Towards a hybrid approach for remote sensing ontology construction

Erosion, flooding and deforestation are natural risks which may affect the environment and therefore have a direct impact on safety and health. Thus, it is important to monitor these phenomena in order to quickly keep track of their variations and take right decisions accordingly. Remote Sensing (RS) is a fundamental source of information that can effectively enable natural risk monitoring. To do that, RS images need to be well represented and interpreted semantically. In this paper, we focus on the representation and interpretation of satellite images using ontologies. We develop a light ontology representing the content of RS images. We use in the ontological development process three resources: a domain-related corpus, existing geographic ontologies, and the lexical thesaurus WordNet.