Generalisation of Geographical Networks

Publisher Summary This chapter reviews approaches to the model generalization of geographical networks. Automating such generalization can be particularly challenging due to the range of possible contextual data and constraints that need to be considered. At its simplest, this task can be viewed as controlled network attenuation in a predictable and repeatable fashion, which preserves important network characteristics. The basic tree-like structure of fiver networks supports a natural generalization based on stream orders. Road networks, in contrast, have more complex structures and semantics, and their generalization has proved more difficult. The use of graph theory in network modeling and analysis to support generalization is first considered. Example applications to road network generalization are described where the functional importance of road segments is inferred from their occurrence in optimal paths.