search

ratt-ru / meqtrees

A library for implementing radio astronomical Measurement Equations
http://meqtrees.net
10 stars 2 forks source link

Solutions not applied to CORRECTED_DATA column? #697

Closed gijzelaerr closed 10 years ago

gijzelaerr commented 10 years ago
at 2009-04-16 19:27:02 George Heald reported:

Solutions not applied to CORRECTED_DATA column?

gijzelaerr commented 10 years ago

Original comment thread migrated from bugzilla

at 2009-04-16 19:27:02 George Heald replied:

When performing a bandpass solution on a simple calibration dataset from WSRT (single 10 MHz subband @ 1420 MHz, 1024 channels, 3C295, ~15x1min samples, 2 pol.) using calico-wsrt.py, I get a very nice solution for B Jones, and a CORRECTED_DATA column seems to be have been created. However, when imaging (with lwimager), the bandpass correction does not seem to have been applied. In fact, producing an image cube from the DATA column and one from the CORRECTED_DATA column gives identical results. Subtracting the two cubes within kvis yields a "Min is not less than max: bad data" error.

I tried using both the "DiagAmplPhase" and "FullRealImag" modules, but neither worked. I am using the most recent svn version of the Cattery.

at 2009-04-20 10:15:10 George Heald replied:

The main problem can be filed under "user error". I flagged the even channels because I am applying Hanning smoothing. Then, I calibrated the even channels. Thus, when I imaged the unflagged (odd) channels, they were not calibrated.

Along the way I realized two things that would be useful: 1) A Hanning smoothing operation which takes an input MS and writes out a new MS, with the Hanning smoothing applied and the inbetween-channels removed. I can look into making such a script. 2) The calico flagger seems to only flag one polarization (in a two-pol dataset) when asked to flag a bunch of channels. I flagged channels 0 to 1022 with stepsize 2, and it did so but only in YY, not in XX. This should be fixed.

I have also noticed that, after fixing the user error, the calibrated amplitudes are close to the right values but still off by (roughly) a factor of four. Possibly additional user error, I'm looking into it.