[daip] TYAPL, and old VLA data

[Eric Greisen]

rperley at nrao.edu wrote:
> Eric:
> 
> In talking with some people here in Garching, a question arose as to the
> weights in AIPS.
> 
> 1) Units.  I told them the weights, when properly computed, are equal to
> 1/(Jy)^2.  Correct?
> 
> 2) TYAPL:  I know what TYAPL does for EVLA data.  But what about old VLA
> data.  If one runs TYAPL on old VLA data, what happens?  I'm confused
> about this, since the on-line system computed the weights directly using
> the values for Tsys provided by the system.  Since there is no new
> information available, what can TYAPL do that's different?
> 
> Their interest is in getting correct weights for some old VLA data
> (provided by Tracy Clarke).  I said that most likely the right thing to do
> is to run REWAY, and compute the dispersions, hence weights, directly from
> the data.  However -- is REWAY recommended for old VLA data? With its 50
> MHz BW and 10 second integration times, I worry that the dispersion in the
> visibilities becomes dominated by gain variations, rather than noise.
> 
> Rick

weights should be after calibration is applied 1/(Jy)^2 yes

TYAPL - the old VLA on-line applied gains to the data but did not 
provide any weights.  FILLM did the weights - initially just integration 
time but eventually also applying the nominal sensitivity numbers 
recorded in one of the columns of the TY table.  TYAPL can remove the 
values of one TY table from the weights and apply the values of a 2nd 
table to the weights (assuming the TY needed editing like SY often 
does).  If the data are from FILLM before weights (or FILLM after with 
the adverbs set to avoid real weights - yes people insisted on and used 
that) then there is no TY to remove but the TY can be applied.  This 
should be done before SPLIT so that the CL values can be applied to the 
weights.

REWAY tends to be less useful in VLA data since the number of channels 
tends to be smaller.

Eric