The Hierarchical Watershed Partitioning and Data Simplification of River Network.

For the generalization of river network, the importance decision of river channels in a catchment has to consider three aspects at different levels: the spatial distribution pattern at macro level, the distribution density at meso level and the individual geometric properties at micro level. To extract such structured information, this study builds the model of watershed hierarchical partitioning based on Delaunay triangulation. The watershed area is determined by the spatial competition process applying the partitioning similar to Voronoi diagram to obtain the basin polygon of each river channel. The hierarchical relation is constructed to represent the inclusion between different level watersheds. This model supports to compute the parameters such as distribution density, distance between neighbor channels and the hierarchical watershed area. The study presents a method to select the river network by the watershed area threshold. The experiment on real river data shows this method has good generalization effect. [ABSTRACT FROM AUTHOR]