SBAS-InSAR Analysis of tectonic derived ground deformation and subsidence susceptibility mapping via machine learning in Quetta City, Pakistan

This study utilized Time-Series Synthetic Aperture Radar Interferometry (TSInSAR) to provide accurate and cost-effective monitoring of ground displacement in Quetta City, Pakistan–a seismically active and rapidly urbanizing region. Investigation into the influence of active fault line networks and lithological composition on ground movements and subsidence susceptibility mapping (SSM) has not yet been revealed, which is crucial for risk mitigation. Employing two years of Sentinel-1 images, this research assesses ground deformation using the Small Baseline Subset (SBAS) technique, while the Logistic Regression (LR) model was employed to assess subsidence susceptibility. Results indicate significant displacement in the central urban area with an average vertical subsidence velocity of–166 mm/yr and an uplift rate of 48 mm/yr in the surrounding hilly terrain. A local Global Positioning System (GPS) station provided validation, confirming an average vertical velocity of–163.3 mm/yr, underscoring the reliability of InSAR data. The LR model owns an accuracy of 0.92 in the Area Under Curve (AUC) approach and predicts the quaternary lithologies, constructed regions, and fault lines are the main triggers of subsidence. In sum, the findings suggest that tectonic activities are the main cause of the ground movement, while human-induced elements contribute significantly as a secondary influence.