Global search algorithm for nondispersive flow path extraction

A new algorithm is developed to extract flow paths from digital elevation data without planar dispersion on the basis of the concept of global search. Widely used nondispersive algorithms, such as deterministic eight-neighbor flow direction retrieval algorithms, suffer serious uncertainty in their determined flow paths because of the lack of variability, i.e., only eight allowed flow directions. Although uncertainty at the local level is an inherent problem residing in the domain discretization, this study shows that more reasonable flow paths at the global scale can be obtained by maximizing the use of all information stored in the given digital elevation data. By utilizing information stored in cells other than those in the direct vicinity, this alternative approach can reduce the uncertainty residing in the extraction of flow paths. The proposed algorithm makes a significant improvement in flow path variability on both theoretical and real landscapes, while it is still simple.