Evaluating the Ability of Multi-Sensor Techniques to Capture Topographic Complexity.

This study provides an evaluation of multiple sensors by examining their precision and ability to capture topographic complexity. Five different small unmanned aerial systems (sUAS) were evaluated, each with a different camera, Global Navigation Satellite System (GNSS), and Inertial Measurement Unit (IMU). A lidar was also used on the largest sUAS and as a mobile scanning system. The quality of each of the seven platforms were compared to actual surface measurements gathered with real-time kinematic (RTK)-GNSS and terrestrial laser scanning. Rigorous field and photogrammetric assessment workflows were designed around a combination of structure-from-motion to align images, Monte Carlo simulations to calculate spatially variable error, object-based image analysis to create objects, and MC32-PM algorithm to calculate vertical differences between two dense point clouds. The precision of the sensors ranged 0.115 m (minimum of 0.11 m for MaRS with Sony A7iii camera and maximum of 0.225 m for Mavic2 Pro). In a heterogenous test location with varying slope and high terrain roughness, only three of the seven mobile platforms performed well (MaRS, Inspire 2, and Phantom 4 Pro). All mobile sensors performed better for the homogenous test location, but the sUAS lidar and mobile lidar contained the most noise. The findings presented herein provide insights into cost-benefit of purchasing various sUAS and sensors and their ability to capture high-definition topography.