Mapping Urban Areas and Infrastructure Through Fusion of Airborne SAR 3-D Images: A Comparative Study With ALS Sensors

3-D urban maps from optical, LiDAR, or synthetic aperture radar (SAR) data are crucial for urban planning, solar panel installation, visibility analysis, and shadow estimation. Digital surface models (DSMs) from airborne laser scanning (ALS) serve as high-quality references but often lack wall information and exhibit gaps in vertical structures. This study explores the effectiveness of airborne SAR in mapping complex urban geometries and compares the results to ALS data, including point clouds and DSMs. We also propose a framework for fusing 3-D SAR and ALS data to enhance the accuracy of 3-D city models. This fusion approach ensures precise alignment, reduces outliers near walls, rooftops, and ground surface (commonly caused by SAR phase noise) and preserves valuable information about walls and vertical structures absent in ALS data. Given the diversity of urban areas, we performed class-specific analyses (ground, trees, buildings, and power lines). Multiaspect SAR was found to be critical for addressing radar shadows and gaps caused by nonbackscattering objects. Using six SAR aspects covering $270^{\circ }$ provided comprehensive 3-D data, minimizing the need to consider building orientation. While SAR and LiDAR provided similar scene information, only 30% of voxels contained the same information from both sources, highlighting their complementary nature. Datasets with more SAR aspects proved more informative than those with fewer aspects and more baselines. Ground and tree reconstructions benefited from multiple baselines due to the resolution of low-backscattering objects, whereas building and power line reconstruction showed minimal improvement. The findings suggest that a combination of ALS and SAR data is essential for a complete understanding of urban environments.