Assessing the Accuracy of Drone-Based Topographic Mapping in Urban Environments

Thesis Abstract

Urban topographic mapping is fundamental to infrastructure development, urban planning, and environmental management, yet traditional surveying methods often face limitations related to accessibility, cost, and time efficiency. Recent advancements in drone technology and photogrammetric processing have introduced new opportunities for rapid and cost-effective spatial data collection. However, the accuracy and reliability of drone-derived topographic data in complex urban environments, characterized by dense built-up areas, tall structures, and variable surface conditions, remain inadequately assessed, necessitating systematic empirical evaluation to establish their suitability for high-precision applications. This study aims to assess the spatial accuracy of drone-based topographic mapping within urban environments and to identify factors influencing data quality. The specific objectives include (1) quantifying the positional accuracy of drone-derived orthomosaics and Digital Surface Models (DSMs) compared to ground-truth reference data; (2) evaluating the influence of different flight parameters—such as altitude, overlap percentage, and camera resolution—on data accuracy; (3) examining the impact of urban terrain complexity on the precision of drone-based measurements; (4) testing the applicability of established spatial analysis theories, notably the Least-Squares Adjustment and the Theory of Error Propagation, in modeling accuracy outcomes; and (5) providing recommendations for optimizing drone surveying protocols in urban contexts. Methodologically, a mixed-methods approach was employed, centering on an empirical, field-based case study conducted within a metropolitan city with diverse urban typologies, including high-rise districts, residential suburbs, and mixed-use areas. A sample of five targeted sectors, each covering approximately 0.5 square kilometers, were selected for detailed analysis. Data collection involved the deployment of a multirotor drone equipped with a high-resolution RGB camera, executing flights at varying altitudes (50 m, 80 m, and 120 m) with different flight overlaps (60%, 80%) and camera settings, resulting in a total of 45 flight missions. Ground control points (GCPs) were established using high-precision Differential GNSS technology, with 10 GCPs per sector, serving as reference control for accuracy assessment. Post-flight, data processing employed Pix4Dmapper software to generate orthomosaics and DSMs, which were subsequently analyzed against GCP coordinates using Geographic Information System (GIS) tools. The primary analysis involved descriptive statistics, Root Mean Square Error (RMSE), and vertical and horizontal positional accuracy evaluations. To model the relationships between flight parameters and accuracy metrics, multivariate regression analysis was conducted, testing the hypotheses derived from the Least-Squares Adjustment and Error Propagation theories. Further, ANOVA tests examined the significance of differences across sectors with varying urban densities, while thematic analysis of field observations provided qualitative insights. The anticipated outcomes of this research include identifying optimal drone survey configurations for urban topographic mapping, with expected RMSE values falling within 10 cm horizontally and 15 cm vertically at specified flight parameters. The study is expected to reveal that increased flight altitude and lower overlap correlate with decreased positional accuracy, particularly in areas with high urban canyon effects. Furthermore, results will demonstrate the applicability of theoretical models in predicting accuracy outcomes, providing a scientific basis for protocol standardization. The findings contribute to the body of knowledge by systematically validating the accuracy thresholds of drone-based topographic data in complex urban environments, bridging the gap between technological capabilities and practical requirements. The research underscores the importance of tailored survey parameters to maximize data quality, thereby enhancing the credibility and adoption of drone-based surveying in urban planning, construction, and environmental management. It concludes with practical recommendations for survey practitioners and policymakers and suggests avenues for future research, including the integration of LiDAR sensors and machine learning techniques to improve urban terrain modeling accuracy.