[ Basic Info | References | User Guide ]
Basic Information on clean
Purpose: Apply Hogbom, Clark or Steer CLEAN algorithm to a map
CLEAN is a MIRIAD task, which performs a hybrid Hogbom/Clark/Steer Clean
algorithm, which takes a dirty map and beam, and produces an output
map which consists of the Clean components. This output can be
input to SELFCAL to self-calibrate visibilities, or input to RESTOR
to produce a "clean" image. Optionally CLEAN can take as one of
its inputs a model of the deconvolved image. This model could be
from a previous CLEAN run, or from other deconvolution tasks
The input dirty map, which should have units of Jy/beam. No
The input dirty beam. No default
An initial model of the deconvolved image. This could be the
output from a previous run of CLEAN, or the output of any of the
deconvolution tasks (e.g. MAXEN). It must have flux units of
Jy/pixel. The default is no model (i.e. a zero map).
The name of the output map. The units of the output will be
Jy/pixel. This file will contain the contribution of the input model.
It also should have a different name to the input model (if any).
It can be input to RESTOR, CLEAN (as a model, to do more cleaning),
or SELFCAL (for self-calibrating visibility data).
The minor iteration loop gain. Default is 0.1.
Extra processing options. Several can be given, separated
by commas. Minimum match is used. Possible values are:
negstop Stop when the first negative component is encounters.
This does not apply when using Steer iterations.
positive Apply a positivity constraint. This constrains the
component image to be non-negative. A side-effect of this
is that CLEAN will stop iterating if it cannot continue
to ensure this. This does not apply when using Steer
asym The beam is asymmetric. By default CLEAN assumes the
beam has a 180 degree rotation symmetry, which is the
norm for beams in radio-astronomy.
pad Double the beam size by padding it with zeros. This
will give you better stability with Clark and Steer
modes if you are daring enough to CLEAN an area more
than half the size (in each dimension) of the dirty
CLEAN finishes either when the absolute maximum residual falls
below CUTOFF, or when the criteria described below is
satisfied. The default CUTOFF is 0.
The maximum number of minor iterations. The default is 250, which
is too small for all but the simplest of images. CLEAN will stop
when either the maximum number of iterations is performed, or the
cutoff (see above) is reached, or if options=negstop was given and
a negative component was found, or if options=positive was given,
and no more positive components could be found.
This specifies the region to be Cleaned. See the help on region
for more information. The default is the largest region that can
be deconvolved safely.
Cornwells prussian hat parameter. When cleaning extended sources,
CLEAN may produce a badly corrugated image. This can be suppressed
to some extent by cleaning with a dirty beam which has had a spike
added at its center (i.e. a beam that looks like a prussian hat).
PHAT gives the value of this spike, with 0 to 0.5 being good
values. Default is 0 (but use a non-zero value for extended
The minimum patch size when performing minor iterations, in pixels.
Default is 51, but make this larger if you are having problems with
corrugations. You can make it smaller when cleaning images which
consist of a pretty good dirty beam.
This is the same as the speed-up factor in the AIPS APCLN.
Negative values makes the rule used to end a major iteration more
conservative. This causes less components to be found during a
major iteration, and so should improve the quality of the Clean
algorithm. Usually this will not be needed unless you are having
problems with corrugations. A positive value can be useful when
cleaning simple point-like sources. Default is 0.
This can be either "hogbom", "clark", "steer" or "any", and
determines the Clean algorithm used. If the mode is "any", then
CLEAN determines which is the best algorithm to use. The default
This sets the relative clip level in Steer mode. Values are
typically 0.75 to 0.9. The default is image dependent.