TSAPW: A Two-Step Adaptive Pixel-Weight algorithm for enhancing an InSAR-based underground goaf locating framework

Illegal and undocumented mining operations pose significant threats to ecological stability and public safety by inducing large-scale ground subsidence. Systematic monitoring of underground mining activities constitutes a critical methodology for governmental authorities to curb illegal mining. Current solutions typically employ an underground goaf locating framework integrating Interferometric Synthetic Aperture Radar (InSAR) technology and deformation transfer theoretical model. This framework estimates goaf parameters by minimizing the full-pixel root mean square error (RMSE) between forward-modelling simulations and InSAR-derived deformation observations. However, this matching approach exhibits inherent limitations in extracting mining-induced deformation information with large gradients, leading to compromised localization accuracy. Addressing this challenge, we propose a novel Two-Step Adaptive Pixel-Weight (TSAPW) algorithm optimized for matching forward-modelling simulations with InSAR-derived deformation. Field verification results from two typical mining areas demonstrate that the proposed TSAPW algorithm improves the accuracy of geometric parameter estimation and centre point localization by an average of 3.5 % and 13.08 m, respectively, confirming its effectiveness for practical underground goaf localization.