Large-scale estimation of roof geometry indicators for wind vulnerability assessment using airborne LiDAR

Roof geometries, such as slopes, orientations and overhangs, play a key role in defining vulnerability to cyclonic winds as it directly informs the pressure and uplift forces applied to buildings. However, these parameters are not available at a large scale especially in French overseas territories (DOM-TOM) particularly exposed to cyclonic winds and where building databases often lack sufficient geometric details. The objective of this work is to establish a workflow to estimate building vulnerabilities to cyclonic winds at a territorial scale using three roof-related parameters.  

Using freely available airborne LiDAR data acquired at a large scale and distributed by the Institut Géographique National (IGN), the proposed approach takes account of the current limitations of the equipment used for the description of buildings. Limitations include: acquisition angle and point density leading to incomplete wall sampling and planimetric uncertainties on the order of fifty centimeters that reduces the object discrimination capacities.  

LiDAR point clouds are used to describe buildings through three classes, walls, roofs and rooftop objects (chimney, technical equipment). LiDAR points are spatially associated with database buildings' polygons IDs, and they are used to reconstruct buildings' footprints to avoid spatial issues during analysis. Wall and roof points are then used to compute parameters. 

• Orientation and slope can be defined by removing walls and rooftop objects using elevation within buildings’ footprints. Statistical analysis can be finally used to describe the roof into categories such as dominant roof orientation, the number of distinct roof orientations, and slope gradient. LiDAR intensity may also provide coarse information on roof material type.  

• Roof overhang estimation remains more sensitive to wall point density and precision of the equipment used. Walls are reconstructed using density-based clustering (DBSCAN) combined with line-fitting (RANSAC and Hough transform) enabling the extraction of geometric features from heterogeneous LiDAR data distribution. Using airborne LiDAR compared to terrestrial increases the number of faces that can be detected but also lowers the global quality of the result.  

Resulting indicators are intended to improve and to complete existing databases at a large scale with relevant details on wind vulnerability. The proposed workflow is meant to be reproducible, scalable to large areas, it is intentionally data-driven and designed to benefit from ongoing improvements in LiDAR acquisition and classification. Current limitations primarily arise from point density and classification quality. Improvements in these parameters would enable more accurate wall reconstruction, roof object discrimination (chimneys, technical equipment), and roof–façade separation, ultimately leading to more reliable vulnerability estimates.