Assessing the Accuracy of UAV-Based Topographic Surveys in Urban Environments

Thesis Abstract

Urban environments present unique challenges to accurate topographic surveying due to complex infrastructure, dense building arrangements, and frequent obstructions, which often compromise traditional ground-based methods and demand innovative approaches to spatial data acquisition. Unmanned Aerial Vehicles (UAVs) have emerged as a promising technology, offering rapid, cost-effective, and high-resolution data collection; however, the accuracy and reliability of UAV-derived topographic data in such settings require thorough empirical assessment to establish their feasibility for urban planning, construction, and environmental management. This study aims to evaluate the positional accuracy of UAV-based topographic surveys within urban environments, specifically focusing on identifying the factors influencing data precision and proposing calibration techniques to enhance measurement reliability. To achieve these aims, the research integrates a quantitative empirical design, structured around field data collection and statistical analysis. The target population comprises UAV survey outputs obtained from five randomly selected urban districts within the metropolitan area, encompassing diverse infrastructural complexities. A stratified sampling approach was employed to select 30 UAV survey datasets—six from each district—collected using a DJI Phantom 4 Pro drone equipped with integrated RTK (Real-Time Kinematic) modules to improve positional accuracy. Data collection instruments involved a high-precision terrestrial LiDAR system serving as the benchmark standard, along with a GNSS receiver for ground control point (GCP) establishment, ensuring the validity of comparative analyses. The validation process involved overlaying UAV-derived point clouds with LiDAR data within a GIS environment, measuring positional deviations across multiple control points, buildings, and terrain features. The analysis employs a combination of descriptive statistics, including mean absolute error (MAE) and root mean square error (RMSE), and inferential techniques such as multiple linear regression to examine the influence of variables such as building density, flight altitude, and obstructions on survey accuracy. The study also applies thematic analysis to contextualize site-specific factors impacting UAV data quality. The findings are expected to reveal that while UAV surveys generally achieve an RMSE within 15 centimeters under optimal conditions, accuracy diminishes significantly in highly obstructed areas or at higher flight altitudes, emphasizing the necessity of GCP deployment and flight planning strategies to mitigate errors. The anticipated results will contribute uniquely to the knowledge base by providing comprehensive empirical evidence on the accuracy thresholds of UAV topographic surveys tailored to urban contexts, highlighting critical factors that influence spatial data fidelity. Moreover, the study aims to establish correction models derived from regression analysis that can be integrated into UAV data processing workflows, enhancing measurement precision in complex urban terrains. Ultimately, the research underscores the practicality of UAV technology as an alternative or supplementary method for urban topographic mapping, contingent on careful planning and calibration. The study concludes that UAV surveys can attain sufficient accuracy for various urban applications when combined with strategic control point placement and appropriate flight parameters. Based on these findings, recommendations include the development of standardized protocols for UAV survey operations in dense urban areas, training modules for practitioners, and further research into adaptive algorithms for real-time error correction. This work advances the operational understanding of UAV applications in urban surveying, fostering more reliable, efficient, and scalable spatial data acquisition methods that could transform urban planning, infrastructure development, and environmental monitoring practices.