Writing Visibility FITS Files

Task fits can also be used to write a visibility FITS file. In addition to the keywords used when reading visibility datasets, fits uses keywords select, stokes and options to control extra processing performed on the visibilities.

Unfortunately fits is more sophisticated at reading FITS files than writing -- note the following caveats carefully.

• fits does not save MIRIAD calibration tables. However fits does give you the option of applying the calibration, as it writes out the data (see the options keyword).

• fits does not create an output antenna table. If you load these data into AIPS , a number of AIPS tasks may have problems with this lack of antenna table. The simplest remedy for this is to add an antenna table to your new AIPS dataset by copying it (using AIPS task TACOP ) from an old visibility file. Make sure, however, that it is an appropriate antenna table!

• fits can cope with only a single pointing of a single source, with a single frequency setup. fits does not issue warnings if you fail to obey this restriction, and quite happily will use the relevant parameters of the first record.

• fits does not write a frequency table (FQ table). fits assumes that the data are a single IF band, with a constant frequency increment between channels. If presented with multiple spectral windows (multiple IFs), an average frequency and increment will be used. This is very undesirable. If you do this, fits will issue warnings about frequencies deviating from linearity. These warnings should be taken seriously. To avoid this, use the select or line parameters to force fits to write out a single spectral window.

• MIRIAD allows you flexibility in the choice of polarisation correlations that you can write out. On-the-fly polarisation conversion can be performed. However AIPS will refuse to read in a number of the possibilities that MIRIAD allows. In particular, AIPS \ UVLOD refuses to load visibility data where there are three different sorts of polarisation correlations (rather than the more usual 1, 2 or 4 sorts).

Typical inputs follow: