The digital replication of critical infrastructure is fundamental to the development of Smart Airports and the implementation of digital twins. Urban aerodromes, such as Congonhas Airport (SBSP) in S˜ao Paulo, face unique spatial constraints due to dense surrounding urbanization, requring accurate geometric models to manage operational risks and evaluate environmental impacts. Despite the potential of combining high-resolution photogrammetry and LiDAR, translating these massive datasets into structured 3D models presents significant methodological challenges. This study proposes an automated, multi-scale hierarchical segmentation and classification workflow to map aerodrome features using 0.25 cm spatial-resolution orthophotos and high-density LiDAR data. The methodology employs the Felzenszwalb-Huttenlocher algorithm across three spatial levels: macro-scale functional zones, elevated infrastructure via normalized Digital Surface Models (nDSM), and ground markings employing LiDAR intensity data. A Random Forest classifier was then applied to categorize thirteen distinct airport classes. The hierarchical approach outperformed traditional single-scale segmentation, achieving an F1-Score of 0.87 compared to the baseline of 0.53. The integration of LiDAR data proved crucial for distinguishing spectrally similar features, such as gray hangars on asphalt pavements. The proposed workflow automates the extraction of structural features, providing a foundational geometry for predictive maintenance, spatial organization, and urban conflict modeling.