Use of network algorithms in spatially distributed models for the study of river basin response

The research concerns the potential applications of data extracted from topographic maps with GIS for the study of the main physical relations between channel network structures and the runoff generation at the basin scale. It deals with the methodology set up for the automatic channel network extraction, according to the link-based approach, using the DEMs. The channel network is obtained as a rooted tree graph of minimum energy, in which the nodes correspond to the cells resulting from the raster scanning of the basin maps. Pixels with no uncertainty in flow direction are subjected to a local optimization, whereas pixels of equal elevation and pits are globally optimized by an algorithm for the minimum cost network flow problem. Achieved is an excellent correspondence between natural and simulated network, thus confirming the natural system tendency to a minimum energy configuration.