Quality control coding

Example: DEFINE A NAME

copied from code--- to be tidied up: Provide a Q flag for every dissipation estimate Q =

      0, if all checks pass
      1, if FOM > FOM_limit
      2, if despike_fraction > despike_fraction_limit
      4, if |log(e_max)-log(e_min)|> 5 sigma

Combinations (sums) are:

      3, FOM and despike fraction fail
      5, FOM and sigma fail
      7, all fail

This allows one to identify the unique ....

Example: Ocean Sites

One option is to follow Ocean Sites for quailty control (QC) coding. The flagging scheme is mostly compatible with the primary level flagging recommended by Intergovernmental Oceanographic Commission of UNESCO (2013).

Flag	Meaning	Comment
0	unknown	No QC was performed.
1	good data	All QC tests passed.
2	probably good data	Data have failed one or more QC tests but detailed examination after processing (e.g. by visual examination) suggests data is good.
3	potentially correctable bad data	These data are not to be used without scientific correction or re-calibration (e.g. uncertain shear sensor sensitivity).
4	bad data	Data have failed one or more tests.
5	-	Not used
6	-	Not used
7	nominal value	Data were not observed but reported (e.g. instrument target depth.).
8	interpolated value	Missing data may be interpolated from neighboring data in space or time.
9	missing value	This is a fill value

Climate and Forecast Metadata Convention (CF) requires that QC flags carry attributes. In netCDF (Network Common Data Form) data files, the following information for quality control flagging should be provided for each data variable <PARAM>.

<PARAM>_QC
<PARAM>_QC:long_name = “quality flag of <PARAM>”;
<PARAM>_QC:conventions = “OceanSITES QC Flags”;
<PARAM>_QC:flag_values = 0, 1, 2, 3, 4, 7, 8, 9;
<PARAM>_QC:flag_meanings = “0:unknown 1:good_data 2:probably_good_data 3:potentially_correctable_bad_data 4:bad_data 7:nominal_value 8:interpolated_value 9:missing_value”