Abstract : There is a need for decision-makers to be provided with both an overview of existing knowledge, and information which is as complete and up-to-date as possible on changes in certain features of the biosphere. Another objective is to bring together all the many attempts which have been made over the years at various levels (international, Community, national and regional) to obtain more information on the environment and the way it is changing. As a result, remote sensing tools monitor large amount of land cover informations enabling study of dynamic processes. However the size of the dataset require new tools to identify pattern and extract knowledge. We propose a model to discover knowledge on parcel data allowing analysis of dynamic geospatial phenomena using time, spatial and thematic data. The model is called Land Cover Change Continuum (LC3) and is able to track the evolution of spatial entities along time. Based on semantic web technologies, the model allows users to specify and to query spatio-temporal informations based on semantic definitions. The semantic of spatial relationships are of interest to qualify filiation relationships. The result of this process permit to identify evolutive patterns as a basis for studying the dynamics of the geospatial environment. To this end, we use CORINE datasets to study changes in a specific part of France. In our approach, we consider entities as having several representations during their lifecycle. Each representation includes identity, spatial and descriptives properties that evolve over time.