Page tree
Skip to end of metadata
Go to start of metadata

Date

Attendees

Goals

  • Receive pilot survey data processing feedback and prioritise development required due to issues arising

Discussion items

TimeItemWhoNotes
5 minPawsey and pipelinesAidan
  • 0.24.4 of askapsoft/askappipeline now available and default on Galaxy
    • Calibration table null entry fix
    • Stokes V MFS image creation (without cubes)
15 minEMU pilot survey feedback and processingAndrew
  • Observations for EMU pilot complete, see table EMU
  • Processing of first block complete, CASDA upload failed due to island catalogue errors
    • James will exclude the island catalogue and re-run the upload today
    • Uploaded data will be visible to EMU team with unreleased status
  • Processing of remaining blocks will proceed once EMU confirm that all is well
    • Initial feedback from Josh is positive
    • Still not hitting the CLEAN RMS threshold, ending on minor cycle iteration limit instead
    • Looks like deeper CLEANing is possible without diverging
    • Smallest scale is smaller than the beam, might not be used
15 minWALLABY Eridanus processing and feedback Karen
  •  Full-field half-band processing complete, but several beams have zebra stripes
    • Caused by occasional channels where data are mostly, but not fully, flagged?
    • askapsoft flagger should flag everything if more than some threshold
    • Continuum subtraction also impacted by single-channel outliers?
  • Data quality difficult to asses due to size of cubes
    • 40 minutes to open in casaviewer
    • Carta visualisation package being tested along with HDF5 conversion
  • Discussion of optimal weighting robustness, tests of single beams with alternative options
10 minGASKAP test observation processingChenoa
  • SMC data has been bandpass-calibrated and visibilities delivered to ANU (first scheduling block)
    • Self-cal failing, but perhaps not needed
  • Second SMC field being transferred
  • Same will be done for LMC data next week, disk space allowing
10minRACS re-processingDave
  • Using scripts designed to keep Galaxy queue filled automatically
    • Estimates number of jobs needed by looking at the number of fields in a scheduling block
    • Queue limited to 10,000 jobs, currently using most of these most of the time
      • Might be good to leave a gap for other users
      • Jobs tend to finish quickly
      • Some blocks need more than 10,000 jobs! Need to split calibration and imaging
      • Mosaicking still the longest step, needs about 40 minutes
    • Dealing with occasional failures is very difficult with this many job numbers
      • 200 / 15,000 failures
  • VAST likely to use a RACS-like pilot survey approach (over several epochs)
  • Roughly half of processing pass two done (includes polar cap)
  • Emil working on re-mosaicking to remove edge effects and HiPS conversion 
  • Stokes V MFS images will be made for processing pass two

Action items

  •