Using a UE5 Platform Based on Point Cloud to Develop the Digital Twin System of a Lianziya Scenic Spot in Xilingxia of the Yangtze River in China

Abstract Addressing the challenges of complex terrain modeling and poor real-time interactivity in geologically sensitive areas, this study presents a digital twin system construction method based on the Unreal Engine 5 (UE5) platform. Taking the Lianziya dangerous rock mass in the Three Gorges area as a case study, high-precision terrain data were acquired using an unmanned aerial vehicle equipped with light detection and ranging, fused with oblique photography imagery. A cascade filtering pipeline was implemented to optimize point cloud quality, and the quadric error metrics algorithm was applied for model simplification. The results show that the method reduced the mesh polygon count by 50% while maintaining level of detail 4 accuracy, with a mean geometric error (Hausdorff distance) of only 0.018 m. The resulting system enables dynamic weather simulation and adaptive path planning, maintaining a stable frame rate of over 60 FPS at 4K resolution, effectively resolving rendering stalls in large-scale scenes. This workflow provides a reproducible, high-fidelity technical solution for the digital preservation and intelligent management of geological hazard sites.