Date: Thu, 12 Oct 2006 12:46:25 -0400 (EDT)
From: Crystal Brogan <cbrogan@cv.nrao.edu>
Subject: My Top Reduction Whamys

Rather than produce a long list at the moment I thought it might be useful
to start with the top three reasons that I have been unwilling to reduce
science data in CASA. All are essential for eVLA and ALMA:

(1) Ability to properly calibrate data weights based on Tsys. Central to
this feature is elevation dependent opacity correction. To get a wide range
of users involved, CASA must recognize Tsys information from: VLA/eVLA, SMA,
CARMA, ALMA, Plateau de Bure, Nobeyama, ATCA.

(2) Ability to flag channel averaged data and apply to spectral line data
set. Note that I require two senses of this feature. (a) In the calibration
stage, it would be sufficient to average the line data itself (on the fly),
flag and then have it apply to current dataset. (b) In the
imaging/self-calibration stage it is ALSO essential that you can flag on a
seperate single channel dataset and then apply those flags to a seperate
line dataset. For example, you will often want to create a continuum only uv
dataset from your spectral line data (NOTE: this continuum will likely need
to be made from many 10s of spectral windows, carefully chosen to exclude
strong lines which does not lend itself to on the fly averaging or imaging -
uv continuum subtraction already exists though I am not positive that it
outputs both the continuum AND continuum subtracted line it may currently
only do the latter), flag the continuum and apply to continuum-subtracted
line data.

(3) It must be possible to self-calibrate on one dataset and apply those
calibrations to another seperate dataset (i.e. along the lines of 2b above).

It is *possible* that some of this functionality has been added - certainly
they have been requested many times and are in both eVLA and ALMA
requirements documents. Note, that I believe that (1) at least has been made
to work in an adhoc way through a script only an expert can write. This does
not meet my personal definition of "working" - most especially for
functionality that is essential for proper data reduction.

I did not send this to Joe because it isn't clear to me in the under the
current organization that I should. Feel free to send it to him.

-Crystal