Digital elevation models (DEMs) are fundamental geospatial data, yet traditional acquisition techniques often fall short in densely vegetated mountainous regions. The advent of unmanned aerial vehicle (UAV) LiDAR technology has enabled novel methods for DEM acquisition in such challenging terrains. This article proposes a method for utilizing UAV LiDAR to acquire laser point clouds and construct DEM through simulated terrain flights. Key technologies such as laser scanning data acquisition, point cloud preprocessing, point cloud filtering, and DEM construction are studied. Taking the pumped storage project in Guangdong Province as an example, the principles and implementation methods of UAV LiDAR simulated terrain flights are thoroughly examined. The laser point clouds are preprocessed, and an incremental and terrain-aware Delaunay triangulation filtering algorithm is proposed to effectively extract the ground points from scattered point clouds. The resulting DEM achieves high precision ranging from 0.07 to 0.31 meters. Experimental results demonstrate that this method can successfully extract ground points and build accurate DEM that meet the requirements of engineering measurement specifications. It provides a new approach and solution for constructing DEM in vegetation-covered mountainous areas, offering practical value and significance for wider application.