[daip] Reference frequencies

[Craig Walker]

We have uncovered a problem with reference frequencies in AIPS data
sets.  We are looking at a geodesy data set with 4 IFs (AIPS
terminology) near 2.2 GHz and 4 IFs near 8.4 GHz.  The observation was
scheduled and processed with the 8.4 GHz channels first, followed by
the 2.2 GHz channels.  AIPS likes the order inverted and FITLD does
indeed invert the order.  On the correlator, the reference frequency
was near 8.4 GHz and the frequency offsets in the FQ table are near
zero for the 8.4 GHz IFs and near -6 GHz for the 2.2 GHz IFs.  In aips,
the reference frequency reported by IMHEAD is near 2.2 GHz and the
offsets in the FQ table are near 0 for the 2.2 GHz IFs and near +6 GHz
for the 8.4 GHz IFs.  So far, so good.  But the reference frequency is
stored in several other tables, including the AN, AT, IM, and MC
tables, and perhaps more.  All of those listed show the reference
frequency to be near 8.4 GHz.  Thus if you want the frequency of a
particular bit of data, you get it wrong by around 6 GHz if you use the
AN table as the source for the reference frequency rather than using
the catalog header.  I forget which is the approved source of the
reference frequency but I think there may be some programs that get it
from the tables rather than the catalog.  If so, they get it very
wrong.

When FITLD reorders the frequencies, it should straighten this out.
One option, perhaps the safest in some ways, would be to keep the
correlator's version and put up with a non-zero, and perhaps large,
offset for IF 1 in the FQ table.  But, especially when users split
their data set by frequency, looking just in the header could cause
considerable confusion.  The other option would be to fix the tables.
But one must be careful that there aren't other types of information in
the tables, especailly the correlator tables, that should also change.

C
raig