Using Scientific Workflow Techniques for Automatic Processing of Environmental Data

Building scientific experiments, and moving from experiments to operational products are two distinct tasks that lack of technical support, especially in the environmental domain, where data sources as well as scientific expertise are distributed among various institutions. Building experiments is made much easier if access to data and programs can be made transparent and if a friendly interface is provided to link them; automation and quality assessment are key issues when designing operational processing chains. This paper presents an approach, that makes use of a middleware (LeSelect) that enables distributed data and programs publication and sharing, and of a workflow engine to support automatic processing. These concepts are presented through their implementation in two European projects: Thetis, for building experiments in oceanography, and Decair, that supports automatic data collection for air pollution prediction systems.