Regularized ensemble Kalman inversion for robust and efficient gravity data modeling to identify mineral and ore deposits

Modeling mineral and ore bodies from gravity anomalies remains challenging in geophysical exploration due to the ill-posed and non-unique nature of the inverse problem, particularly under conditions of noisy or sparse data. Established inversion methods, including local optimization and metaheuristic algorithms, often require extensive parameter tuning and may yield unstable or poorly constrained solutions. This study proposes a regularized ensemble Kalman inversion (EKI) framework enhanced by Tikhonov regularization to improve numerical stability and mitigate sensitivity to ensemble degeneracy, thereby enabling efficient uncertainty quantification through ensemble statistics. Controlled numerical experiments show that the ensemble size is larger than [Formula: see text] with moderate regularization, we can achieve an optimal balance between convergence stability and model resolution. Benchmarking against established metaheuristic algorithms (PSO, VFSA, and BA) suggests superior computational efficiency and stable convergence. Synthetic and real gravity data inversion (chromite, Pb-Zn, sulphide, and Cu-Au deposits) suggests that the regularized EKI yields stable, geologically consistent results with prior interpretations and drilling data. These results highlight the regularized EKI framework as a robust and efficient tool for mitigating mining risks and supporting strategic decision-making in mineral exploration.