APPLYING THE CALIBRATION TO THE VISIBILITIES

      

After you are satisfied with the calibration, you may not want to worry about remembering about all these calibration tables, but would rather simply apply them directly to the data and write a corrected file. This step is also necessary should you want to combine multi-configuration data (see § 14). This is done with the task SPLIT. In addition to applying the calibration, SPLIT extracts individual sources from the multi-source file and writes them to single-source files. The main reason for this is that the most important imaging routines, MX and UVMAP function only on single-source files. Note that there is an imaging task called HORUS (for those who understand NRAO, HORUS was the son of ISIS) which functions on multi-source files (see § 15) as well as single-source files, so that you may not necessarily need to run SPLIT.

• You can apply SPLIT either to the channel 0 data base or to the data base containing all channels if you are interested in the spectral information.
• If you give SPLIT a list of sources, then you get one output file per source (the file name is given by the source name).
• The stokes adverb allows you to convert to Stokes parameters after applying the calibration. However, the only Stokes parameter that you can get from ATCA data with AIPS is that of total intensity (I) (see § 3.2).

  If you are working with a large spectral-line visibility file now is not an unreasonable time to set stokes='i' as this converts XX and YY into I thus producing an output file half as big as the input file.

  If you are working with continuum data it doesn't matter much in terms of disk space. You can always ask for Stokes I when you image the data. If the source is unpolarized, and you intend to self calibrate, you are best to leave this adverb blank and pass the correlations in their raw polarization form. However, if the source is polarized and you intend to self calibrate, then you should convert them to Stokes I (this can be done during the self calibration stage but the switch is hidden in one of those aparm arrays).

• As usual, you can only get one freqid at a time.
• If you are desperate conserve disk space, you might consider dropping channels at the beginning and end of the band with bchan and echan where the filter response has rolled off. However, note that if you are going to do visibility-based continuum subtraction of spectral-line data (see § 13) it is important to have as many line-free channels as possible.
• It is generally safer to split one IF at a time, as from time to time, there have been problems imaging other than the first IF. This has especially been true if you concatenate multi-IF, multi-channel files with DBCON and then try to image the second IF (see § 14; is fixed in ATNF AIPS only so caution is advised). So use bif and eif to select a single IF at a time.
• Now, to apply the calibration, set docalib=1, and set gainuse to the version of the CL table that you wish to apply. If you have several CL tables, then make sure that you apply the correct CL table with the correct source selection criteria.
• Leave dopol and blver at zero as you have not done any polarization or baseline calibration.
• Set flagver to point at the desired flagging table.
• If you wish to apply a bandpass calibration, then put doband to 1 (average all bandpass solutions in time for each antenna) or 2 (take the nearest bandpass solution in time). The option to set doband=3 does not nor will it ever work. Set bpver to point at the desired bandpass table version. You should do this if you are planning to make a multi-frequency synthesis image from all your channels, or if you are doing spectral-line work.
• If you wish to either retain or write compressed data, set douvcomp=1.

Leave the rest of the adverbs at 0 or blank and GO SPLIT. Always check the integrity of the output file by examining the header with IMHEAD to make sure that the axes are labelled as expected (RA, DEC, frequency increments and sign and so on).

inname,inclass,inseq,indisk Fill in for spectral or channel 0 data base

sources='1123-77','1333-33',' ' Specify sources to split or select subset

qual=-1 All qualifiers or specify

calcode=' ' All calcodes or specify

timerang=0 All times

stokes=' ' Pass Stokes parameters as they are

stokes='i' or ask for Stokes I only

freqid=3 One freqid per run

bif=1 Select IF eif=bif

bchan=0 All channels echan=0

docalib=1 Calibrate data

gainuse=2 Specify CL table to apply

dopol=-1 No polarization calibration

flagver=0 Apply highest version FG table or specify

doband=1 Apply bandpass by averaging all times

doband=2 Apply bandpass by finding nearest time

bpver=0 Apply highest BP table version

outclass, outseq, outdisk Specify out class, seq and disk of each output file, named by source

douvcomp=-1 Select un-compressed data

aparm=0 Defaults OK

chansel=0 Select all channels or specify

aparm(1)=1 channels to AVERAGE with chansel

chansel=8,28,1 average channels 8 through 28

Last update : 27/11/93

1