Abstract: The extraction of river catchment basin is one of the fundamental problems in automatic cartographic generalization of river networks, the catchment basin area is a mostly used index to quantify the importance of river in the networks. There are two commonly used methods to extract catchment areas. One method is to split regions between the rivers medially without considering topographic information, but with this method the extracted catchment basin areas often deviate from the actual situation. Another method is to extract rivers and watersheds from the DEM data, and then to compute the catchment basin areas, however, the extracted rivers is different from the real river objects after all, so the extracted catchment basin areas are not corresponding with the real river objects completely. This paper proposed a catchment basin extraction algorithm based on the cooperation of contour clusters and river networks according to their natural spatial relation. The constrained Delaunay TIN（CD-TIN） was constructed to organize the topographical information of contour clusters and river networks, then the triangles were classified and two watershed extraction method was discussed separately. The watershed segments extracted from the typical triangles with the specific rules were subsequently connected to form the watershed network in order to build the catchment basins. The algorithm’s effectiveness was confirmed with the experiment and the result was in accordance with the terrain characteristics.

Key words: Contour, River Network, Watershed, Catchment Basin, CD-TIN