Assessing the Accuracy of UAV Photogrammetry for Urban Topographic Mapping

Thesis Abstract

Urban topographic mapping is vital for effective urban planning, infrastructure development, and disaster management; however, traditional surveying methods are often limited by high costs, extensive time requirements, and operational hazards. Recent advancements in unmanned aerial vehicle (UAV) photogrammetry offer a promising alternative due to its cost-effectiveness, rapid data acquisition, and high-resolution outputs. Nevertheless, the accuracy of UAV-derived topographic data in complex urban environments remains a subject of ongoing investigation, necessitating comprehensive validation to foster wider adoption in geospatial applications. This study aims to assess the positional accuracy of UAV photogrammetry in urban topographic mapping through a rigorous empirical field-based investigation. The specific objectives are to evaluate the horizontal and vertical accuracy of UAV-generated digital surface models (DSMs) against ground truth data, identify factors influencing accuracy such as flight altitude, viewing angle, and building density, and provide calibration guidelines for practical implementation in urban contexts. Employing a quantitative research design, the study was conducted in a metropolitan city with diverse urban morphologies. The target population comprises urban survey points and structure from motion (SfM) imagery datasets. A stratified random sampling technique selected 100 representative control points across different urban zones, including densely built districts and open spaces, ensuring comprehensive spatial coverage. Data collection involved UAV flights conducted at three different altitudes (50 m, 75 m, and 100 m) using a DJI Phantom 4 Pro equipped with a high-resolution RGB camera. Ground control points (GCPs) were surveyed with a differential GNSS receiver, establishing a benchmark for accuracy assessment. UAV imagery was processed through Agisoft Metashape software to generate DSMs, which were then compared to GCP coordinates. The analysis employed descriptive statistics to summarize the data, followed by accuracy assessment through root mean square error (RMSE), mean absolute error (MAE), and horizontal/vertical linear regression analysis to quantify spatial discrepancies. ANOVA tests examined the significance of differences in accuracy across flight altitudes and urban land use types. Results are anticipated to indicate that UAV photogrammetry achieves horizontal accuracy within 0.2 meters and vertical accuracy within 0.3 meters at optimized flight parameters, demonstrating its suitability for detailed urban topographic mapping. Variations in accuracy are expected to correlate negatively with increased flight altitude and higher building density, underscoring the importance of flight planning tailored to urban complexity. Theoretically, the study is informed by Tobler’s first law of geography and the principles of volumetric measurement theory, establishing a foundation for understanding spatial error propagation in photogrammetric processes. Practically, the research fills a critical gap in the validation of UAV-based urban mapping, providing empirical evidence and calibration standards to enhance reliability. It is expected that the findings will contribute to the development of a procedural framework for UAV data acquisition and processing in urban environments, supporting policymakers, urban planners, and geospatial practitioners. The study concludes that UAV photogrammetry can deliver highly accurate topographic data suitable for a range of urban applications when operated under optimal parameters. Recommendations include standardized flight altitude guidelines, calibration protocols, and suggestions for integrating UAV data with existing GIS tools. Further research is proposed to explore the application of multispectral sensors and machine learning techniques to improve accuracy and automate feature extraction in urban mapping workflows.