Building information extraction and earthquake damage prediction in an old urban area based on UAV oblique photogrammetry

• Published in: Natural hazards (Dordrecht) 2024-10, Vol.120 (13), p.11665-11692
• Main Authors: Zhao, Yu, Huang, Bo, Zhu, Zhizhong, Guo, Jiachen, Jiang, Jianqun
• Format: Article
• Language: English
• Subjects: Aerial photography; Building damage; Buildings; Civil Engineering; Damage detection; Earth and Environmental Science; Earth Sciences; Earthquake damage; Earthquake prediction; Earthquake resistance; Earthquakes; Elastoplasticity; Environmental Management; Environmental risk; Geophysics/Geodesy; Geotechnical Engineering & Applied Earth Sciences; Hydrogeology; Image acquisition; Information retrieval; Learning algorithms; Machine learning; Natural Hazards; Neighborhoods; Original Paper; Performance assessment; Performance testing; Photogrammetry; Seismic activity; Seismic response; Unmanned aerial vehicles; Urban areas
• ISSN: 0921-030X; 1573-0840
• DOI: 10.1007/s11069-024-06639-5

Seismic performance investigation and earthquake damage prediction of buildings in old urban areas are of great significance for the resistance to earthquake risks. Building attributes are fundamental to earthquake damage prediction, but the information of buildings in old urban may be insufficient and outdated. In this paper, UAV (Unmanned Aerial Vehicle) based oblique photogrammetry is used to the building-scale seismic performance assessment of the old urban area in Jaxing, China. Based on obtained UAV data, the building footprint is primarily detected, and direct attributes of building that determine the seismic performance are extracted, including number of storeys, nearborhood attributes and color attributes etc. Then indirect attribute, the building age, is predicted by machine learning. Base on the attritubes obtained, the elasto-plastic time history analysis of each detected building is carried out in the simplified model of the floor shear model, and the building damage distribution in study aera is finally obtained. Highlights • A method based on unmanned aerial vehicle (UAV) tilt photogrammetry was initially used for the building-scale seismic performance assessment in Chinese old urban area. • Building attributes that determine the seismic performance are directly recognized or indirectly acquired through UAV tilt photogrammetry. • The method of analyzing building age in Chinese old urban areas by machine learning using spatial features, neighborhood features and image features is proposed.