Determine the imaging parameters needed for a full-field image of the continuum. Set the BPARMs to cover the full single-dish beam area (or more at 21-cm wavelength).
Make a CH0 image with a large field (set by IMSIZE and CELLSIZE returned by SETFC) to look for strong continuum sources. Use uniform weighting (UVWTFN ’ ’) with ROBUST of 0 or -1 if resolution is more important than detecting significantly extended sources; set ROBUST to 2 or more if the converse is true.
Then make a dirty image cube of your LINE data in the center field only (unless more are needed for the line source).
Examine the cube on the TV to determine which channels are free of line emission.
Calculate the noise in a few of the line-free channels.
Select a large window that contains no continuum sources.
Calculate the RMS noise inside the window. The verb TVSTAT is helpful if there are no large rectangular windows free of continuum emission.
Having an initial guess of the signal-free RMS, you may attempt to refine that estimate (see message generated) and also produce a histogram plot.
Remove the continuum emission by fitting a baseline in the uv plane to the line-free channels; see EXPLAIN UVLSF and §8.3. If you don’t want this task for flagging, but only for continuum subtraction, the important parameters are
Choose your line-free channels, normally avoiding the channels at the edges of your bandpass since they are usually quite noisy.
ICHANSEL < begin1 >,< end1 >,, 0, 0, < begin2 >,< end2 >
This task will also write a new continuum data set which is a better estimate of the continuum than the CH0 file which contains contributions from the line signals. If there is a bright continuum source far away from the phase center, you will want to use SHIFT to center it while running UVLSF. Multiple interfering sources might even require UVSUB with a good continuum model rather than UVLSF.
Make a dirty cube containing only the line emission. Use the same parameters as in Step 22, including usually only one field — but consider multi-scale Clean.
Calculate the noise in the line-free channels as in Step 24.
Make a contour plot of the beam.
Select a window containing the source (or beam) you want to plot.
Display the contour plot on the TV. For a beam plot, good parameters are
When you are happy with the plot, generate a plot (PL) extension file.
Send the plot file to the default printer.
Measure the beam diameter at half-power (FWHP) to get beam parameters (BMAJ, BMIN, BPA) for the Cleaning process. If you do not specify these, IMAGR will choose them automatically by fitting a Gaussian beam with an elliptical cross-section. This is usually fine for uniformly-weighted images (ROBUST around 0), but may not be desirable for naturally-weighted images for which the beam is often rather non-Gaussian.
Select boxes containing all of the line emission in the cube. Use as many boxes as necessary (up to 50). The better that you constrain the locations of real emission, the faster the Cleaning process will go. Note that IMAGR allows you to set the boxes interactively when you run with DOTV 1, which is important when Cleaning reveals emission initially lost in the sidelobes of the stronger objects.
Make a Clean cube of the line emission. Start by Cleaning just one or two channels (set with BCHAN and ECHAN) to ensure that your inputs are set correctly. If you don’t want to waste time, it is a good idea to Clean only those channels with emission that is ≥ 4 times the RMS noise (Step 20). Use SUBIM and MCUBE to construct a full cube of the images later. Use the same parameters as in previous runs of IMAGR, but specify the beam parameters (BMAJ, BMIN, BPA) and control the Cleaning depth with FLUX and NITER.
Examine the Cleaned cube and do progressively deeper cleaning with IMAGR until you are satisfied with the result. The DOTV 1 option in IMAGR will help you reach this state more quickly. You can examine the Clean components with PRTCC. You should Clean until the total Cleaned flux converges.
Correct for attenuation away from the center of the primary beam. Use your Cleaned cube as the input, the default parameters should be adequate, and the output is a corrected cube.
Print the contents of the tapes for your records.