Semi-automated detection of rangeland runoff and erosion control berms using high-resolution topography data

Abstract

An inventory of topographic modifications is essential to addressing their impacts on hydrological and morphological processes in human-altered watersheds. However, such inventories are generally lacking. This study presents two workflows for semi-automatic detection of linear earthen runoff and erosion control berms in rangelands using high-resolution topographic data. The workflows consist of initial object identification by applying either morphological grayscale reconstruction (MGR) or the Geomorphon (GEO) method, followed by identification refinements through filters based on objects’ horizontal and vertical information. Three sites were selected within the Altar Valley, Arizona, in the southwestern United States. One site was used for developing workflows and optimizing filter thresholds, and the other two sites were used to validate workflows. The results showed that: 1) The MGR-based workflow methodology could produce final precision and detection rates of up to 92% and 75%, respectively, and take less than 5 s for a 10.1 km2 site; 2) The workflow based on the MGR method yielded greater identification accuracy than did the GEO workflow; 3) Object length, orientation, and eccentricity were important characteristics for identifying earthen berms, and are sensitive to general channel flow direction and berm shape; 4) Manual interrogation of topographic data and imagery can significantly improve identification precision rates. The proposed workflows will be useful for developing inventories of runoff and erosion control structures in support of sustainable rangeland management.

Keywords

• Earthen berm
• Rangeland
• Grayscale morphological reconstruction
• Geomorphon
• LiDAR