Imaging multiple freqids and multiple channels

 

The final category of continuum data that you might wish to image is multiple bands of channels.

1. It will be necessary in this case to use SPLIT\ (§ 12) for each freqid of interest (or if you combined data from different observations you may already have done this). It may then be necessary to sort each visibility file into XY order with UVSRT as described in § 15.2 if you are making very large images (but try the internal sort of MX and HORUS first).
2. Now, we would like to concatenate the data with DBCON, but there is a problem. Each individual file will have a different reference frequency (use IMHEAD to see this). DBCON will create an output file with just one reference frequency, and it will be the reference frequency of the first freqid file used in the concatenation process. This is no good, because MX and HORUS will use this reference frequency for all the data when they try to work out the variation of u and v across the band so that the different channels can be gridded in the correct place.

   The solution is to image each file individually, and then sum the resultant images and beams. Since the Fourier transform is a linear procedure, there is no substantial difference in making the image in this way. The only caveat to this is that this statement is really only true for natural weighting (uvwtfn='na' in the imaging tasks; see § 15.2). You should image the individual files with MX or HORUS just as described in § 15.4.

   Again MIRIAD performs much better in the multi-frequency synthesis imaging area than AIPS.

3. Now average the images and beams with COMB. This image can be deconvolved with the averaged beam.

   doalign=1 Check images for alignment

   blc=0 Combine all of images trc=0

   opcode='sum' Sum images aparm=0

   aparm(1)=0.5 Divide input image 1 by 2

   aparm(2)=0.5 Divide input image 2 by 2 bparm=0