Integrated seismic damage scenario and resilience assessment using multimodal data and support vector machine, a case study of Jiangyou City, Southwestern China

Seismic damage scenario and resilience assessment are essential for disaster prevention and urban planning, but traditional in-situ fieldwork is resource-intensive and costly at large scales. This study introduces an integrated geospatial framework using support vector machine (SVM), multimodal data (high-resolution remote sensing, street-view imagery, nighttime light, socio-economic statistics, urban planning data, and active-fault maps), and a locally calibrated Vulnerability Index Method for efficient seismic vulnerability and resilience evaluation with minimal fieldwork. Applied to Jiangyou City, southwestern China—a county-level city devastated by the 2008 Wenchuan earthquake—the framework attains 81% SVM accuracy, producing city-wide vulnerability maps, damage scenarios for intensities VI-XI, and a six-indicator resilience index. Key contributions include: (1) simultaneous quantification of seismic risk and urban resilience; (2) reduced fieldwork (only 26% of buildings surveyed) via machine-learning extrapolation; and (3) spatially explicit policy recommendations for retrofitting priorities in small- to medium-sized cities. Results show moderate building vulnerability (EMS-98 classes C-D), severe-to-catastrophic damage under intensities IX-XI, and low resilience with concentric decay from the center. Limitations involve static indices lacking real-time monitoring, incomplete socio-economic factors, and single-city validation; future work will integrate dynamic IoT data and expand regionally.