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

Thesis Abstract

Urban topographic mapping is integral to urban planning, infrastructure development, and environmental management; however, traditional surveying methods often face limitations related to cost, time, and accessibility in complex urban terrains. The emergence of Unmanned Aerial Vehicles (UAVs) has revolutionized geospatial data collection by enabling rapid, cost-effective, and high-resolution topographic mapping in urban environments. Despite widespread adoption, the accuracy and reliability of UAV-derived topographic data relative to conventional ground-based survey techniques remain under-investigated, particularly in areas with dense infrastructure and variable surface conditions. This study aims to empirically assess the positional and elevation accuracy of UAV-based topographic mapping in urban settings, with a focus on identifying factors influencing data quality and proposing calibrated correction models to enhance accuracy. The primary objectives of this research are to (1) quantify the positional and elevation accuracy of UAV-derived point clouds compared to ground control surveys, (2) evaluate the influence of urban surface complexity, building density, and atmospheric conditions on data accuracy, (3) develop statistically validated correction and calibration models to mitigate identified errors, and (4) provide comprehensive guidelines for optimizing UAV-based topographic surveys in urban contexts. To achieve these objectives, a mixed-methods research design combining quantitative accuracy assessment and qualitative analysis was adopted. The study population comprised urban areas within the metropolitan city of 1.2 million residents, selected to represent diverse urban morphologies, including high-density downtown districts, mixed-use neighborhoods, and peri-urban zones. A stratified purposive sampling technique was employed to select five representative sites, each covering approximately 2 square kilometers, for detailed survey analysis. Data collection involved UAV flights conducted with a DJI Phantom 4 RTK drone equipped with centimeter-level GNSS-enabled RTK modules, and conventional ground control point (GCP) surveys executed using high-precision total stations and differential GPS equipment with an accuracy of ±5mm. A total of 150 GCPs were strategically distributed across the sample sites. The UAV surveys produced point clouds and orthomosaics processed through Pix4Dmapper, while GCP data provided a benchmark for accuracy validation. Data analysis incorporated geo-comparative accuracy assessments using Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and vertical/horizontal deviation analyses. Regression and ANOVA tests were applied to evaluate the influence of urban surface variables on errors, and multivariate calibration models were developed using statistically significant predictors. The validity and reliability of the data were ensured through calibration checks, repeat surveys, and cross-validation techniques. Anticipated findings include quantifiable measures of deviation between UAV-derived and ground-truth data, with expected RMSE values below 0.10 meters for elevation and 0.15 meters for horizontal positions, contingent on site conditions. It is hypothesized that densely built areas and rough surface textures will contribute significantly to positional and elevation errors, with atmospheric variables such as sunlight and wind also impacting data quality. The study is expected to develop correction models that can reduce systematic errors by up to 80%, thereby enhancing the practical utility of UAV mapping in urban settings. This research contributes to the body of knowledge by providing a rigorous, empirically validated assessment of UAV data accuracy in complex urban terrains, a critical step toward standardizing UAV survey procedures. It advances understanding of environmental and infrastructural factors influencing data quality, offering a foundation for developing robust correction protocols and survey guidelines. Practical implications include improved accuracy assurances for urban planners, engineers, and GIS professionals implementing UAV technology, as well as policy recommendations for integrating UAV-based data into official geospatial frameworks. In conclusion, with the integration of advanced statistical modeling and empirical validation, this study offers concrete, field-tested strategies for optimizing UAV-based topographic surveys. The results advocate for adopting calibrated correction models and strategic site selection to maximize data accuracy, thereby facilitating wider acceptance and reliable application of UAV technology in urban geomatics. Future research directions recommend expanding the scope to include multi-temporal surveys, diverse urban typologies, and the integration of LiDAR data for enhanced surface representation.