solutions for all antennas with SNPLT or LISTR (OPTYPE=’GAIN’). If you have multiple primary or secondary calibrators you will have to run CALIB separately for each, using models where they are available and restricting the UVRANGE and ANTENNAS where they are not. You can either write into the same SN table by setting SNVER to a table number or to different SN tables by setting SNVER = 0. Then you you can proceed as normal flagging and editing your data and proceed to final calibration as described in §4.3.10.

4.3.9.3 Flux calibration without calibrator models

Normally, the primary flux calibration sources are located too far from the target sources to be used to track time variations in phase and amplitude. Therefore, secondary calibration sources whose fluxes and spectral indices are not known in advance.

Once you have read in procedure VLACALIB (see §4.3.9.1), you may use it to invoke CALIB. You will have to do this once for each calibrator, unless you can use the same UVRANGE for more than one of them. Thus,

This procedure will first run CALIB, then print any messages from CALIB about closure errors on the line printer, and finally run LISTR to print the amplitudes and phases of the derived solutions. Plots of these values may be obtained using task SNPLT.

If the secondary calibrators require different values of UVRANGE, then CALIB must be run until it has run for all calibration sources. Attached to your input data set is a solution SN table. Each run of CALIB writes in this table (if SNVER = 1), for the times of the included calibration scans, the solutions for all IFs using the flux densities you set for your calibrators with SETJY or GETJY. (CALIB assumes a flux density of 1 Jy if no flux density is given in the SU table.) If a solution fails, however, the whole SN table can be compromised, forcing you to start over. It is possible to write multiple SN tables with SNVER = 0. Later programs such as GETJY and CLCAL will merge all SN tables which they find (if told to do so). Tables with failed solutions must be deleted.

4.3.10 Bootstrapping secondary flux-density calibrators

Task GETJY can be used to determine the flux density of the secondary flux calibrators from the primary flux calibrator based on the flux densities set in the SU table and the antenna gain solutions in the SN tables. The SU and SN tables will be updated by GETJY to reflect the calculated values of the secondary calibrators’ flux densities. This procedure should also work if (incorrect) values of the secondary calibrators’ flux densities were present in the SU table when CALIB was run. Bad or redundant SN tables should be deleted using EXTDEST before running GETJY, or avoided by selecting tables one at a time with adverb SNVER.

To use GETJY:

GETJY will give a list of the derived flux densities and estimates of their uncertainties. These are now found by “robust” methods and additional information about numbers of aberrant solutions are given. If any of the uncertainties are large, then reexamine the SN tables as described above and re-run CALIB and/or GETJY as necessary. Multiple executions of GETJY will not cause problems as previous solutions for the unknown flux densities are simply overwritten. You may wish to run the task SOUSP to determine the spectral indices of your calibrators from their fluxes in the SU table. You can even replace the values in the SU table with the curve fit by SOUSP and correct the gains in one or more SN tables with the newly determined fluxes. These spectral index parameters may be useful in running BPASS and PCAL. However, BPASS knows the flux coefficients for all standard calibration sources and can fit the spectral index of other calibration sources from the SU table.

4.3.11 Editing visibility data with EDITA

The task EDITA uses the graphics planes on the  TV display to plot data from tables and to offer options for editing (deleting, flagging) the associated uv data. Only the TY (VLB and historic VLA system temperature), SY (modern VLA SysPower table), SN (solution), and CL (calibration) tables may be used. We recommend using EDITA with the SN table updated by GETJY as a quick and easy way to examine the quality of the solutions. If a small amount is disturbed, use EDITA to flag the relevant data. If the solutions for most antennas are disturbed, make a note of the time and source and use other flagging tools to find and flag the cause of the disturbance. If you flag more than a little, you should delete the SN table(s) and re-run VLACALIB and GETJY. Try:

If you make multiple runs of EDITA, it is important to make sure that the flagging table entries are all in one version of the FG table. The easiest way to ensure this is to should set FLAGVER and OUTFGVER to 0 and keep it that way for all runs of EDITA. This may create an excessive number of flag tables, but unwanted ones may be deleted with EXTDEST. If you make a mistake two flag tables may be merged with the task TAPPE. A sample display from EDITA is shown on the next page.

The following discussion assumes that you have read §2.3.2 and are familiar with using the  TV display. An item in a menu such as that shown in the figure is selected by moving the TV cursor to the item (holding down or pressing the left mouse button). At this point, the menu item will change color. To obtain information about the item, press  TV “button D” (usually the D key and also the F6 key on your keyboard). To tell the program to execute the menu item, press any of  TV buttons A, B, or C. Status lines around the display indicate what is plotted and which data will be flagged by the next flagging command. In the figure below, only the displayed antenna (2), and time range will be flagged. You must display at least a few lines of the message window and your main AIPS window since the former will be used for instructions and reports and the latter will be needed for data entry (e.g., antenna selection).

The first thing to do with EDITA is to look at all of the polarizations, IFs, and antennæ, in order to make notes about disturbances in the solutions. Use SWITCH POLARIZATION to switch between polarizations and ENTER IF to select the IF to edit. Alternatively, NEXT POL/IF will cycle through all polarizations and IFs. If CROWDED was set to 1 or 2, SWITCH POLARIZATION will cycle through displaying both polarizations as well as each separately, and ENTER IF will accept 0 as indicating all. These options appear only if there is more than one polarization and/or more than one IF in the loaded data. Use ENTER ANTENNA to select the antenna to be flagged and ENTER OTHER ANT to select secondary antennæ to be displayed around the editing area. If the secondary antennæ have no obvious problems, then they do not have to be selected for editing. EDITA will plot all of the times in the available area, potentially making a very crowded display. You may select interactively a smaller time range or “frame” in order to see the samples more clearly. It is necessary to select each frame in order to edit the data in that frame so it helps to make the TV screen as big as possible with the F2 button or your window manager. Note that the vertical scales used by EDITA are linear, but that the horizontal scale is irregular and potentially discontinuous. Integer hours are indicated by tick marks and the time range of the frame is indicated. Use FLAG TIME or FLAG TIME RANGE to delete data following instructions which will appear on the message window. While you are editing, the source name, sample time and sample value currently selected will be displayed in the upper left corner of the TV screen.

Having noted all obviously bad points, you may now selectively flag visibility data. Select SWITCH ALL IF, SWITCH ALL TIME, SWITCH ALL ANT, SWITCH ALL POL, and SWITCH ALL SOURC so that the next flag command(s) apply to the portion of the data that should be flagged. The current setting of these toggles is displayed at the lower left corner. The various FLAG options allow you to flag single times, time ranges, all samples above or below a selected level, and single points either carefully or quickly. Finally, apply your flagging to your uv data set by selecting EXIT or throw it all away by selecting ABORT..

Note that it may also be profitable to run EDITA using the SY table. Unlike complex gains which reflect the values of all antennas, the SY values are strictly dependent only on the specific antenna. They are however affected by RFI in ways that may not affect the visibilities, so flagging uv data based on the SY values should be done with caution. EDITA, when run on SY or TY tables, has the option to display the difference of the current parameter with a running mean of that parameter. Such displays are a powerful editing tool.

4.3.12 Applying solutions to the calibration table with CLCAL

At this point you should have gain and phase solutions for the times of all calibration scans, including the correct flux densities for the secondary calibrators. The next step is to interpolate the solutions derived from the calibrators into the CL table for all the sources. CLCAL may be run multiple times if subsets of the sources are to be calibrated by corresponding subsets of the calibrators, unless you limit it to one or more tables with SNVER and INVERS, CLCAL assumes that all SN tables contain only valid solutions and concatenates all of the SN tables with the highest numbered one. Therefore, any bad SN tables should be removed before using CLCAL. For polarization calibration, it is essential that you calibrate the primary flux calibrator (3C48 or 3C286) also so that you can solve for the left minus right phase offsets and apply PCAL.

CLCAL has caused considerable confusion and user error because it implements two somewhat contrary views of its process. The older view, represented by previous versions of this ookook, had the user gradually building a final CL table from multiple runs of CLCAL, each with a selected set of calibration sources, target sources, antennas, time ranges, and so forth. In this scheme, the user had to take great care that the final CL table actually contained information for all antennas, sources, and times for which it would be needed. It was easy to get this wrong! The second and now prevailing view is that every execution of CLCAL should write a new CL table containing all sources, antennas, and times, but with a selected subset modified by the current execution. This leads to there being a potentially large number of CL tables, but no data will be flagged due to the absence of data in the CL table. The user will still have to be careful to insure that all CL records have received the needed calibration information.

To use CLCAL:

Calibrator sources may also be selected with the QUAL and CALCODE adverbs; QUAL also applies to the sources to be calibrated. Note that REFANT appears in the inputs because  references all phases to those of the reference antenna. If none is given, it defaults to the one used in the most solutions.

The smoothing and interpolation functions in CLCAL have been separated into two adverbs and the smoothing parameters are now conveyed with BPARM and ICUT. In smoothing, the DOBLANK adverb is particularly important; it controls whether good solutions are replaced with smoothed ones and whether previously failed solutions are replaced with smoothed ones. One can select either or both.

Note that CLCAL uses both the GAINUSE and GAINVER adverbs. This is to specify the input and output CL table versions, which should be different. If you are building a single CL table piece by piece, then these must be set carefully and normally held fixed. In the more modern view, they are set to zero and the task takes the latest CL table as input and makes a new one. CL table version 1 is intended to be a “virgin” table, free of all injury from any calibration you do using the  package. It may not always be devoid of information, as “on-line” corrections may be made and recorded here by some telescope systems, e.g., the VLBA. The VLA, through tasks FILLM, BDF2AIPS, or INDXR, can now put opacity and antenna gain information in this file. CLCAL and most other  tasks are forbidden to over-write version 1 of the CL table. This protects it from modification, and keeps it around so that you may reset your calibration to the raw state by using EXTDEST to destroy all CL table extensions with versions higher than 1. Be careful doing this, since you rarely want to delete CL version 1. Should you destroy CL table version 1 accidentally, you may generate a new CL table version 1 with the task INDXR. This new CL table may contain the calibration generated from the weather and antenna gain files.

If you have any reason to suspect that the calibration has gone wrong — or if you are calibrating data for the first time — you should examine the contents of the output CL table. The task EVASN may help you determine the degree of phase and amplitude coherence in your calibration table. A lack of coherence suggests that the calibration is rather uncertain. Task SNPLT will provide you with a graphical display:

Note that SNPLT may be used to examine SN, SY, and TY tables as well. OPCODE controls whether 1 IF or all IFs and 1 polarization or all polarizations are displayed. Task SNBPL will plot the calibration values on a baseline basis which can help to isolate the source of incoherence. SNFLG can be used to flag incoherent baselines rather than deleting whole antennas.

To test that the calibrators are actually properly calibrated by the new CL table, use UVPLT Try

You may need to limit which IFs are displayed or color the IFs with DO3COL 1. The phases should be near zero and the amplitudes should be nearly constant with baseline length. A decrease in the amplitude at the longest spacings indicates that the calibrator is not strictly a point source.

ANBPL converts baseline-based data before or after calibration into antenna-based quantities. In particular, the calibrated weights are very sensitive to problems with amplitude calibration.

If the previous steps indicate serious problems and/or you are seriously confused about what you have done and you want to start the calibration again, you can use the procedure VLARESET from the RUN file VLAPROCS to reset the SN and CL tables.

4.3.13 Flagging RFI with RFLAG

A very promising relatively new tool flags RFI on the assumption that it is either quite variable in time or in frequency. This task, called RFLAG, computes the rms over short time intervals in each spectral channel and IF individually and flags the interval whenever the rms exceeds a user-controlled threshold. Optionally, it will also use a sliding median window of user-specified width over the spectral channels to the real and imaginary parts of the visibility separately. Any channel deviating from the median in either part by more than a user-specified amount will also be flagged. If DOPLOT> 0, RFLAG will make plots of normal and cumulative histograms and of the mean and rms of the time and spectral computations as a function of channel. It will also make a flag table only if requested (DOFLAG> 0). These plots will suggest threshold parameters and allow you to choose values to use. A flag table is made for any value of DOFLAG if no plots are requested (DOPLOT≤ 0).

In detail, RFLAG is run using

This will produce plots and set cutoff levels in adverbs NOISE and SCUTOFF. An example of the spectral plot is shown in Figure 4.3. Another run, with DOPLOT = 0 will apply these cutoffs and create a new flag table. Note that the flux cutoff levels may depend on the source flux, calling for different levels for strong calibrators, weak calibrators, and very weak target sources. Different cutoff levels for STOKES=’RRLL’ and STOKES = ’RLLR’ may also be needed. A strong, resolved target source may require different levels for different UVRANGEs. UVRANGE is now applied on a per spectral window basis so you no longer have to break up wide bandwidth data into multiple files when UVRANGE is required. RFLAG is a new task, so experiment a bit. Note that, if you set DOFLAG=1, the creation of a new flag table will happen after the plots in the same execution of RFLAG. If a channel is found bad at a time in any one polarization, all polarizations are flagged. If you have a significant spectral line signal in your data, use DCHANSEL to have the affected channels ignored throughout RFLAG or set FPARM(4) = 0 so that the spectral function is not performed.

There are a lot of adverbs to RFLAG. FPARM(5) allows you to speed up the spectral part of the flagging by testing more than just the central channel in the sliding median filter. FPARM(6) allows you to expand all flags to adjacent channels. FPARM(7), (8), (11), and (12) control the extending of flags to additional channels, baselines, or antennas if too large a fraction of channels, baselines, or baselines to an antenna are flagged in the basic time and spectral operations. Similar adverbs also occur in the new task REFLG whose job it is to compress the enormous flag tables generated generated by RFLAG. REFLG does not handle flags generated by CLIP, TVFLG, and SPFLG since they vary with polarization. REFLG can extend a flag to all times if too large a fraction of time is flagged for a given channel, baseline, etc. REFLG may not correctly compress all flags, but its other options are useful. The application of 10 million flag entries to a data set repetitively is rather expensive. Copying the data, applying the flags once and for all, is the best solution. RFLAG may now do this; set DOOUTPUT to true in 31DEC22 and later, along with output name adverbs OUTNAME, et al. UVCOP has been the traditional method to do this. However, TYAPL which needs to be run next and must make a new copy of the data has been given the option of applying a large flag table to avoid having to copy the data set twice. Task FGCNT lets you see how much of your data is flagged by any particular flag table. Task FGPRT will plot a matrix by baseline of the number of flags applying in a user specified range of time, spectral channel and IF, source, etc.

4.3.14 Restart the calibration

Having done a more careful job with your editing, it is now time to discard with EXTDEST the bandpass (BP) tables and all CL tables after the one written by VLANT. Discard all SN tables, but keep the highest numbered flag (FG) table.

Be very careful with the flag tables to save the last one - use IMHEADER to show you how many there are.

4.3.15 Calibration with the SysPower table

Because of its wide dynamic range, the EVLA does not normalize its output visibilities. To calibrate gains it records the total power when the switched noise tube is on and when it is off. These data, taken in synchronism with the visibilities, are recorded in the SysPower table of the ASDM. The OBIT program BDFIn, available to  users in the verb BDF2AIPS, reads this table and creates an  SY table. The columns of this table contain POWER DIF (Gain × (P_on - P_off), POWER SUM (Gain × (P_on + P_off)), and POST GAIN (Gain) columns for right and left polarizations with values for each IF.

This table is accessible to  users with a number of tasks. Begin with task PRTSY to view the table statistically over time on a per IF, per antenna basis or to view scan or source median averages of one or more of the SY table parameters. A more detailed look is available from SYPRT. Then to examine its contents graphically in various ways, use SNPLT with OPTYPEs ’PDIF’, ’PSUM’, ’PGN’, ’PON’, ’POFF’, ’PSYS’, ’PDGN’, or ’PSGN’. You could use OPTYPE = ’MULT’ to examine more than one of these at one time, comparing any oddities in e.g., Psum and Pdif. Note that PSYS is especially interesting since T_cal * P_sum∕(2 * P_dif) = T_sys, the system temperature. It should reflect changes in elevation and strength of the observed source, but should be immune to adjustments to the gain of the telescope. It determines data weights in TYAPL while divides into the visibilities. You may use EDITA (§4.3.11) to edit your uv data on the basis of the contents of the SY table. Editing may be based on Psum, Pdif, Pgain, Tsys, and on the differences between these parameters and a running median of these parameters. One may also edit the SY table itself with SNEDT; the same parameters are available. LISTR can even display the SY table Psum, Pdif, system temperatures, and gain factors with OPTYPE ’GAIN’ and DPARM(1) set to 17, 18, 15, or 16, respectively.

More importantly, the SY table can be used to do an initial calibration of the visibility data. Use the display programs to decide if your SY table is fine as is or needs editing. SNPLT may be the best way to examine The SY table. In 31DEC23, SNRMS may be used to determine the average values in the table and their rms. The following is somewhat controversial; there are users who believe that the SY table is never adequately reliable for all antennas because of RFI and other issues. Since the value of Tcal are not perfectly known, nor are the antenna efficiencies, the corrections made by TYAPL can never be perfect. Nonetheless, in many cases, they are a great help. CALIB will still be needed to finish the amplitude calibration.

The tasks TYSMO and TYAPL may be used with EVLA data having an SY table. TYSMO flags SY samples on the basis of Pdif, Psum, Pgain, and Tsys and then smooths Psum, Pdif and Pgain to replace the flagged samples and/or reduce the noise. You will want to do this to remove outlying bad points and to reduce the jitter in these measurements. TYSMO even applies a flag table to the SY table before its clipping and smoothing operations. Be sure to plot the results to make sure that the task did what you wanted. It is both powerful and very helpful. Then use TYAPL to remove a previously applied SY table (if any) and to apply the SY table you have prepared. The result should be data scaled nearly correctly in Jy and weights in 1∕Jy² in all IFs. The CUTOFF option allows you to use obviously good values from the SY table while passing the data from antennas with poor SY values along unchanged. If the SY values from some polarizations or IFs are bad due to RFI while others are good, you may copy the good values to replace the bad values using task TYCOP. The wide-band, 3-bit mode of the EVLA has serious non-linearities in the Pdif measurements which are not yet fully understood. For such data, use OPTYPE=’PGN’ to apply only the post-detection gains to the data. This will remove abrupt jumps due to changes in those gains (which will be more common with 3-bit data). Note that TYSMO and TYAPL also require a table of the Tcal values which OBIT provides in an  CD table. Amplitude calibrations are not applied to EVLA data weights until they have been made meaningful by TYAPL or REWAY. Set FLAGVER in TYAPL if you want to apply your flag table once and for all. TYAPL can handle flag tables much larger than those that many other tasks can handle. Note that TYAPL must copy the data set since it makes corrections to the antennas on very short intervals (not suitable to a CL table).

If you have observations of the Sun, do not use TYAPL; use SYSOL instead. This task does the usual TYAPL operation on non-Solar scans, but, for Solar scans, it must determine the average gain and weight factors on those antennas having solar Tcals and then apply those averages to all antennas not equipped with Solar Tcals. Use a well-edited SY table if possible. This Solar capability was made available in OBIT (version 567) and  in mid-June 2017. The revised format of the SY table includes a column to tell tasks whether to use normal or solar Tcals for that row of the table.

4.3.16 Calibrating polarization

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  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. Note that we discuss here the calibration of both linear and circular polarization with tasks TECOR, RLDLY, and PCAL. At the phase difference stage, however, the process for linear feeds is different than for circular feeds.

The calibration of wide-band visibility data sensitive to polarization involves three distinct operations: (1) determining and correcting the offset in delay between the two parallel-hand polarizations, (2) determining and correcting the data for the effects of imperfect telescope feeds, and (3) removing any systematic phase offsets between the two systems of orthogonal polarization. These three components of polarization calibration will be considered separately.

Faraday rotation in the Earth’s ionosphere can add a phase difference between the two parallel-hand polarizations which is both time and direction dependent. These are particularly important at lower frequencies (< 4GHz), but may rarely affect higher frequency-data as well. One way to remove at least some of the ionospheric phase offsets is by applying a global ionospheric model derived from GPS measurements. The  task TECOR processes such ionospheric models that are in standard format known as the IONEX format. These models are available from the Crustal Dynamics Data Information System (CDDIS) archive. There is a procedure which is part of VLAPROCS, called VLATECR that automatically downloads the needed IONEX files from CDDIS and runs TECOR. It will examine the header and the NX table and figure out which dates need to be downloaded, so the observation date in the header must be correct and an NX table must exist. See EXPLAIN VLATECR for other requirements. The default TECRTYPE is jplg which seems to work well for the VLA.

TECOR applies its corrections to the highest numbered CL table and writes a new one. A significant bug fix was made in January 2023 and applied retroactively to the 31DEC21 and 31DEC22 releases. TECOR now writes a “TE” table which may be plotted with TEPLT to examine all details of this correction. The new task MFIMG may also be used to make an image of the TEC data or the model used for the Earth’s magnetic field in TECOR. All standard image displays may then be used to view the results.

Frequently, the delay difference between right- and left-hand, or V and H, polarizations must be determined even if FRING was not required for the parallel-hand data. Use POSSM to plot the RL and LR (or VH and HV) spectra to see if 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

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.

To begin with, 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:

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.

Use TASAV to copy all your table files to a dummy uv data set, saving in particular the CL table with the results of the amplitude and phase calibration. This step is not essential, but it reduces the magnitude of the disaster if the the next step fails in some way. (Note - this may be a good idea at several stages of the calibration process!)

The task TACOP may be used to recover any tables that get trashed during later steps. Note that the next step writes a new CL table, but changes the AN and SU tables in place.

Having prepared a continuum solution for Q and U, you must also correct it for the difference in phase between R and L (or V and H) polarizations which normally varies considerably between IFs. For circularly polarized feeds only, 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.