Abstract : Early models of Geographical Information Systems (GIS) have been deeply influenced by quantitative models and geometric representations of space. Despite the interest of these approaches for cartographical applications, they do not completely reflect the way a human being perceives and describes his environment since he preferably stores and processes qualitative information. The study of cognitive processes has raised fundamental issues such as how spatial knowledge is acquired by human beings, and to which degree the language people manipulate affects their ability to interact effectively with spatial information. First results have particularly reported that the modelling of space is a cognitive activity that still requires appropriate spatial representations. We have provided a structural categorization for descriptions that result from the perception of a scenery. The model illustrates the spatial, relational and semantic constructs that emerge from these descriptions. It helps for the identification of the described sceneries and is a support for cross-cultural studies. The model is extended by a salience-based approach that reflects the particularities of the entities identified, such as their linguistic properties and their spatial characteristics.