Data Quality Control
All marine mooring data are provided with a quality descriptor, and data undergo several checks, both preliminary real-time tests and further review by data managers. The 3-digit qualifier code that accompanies each data point describes both the quality and the level of quality control (QC) that has been completed.
The first digit denotes the status and overall data quality:
(1 . . ) = Finalized – Good – passed secondary review by data manager. This step is done on a semi-annual basis.
(2 . . ) = Provisional – Good – passed initial automatic quality control check.
(3 . . ) = Questionable/suspect data. Use data with much caution.
(4 . . ) = Poor/bad data. Avoid using this data.
(9 . . ) = Missing data.
For example, during data download, if you select “Finalized and provisional data only,” then you’ll receive only data that are qualified by (1..) or (2..). Any suspect or poor data will be blank.
The remaining digits provide more information about which automatic QC flags were assigned and whether the QC flag was applied after final review by a data manager.
The second digit shows if the data passed or failed any automatic tests:
(. 0 . ) = Quality is unknown for a specific parameter and requires calibration checks, for example, for pH measured internally by the SeaFET. Avoid using this data without further review.
(. 1 . ) = Data passed automatic quality control checks.
(. 2 . ) = Sensor failed end check validation. The end check is a delayed quality control step completed after sensors are recovered from the water, and helps ensure that the sensors remained within acceptable calibration limits throughout the deployment period. YSI sondes are compared to a known standard in the lab, and SBE HydroCAT-EPs (HCEP) are compared to the profiling CTD in well-mixed deep seawater. See the end check range values in the table.
(. 3 . ) = Value is outside of climatological range by site, and data are flagged in real-time as suspect and require further review to verify. Contact a data manager for details on ranges.
(. 4 . ) = Value is outside of measurement range for sensor, and are flagged as bad data in real-time. See the ranges in the table.
(. 9 . ) = Missing data; instrument is not deployed or unable to communicate.
The third digit shows whether the data have been reviewed and the level of quality determined by a data manager:
(. . 0) = No secondary data review has been completed. This is the default value.
(. . 1) = Good data, no adjustments made.
(. . 2) = Probably good data, no adjustments made. For example, if the sensor failed the end check validation at the end of the deployment, but the previously collected data looks good, then it is qualified at this level.
(. . 3) = Questionable/suspect data. For example, if the sensor shows signs of drift or biofouling, and data cannot be corrected or verified.
(. . 4) = Poor/bad data. For example, the sensor malfunctioned or was out of the water.
(. . 5) = Value corrected/changed manually. If calibration samples are available during the deployment, then some sensor data may be corrected for drift, such as the SeaFET pH data.
(. . 8) = Interpolated value, or used an interpolated input value in the calculation. For example, salinity data are used in the SeaFET external pH calculation. If salinity data are bad or missing, then an interpolated salinity value was used to calculate pH and flagged with this qualifier.
(. . 9) = Missing data. Assigned automatically.
For the list of sensor data ranges and end check validation ranges, see the table.