Developing a Social Vulnerability Index to Assess Landslide Risk in Hong Kong: A Longitudinal Study Using Geographically Weighted Regression and Principal Component Analysis

This study proposes a multi-dimensional Social Vulnerability Index (SVI) framework, integrating spatiotemporal exposure dynamics, socio-economic vulnerability, and community resilience to assess landslide risk in Hong Kong. Drawing on environmental data from 2011 to 2022 (e.g., slope, rainfall intensity, and impervious surfaces), we applied Geographically Weighted Regression (GWR) to model the annual landslide exposure probability (standardized from 0 to 1). Census data was further used to construct the vulnerability and adaptability indices through principal component analysis (PCA), focusing on 10 variables including aging population, low-income households, and minority density, etc. Adaptability indicators include education level, infrastructure, and household resilience. The final SVI is aggregated at the District Council Constituency Areas' scale and calculated as SVI = Exposure + Vulnerability - Adaptability. The spatio-temporal analysis from GWR identifies chronic high-risk districts (e.g., Po Tin, Kwong Tak...), where high exposure overlaps with high vulnerability and low adaptability, largely attributed to steep topography and socio-economic marginalization. While districts like western area of Hong Kong Island demonstrated resilience due to robust infrastructure and economic capacity. By deconstructing vulnerability and adaptation, this framework offers new insights for landslide risk management, prioritizing slope stabilization in high-exposure areas and formulating contingency plans for highly vulnerable communities.