Delayed mode qualification of ARGO floats: CORIOLIS analysis system

E. Autret, F. Gaillard

Laboratoire de Physique des Océans, Ifremer, BP70, 29280 Plouzané France

Coriolis data center collects, checks for quality and distributes in real time a great number of global in situ data. Managed data sets are mainly temperature and salinity profiles from ARGO floats, XBTs, CTDs, drifting and moored buoys. In order both to qualify the dataset and provide temperature and salinity analyzed fields a data analysis system based on optimal estimation methods has been developed. Using the measurements from the CORIOLIS data base the analysis computes temperature and salinity 3-D fields on 60 levels from the surface down to 2000 meters. In addition to the estimate of temperature and salinity at each grid point, the method calculates at each data point, a residual corresponding to the difference between the estimate and the measurement there. A quality control tool based on the behavior of the residuals has been developed.
The operational system implemented is exploited in real time to produce global analyzed fields and to detect possible gross errors. In delayed mode, re-analysis are performed and a method using the statistical behavior of the residuals has been developed to detect possible sensor drifts or bias particularly for ARGO floats. The method has been used to diagnose the behavior of the whole fleet of ARGO floats in the Atlantic. This test has shown that this operational method is a fast and easy mean to estimate offsets and to point out possible drifts. In this particular case, an expertise is needed to decide whether it is due to the analysis (inconsistency with climatology) or to a sensor drift and each case should be looked at individually and compared with other methods.