Talk:Raw data review (QA1)

From Atomix

If I was new to ADCP data processing for turbulence, I think I would find this list a little confusing in the way it is currently organized. It might be easier list the QC criteria by the issue instead of data type. e.g.

It is recommended that velocity data should be flagged or removed if the following are observed:

  1. Data quality is poor
    • Low correlation values
    • Echo intensity anomalies
    • Low percent good values
    • Data return rate varies
  2. Unrealistic velocity values
    • Evidence of phase wrapping
    • Velocities outside nominal measurement range
    • Velocities outside expected distribution
  3. Significant instrument motion and orientation
    • High variability in pitch, heading and roll
    • Orientation deviates from expected values
  4. There is wave or periodic motion
    • Periodic motion at wave frequencies observed in velocity data
  5. Velocity shear is too large
    • earth velocities indicate significant horizontal shear
  6. Stationary assumption may be violated
    • variance of velocity bursts shows spatial or temporal trends [ALTHOUGH HARD TO DETERMINE WHAT IS REAL VARIABILITY]
  7. Stratification is too large
    • Temperature and salinity (if available) indicate local stratification

Brian scannell (talk) 15:58, 12 November 2021 (CET) Agree that this is better. I guess users would like to be told the “right" criteria thresholds and “expected values” to use, but is best practice sufficiently well defined to make pronouncements on this? Personally, I wouldn’t feel comfortable doing so. Probably best not to refer to removing data - the netcdf approach just flags it as being bad data - is this the place to describe the QC criteria? Brian

Jmmcmillan (talk) 21:31, 12 November 2021 (CET) I don't think we need to give the thresholds because it is instrument dependent and configuration dependent. We are just telling them what things might possibly contaminate the data. We can link to other pages if we want to include more description and/or references. Since we are in agreement of the format, I will drop the "remove" and put this list directly on the page.