In this work, we consider that the modelling of complex domains can be performed using Domain Specific Languages (DSL). The main principle of this approach consists in developing DSL primitives and to assemble them to model a certain domain. The ability to add new primitives into an existing model and to fine-tune it by replacing some of them provides a flexibility that is highly desirable in simulation intense fields.We have designed such a language, named Ocelet, which is tailored for dynamic landscape modelling.We consider that three important components may influence the adoption of this approach: a graphical user interface to build models in an efficient and user-friendly way, a solution to reason, e.g. consistency checking, about model primitives and a tool to facilitate the development of primitives repositories. In this paper, we emphasize that an ontology-based approach is adapted to design all these components. Moreover, a mapping between ontology and Ocelet elements is sufficient for its achievement and supports automatic transformations from one model to the other.