Assessing landslide susceptibility with dynamic deformation monitoring and explainable machine learning: a case study in Longhua County, China
Year: 2026
Type: article
Abstract: Landslide susceptibility assessment (LSA) is still dominated by static conditioning layers with limited interpretability, and it under-utilises kinematic deformation for operational prioritisation. To address these gaps, we develop a Landslide Optimised Susceptibility Assessment (LOSA) that dynamically fuses Small Baseline Subset Interferometric Synthetic Aperture Radar (SBAS-InSAR) time-series deformation with a Random Forest (RF) susceptibility surface via a spatiotemporal weighting matrix, and explains drivers with SHapley Additive exPlanation (SHAP). Using 186 Sentinel-1A scenes (2019–2022) over Longhua County, SBAS-InSAR resolves millimetre-scale motion (peak 104.95 mm/y), while Random-Forest attains Area Under Curve (AUC) = 0.838. The integrated LOSA delivers a 70.66% contraction of the extremely high risk footprint and indicates more targeted mapping under the conjunctive fusion, with field audits confirming hotspot activity. LOSA is a transparent, adaptive, and rapidly updatable tool that links kinematics with interpretable modelling to support landslide-risk governance.
Language: English
Subfield: Management, Monitoring, Policy and Law
Field: Environmental Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $1,790
Awards: G1601466381, G5584701337
OpenAlex ID: W7124470715