Problematic:
Following research thematics presented previously, the building of software tools is a logical added-value of this project, which implies the definition of systems able to facilitate models and data integration, from a pre-operational point of view.
The previous experience of the AIR project in that domain allows a correct assessment of the stakes and the technological resources needed to accomplish such an integration task (collaboration with SMIS project on LeSelect, European projects THETIS and DECAIR). The problematic is very wide and lies on the most recent tools available from information technology (indexation, middleware, distributed databases, high-bandwith networks). The goal is setting up a system capable of integrating data and models ``from inside''. Any user that wants to make use of a complex system must have access to modeling tools not statically deposited at any particular server, but on the contrary dynamically assembled from models accessible in a distributed manner. That user may be able to choose visualization methods and interpretation tools which can vary depending on the problematic. Data itself must be dynamically chosen from selection criteria (e.g. the most recent satellite data, or satellite data having specific spatial or spectral or quality requirements). The system must be able to perform quality assessment of results or of data. Lastly, it is desirable to assess the global quality of a dynamic assemblage of models, and compare the results coming from different workflows.
Objectives:
Different technologies nowadays exist to fulfill these objectives. They are indexing techniques, data and model publication in distributed environments, and dynamic workflows:
These technologies have been experimented, at various levels, in the collaboration with SMIS project (and MEDIENCE) on LeSelect. In this proposal, we want to put data and model integration to another level (multidimensional data, parallelization, workflow definition systems etc.), and systematically experiment data, models and results quality evaluation.
Our goal is to bring to such distributed systems the necessary functionalities and also to integrate data and models in these systems, in order to get powerful tools adapted to external institutions.
The problem of visualizing output results from the software tools is crucial, notably for applications (e.g. atmospheric applications) where output vector fields are 
4D (time, space, 
physical fields). The visualization should tailor the needs of the scientific community (better understanding the processes) or operational specialists (complex geometry, better representation etc.). From this point of view, the coupling with virtual reality is a necessary step.
Typical examples of such software tools are already under consideration among the CEREA on-going projects (e.g.: ozone and radionuclide forescating).