Automated identification of earthen berms in western US rangelands from LiDAR-based digital elevation models

Authors: XU, H. - University Of Arizona; Nichols, Mary; Lapides, Dana; Crompton, Octavia

Submitted to: Earth Surface Processes and Landforms
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 9/29/2024
Publication Date: 11/3/2024
Citation: Xu, H., Nichols, M.H., Lapides, D.A., Crompton, O.V. 2024. Automated identification of earthen berms in western US rangelands from LiDAR-based digital elevation models. Earth Surface Processes and Landforms. 49(15):5012–5026. https://doi.org/10.1002/esp.6009.
DOI: https://doi.org/10.1002/esp.6009

Interpretive Summary: Across the western US soil and water conservation structures, such as earthen berms, have been constructed since the late 1800s as ranching spread throughout the area. Berms alter hydrologic, geomorphic, and ecologic processes by intercepting runoff and altering the pre-existing pattern of water availability on the landscape. This research was conducted to develop an automated way to map these structures. We conducted our study in four rangeland sites in Arizona and one site in Colorado. Geomorphon, a computer vision tool, was used to classify landforms and identify berm-like landforms, including summits and ridges. Ten geomorphic and geometric attributes of each object were used to develop a machine learning model for distinguishing berms from natural summits and ridges. The model was trained and applied to independent test sites to identify and map berms. The mapped berms were compared with manually identified reference berms for accuracy assessment. The best identification result achieved 0.87 recall and 0.89 precision. The automated framework has the potential to be scaled up to larger areas in semi-arid environment.

Technical Abstract: Earthworks such as earthen berms have been constructed across the western US since the late 1800s to mitigate erosion in these landscapes where water is the dominant driver of erosion and the limiting resource for biota. Berms alter hydrologic, geomorphic, and ecologic processes by intercepting runoff and altering the pre-existing pattern of water availability on the landscape. Understanding site-specific changes in process dynamics requires accurate mapping of berm locations and knowledge of their condition. This paper presents an automated, object-based framework for identifying earthen berms from 1 m LiDAR derived digital elevation models at four rangeland sites in Arizona and one site in Colorado. Geomorphon, a computer vision tool, was used to classify landforms and identify berm-like landforms, including summits and ridges. Ten geomorphic and geometric attributes of each object were used to develop a machine learning model for distinguishing berms from natural summits and ridges. The model was trained and applied to independent test sites to identify and map berms. The mapped berms were compared with manually identified reference berms for accuracy assessment. The best identification result achieved 0.87 recall, 0.89 precision, and 0.88 F-measure. We also explored the influence of training sample selection on model performance and conducted an analysis of attribute relative importance. The automated framework has the potential to be scaled up to larger areas in semi-arid environment.