Low-Cost UAV in Photogrammetric Engineering and Remote Sensing: Georeferencing, DEM Accuracy, and Geospatial Analysis

Conventional land surveying, while reliable, is limited by factors such as restricted area coverage, time-intensive procedures, and high costs. In contrast, consumer-grade unmanned aerial vehicles (UAVs) emerge as a promising and cost-effective alternative. This study explores the potential of harnessing an exceptionally low-cost consumer-grade UAV to deliver reliable measurements and geospatial products in photogrammetric engineering and remote sensing. The Mavic 2 Pro was flown at an altitude of 165 m above ground in a single grid configuration with an 80% forward overlap and 72% side overlap to achieve an average ground sampling distance (GSD) of 3.7 cm at two separate sites. At site A, 300 nadir-looking RGB images were taken, while at site B, 9899 images were captured. Site A was designed for comprehensive assessment whereas site B was selected for scalability testing. Our research encompasses an in-depth investigation into georeferencing methods based on structure-from-motion (SfM), focusing particularly on digital elevation models (DEMs) and their accuracy. For site A, the root-mean-square error (RMSE) in elevation improved from 180 x GSD to 3.3 x GSD for direct and indirect georeferencing respectively on independent check points (CPs). For site B, horizontal and vertical RMSE remained within 2 x GSD on CPs for indirect georeferencing. The evaluation of georeferencing methods highlights the importance of incorporating ground control points when exclusively relying on the low-cost UAV's onboard positioning and orientation data. Additionally, our research underscores the indispensable role of linear rolling shutter adjustments in enhancing overall accuracy for non-mechanical electronic shutter sensors.