Understanding spatially nonstationary effects of natural and human-induced factors on land subsidence based on multi-temporal InSAR and multi-source geospatial data: a case study in the Guangdong-Hong Kong-Macao Greater Bay Area

Abstract

Land subsidence, a common geological phenomenon in deltaic regions, poses significant risks to infrastructures, environments, and human lives. Monitoring and understanding land subsidence are crucial for establishing resilient, adaptive, and sustainable environments. In this study, a robust multi-temporal interferometric synthetic aperture radar (MTInSAR) method was employed to monitor land subsidence in the Guangdong-Hong Kong-Macao Greater Bay Area (GBA) using 541 Sentinel-1 SAR images. Spatial auto- and cross-correlation analysis were applied to select the most reasonable explanatory variables from nine candidates in each city of the GBA. City-level geographically weighted regression (GWR) models were then constructed to explore the spatially nonstationary effects of natural and human-induced factors. The findings revealed subsidence velocities ranging from 0 to 83.7 mm/yr in the GBA from 2015 to 2021, with the dominant factors affecting subsidence varying among regions. Natural and human-related factors accounted for 51.13% and 48.87%, respectively. Further analysis of representative subsidence areas highlights the importance of continuous monitoring of ground deformation using MTInSAR, region-specific land subsidence mitigation strategies, and regular maintenance of urban infrastructure. These insights are valuable for policymakers and urban planners in comprehending the complex processes underlying land subsidence and informing decision-making processes for sustainable and resilient urban development.

Keywords

• land subsidence
• resilient and sustainable environments
• spatial heterogeneity
• multi-temporal insar
• geographically weighted regression