Quality Control

Antonia Rowlinson

Transcript

1. Quality Control
   Antonia Rowlinson and Gijs Molenaar
   

   View Slide

2. Making bad images is easy...
   

   View Slide

3. ...but ruin transient searches
   An extreme example with the lightcurve of 3C295
   before and after removal of bad images...
   Time Time
   

   View Slide

4. Bad images cause:
   Inaccurate source
   fluxes
   Large position offsets
   Elliptical restoring
   beams which can
   make source
   association difficult
   Artifacts in images
   

   View Slide

5. Problem: Too many images to
   check by eye...
   Need automated
   quality control
   within the TraP
   that can quickly
   identify bad
   images and skip
   over them in
   the following
   processing steps
   

   View Slide

6. Pointing direction issues
   Problems may occur if pointing direction is near:
   A-team sources - need demixing
   Sun - e.g. solar flares
   Galactic plane - source confusion
   Other things that might be important:
   Time of day - particularly for LBA
   Low elevation - prone to more RFI
   Lack of sources in skymodel - not an issue if using a calibrator
   = Version 1
   First identified and studied by
   A. Stewart & R. Breton (MSSS
   report week 41 and talk
   yesterday)
   

   View Slide

7. Flagging
   Too much data is flagged:
   total - likely to be suffering from severe RFI
   particular baselines - causes poor uv coverage
   particular station types - causes poor uv coverage
   = Version 1
   See work by Y. Cendes
   for more details about
   flagging and transient
   searches
   

   View Slide

8. Calibration
   Primary beam problems
   Amplitude spikes
   Zero amplitude baselines
   Bad weight spectrum values
   Features in images such as stripes due to bad calibration -
   possibly use the stripy image detection script developed for
   MSSS
   = Version 1
   MSSS report Week 42:
   R. Breton, D. Carbone, V. Heeson
   

   View Slide

9. Imaging
   RMS noise in centre of image:
   RMS significantly above/below the theoretical noise
   Restoring beam:
   Beam undersampled or significantly oversampled
   Beam highly elliptical
   = Version 1
   Already
   Im
   plem
   ented
   in
   TraP
   Over sampled
   Under sampled
   Normal
   

   View Slide

10. Flux scale and positions
    Use the brightest catalogued sources to check flux
    scale and position offsets
    Extract brightest sources from image
    Basic source association using the beam FWHM and sky model
    Calculate average position offset - systematic position errors
    Calculate average flux ratio - incorrect flux scale in image
    = Version 1
    MSSS report Week 28:
    A. Rowlinson
    

    View Slide

11. Source extraction
    Source finder problems include:
    Large errors on gaussian fits
    FWHM of fit is smaller than restoring beam
    Spectrum unrealistic
    Not flagging sources in the catalogue but when they
    are extracted, therefore thresholds are all user
    defined
    = Version 1
    

    View Slide

12. Transient Specific
    Relative to bright sources:
    Specific transient is very close to a bright source
    Over density of transients close to bright sources - further
    study required
    At peaks in the PSF for bright sources
    Systematic effects:
    Specific transient is very close to image edge
    Over density of transients near image edge or with specific RA/
    Dec - further study required
    = Version 1
    

    View Slide

13. These tests should mean all our
    images will look like:
    

    View Slide

14. Contribute to discussion online:
    http:/
    /www.lofar.org/operations/doku.php?
    id=tkp:qualitycontrol
    

    View Slide

15. http:/
    /www.lofar.org/operations/doku.php?
    id=tkp:qualitycontrol
    

    View Slide

16. View Slide