[daip] APCAL opacity difficulties.

[Craig Walker]

I have run APCAL on a significant number of 43 GHz VLBA observations 
lately and the opacity fitting has worked fine, with remarkably few if 
any glitches.

I just tried to run it on a 43 GHz global observation from 1999 out of 
the archive.  I am not trying to do a full calibration of this 
observation - just a casual, VLBA only calibration to do a sanity check 
on some amplitude issues.  The calibration tables do not have data for 
the non-VLBA stations so I just want to ignore them.

First I copied the file, specifying only the VLBA antennas in ANTENNAS 
and BASELINE.  That presumably did what I wanted, except that the AN 
table still has all of the antennas and the antenna numbering was not 
changed.  Normally this shouldn't be a problem.

Then I try to run APCAL.  I have tried many variations and have not yet 
succeeded in getting a usable calibration of any antennas.  If I just 
run with defaults, it dies because of missing GC information on the 
first antenna - one of the global ones that has not been copied.

I then try being selective with lots of combinations of setting ANTENNAS 
and DOFIT.  Some combinations appear to be doing fits, but they get Trec 
of zero and an opacity of 0.201.  The plotted fits are really bad. 
These happen, for example, if I select the antennas, set trecvr, set 
opcode to 'grid' and ask for fits on all but the first of the listed 
antennas.  The data are good and should be easy to fit.

If I try opcode 'opcr', it dies claiming whatever station it is doing 
first has and unreasonable Trec and gives a value, which is actually 
very reasonable.  Looking into the code, I think that means that the 
Trec is higher than the measured Tsys's, which it is not.

So I tried not using ANTENNAS.  If I asked for a fit on the VLBA 
antennas, I got the Trec=0, opacity=0.201 result.  If I try to just 
force Trec and Tau0 by setting DOFIT=0 for the VLBA antennas (-1 for the 
global antennas), it sort of looks like it is running.  But the plotted
opacity values are very high, approaching 1 in the middle of the day. 
In fact, they peak just when the Tsys is lowest.

Something is very wrong.  I have been completely defeated in making any 
sort of opacity correction to this data set.

I note that the elements of DOFIT apply to the antennas as listed in 
ANTENNAS rather than as listed in the antenna table.  Other arrays, like 
TRECVR and TAU0 are not specified.  I've assumed that they correspond to 
the antenna table numbers but I could be wrong.  I'm wondering if there
is some internal bookkeeping confusion about which antenna's data is 
being examined at various steps in the process.  For example, the 
unreasonable Trec complaint might have a Tsysmax value from the wrong 
antenna (I did not go that deep in looking at the code).

The data involved are on noatak disk 1, user 2438.
AIPS 1:    4 2438 GJ009-VLBA  .MULTI .    1 UV 12-SEP-2006 21:46:46

I have made a TASAV copy that might be easier to deal with (assuming you 
don't really need the actual data).  That is
AIPS 1:    5 2438 GJ009-VLBA  .TASAV .    1 UV 12-SEP-2006 21:47:13

Cheers,

Craig

---------------------------------------------------------------------
     R. Craig Walker            Array Operations Center
     cwalker at nrao.edu           National Radio Astronomy Observatory
     Phone  505 835 7247        P. O. Box O
     Fax    505 835 7027        Socorro NM 87801   USA
---------------------------------------------------------------------