Two different IF bandwidths enter into the observing strategy for spectroscopy.
that is spanned by all the line channels.
determines the range of velocities included by the experiment,
which should span not only the interesting parts of the line profile,
but also an adequate number of line-free channels (allowing for any
band-edge effects in autocorrelation spectrometers).
. This sets the velocity resolution of the experiment.
It also determines the chromatic aberration effects for the individual
channel images.
As indicated in Figure 3, the choices of
and are usually tightly coupled by the design of the
spectral line correlator, and velocity resolution and velocity span
must often be traded against one another.
The need for velocity resolution usually drives the
channel bandwidth to a small enough value that individual channel
images are negligibly smeared by chromatic
aberration. This does not mean, however, that spectroscopists can
ignore chromatic aberration in synthesis imaging. The
synthesized beam pattern is different for every channel image when
the channels are gridded separately with their correct frequencies.
Any data reduction schemes that do not separately deconvolve the
synthesized beam from each of the individual channel images may
therefore ``rediscover" effects related to the frequency dependence of
the beam. As Lecture 12 describes, such separate deconvolution of all
the channels is not always a good strategy. If you estimate the
continuum distribution by summing the ``dirty" images over a range of
line-free channels, or if you difference two ``dirty" line channel
images, the variation of the synthesized beam with frequency may
become apparent. If you try to avoid this effect for some channels by
artificially assigning them the same frequency (to make the ``dirty
beam" the same for all images), you will restore the chromatic
aberration corresponding to the widest channel separation
used in such processing.