Chapter 9
Reducing VLBI Data in AIPS

This chapter describes the reduction of VLBI data in AIPS . A Step-by-step recipe, covering both simple and more difficult situations, is presented. See Appendix C for simpler and shorter recipes suitable for straightforward observations. Procedures to simplify some of the VLBI reduction steps are mentioned and become available to you after you enter the command RUN VLBAUTIL  C R. There is also a VLBA pipeline, VLBARUN, which is useful for simple datasets, see 9.2. If you have very old VLBI data, in Mk III or Mk IV format, see O.2. We also include here some background information concerning the structure of VLBI data sets, the data reduction philosophy and a description of some of the effects for which corrections must be determined and applied. It is important to understand these aspects if you wish to reduce your data reliably. For more background information on VLBI data reduction consult VLBI and the VLBA, Astronomical Society of the Pacific (ASP) Conference Series No. 82, 1995.

Programs of particular interest for VLBI may be found in Chapter 13 or displayed from inside AIPS by typing ABOUT VLBI C R or APROPOS VLBI  C R. Remember, the best and most complete information available on all AIPS verbs and tasks may be found in their EXPLAIN files. A 15APR97 or later version of AIPS is required to support full Space VLBI data reduction.

Most types of VLBI data, once read into AIPS, appear very similar in structure as far as the user is concerned. We shall concentrate on describing the reduction path for data produced by the VLBA correlator, but most operations also apply to MkII, MkIII, and MkIV data. See O.2 where we draw the reader’s attention to any differences. The few extra steps necessary for calibrating phase-referencing observations are described in Note that successful phase-referencing observations require careful planning before the observations. See VLBA Scientific Memo No. 24 by J. Wrobel, C. Walker, J. Benson, and A. Beasley.

Some of the VLBI-related tasks require the ability to read files resident outside AIPS . To communicate to AIPS the directory in which these files exist it is necessary to define a logical pointer or environment variable. Please refer to 3.10 to see how this is done.

While the majority of VLBI observations are continuum observations, more sophisticated data reduction techniques are increasingly common. Continuum VLBI observers sometimes also apply spectral-line VLBI techniques to improve the dynamic range of their data sets. For these reasons, this chapter is organized to make the discussion of data reduction techniques more uniform. The overview in this section of the steps involved for several type of VLBI data reduction is meant to guide the user through the rest of the chapter. It is strongly recommended that you read the overview carefully before proceeding.

The expected size of the output uv data file can be an important consideration in VLBI data reduction. The disk space required by AIPS for a compressed dataset is given by the relation:

                                    Nant         Texpt
Disk Space = 4x 10-6 NStokes Nchan NIF-2-(Nant + 1)-△T-M Bytes
where Texpt is the total observing time, T the correlator integration time, NStokes the number of polarization correlation pairs (RR,LL,RL,LR), Nchan the number of spectral channels per IF, NIF the number of IF’s, and Nant the number of antennas in the network. Space VLBI (SVLBI) data can have different integration times for the ground and space baselines of Tg and Ts, respectively, and therefore the total disk space requirement is larger. If Nant is the number of ground telescopes,
                                    [                             ]
Disk Space = 4 x10-6 NStokes Nchan Nif 1Nant(Nant - 1)Texpt-+ NantTexpt M Bytes
                                     2             △Tg        △Ts
In uncompressed format the same data set will require two-three times the disk space. Be forewarned that some tasks attempt to create uncompressed scratch files which may not fit into the available disk space. The amount of available free disk space can be determined using the AIPS command FREE. The blocks referred to in the FREE output are equal to 1024 bytes.

Note that certain operating systems are still subject to a 2-Gigabyte limit for any individual file, as a result of their 32-bit file systems. Larger AIPS files are supported on DEC Alpha, SGI (running XFS), HP, Solaris (revision 2.6), and Linux (kernel 2.4.2). The last three require all AIPS’ C code to be compiled with an additional option. The size of the output file can be reduced by IF selection or limited concatenation in FITLD or by time- or spectral-averaging later using UVAVG, SPLAT or AVSPC.

It is possible to construct data sets on disk that cannot be written to a single tape using FITTP because FITTP uncompresses the data when writing to tape. The task FITAB is designed to address this problem. FITAB writes data in compressed form to tape and can write data in pieces to multiple tapes. Note that FITAB is only available in 15APR99 and later releases and that versions of FITLD from earlier releases cannot read such data. Packages other than AIPS may also be unable to understand these files.

One large point of divergence in the reduction of continuum polarization VLBI data is the question of whether or not to determine separate LL and RR phase solutions. The polarization-specific portions of the recipe given below are based upon the premise that L and R phase solutions should always be determined separately on the grounds that it is safer and should work with data from a wide variety of antennas. If the L-R phase offsets for antennas in your data set are small and constant in time, you may consider modifying the recipe in 9.1 by determining averaged LL,RR phase solutions everywhere except in step 7.

 9.1 VLBI data calibration recipe
 9.2 Pipeline for the VLBA
 9.3 Loading, fixing and inspecting data
  9.3.1 Loading data from the VLBA correlator
 9.4 Tools for data examination
  9.4.1 Textual displays
  9.4.2 Spectral displays: POSSM
  9.4.3 Time displays: VPLOT, CLPLT, and CAPLT
  9.4.4 EDITR
  9.4.5 SNPLT
  9.4.6 COHER
  9.4.7 FRPLT
 9.5 Calibration strategy
  9.5.1 Incremental calibration philosophy
  9.5.2 Processing observing log and calibration information
  9.5.3 Data editing
  9.5.4 Amplitude and instrumental delay calibration
  9.5.5 Spectral-line Doppler correction
  9.5.6 Spectral-line amplitude calibration
  9.5.7 Phase calibration
  9.5.8 Complex Bandpass
  9.5.9 Baseline-based errors
 9.6 After initial calibration
  9.6.1 Applying calibration
  9.6.2 Time averaging
  9.6.3 Verifying calibration
 9.7 Self-calibration, imaging, and model-fitting
  9.7.1 CALIB
  9.7.2 IMAGR, SCIMAG, and SCMAP
  9.7.3 Non-conventional methods of imaging
 9.8 Summary of VLBI calibration tables
 9.9 Additional recipes