Delayed time data are processed through a quality control procedure before they can be used for research purposes. We used for that a dedicated software called TSGQC that we developed under Matlab©, which is freely available on the internet (http://www.ird.fr/us191/spip.php?article63).
Details about delayed time processing can be found in: Alory G. et al., 2015. The French contribution to the Voluntary Observing Ships network of Sea Surface Salinity. Deep Sea Research, 105, 1-18, doi:10.1016/j.DSR.2015.08.005.
The distribution of quality flags differs for each ship. Overall the ratio of Good and Probably Good data relative to the total amount of SSS underway data collected in delayed time is around 92% in 2018.
A correction based on comparisons with daily water samples is applied on TSG SSS data. The distribution of these corrections shows a distinctive peak around zero and is largely skewed towards positive values, to account for negative drifts or shifts in TSG measurements.
An error based on the dispersion of differences between TSG and water samples is associated with corrected TSG SSS data. The distribution of these errors shows values generally smaller than 0,05 pss.