[daip] EDITR "features"

[James M Anderson]

Hello.  I am having some issues with EDITR.  I frequently use EDITR to
flag my VLBA data.  My sources are generally rather weak (a few mJy),
so I normally run EDITR with CROWDED = 1 to display all IFs at once.
As soon as EDITR starts, I turn on flagging of all IFs and all
polarizations, as most of the problems I see show up in all IFs.
Normally, everything works ok, but occasionally EDITR will not allow
me to flag data.  I can see datapoints in the window which need to be
flagged, set a time range over which to flag, and EDITR will do
nothing.

If I inspect the data more closely, this problem appears to happen
when IF 1 is absent.  The other IFs are still there, but EDITR will not
flag them if IF 1 is missing.  If I manually ask for IF 2 (for
example), then all of the other IFs disappear and I can flag that IF.
But then I have to restart thing in order to get the CROWDED condition
back.  I think that it would be nice, if I have checked the all IF
flagging option, if AIPS would actually check if there are other IFs
present before deciding to not flag.


My second problem is perhaps less easy to fix.  Because my VLBA
targets are very weak, I frequently find it useful to smooth out the
data to reduce the apparent noise level so that I can find more subtle
problems.  Suppose I have data with 2 second integration times and
want to smooth over 10 second intervals.  The easiest way to do this
appears to be using SOLINT = 1/6.  This produces data in the EDITR
screen which for the most part are less scattered and closer to zero
amplitude (I have mJy sources).  However, there are occasional spikes
which show up.  For instance, if the 2 s data are scattered between 50
and 300 milli units, most of the 10 s averaged data will be around 75
milli units or so, but occasionally there are time locations with the
IFs still scattered between 50 and 300.  

I suspect that what is happening is that SOLINT is dividing the data
up into discrete 10 s chunks with SOLINT = 1/6.  Most of the time,
this will be fine as there will be 5    2 s integrations to average
over.  However, because of different antenna slew times some antennas
may only have 1 point in the first 10 s chunk, or because the scan
length was 42 s, there is a single 2 s integration left over for all
telescopes.  I am guessing that the SOLINT averaging takes this
single datapoint, and because it is not averaged with other noisy
points, it remains more noisy that chunks with 5 datapoints.  This is
probably not so apparent with high S/N data from bright targets, but
for my weak sources with short scan times for phase referencing, it
makes it difficult to spot the actually corrupted data.  I can always
turn up SOLINT large enough to cover an entire scan length, but
sometimes this is not what I want either.

Is there a simple way to tell EDITR to smooth the data, but make sure
that all points get averaged with their neighbors and not left
isolated?  Perhaps an option to boxcar smooth the data over some time
width would be useful.  I suspect that there must be a way to get AIPS
to do this to the UV data, but it would be nice for EDITR to do this
for me so that I didn't have to produce a new UV file, flag that, and
copy the flag table over.

Thanks for your time.


James M Anderson

-- 
//****************************************************************************
James M Anderson                    New Mexico Tech/NRAO
Physics Graduate Student            P.O. Box 0
e-mail: janderso at aoc.nrao.edu       1003 Lopezville Rd
Telephone: 505-835-7094             Socorro NM 87801