[daip] fix for channel reversals (fwd)

[Lorant Sjouwerman]

FYI



Regards,

Loránt Sjouwerman - Scientific Services - lsjouwerman at nrao.edu
--------------------------------------------------------------
c/o NRAO Array Operations Center       Phone:  +1-505-835-7332
     P.O. Box 0 (1003 Lopezville Rd)    Switch: +1-505-835-7000
     Socorro NM 87801                   Fax:    +1-505-835-7027

---------- Forwarded message ----------
Date: Wed, 17 Oct 2007 09:20:20 -0400 (EDT)
From: Neal Miller <nmiller at skysrv.pha.jhu.edu>
To: lsjowerm at nrao.edu
Cc: cchandle at nrao.edu
Subject: fix for channel reversals

Hi Lorant.

I've come up with a fix for the channel reversals seen
on occasions in multichannel continuum data. It's tedious,
but uses existing AIPS tasks and solves the problem. It
could be scripted pretty easily, too. I bounced it off of
Ed, and he let me know you were the contact person for
fixing the existing data in the archive that had this
problem.

Here's what I've done:

1) Use UVCOP with channels and timerange set to create
    subsets of the data by channel.
    a) Take the normal channel 1 data: bchan=echan=1,
       no problem doing both IFs and pols together with
       this step, set timerang to select data where
       channel indexing and data are fine
    b) Take the reversed data as in (a) but with bchan=echan=7
2) Conjugate the phases of the reversed data (the uvcop
    output of step 1b) using PHNEG
3) Use DBCON to put together the output files from steps
    1a (the normal data for channel 1) and 2 (the fixed
    data that had been channel 7 but should be channel 1)
4) Use BLOAT to make the output from step 3 look like
    normal multichannel continuum data. In this case, set
    APARM(1)=7, and APARM(2)=1 (set aparm2 to be the
    proper channel number in question).
5) Repeat steps 1-4 for each channel (i.e., channel 2 normal
    data paired with channel 6 flipped data, 3 with 5, 4 with
    4, 5 with 3, 6 with 2, and 7 with 1)
6) DBCON the 7 "bloat" files to create an output single
    multichannel continuum data set.

I had to do an INDXR on the output data set, and if you're
particular you could update the header if necessary (my output
had the frequency axis starting at pixel 1 and its associated
frequency, instead of the usual pixel 4 and frequency corresponding
to the center of IF1).

I've done this for our September 12 AM889 data and it works.
This same problem cropped up for us on Aug 23 and Sep 6, and
I'll handle those data shortly.

Take care,
Neal