0000216: Wrong value filled into the Measurement Set Feed table parameter SPECTRAL_WINDOW_ID

  

Viewing Issue Simple Details Jump to Notes ] Issue History ] Print ]
ID Category Severity Reproducibility Date Submitted Last Update
0000216 [TMS] minor always 18-01-10 10:05 12-04-10 07:12
Reporter Arno Schoenmakers View Status public  
Assigned To Arno Schoenmakers
Priority normal Resolution open  
Status closed  
Summary 0000216: Wrong value filled into the Measurement Set Feed table parameter SPECTRAL_WINDOW_ID
Description The parameter SPECTRAL_WINDOW_ID in the WSRT MSes contains the value '0', whereas this should be some other value (-1 is the defined default, for instance).

See 'Additional Information' section.
Additional Information Hi,

these people are right; the SPECTRAL_WINDOW_ID in the FEED table is filled with the value '0' by the WSRT, but this should be set to -1 as is stated in the MS2 definition. However, in the case of frequency mosaicing with one band being the 13-cm band, this will be more complicated (as we then use a mix of XY and RL receivers). I do not readily see how we can do that in that case.

  Cheers, Arno

>>> On Woensdag 13 Januari 2010 at 22:26, <jozsa@astron.nl> wrote:
> Hi Arno and Hans,
>
> I have got an email from a WSRT user, who forwards information from Rob Reid
> concerning a data reduction of WSRT MS with casapy. Apparently, the WSRT MSs
> do not have the correct format. Is this true, and if it is true, is there a
> simple solution to it?
>
> Cheers,
> Josh
>
> -------- Original Message --------
> Subject: Fwd: [Fwd: [JIRA] Resolved: (CAS-1875) split on Westerbork data
> doesn not work if spw 0 is not included]
> Date: Wed, 13 Jan 2010 22:21:10 +0100
> From: Reinout van Weeren <rvweeren@strw.leidenuniv.nl> (
> mailto:rvweeren@strw.leidenuniv.nl )
> To: Gyula Jozsa <jozsa@astron.nl> ( mailto:jozsa@astron.nl )
> References: <4B4E00FF.3000304@nrao.edu> ( mailto:4B4E00FF.3000304@nrao.edu )
>
>
> Hi Josh
>
> Apparently, there is a bug in the WSRT MS being written (see the forwarded
> text). Basically in the Feed table under the MS, SPECTRAL_WINDOW_ID should
> be set to -1 instead of 0. The incorrect value of SPECTRAL_WINDOW_ID causes a
> casapy crash (due to the uncorrected MS format from the WSRT). This may
> happen for example if the first spectral window is not included in the data
> (for example if you only want to work with a spectral window (or IF) larger
> than 0.
>
> Can you forward this to the appropriate person(s) ?
>
> Thanks, Reinout
>
>
>
> .
>
>
>
>
>
> From: Ed Fomalont <efomalon@nrao.edu>
> Date: January 13, 2010 6:21:03 PM GMT+01:00
> To: Reinout van Weeren <rvweeren@strw.leidenuniv.nl>
> Subject: [Fwd: [JIRA] Resolved: (CAS-1875) split on Westerbork data doesn
> not work if spw 0 is not included]
>
> Hi Reinout,
>
> Rob found the problem. In the Feed table under the MS, change
> SPECTRAL_WINDOW_ID from 0 to -1. You can do this in the browser. Perhaps,
> Rob's suggestion about 'ignoring' the value is a reasonable solution,
> although the WSRT code making the ms should put in -1 instead of 0.
>
> Thanks, Ed
>
>
>
> Robert Reid resolved CAS-1875.
> ------------------------------
>
> Fix Version/s: Release 3.0 (Public Release)
> Resolution: Won't Fix
>
> split is working properly. The problem is that the input MS says that the
> feeds are only valid for spw 0. The Measurement Set definition for the FEED
> table says to use -1 for SPECTRAL_WINDOW_ID if you want the feed to be valid
> for all spws:
>
> http://casa.nrao.edu/Memos/229.html#SECTION00066000000000000000 [^]
>
> The error message was certainly misleading, and I will try to fix that, and
> look into whether the SPECTRAL_WINDOW_ID column of FEED can be ignored if all
> of its values are the same.
>
> >
Attached Files

- Relationships

- Notes
(0000230)
Arno Schoenmakers
19-01-10 11:15

Hans has changed a MS. The MS could be converted to j2000 and uvfits without errors from the converters. Josh will look at the UVFits file to check if it is still ok.

I asked Josh to find out what the CASA people did exactly, so we can test if their problem has been solved:

i Josh,

Hans changed the FEED table in one of the MSes. I converted it to UVF now, and the UVfits converter does not seem to have a problem with this, but perhaps you can have a closer look at the UVFits file to make sure the administration is still ok. If this is OK, we can most likely change the MS writer without impacting export conversions.

The file is on wop24, in /wop24_1/schoenma/feedtable/10904201_S0_T0.UVF

Perhaps you can also ask these guys how we can test if we did the right thing (which commands did they use to raise the error).

Cheers,

   Arno


 
(0000250)
Arno Schoenmakers
09-02-10 08:23

Hi,

I verified that the correction in the feed table is sufficient to prevent the error reported by these guys to occur. So, now we:
- Need to build this into MSFiller and put it into production; I will look into that with HvSG
- Possibly correct the archive MSes with a simple script.

Cheers, Arno

 
(0000275)
Arno Schoenmakers
25-03-10 10:33

Hi,

we (finally...) put into production the requested Feed table change in the WSRT Measurement Set. Now, the SPECTRAL_WINDOW_ID is set to -1, always. This is effective as of Seq.nr. 11001869.

There is a Wiki page describing this change: http://www.astron.nl/wsrt/wiki/doku.php?id=wsrtinfo:msdefinitionmar2010 [^]

We still need to adapt all MSes in the archive.

  Arno
 

- Issue History
Date Modified Username Field Change
18-01-10 10:05 Arno Schoenmakers New Issue
19-01-10 11:15 Arno Schoenmakers Note Added: 0000230
19-01-10 11:15 Arno Schoenmakers Status new => testing
21-01-10 10:05 Arno Schoenmakers Assigned To  => Arno Schoenmakers
09-02-10 08:23 Arno Schoenmakers Note Added: 0000250
25-03-10 10:33 Arno Schoenmakers Note Added: 0000275
12-04-10 07:12 Arno Schoenmakers Status testing => closed