Challenges of UAV Photogrammetry as a Tool for Aerial Mapping and Digital Elevation Models: An Applied Study of the Ghiran Area - Misrata"

Abstract

Unmanned Aerial Vehicle (UAV) photogrammetry has emerged as one of the most significant and modern methodologies for large-scale geographical mapping and high-precision Digital Elevation Model (DEM) extraction, among various other engineering applications. It offers substantial economic advantages, including cost-effectiveness, significant time savings, and the reduction of field labor to a minimum. This paper highlights the technical and field challenges associated with UAV surveying in the "Ghiran" area of Misrata, a region that has lacked updated maps for over 15 years. The primary objective of this study is to diagnose the difficulties encountered during the surveying and digital processing phases. Specifically, it clarifies the reasons behind the failure of automated classification algorithms and deep learning models within Pix4D Mapper and Geographic Information Systems (GIS) to accurately differentiate between various building rooftops and ground-level features. The study's findings reveal that spectral similarity and the visual resemblance of building surfaces—often covered with dust or bitumen, to pavements and natural soil led to the generation of distorted Digital Elevation Models (DEMs). This necessitated extensive and arduous manual intervention to correct classifications and ensure spatial accuracy, thereby increasing the duration of office processing. Despite these hurdles, the research successfully extracted point clouds and DEMs with centimeter-level accuracy (< 5 cm), which is sufficient for infrastructure design. Ultimately, the study proposes robust solutions to mitigate spectral overlap in future projects, ensuring efficiency and the high-precision topographic data required for modern urban planning.