Global Wire Following Profiler Salinity Drift Correction

There is an offset in the observed conductivity and salinity which occurs at the start of some of the deployments of Global Wire Following Profilers (GWFP). This offset decays over time in an exponential fashion and is observed observed across arrays, profilers, and deployments but with no predictability: for example, it occurs on the third deployment of the lower GWFP of the Global Southern Ocean Profiler Mooring, but not on the two preceding deployments. Conversation with Sea-Bird Instruments indicated that this is a previously identified problem and was first observed in Argo float data (Sea-Bird Application Note 97) and is likely due to fouling of the conductivity cell by the Tributyl Tin (TBT) antifouling capsule. Sea-Bird suggests that if the TBT capsule is not thoroughly dried before shipping, the TBT may leak from the capsule and may form a viscous film over the conductivity cell that interferes with the conductivity measurement, which is then gradually washed away once the instrument is deployed.

The approach to correcting the salinity offset is described in OOI White Paper 3408-30001 (
3408-30001_Global_Wire-Following_Profiler_Salinity_Drift_2022-09-10_Ver_1-00.pdf (708.1 KB)
), with associated example data processing notebook (available on gitHub), using data from deployment #5 of the Station Papa lower GWFP. First, the salinity data is binned along isotherms. Then, for each isotherm, a model of the interference due to the TBT is fit to the data, and an estimate of the length of time it takes to wash away the TBT film is calculated. Next, the data identified by the model with little TBT interference is used to calculate a gain correction following the method outlined in Sea-Bird’s Application Note 31. The gain correction is then applied to the timeseries to derive the expected salinity time series absent any TBT interference. The gain correction method is then further expanded upon to include possible trends in the data and explicitly derive the expected salinity offset. Including a linear trend in the gain correction method results in a “corrected” salinity time series that closely matches the observed salinity where we expect the TBT interference to be negligible, providing confidence that the corrected salinity where the offset is large is a reasonable reconstruction. The methods and results from the example time series are not assessed for accuracy of the salinity measurements; that may be done by the user utilizing CTD and discrete bottle samples associated with the deployment and recovery of the profiler, comparison with nearby sensors or the preceding/following deployment, or another approach.

Additionally, the oxygen concentration data product in [μmol/kg] from the GWFP oxygen sensors are affected by the drift and should also be recomputed with the corrected salinity data. The oxygen data in μmol/L are unaffected.