© Crown copyright Met Office The EN QC system for temperature and salinity profiles Simon Good.

© Crown copyright Met Office

The EN QC system for temperature and salinity profilesSimon Good


Quality control tests

Manual exclusions

Track check

Profile check (spikes etc.)

Thinning (informational)

Stability check

Background checks

Buddy checkMulti level check

Argo delayed mode flags

Argo grey list

Argo altimetry quality control

External quality information

Automatic quality checks

Bathymetry check

Measurement depths check

Waterfall check

Near surface and deep BTs , zero lat and lon

Range check


Rejects based on external information• Bathythermograph levels shallower than 4 m or

deeper than 950 m are rejected.• Done because of start up transients in XBT

data and because Ingleby and Huddleston (2007) found quality problems with deep XBT data.

• Profiles with latitude and longitude exactly 0° are rejected.

• Based on feedback from a user.• [Not done for the IQuOD data].


Range check

• Very broad check to ensure that temperature and salinity values are not ridiculous.

• Temperature has to be in the range -4 – 40°C.

• [Salinity in the range 0 – 45.]


Bathymetry check

• Compares location of profile to the ETOPO1 global relief dataset.

• Tries to be conservative – a profile is only rejected if all surrounding bathymetry points are land.


Profile check

• Detects constant values (90% of levels have the same value, profile extends at least 100 m).

• Detection of spikes and steps in the data.• First checks for large spikes where a value disagrees

with levels above and below by a depth varying tolerance (5°C at surface – 1.5°C below 600 m) to reject just the spike.

• There is a further check for smaller spikes that also uses the vertical gradient.

• Steps exceeding the tolerance are rejected but with some conditions e.g. no rejection if the step may be due to the thermocline.


Track check

• Finds instances where platform speed exceedsa threshold (15 m/s for ships and 2 m/s for buoys) or if greater than a (smaller) threshold and with a sudden angle change to the track.

• Detailed set of tests to determine which point to reject.

• If more than half the track is rejected the whole set is rejected.


Stability check

• For profiles with temperature and salinity data.

• Potential density inversions < 0.03 kg/m3 are rejected.

• Spikes in potential density are detected.

• If number of inversions is > ¼ of the number of levels in the profile, the whole profile is rejected.


Measurement depths check

• Checks that depths are reasonable (e.g. not above the surface) and that they increase throughout the profile.

• Works by comparing each level to the rest of the profile to determine which is most suspect.

• Ensures that the depthsof all non-rejected levelsincrease monotonicallythroughout the profile.


Background checks

• Profiles are compared to a persistence forecast of the ocean state.

• This is done twice – once on observed levels and once on standard levels (with the check more strict).

• A probability of gross error is calculated using Bayesianprobability theory.

• Results of the background check on standard levels are combined with the buddy check.

• If a background value is not available (e.g. because of error in profile location or incomplete background grid, a level is rejected.)


Buddy check

• Profiles within 400 km but with different identifiers are compared to each other.

• Increases/decreases the probability of gross error calculated during the background check.

• Levels with probability of gross error ≥ 0.5 are rejected.


Multi level check

• If more than 50% of the levels have been rejected, all are rejected.

• Levels failed by the background/buddy check can be ‘unrejected’ if they agree with levels above/below (done for about a tenth of levels rejected by the background/buddy checks).


Statistics of checks rejecting whole profiles in EN4

• 6.6% had the whole profile rejected.• Relative proportions of checks applied to IQuOD data

are:


Statistics of checks rejecting levels in EN4

• 3.2% of levels were rejected.• Relative proportions of checks applied to IQuOD data

are:


More information

• Full detail of many of the quality control checks are given in Ingleby and Huddleston (2007), Journal of Marine Systems, 65.

• Summaries and details of some of the tests are available in Good et al. (2013), JGR-Oceans, 118.


Questions and answers

© Crown copyright Met Office The EN QC system for temperature and salinity profiles Simon Good.

Documents