The raw observations were quality-controlled, accounting for known sensor responses and deviations from factory calibrations. Detailed descriptions of the data processing, quality control metrics, and calibration methods are provided by Garcia et al. (2023; https://doi.org/10.5281/zenodo.10247019). Quality control steps included a pitch angle test (to detect density inversions), global and regional range tests, stuck value test, biofouling evidence test, and visual inspection. If a test passed, the data were assigned a QC value of 1 (‘good data’). If a test failed (e.g. out of range or density inversion tests), data were flagged, either as ‘questionable data’ (QC = 3) or as ‘bad data’ (QC = 4). All quality control flags and adjustments (i.e. calibration and correction parameters) are preserved in each mission’s file. Code for all processing steps is available on GitHub: https://github.com/cathygarcia/SeagliderDataprocessing.
Data processing steps for each variable are summarized below:
Temperature, Conductivity, Salinity, and Potential Density Anomaly
- - Both temperature and conductivity profiles were lag corrected.
- - Practical salinity was calculated using the Gibbs Seawater Toolbox (gsw_SP_from_C.m), and then converted to absolute salinity (gsw_SA_from_SP.m).
- - Potential density anomaly was calculated with respect to a reference water pressure of 0 db using the Gibbs Seawater Toolbox (gsw_sigma0.m).
Dissolved oxygen concentrations
- - Raw optode phase values proceeded through a series of corrections to account for the effects of temperature, salinity, pressure, and time response in addition to sensor drift (Bittig et al., 2018, Barone et al., 2019).
- - Optode phase values were converted to dissolved oxygen concentrations, and re-calibrated using discrete Winkler measurements.
Chlorophyll a
- - Factory-calibrated chlorophyll a observations were re-calibrated using discrete measurements of either HPLC chlorophyll a (16 missions) or fluorometric chlorophyll a (2 missions).
- - Daytime chlorophyll a values are not quench corrected, and may be lower than actual values. It is recommended to use nighttime profiles near the surface.
- - Additionally, a spike flag is included based on published protocol (Briggs et al., 2011).
Backscattering coefficient due to particles (bbp)
- - Factory-calibrated bbp values could have a large offset, that was not expected based on natural variability.
- - A mission-specific deep dark correction (1st percentile of bbp at 190-200 m) was subtracted for each bbp dataset. Both the uncorrected and corrected data are available.
- - Additionally, a spike flag is included based on published protocol (Briggs et al., 2011).