The project aims at providing end-users with high-skill operational coastal ocean forecasts, for the (trans-frontier) northwestern part of the Iberian peninsula. This objective fits very well within the current needs of multiple communities (coastal and open sea management, ecosystem protection, fishing industry, maritime transport, catastrophe management, etc).The project will take place at CIMAR, Portugal, within the framework of the RAIA project (http://www.observatorioraia.org). RAIA partners provide different forecasts for the same geographical region. More specifically, different versions of the ROMS model are run concurrently. However, data is under-used: (remote-sensed and in situ observational) data assimilation is not implemented operationally, and no model fusion, ensemble modeling, or super-ensemble technique is implemented at all.The proposed project aims at replacing one of the models by an ensemble of models. Methods for generating random but physically consistent perturbations of the oceanographic variables will be researched and implemented. This will allow obtaining more reliable model error estimations; and furthermore an Ensemble Kalman filter with a realistic error covariance matrix will be implemented to assimilate observations.It has been shown that a super-ensemble (SE) of models (i.e. a weighted combination of individual models) provides forecasts with higher skill and reduced uncertainty. SE techniques are relatively new in the ocean modeling community, but their usage is expected to increase together with the number of seas covered by concurrent models. In the proposed project, SE techniques will be further developed, and new filters will be tried out to evolve the SE combination in time. SEs will be tried out on full 3D model fields.The project will be favorable to the fellow's scientific career, will provide new research results, and will also improve the operational forecasts of the studied region.