Photogrammetry- and LiDAR-based Multi-temporal Point Cloud Models and Digital Elevation Models for Landslide Investigation in Hong Kong - Feasibility and Challenges

In adapting to rapid urban development and changing climate, the geotechnical industry is shifting towards harnessing digital technologies in Natural Terrain Hazard Study (NTHS) for landslide investigation. In this paper, we adopted a new digital method using multi-temporal point cloud models and digital elevation models derived from various available resources for the assessment of landslide source volume and dimensions. These resources include (1) historical aerial photographs from territory-wide aerial photographic survey carried out by the Lands Department, (2) project-specific UAV photographic and video surveys, and (3) the territory-wide airborne LiDAR surveys data. Two case studies from the Fei Ngo Shan area, Hong Kong, were carried out. Case 1 involves two recent landslides that occurred in 2005, and Case 2 involves a cluster of eight recent landslides that occurred in 2020. All these ten landslides were carefully investigated using conventional methods (e.g., field measurement or API) by GEO and GeoRisk Solutions, respectively. These investigation results were taken as legacy records for a comparison with the results derived from our adopted digital method. The comparison shows that the landslide source volume derived from the digital method is similar to the legacy record. This paper assessed the feasibility and accuracies of aligning and comparing digital models derived from multi-sources for landslide studies.