Assessment of methods for extracting low-resolution river networks from high-resolution digital data

Abstract

A range of methods for deriving lower-resolution river networks from higher-resolution Digital Terrain Model (DTM) data are assessed at various spatial scales. Derived flow networks are compared to fine-scale hydrologically corrected rivers using a range of performance criteria. These include a measure of spatial distance between fine-scale and derived rivers, and criteria which assess errors in derived catchment area. The Network Tracing Method (NTM), a vector-based network scheme, and the COTAT+ method, a raster-based scheme, are shown to produce river networks that most closely resemble the base fine-scale river networks. COTAT+ is better at preserving catchment areas while the NTM method is often spatially closer to the base river network, especially when applied at lower resolutions, due to a higher percentage of diagonal flow paths. Automatically derived river flow networks will only ever be as good as the base DTM or flow direction data set from which they are derived.