Imaging - INVERT

Most of the inputs to invert are the same as with conventional imaging. Only mosaic-specific considerations will be mentioned here. See Chapter 13 for more information. Note that invert supports multi-frequency synthesis, Stokes and spectral imaging.

• options=mosaic: The most important thing to remember is to invoke invert's mosaic mode! This causes invert to expect multiple pointings in the input visibility data, and to perform the linear mosaicing steps and geometric corrections.
• options=double: If you intend to deconvolve, options=double should always be used. This is because the full field of the each individual pointing is potentially filled with emission.
• vis: When mosaicing, invert handles input datasets which contain multiple pointing centres.
• imsize: In mosaic mode, this is interpreted as the image size, in pixels, of each subfield. There are two constraints that are important if you wish to deconvolve. These can be relaxed, with corresponding degradation in deconvolution.
  • Ideally imsize should be large enough to contain all emission in the main lobe of the primary beam (as Miriad only models the main lobe, making it larger has no beneficial effect). Table 21.2 gives these sizes as a function of frequency for the ATCA.
    

    Table 21.2: Size of Main Lobe of the Primary Beam for ATCA dishes

    Frequency	Primary Beam Main Lobe Size
    $\nu$ (GHz)	$\theta$ (arcmin)
    1.384	70.0
    2.496	42.4
    4.800	20.6
    8.640	12.2

    
    

  • The image size should not be a power of 2, or a number within the range (approximately) $[0.9\times2^n,2^n]$ (note that just 1 pixel more than a power of 2 is fine - and indeed good). This restriction is to help reduce the effects of the aliasing caused by a ``grid-and-FFT'' imaging algorithm which invert uses.
• sup: Recall that sup gives the sidelobe suppression region, and is a way of setting uniform, super-uniform and sub-uniform weights. In mosaic mode, invert's default is to suppress the sidelobes in a region the size of each sub-image. However suppressing sidelobes in an individual sub-images is not the same as suppressing them in the linear mosaic of the sub-images. This is because the linear mosaicing process weights down sidelobes some distance from the pointing centre. In mosaicing, minimum sidelobes will be achieved by setting sup equal to about twice the pointing separation (which should be about the side of the primary beam FWHM).

  As always, best point-source sensitivity results by setting sup equal to 0.

• offset: This has a different meaning in mosaic mode. It gives the position on the sky (the so-called tangent point), in RA and DEC, which is used for geometry calculations. The value is given in the form hh:mm:ss,dd:mm:ss, or as decimal hours and degrees. Normally you can allow this to default, and invert will choose a central pointing centre as the tangent point.

Typical inputs to invert would be:

INVERT
vis=lmc.uv	The input multi-pointing dataset.
options=mosaic,double	Use mosaic mode and make large beam.
offset	Usually can leave blank.
map=lmc.map	Output image name.
beam=lmc.beam	Output beam name.
cell=	Set cell size.
imsize=	Set output image size.
sup=2000	Set to approximately twice the pointing separation for
	minimum sidelobes, or
sup=0	set to zero for best point-source sensitivity.