[daip] [!8171]: AIPS - REWAY

[Lynn Matthews]

Lynn Matthews updated #8171
---------------------------

REWAY
-----

           Ticket ID: 8171
                 URL: https://help.nrao.edu/staff/index.php?/Tickets/Ticket/View/8171
           Full Name: Lynn Matthews
               Email: lmatthew at haystack.mit.edu
             Creator: User
          Department: AIPS Data Reduction
       Staff (Owner): -- Unassigned --
                Type: Issue
              Status: Open
            Priority: Default
                 SLA: NRAO E2E
      Template Group: Default
             Created: 16 March 2016 07:34 PM
             Updated: 18 March 2016 04:29 PM
      Resolution Due: 24 March 2016 07:34 PM (6d 3h 5m)



I have done a few more tests, including data from two other projects, to be sure they are not affected by something pathological.

Of course you are right about REWAY not changing the amplitudes per se, and that is clearly verified with UVPRT.

Also I find that the RMS in the image goes down when I image the data after REWAY, versus taking all the weights to be equal. 

What seems potentially more dangerous is if one uses REWAY in a case there are some bad data present that one has missed. 

REWAY does downweight the bad data. For example, in one test case, baselines to a bad antenna have weights of >100 while other baselines are ~2-3. But while the resulting images might not be very different to the eye, they can have very different quantitative properties. I find that systematically, the apparent RMS of an image from REWAYed data that contains bad samples is always *lower* than an image of REWAYed data without the bad points.  Plus the flux densities of sources seem off.

Example:

REWAY, GOOD FLAGGING
Source flux density 1.6652E-04 +/- 8.12E-06 JY, image RMS=4.3717E-06

no REWAY, GOOD FLAGGING
Source flux density 1.5958E-04 +/- 8.29E-06 JY,  RMS=4.5027E-06

REWAY, bad flagging
Source flux density 1.3046E-04 +/- 1.00E-05 JY, RMS=3.3536E-06

no REWAY, bad flagging
Source flux density 1.5603E-04 +/- 1.10E-05, RMS=5.7796E-06 

I tried different APARMs without seeing much effect, but I will experiment some more.

Of course one should always flag carefully, but the results of these tests surprised me. 

Lynn

------------------------------------------------------
Staff CP:  https://help.nrao.edu/staff