LOV, Villefranche-sur-Mer, France

The impact of Argo float trajectory assimilation on the quality of ocean analyses is studied by means of an operational oceanographic model implemented in the Mediterranean Sea and a 3D-Var assimilation scheme. For the first time, both Argo trajectories and vertical profiles of temperature and salinity (TS) together with satellite altimeter data of sea level anomaly (SLA) are assimilated to produce analyses for short term forecasts. The study period covers 3 months during winter 2005 when four Argo trajectories were present in the northwestern Mediterranean Sea. The scheme is first assessed computing the misfits between observations and model forecast and analysis. The misfit statistics appear improved for float trajectories, while they are not degraded for the other assimilated variables (TS profiles and SLA). This indicates that the trajectory integration is consistent with the other components of the assimilation system and provides new information on horizontal pressure gradients. Comparisons between analyses obtained with and without trajectory assimilation suggest that trajectory assimilation can have an impact on the description of boundary currents and their instabilities, as well as mesoscale activity at regional scales. Changes are depicted by intermediate water mass redistributions, mesoscale eddy relocations, and net transport modulations. These impacts are detailed and assessed considering historical and simultaneous in situ data sets. The results motivate the integration of Argo trajectories in the operational Mediterranean Forecasting System.

The Mediterranean Forecasting System (MFS) has been operational for a decade, and is continuously providing forecasts and analyses for the region. These forecasts comprise local- and basin-scale information of the environmental state of the sea and can be useful for tracking oil spills and supporting search-and-rescue missions. Data assimilation is a widely used method to improve the forecast skill of operational models and, in this study, the three-dimensional variational (OceanVar) scheme has been extended to include Argo float trajectories, with the objective of constraining and ameliorating the numerical output primarily in terms of the intermediate velocity fields at 350 m depth. When adding new datasets, it is furthermore crucial to ensure that the extended OceanVar scheme does not decrease the performance of the assimilation of other observations, e.g., sea-level anomalies, temperature, and salinity. Numerical experiments were undertaken for a 3-year period (2005–2007), and it was concluded that the Argo float trajectory assimilation improves the quality of the forecasted trajectories with ~15%, thus, increasing the realism of the model. Furthermore, the MFS proved to maintain the forecast quality of the sea-surface height and mass fields after the extended assimilation scheme had been introduced. A comparison between the modeled velocity fields and independent surface drifter observations suggested that assimilating trajectories at intermediate depth could yield improved forecasts of the upper ocean currents.