Multisource geoscience data-driven framework for subsidence risk assessment in urban area

Abstract

Land subsidence, especially in developed cities, poses significant risks to human life, social property, and urban sustainability. Taking Liwan District in southern China as an example, this study proposed an acceptable framework for regional land subsidence risk assessment while complying with current national assessment system. With integrating the multi-source geospatial data from remote sensing and various geology surveys into ArcGIS, the subsidence risk assessment was carried out based on the subsidence susceptibility mapping, hazard and vulnerability surveying by using a series of data-driven methods. The results showed that, (i) although not all surface deformations detected by InSAR technology were caused by subsidence, they were instrumental in updating subsidence records; (ii) with the help of spatial correlation analysis using weight evidence as well as multi-source data fusion in high spatial resolution, the Random Forest-based classification models effectively identified the land use types and accurately mapped the land subsidence susceptibility; (iii) the hazard and vulnerability surveying based on a series of newly developed combined weight methods, improved the reliability of risk assessment; (iv) the extremely high- and high-risk areas from the zoning of the land subsidence, provided target areas for further management and prevention of land subsidence. This comprehensive and quantitative assessment framework highlights the need for continued monitoring in subsidence-prone regions, helping to propose strategies for risk mitigation and adaptive planning in urban areas.