Emergency management systems reason about contextualized information, i.e., domain entities which have a precise position in a space. Often, domain entities lie on different spaces, e.g., river basins on a geographical space and routes on a topological one. Sometimes the same domain entity lies on two or more different spaces, e.g., a rescue squad may be localized both in a geographical and in a competence space. In such a complex scenario, emerges the need of a general model to specify localized domain entities that preserves the separation between 'what' (domain entity) and 'where' (its location in a space) and that encapsulates the intrinsic structure of the space. The paper presents such a model which relies on a unified paradigm for the definition of spaces. The model has been reified in an exemplified scenario dealing with a flooding prediction and risk management system.