Semantic classification and annotation of satellite images are of great importance and require knowledge resources. The complexity of satellite scenes makes its classification and annotation hard tasks and we are still far from totally resolving the semantic gap problem. There are several knowledge resources such as semantic networks, taxonomies and ontologies. In this paper, we propose to enrich the SatelliteScene-Ontology using real hyperspectral scenes, the USGS spectral library and the WordNet lexical database. The resulting ontology would be published online for further exploitation by researchers.