There is an increasing need to provide scientists and researchers as well as policy makers and the general public with valueadded services integrating information spread over distributed heterogeneous repositories. In order to incorporate available data sets and scientific programs into a powerful information and computational system it is mandatory to identify and exploit their semantic relationship. For this purpose, we advocate an ontological framework that captures these relations and allows the inference of valid combinations of scientific resources for the production of new data. We show how a knowledge base that commits to an ontology can be used to generate workflows on demand for the multiplicity of resources known to the system. To validate our ideas, we are currently developing a prototype for the area of Coastal Zone Management.