Road networks, roads, and junctions are examples of natural language terms whose semantics can be described by affordances of their physical referents. In order to define affordances in such a way that they can be used for classifying and describing instances in a geographic database, one has to deal with the problems of informational incompleteness and limited definability. In this paper, we propose an affordance-based theory of channel networks, based on the work of Hayes [4], as a means to derive necessary conditions for database representations of road networks. By exploring this example, we show that affordance-based logical definitions are a convenient method to capture essential properties of physical objects usually not present in their database representation, but appropriate to explain and define its structure.