Arbitrary Dipole-Dipole Observation Systems and High-Precision Resistivity Imaging Algorithms for Complex Survey Areas

Abstract

Electrical resistivity tomography (ERT) plays a crucial role in resource development and hazard assessment in both urban and mountainous areas. However, conventional 3-D ERT, which relies on a regular grid layout, often faces limitations in acquiring sufficient and effective observational data in complex survey areas. Moreover, due to the influence of electric field volume effects, traditional gradient-based inversion algorithms struggle to achieve high-precision imaging results and interpretations. To address these challenges, we propose a new scheme that combines an arbitrary dipole-dipole acquisition system with a high-precision inversion imaging technique based on the supervised descent method (SDM). The arbitrary dipole-dipole acquisition system offers enhanced flexibility in electrode deployment, enabling the collection of a larger volume of observational data with richer polarization information, thereby laying a solid foundation for high-resolution exploration imaging. The high-precision SDM inversion technique integrates the powerful nonlinear fitting capabilities of neural networks with the physical laws governing electric field propagation, significantly improving the resolution of inversion imaging results. Numerical simulations confirm that, compared to conventional ERT, the arbitrary dipole-dipole acquisition system enables the gathering of more abundant and effective observational data in complex measurement environments. In addition, the simulation results demonstrate the superior performance of SDM over the Gauss-Newton (GN) method and the pure data-driven network inversion (PDNI) method in terms of imaging quality, computational efficiency, and generalization ability.