Automatic mapping of river canyons using a digital elevation model and vector river data

Abstract

River canyons mapping plays an important role in water conservation project construction, tourism resource development, and analysis of fluvial processes. However, the extraction of river canyons via manual interpretation or semi-automatic methods is inefficient and expensive, especially at large spatial scales. Therefore, the objective of this study is to propose a novel method for automatic extraction of river canyons. The method mainly involves (1) extracting the indegree of river segments and generating river buffers based on the indegree, (2) generating topographic profiles at the two riversides of each river segment based on a digital elevation model, (3) extracting peaks from the topographic curves with the assistance of depth curves, (4) matching the peaks from different sides of each river segment based on a distance-priority strategy and then generating peak pairs based on the results, and (5) extracting the geographic range and attributes of river canyons and mapping them into a layer. Results of cases in the Three Gorges and Yarlung Zangbo areas in China illustrate effectiveness and accuracy for the extraction of river canyons. In this case, the false alarm rate and the miss alarm rate of this approach are both no higher than 17%.