AIPS HELP file for SMOOTH in 31DEC13
As of Wed May 22 21:20:53 2013
Type: Adverb (REAL(3))
Use: SMOOTH specifies the type of spectral smoothing to be applied to
a uv database or other spectra. The default is not to apply any
smoothing. The elements of SMOOTH are as follows:
SMOOTH(1) = type of smoothing to apply:
0 => no smoothing
1 => Hanning applied before bandpass calibration
2 => Gaussian applied before bandpass calibration
3 => Boxcar applied before bandpass calibration
4 => Sinc (i.e. sin(x)/x) applied before bandpass cal.
5 => Hanning applied after bandpass calibration
6 => Gaussian applied after bandpass calibration
7 => Boxcar applied after bandpass calibration
8 => Sinc (i.e. sin(x)/x) applied after bandpass cal.
SMOOTH(2) = the "diameter" of the function, i.e. width
between first nulls of Hanning triangle and sinc function,
FWHM of Gaussian, width of Boxcar. Defaults (if < 0.1) are
4, 2, 2 and 3 channels for SMOOTH(1) = 1 - 4.
SMOOTH(3) = the diameter over which the convolving function
has value - in channels. Defaults: 1, 3, 1, 4 times
SMOOTH(2) used when input SMOOTH(3) < net SMOOTH(2).
When applied to uv data, it is important to decide whether the
smoothing should be applied before or after application of the
bandpass calibration. If SMOOTH was used in BPASS (or CPASS),
then the same SMOOTH must be used with all the data (and
SMOOTH(1) must be 1 through 4). If SMOOTH was not used in
BPASS, then SMOOTH(1) must be 5 - 8 for correct results.
Null Value: 0
A null value of SMOOTH(1) means no smoothing is desired;
the useage of null values for other parameters depend on
the application. The usual defaults are listed above.
Frequency smoothing of UV data interacts with data flagging in
ways that are not entirely intuitive. In fact, it is strongly
recommended that any data smoothing to be applied before the
bandpass be done, once and for all (with e.g. SPLAT), before any
channel-dependent flagging is done. Data with RFI frequently
requires a Hanning smooth up front to reduce ringing. Do this
with SPLAT once and for all eraly in your data reduction.
Post bandpass smoothing also interacts with flags and the
computatin of Stokes parameters. A SMOOTH of, e.g. 6,0,0
functions as follows:
1. Smooth each polarization separately ignoring flagged
channels and maintaining the flagging.
2. Combine polarizations to make the Stokes parameters.
This can be particularly odd with Stokes I - which is taken to
be (RR+LL)/2 unless one of RR and LL is flagged, in which case
the other is taken. Thus smooth values of RR and LL are
averaged but isolated channels of only RR or LL also appear.
These isolated channels will have smoothed values appropriate
to RR or to LL but not to the average unless RR and LL are
nearly identical. Thus single channels will stick out from the
otherwise smooth spectrum.
Even with Stokes F (formal I), Q, U, and V, odd results may
arise. Flagging of one of the two observed polarizations for
these parameters leads to flagging of the result at the
particular flagged channel. But, if one of e.g. RL and LR is
not flagged, then those unflagged values will affect the
adjacent channels in the smoothing that preceeds the Stokes
computation. Thus, at a good channel, the results will be
different from the results if all polarizations rather than
just 1 are flagged at the flagged channels.
POSSM has a special option where the spectra are averaged with
no smoothing and then smoothed just before displaying them. In
this way, most of the complications described above are