9.4 Tools for data examination

Before proceeding further it is important to examine the data, to make sure they are all loaded, and (especially if this is the first time you have reduced VLBI data) to familiarize yourself with the data structure. As processing continues it is also important to inspect the data periodically to check on the progress of the calibration. Use the verb IMHEAD regularly to check the uv-data header, particularly the list of tables (as seen in

Some tasks that can be used to examine the data and the associated tables are LISTR, DTSUM, POSSM, VPLOT, CLPLT, CAPLT, EDITR, TVFLG, SPFLG, SNEDT, SNPLT, PRTAB, FRPLT, PRTAN, COHER, OBPLT, and SHOUV. Some of these tasks are described in the next few pages.

9.4.1 Textual displays

As a first step, use the procedure VLBASUMM to print out the essential contents of your data set:


to acquire the procedures; this should be done only once since they will be remembered.

> INDISK n ; GETN ctn  C R

to specify the input file.

> DOCRT -1  C R

to direct the output to the line printer.


to review the inputs.


to run the procedure.

This will make a listing of the scans, sources, frequency structure, and antennas found in your data set. You should run this procedure after “fixing” the data with VLBAMCAL, VLBAFQS, VLBASUBS, and VLBAFPOL, but you may also find it useful on the initial dataset.

VLBASUMM runs the task LISTR to give a listing of the scans, with source names, time ranges, frequency ID’s and total number of visibilities per scan for each of your output files. It is often useful to print out a paper copy of this to facilitate later data plotting/editing. If you did not do VLBASUMM, use:


to review the inputs.

> INDISK n ; GETN ctn  C R

to specify the input file.


to request printing of scan summaries.

> DOCRT -1  C R

to direct the output to a printer.

> OUTPRINT ’ ’ ;  C R

to have the output printed immediately.

> GO  C R

to run the program.

Note that at the end of the above LISTR output is useful information about the frequency structure of your data set. In 31DEC23, LISUN produces a similar listing plus the angle of each scan to the Sun with fewer adverbs to consider. LISTR with OPTYP = ’LISTand DPARM(1) = 1 is also a good way to look for phase coherence. If LISTR fails at this point, you may have forgotten to run INDXR and/or MSORT (see and

Other verbs/tasks for inspecting your data include;

  1. IMHEAD lists the file header including information on the number of frequency channels and Stokes parameters in the data and gives a list of all the extension tables.
  2. PRTAB can be used to print the contents of any of these extension tables, for instance the SU or Source table contains information about each of the sources observed. Some tables have very many columns. You can use input parameter BOX to select the list of the columns you want to print.
  3. PRTAN, run by VLBASUMM, provides a listing of the antenna names and their associated antenna numbers. This is useful because antennas are generally specified by their antenna numbers. (Procedure ANTNUM in the VLBAUTIL set allows you to translate station names into numbers.)
  4. DTSUM produces a matrix showing the number of visibilities on each baseline for each scan allowing a check to be made that all baselines have been loaded. It also tells you the data integration time. DTSUM also has a mode (triggered by setting APARM(1) = 1) that will produce a useful matrix summary of your whole data set. DTSUM does not properly report the integration times when there are multiple integration times in the data set. Verb GETITIME in 31DEC19 returns the integration time as INTTIME but has the same problem with multiple integration times.

9.4.2 Spectral displays: POSSM

Your data file will probably contain a number of IFs, observed at different frequencies, corresponding to the separate “IF channels” used during the observations. Use IMHEAD to find the number of IF channels and the number of spectral channels per IF channel, or examine the LISTR or LISUN output. The reason that the data must be stored in narrow spectral channels, even for continuum applications, is that, in VLBI, the geometrical and propagation errors affecting the data can be large enough to cause significant phase changes across an IF channel bandwidth, preventing a coherent integration over the full bandwidth.

The frequency structure of the data can be inspected using POSSM, which provides a plot of visibility data as a function of frequency as integrated over a specified time interval. Optionally, data from up to nine baselines can be plotted on a single plot page. Initially it may be interesting to view the frequency structure of data on a bright calibrator source, as in the example below. Because, prior to calibration, the phases in each IF channel are likely to vary rapidly with time, it is important to average data coherently only over a short time interval. In general, you will see phase slopes and offsets affecting the data; these phase errors must be determined and removed before the data can be averaged in frequency and/or time. See 9.5.7 for more information and a sample plot. There is a procedure simplifying the use of POSSM:


to acquire the procedures; this should be done only once since they will be remembered.

> INDISK n ; GETN ctn  C R

to specify the input file.


to plot all sources.


to plot all times.


to plot all subarrays.


to plot the cross-power spectrum for baselines with antenna n.


to plot Stokes I.


to apply CL table CLin to the data before plotting.

> DOTV 1 C R

to plot the data on the TV; -1 to make a plot file.


to plot the data.

To use POSSM directly to display the visibility spectrum of a source on the TV, use:


to review the inputs.

> INDISK n ; GETN ctn  C R

to specify the input file.

> SOURCES ’OQ208’, ’ ’  C R

to specify a single source name.

> TIMER 1 2 15 0 1 2 15 30  C R

to define a time range.


to plot all baselines to antenna 8.

> DOCAL -1  C R

to plot the data without calibration.

> BIF 0 ; EIF 0 ; BCHAN 0 ; ECHAN 0  C R

to include all IFs and spectral channels.


To plot RR and LL separately.


to plot amplitude and phase.


to average over the full time range.

> APARM 1 , 1 , 0 , 0 , -180 , 180 , 0 , 0 , 3 , 0  C R

to control the plot: APARM(1)=1 to use vector averaging, APARM(2)=1 to use fixed scale plots, APARM(5) and APARM(6) to set phase range, APARM(9)=3 to plot all IFs and polarizations together in one diagram.


to have 9 plots per page without division by “channel 0” and without writing the spectrum to a file.

> DOTV 1  C R

to plot on the TV, else create plot extension.


to use all disks for scratch.

> GO  C R

to run the program.

Note that the amplitudes are totally uncalibrated at this stage and are in units of “correlation coefficients”; these will generally appear on plots mislabeled as mJy (representing multiples of 10-3 in correlation coefficient). POSSM can produce text output into the file given by OUTTEXT.

Sample POSSM displays are given in Figure 9.1, Figure 9.2, and Figure 10.2.

Task SHOUV with OPTYPE SPECwill display the data from a number of channels on the printer with optional time averaging.

9.4.3 Time displays: VPLOT, CLPLT, and CAPLT

The task VPLOT can be used to view the visibility data as a function of time (or other variables). Again, data from several baselines can appear on one plot page. Plots of amplitudes and phases and several other quantities can be made, although, to view closure phase and amplitude, you must use tasks CLPLT and CAPLT. Note that VPLOT can average spectral channels or plot them individually under control of AVGCHAN and can plot spectral channels and IFs in separate panels or all together under control of CROWDED. Data points are plotted with a user-selected SYMBOL and may be connected by lines under control of FACTOR. Calibration can be applied to the data before they are plotted. The data can be averaged in time and the max/min within the interval plotted along with the average. Also, if desired, a model can be plotted against the data. The model can either be displayed at the times of the data samples or, with somewhat less accuracy, continuously, even at times for which there are no associated data or recorded uv coordinate values.

The following parameters will display uncalibrated amplitudes and phases from a single spectral channel of a single IF channel for a short scan on a bright calibrator:


to review the inputs.

> INDISK n ; GETN ctn  C R

to specify the input file.


to ensure no model is plotted.

> SOURCES ’OQ208’ ’  C R

to specify the source name.

> BIF 4  C R

to give first included IF channel.

> EIF 4  C R

to give last included IF channel;, if EIF > BIF then IFs are averaged.

> BCHAN 8  C R

to set the lowest spectral channel to include in average prior to plotting.

> ECHAN 8  C R

to set the highest spectral channel to include in average prior to plotting; no averaging in this case.

> TIMER 1 2 15 0 1 2 25 00  C R

to define a 10-minute time range.

> DOCAL -1 ; DOBAND -1  C R

to apply no calibration tables or bandpass tables to data.

> OPTYP ’ ’

to display cross-correlations; ’AUTO’ to get auto-correlations.


to do no time averaging of the data before plotting.

> XINC 1  C R

to plot every record.

> BPARM 0 , -1  C R

to set x-axis type (BPARM(1) = 0 plots time (Hrs, min, sec)), y-axis type (BPARM(2)= -1 plots both amplitude and phase), and to use self-scaling. See EXPLAIN VPLOT  C R for other options.

> NPLOT 4  C R

to plot 4 baselines/page.

> GO  C R

to run the program.

An example VPLOT output appears as Figure 9.1. It may be useful to make a plot of data “weight” versus time on the autocorrelation data from each antenna (set BPARM(2)=16 and OPTYP = ’AUTO’). The weight depends on the number of valid bits correlated and is a good indication of tape playback quality.

9.4.4 EDITR

Another task which can display data as a function of time is EDITR. This program is used primarily to edit data interactively (see 5.5.2), but its interactive aspects (i.e., allowing the user to “zoom” in on certain time periods) make it useful for pure data inspection. EDITR has been much improved in recent releases to offer options reminiscent of difmap, the VLBI data reduction package from CalTech. In particular, the CROWDED adverb allows displays of all IFs and/or all polarizations at the same time (and color may be used to separate them visually) and the FLAG QUICKLY run-time option allows fast sample deletion with only quick mouse clicks. It is well worth exploring the abilities of this powerful program. Tasks TVFLG (O.1.6), SPFLG (10.2.2), and WIPER are also useful in this way.

9.4.5 SNPLT

Supplied with your VLBI data will be a number of important tables used for calibration, and many more are generated as calibration proceeds. 9.8 summarizes the contents of each of these tables. Two of the most important tables are the calibration or CL tables and the solution or SN tables.

The task SNPLT should be used periodically to inspect the contents of the latest CL and SN tables. There is a simplified procedure for making these plots:


to acquire the procedures; this should be done only once since they will be remembered.

> INDISK n ; GETN ctn  C R

to specify the input file.


to plot a CL table.


to plot the highest version.


to plot all sources.


to plot all times.


to plot both R and L solutions.


to plot all subarrays.


to look at amplitudes; PHAS, DELA, and ’RATE’ are other useful choices.

> DOTV 1 C R

to plot the data on the TV; -1 to make a plot file.


to plot the data.

Using SNPLT directly, the example below plots the antenna-based amplitude corrections stored in CL table m.


to review the inputs.

> INDISK n ; GETN ctn  C R

to specify the input file.


to plot amplitudes.


to plot both polarizations in each plot; selected IFs are plotted separately.


to choose CL table version.


to plot on the TV screen including failed solutions; otherwise, create plot extension files.


to plot 5 antennas/IFs per page.

> DO3COL 2  C R

to separate the calibration sources by color.

> GO  C R

to run the program.

SNPLT can also be used to plot quantities from other tables generated by the calibration process including the contents of TY (“system temperature”), and PC (“phase-cal”) tables. OPTYPE = ’MULTallows you to compare values of two or more parameters, plotting them at the same time in separate panels. Task SNIFS is similar and may be used to compare values across IFs. SNBPL is similar to SNPLT, but plots calibartion parameters on a baseline basis. Unstable phases may be stable on some baselines, but not others. The antenna-based plots of SNPLT may not differentiate these adequately.

9.4.6 COHER

The task COHER can be used to determine the coherence time in a uv data set broken down both in time and by antenna and baseline. The coherence time is estimated by comparing vector and scalar averaged amplitudes over increasing time averaging intervals. Averaging is not performed over source scan boundaries. The coherence time is defined as the averaging interval over which the ratio of vector and scalar amplitudes falls below a pre-assigned level. This can be set under user control. In addition, data that fall below a specified signal-to-noise ratio can be excluded from the coherence time estimates. Provision is made for selection by source name, time range, IF channel, antenna and frequency ID, and the ability to average over a subset of individual frequency channels. The input parameters to task COHER take the form:


to review the inputs.

> INDISK n ; GETN ctn  C R

to specify the uv input file.

> APARM 0  C R

SNR cutoff 5; vector to scalar cutoff 0.8.

> TIMERANG 0, 10, 5, 0, 0, 10, 15, 0  C R

time range selection.

> BIF 5 ; BCHAN 1 ; ECHAN 8  C R

IF and channel selection.

> SOURCES ’DA193’ , ’   C R

source selection.


Frequency ID selection.

> GO  C R

to run the program.

Be warned that COHER can take quite a long time to run. A simpler, though less rigorously correct method for determining coherence intervals is to examine the data on different baselines using EDITR.

9.4.7 FRPLT

A preliminary examination of the coherence of individual scans or time segments in the data can also be performed using task FRPLT, which allows time series or fringe-rate spectra to be plotted for one or more baselines and time intervals. This task allows data to be selected by the usual criteria, including time range, source name, and frequency parameters, amongst others, and will then plot the individual time series in amplitude and phase or the associated fringe-rate spectrum. Provision is made for averaging over frequency channels within each IF, for varying degrees of padding in the FFT, and for division by the pseudo-continuum average “channel zero” before plotting. Typical input parameters to FRPLT are:


to review the inputs.

> INDISK n ; GETN ctn  C R

to specify the input file.

> SRCNAME ’DA193’  C R

to select a source.

> TIMERANG 0, 10, 5, 0, 0, 10, 15, 0  C R

to select a time range, strongly recommended.


to plot fringe-rate spectra for every 2-minute interval in TIMERANG.


to do 6 plots per page.


to select a single Stokes.

> BIF 1 ; EIF 0  C R

to select the IFs included.

> BCHAN 3 ; ECHAN 12  C R

to select the range of frequency channels averaged within each IF.

> ANTENNAS 3, 4  C R

to select baseline 3-4, 3-5, and 4-5.

> BASELINE 4, 5  C R

to select baseline(s) plotted in the familiar way.

> DOCAL -1 ; DOPOL -1  C R

to apply no continuum calibration.

> DOBAND -1  C R

to do no bandpass calibration.

> APARM(1) 1 ; APARM(2) 0  C R

integration time 1 sec; self-scale the plots.

> APARM(7) 0  C R

plot fringe-rate spectra with no padding.

> BPARM 0  C R

to do no division by “channel zero.”

> DOTV 1  C R

to plot directly on the TV device.

> GO  C R

to run the program.

The underlying time series in amplitude and phase can be plotted by setting APARM(7)=1; otherwise the fringe-rate spectrum is plotted. Note that the baseline(s) shown are selected with ANTENNAS and BASELINE in the usual way and a time range must be selected with TIMERANG. APARM(1) sets the integration time to be used before doing the FFT over the selected time range. SOLINT may be used to break the time range into intervals. Separate plots are produced for each IF, baseline, and time interval.