Synopsis
mosaicwizard()
include "mosaicwizard.g"
Description
mosaicwizard provides a GUI for making a deconvolved image from a
multi-field Measurement Set or FITS file with the minimum of user intervention. It is based on
table, imager, ms,
and wizard. Display is performed using the
viewer.
In mosaicwizard, the user is walked through a number of steps
using a simple GUI to create first a low resolution image, and
then higher resolution images until the full resolution is achieved.
At each step of increasing resolution, the previous low resolution image
is regridded and used as a starting model, so only incremental changes
from this model are solved for. This technique is not very effective
for dealing with point sources, but it is excellent at imaging extended
structure. The steps which the mosaicwizard lead you through include:
- Either a MeasurementSet or a FITS file can be specified.
If the latter, it is converted to a MeasurementSet using
the fitstoms constructor of the
ms tool.
- The user can provide an optional initial model.
- The spectral windows are listed and the user asked to
select one for imaging.
- The positions of the fields are plotted, and the user
is asked to select several for imaging. The extent of the image
mosaicwizard makes is initially set by the fields selected and the
calculated extent of the primary beam.
- The main imaging loop is described below:
- The user is asked to select the deconvolution algorithm and the values
for certain control parameters. The most important of these is the
(u, v) scaling parameter (which controls the maximum baseline used as
a fraction of the maximum baseline present in the data, hence
determining the resolution of the mosaic image).
Some deconvolution algorithms, such as MEM and Multi-Scale
Clean, require additional specialized parameters. Note that the image
size and cell size are not directly under user control. However, they
are indirectly controlled by the region size (see below) and the
(u, v) scaling parameter.
- A low resolution deconvolved mosaic of all selected fields is made. In this
stage, the data are weighted as determined by the weighting parameter,
plus an additional Gaussian taper if the image is not at full
resolution. The Gaussian taper will reduce PSF sidelobes caused by
the sharp cutoff in the Fourier plane coverage when the (u, v)
scaling parameter is less than 1. If you set the displayprogreess
parameter to be true, you will get a real time display showing the residuals
and the cleaned flux as a function of iteration number. This progress
display can be very helpful in understanding what is happening in the
deconvolution, and in analizing any failures which may occur.
- The resulting image is
displayed and the user asked to zoom the region of interest. Either a
rectangular or a polygonal region can be selected. The region will
have two effects on the imaging: first, the angular size of the next
iteration's image will be smaller, matched to this region (plus an
appropriate guard band as required by the PSF; note that mosaics with
many pointings don't need the traditional factor of 2 PSF guard band);
second, the region of interest will also be turned into a mask image which will
be used like a CLEAN box in the next iteration's deconvolution, precluding deconvolved
emission from the region outside the mask. Note that the details of the mask
generation are not included in the generated script, as they require interactive
user input. Making a good
tight mask can be very important in some mosaics, especially if you
lack total power. For example, in order to get a good image out of
the test Cas A MeasurementSet, you need to supply a good mask early on in the
process.
- The imaging loop is automatically repeated, with the (u, v) scaling
parameter increasing by a factor of 2 by default, though the user may
wish to change this. The imaging loop stops when the scaling
parameter is greater or equal to 1, or when full resolution has been
reached. The model image generated from the iteration of the loop
we just completed will be regridded to the next iteration's resolution,
clipped at zero, and used as the initial model for the next iteration.
Typically, mosaicwizard works well for a quick look at the mosaic
data, but you may want to use imager directly for the
final image if you find that mosaicwizard does not give you enough
control over the imaging process.
During the deconvolution stage, many of the relevant Glish commands for
imager and the other tools are displayed. Note
that these are slightly more verbose than you would type but should be
executable. Not all commands are displayed, obviously. In particular,
we have not shown those related to the operation of mosaicwizard or those
related to the generation of the mask (which requires interactive user
input).
Hence all the commands you see are ones that you might plausibly
use during data reduction. The commands are written to a file
called 'scripter.log_*' in the current directory. A more complete
listing of the commands will be forthcoming in the next release.
Note that if you don't have a data set, then the function
imagermaketestmfms() can be used to
make a copy of the standard mosaicing test MeasurementSet XCAS.ms, a
seven pointing VLA X band D array data set on Cas A.
If you leave the MeasurementSet parameter unset in step 1 of the
mosaicwizard, the mosaicwizard will create a MeasurementSet called
mosaicwizard.ms from the Cas A data.
News 
