[ Basic Info | References | User Guide ]
Basic Information on uvlin
Task: uvlin
Purpose: Separate continuum and lines in a spectral data-set.
Categories: uv analysis
UVLIN separates line and continuum in a spectral visibility
data-set. It does this by fitting a low order polynomial to the
real and imaginary parts of the line free channels of each spectrum.
This polynomial function is taken to represent the continuum, and
any deviation from this is taken to represent the line data.
Optionally a constraint can be added that the line shape is constant,
and its strength is proportional to the continuum level. This is a
good approximation in many recombination line experiments. In this
case, an iterative algorithm is used to estimate the line shape and
to optimally disentangle the line and the continuum.
See the related tasks CONTSEN and CONTERR, which give estimates
of the noise amplification and the residual continuum.
Key: vis
The name of the input uv data set. No default.
Key: select
The normal uv selection commands. The default is to select everything.
Key: line
The normal uv linetype in the form:
line,nchan,start,width,step
The default is all channels. Note that if there are multiple
spectral windows, the fitting process is performed on each window
separately.
Key: chans
This specifies the channel ranges that contain only continuum
(line free). It consists of a number of pairs, each pair giving
a start and end channel. These are relative to the channels
selected with the line parameter. The default is that all channels
are line-free, which is quite a good approximation if the line is
weak compared to the continuum.
Key: out
The name of the output uv data-set. There is no default name.
The output can be either the line, continuum, or fitted line shape,
as specified by the options keyword. The default is to write the
line data.
Key: order
Order of polynomial used to fit the continuum. The default is 1
(i.e. a linear fit). Possible values are 0 to 5. Also see the
"relax" option.
Key: offset
An offset (in arcsec) that would shift the data to the center of
the dominant emission. Giving this shift allows UVLIN to reduce
the error involved in approximating the continuum as a polynomial.
The shift information is only used in the fitting process -- the
output data is not shifted. The default is 0,0 (no shift).
The following options determine what is written to the output
data-set.
Key: mode
This determines what is written out. The default is to write out
the line data. The following are possible values:
line The output data-set is the line data. This is
the default.
fit The output data-set is the fitted line shape data.
This is valid only when using the 'lpropc' option.
chan0 The output data-set is an average of all the
continuum data in a visibility record.
continuum The output data-set is the continuum data. This
has the same number of channels as the input
data-set.
Key: options
This gives extra processing options. Several options can be given,
each separated by commas. They may be abbreivated to the minimum
needed to avoid ambiguity. Possible options are:
sun With this options, the OFFSET keyword is ignored,
and a shift appropriate for the Sun is determined
This can be useful to eliminate solar interference.
twofit If either the OFFSET parameter is set, or the SUN
option invoked, this options instructs UVLIN to
perform a simultaneous polynomial fit for a emission
both at the phase centre and at the appropriate
offset.
nowindow Normally uvlin processes the different windows separately.
This will usually give the best result when the different
windows are at quite different frequencies. However when
the windows are at similar frequencies, or perhaps even
abutting, then it is better to perform a single fit over
all windows simultaneously. The "nowindow" option causes
uvlin to do a single fit over all windows.
lpropc The line shape is constant throughout the data-set,
and the intensity is proportional to the continuum
strength. This cannot be used together with
the "twofit" option.
relax Normally UVLIN attempts to avoid overfitting the
data by reducing the order of the fit. It does this
if a significant number of channels are flagged
(more than 40% bad) or if it believes there is
inadequate data to perform the fit (less than
5*(order+1) channels). You can overwrite this
conservatism, if you know what you are doing, with
the "relax" option.
The following options can be used to turn off calibration corrections.
The default is to apply any calibration present.
nocal Do not apply the gains table.
nopass Do not apply bandpass corrections. It is unwise
to turn off bandpass correction, as the continuum
estimation process will be confused by a bandpass
which is not flat.
nopol Do not apply polarization corrections.