Catchment specific thresholds for fully automated flow accumulation modeling

Abstract:

Flow accumulation modeling is applied in disciplines including emergency management, ecology, engineering and cartography. A challenge to broad application of flow accumulation modelling is the requirement to fit flow accumulation thresholds, needed for constraining the derived channel network extent, to different landscapes. Here we present the recently developed stream cross-section analysis (SCA) strategy for fully automated catchment specific pruning of flow networks. SCA determines the initiation of channel formation by identifying a corresponding jump in cross-section standard deviation of curvature (sK) for headwater channels. Headwater channels in an over-extracted elevation-derived drainage network are systematically thinned until SCA identifies significant stream heads. Cross-sections are evenly spaced along flow accumulation channels, sampling relative elevations in the digital terrain model. A catchment’s specific sK value is a function of the total sK distribution and is defined as 05*Tukey’s upper inner fence. The SCA method performs comparably to methods with manually determined thresholds when tested on catchments ranging in size between 2.8 and 7.5 square kilometers with varied topographic relief and land cover. Results of this work suggest that the SCA method can be applied to generate channel network estimates for large areas in an automated process, although further testing in other terrain conditions is still necessary. The method is also potentially useful for cataloging and change detection of sensitive ecosystems associated with riparian zones of headwater streams.