Automatic extraction and analysis of the hierarchical valley network - Application to a numerical model derived from two stereoscopic SPOT images

An automatic process flow is described, from the computing of a digital elevation model (DEM) to the extraction and analysis of the hierarchical valley network. The DEM is calculated by automatic correlation of two stereoscopic images acquired with the SPOT satellite. A new algorithm allows the extraction of a complete and connected hierarchical valley network from this DEM. Results show that the six statistical measurements used (curvature, dimension, length, confluence angle, density, and parallelism index) are principally controlled by the topography.