Simulation of landscape evolution using a global flow path search method

Abstract

The choice of a flow path extraction method is a key issue in numerical simulation of fluvial landscape evolution. Most whole landscape evolution models use the deterministic eight-neighbor flow direction retrieval method (D8), which generates angular uncertainty in determined flow paths. In the modeling of landscape evolution, the uncertainty generated at each local flow direction accumulates over both space and time. Recently, a new method which searches for flow paths over global scale, called global D8 or GD8, was proposed as an alternative to D8. GD8 relaxes uncertainty generated at a local level over an entire flow path while still defining specific flow paths without artificial dispersion. On the basis of these advantages of GD8 demonstrated for static landscapes, this paper presents the first landscape evolution model that uses GD8, i.e., a new Landscape Evolution model using Global Search (LEGS). Using LEGS, the difference between D8 and GD8 simulations is investigated with focus on both evolution rate and resulting topography. Theoretical landscapes simulated by LEGS are evaluated using typical characteristics of natural river networks such as concave profiles, Hack's law, the power-law exceedance probability distribution of contributing areas, and the power-law variogram.