Self-Aware Joint Inversion of Multidisciplinary Geophysical Data in Mineral Exploration Using Hyperparameter Self-Adjustment: A Preliminary Study

This paper introduces a novel methodology for subsurface characterization in mineral exploration, through the simultaneous joint inversion of seismic and geoelectrical data. By combining complementary information provided by multidisciplinary geophysical data, the joint inversion yields a more accurate and consistent representation of subsurface properties. Furthermore, the joint inversion algorithm is empowered by dynamic hyperparameter self-adjustment. Hyperparameters are settings or configuration values that control the behavior of the inversion algorithm but are not directly learned from the data. Examples include regularization weights, coupling parameters, learning rates (if using gradient-based methods), and number of iterations. In traditional approaches, these values must be manually selected or tuned, often through trial and error, which is time-consuming and may lead to suboptimal results. Instead, in the approach here introduced, a self-adaptive mechanism monitors the evolution of the cost function and optimization performance, automatically tuning hyperparameters to enhance convergence toward an optimal (global) solution. For the purposes of this preliminary study, the method is tested on synthetic 2D geophysical scenarios featuring resistivity and seismic velocity anomalies representative of potential mineral targets. Results show the effectiveness of the approach in accurately identifying these subsurface anomalies. Finally, we show that this joint inversion technique holds significant promise for mineral exploration, particularly in detecting geological features such as ore bodies and mineralized zones, which can manifest as contrasts in seismic velocity and resistivity.