[daip] FW: Some data reduction challenges

[Eric Greisen]

Dougherty, Sean wrote:
> Eric,
> 
> I reduced some data recently at the DSOC using AIPS. Below are a few comments/questions I have about using the Appendix E of the AIPS manual to calibrate the data. I attempted to not resort to asking any in-house experts, though Rick did suggest the use of POSSM to look at band passes - which was a *very* helpful tip. 
> 
> I sent these to Gustaaf originally, but he advised I send the AIPS-specific ones to you directly.
> 
> All the best, 
> 
> Sean
> 
> -----Original Message-----
> From: Gustaaf Van Moorsel [mailto:gvanmoor at aoc.nrao.edu] 
> Sent: December-20-11 1:50 PM
> To: Dougherty, Sean
> Subject: Re: Some data reduction challenges - as promised
> 
> Hi Sean,
> 
> Thanks again.  Most of your questions are AIPS specific and outside my
> control.  Do you mind sending your points 4 - 7 to Eric Greisen instead?
> 
> 
> Gustaaf
>> 4.      Initial calibration. The AIPS manual suggests using DOCALIB=1 (end of section E.2). What about applying weights (i.e. DOCAL=2)? Should make a note of the need to use DOCAL=1 as opposed to some other value if it is an important distinction at this stage of the process. (Not clear from the manual if there are any appropriate weights associated with the data at this stage of data processing).
>>
>> 5.      Initial calibration. The use of POSSUM to examine the derived band pass's should be mentioned somewhere in section E.4 - some users don't know of this very useful utility (this was something I did glean from a local expert).
>>
>> 6.      It is not clear why ICHANSEL in CALIB "must" be set the same as in BPASS (see bottom of page E-7). The use of "must" makes it sound like the outcome of CALIB will be in jeopardy otherwise - should be at least a brief explanation of why if this is indeed the case.
>>
>> 7.      Applying the Sys Power (SY) table values. There should be some note here to say that this should only be applied by people that know what they are doing - especially true during the commissioning phase of the project when certain capabilities are not completely debugged. For example, if values of Pdiff are missing (for some reason), the resulting calibration may be flawed if the SY tables are applied.
>>
>> Hope these are in some way helpful.
> 

Thanks - I will consider these.

Note DOCAL> 0 and < 100 now calibrates weights if there are meaningful 
weights.  EVLA weights are not calibrated unless a specific keyword 
appears in the header.  That keyword is added by TYAPL and REWAY.

POSSM is mentioned widely in the Appendix but not in conjunction with BP 
tables - good suggestion.

The ICHANSEL "must" has to do with normalizing the same way.  If it is 
very different from one to the other one may end up with things not 
scaled properly.

If you encountered a case with SY values missing we should be told about 
that (Michael and other testers).  The SY table portion requires 
examination of the values - TYSMO, SNEDT etc.  If you have very strong 
sources it will do the wrong thing at present (a non-linearity we do not 
understand).  But the SY values once the SY table is clean do not reuire 
great expertise.  I will look at extending the wording some however.

Thanks

Eric Greisen