Bayesian shrinkage estimation for extreme values in 3D satellite-based geospatial modeling of oil and gas systems

Abstract

This study proposes an integrated methodological framework for modeling extreme geospatial parameters in oil and gas system design using 3D satellite-derived data. The approach combines extreme value theory with Bayesian shrinkage estimation under asymmetric loss functions, employing the Fréchet distribution to capture heavy-tailed behavior and rare extreme events commonly observed in geospatial variables. The proposed model enables stable and realistic estimation of extreme values by balancing sensitivity to genuine extremes with robustness against noise and outliers.
Simulation-based analysis demonstrates that Bayesian shrinkage estimation improves the reliability and interpretability of extreme-value parameters compared to classical estimation methods. The resulting posterior uncertainty measures provide valuable decision-support information, allowing engineers to assess confidence levels associated with extreme geospatial predictions. The findings highlight the practical relevance of incorporating decision-theoretic principles into geospatial modeling workflows.
Overall, the study contributes a robust and flexible framework for uncertainty-aware 3D geospatial modeling and supports risk-sensitive decision making in complex geological environments relevant to oil and gas applications.