MinMax QC approach: Characterization of alert statistics and configuration for CMEMS In Situ TAC Nov. 2023 EiS - Case of ARGO T/S dataset

Since 2019, the NRT operations at CORIOLIS data center apply the MinMax approach as a local range test to the Temperature and Salinity ARGO profiles. The test detects suspicious profiles that are visualized by the NRT operator in order to take a quality decision. While updated by early 2020, the method has demonstrated its capacity to early detect ARGO platforms suffering from Salty drift (see "Abrupt Salty Drift (ASD) in SeaBird CTDs" section in https://argo.ucsd.edu/data/data-faq/ #fsd). Since early 2023, the method is also applied to all non-ARGO profiles distributed in the CMS product. In this technical note, we document the recent developments performed at POKaPOK in order to extend the domain of applicability and improve the method robustness without degrading its performance. Main updates concern 1) a new reference dataset with 4,5 years of additional ARGO profiles, new historical CTD dataset (from ARGO DMQC operations) and latest MEOP (Sea Mammals) dataset, 2) a vertical extension of the reference fields from 0-2000 dbar to 0-5500 dbar, 3) revisiting the validity interval widening strategy that aims at improving the robustness through accounting for the variability not sampled in the reference dataset. In a last step, from the results of the widening study mentioned above, we present the configuration choice made for the CORIOLIS operational chain update.