Inversion of IP-affected TEM data with full parametrization of dispersive resistivity

Chargeable materials in the mining industry cause the induced polarization (IP) phenomenon when exposed to an electromagnetic field, which affects the electromagnetic response in transient electromagnetic (TEM) surveys. This distortion of TEM data is a challenge for traditional inversion methods, which typically focus on resistivity and may fail to provide reliable results when IP effects are significant. To address this limitation, this paper introduces a Bayesian inversion framework that incorporates full dispersive resistivity, using the Cole-Cole model to simulate both electromagnetic induction and IP phenomena. This approach allows for the effective recovery of Cole-Cole parameters from TEM data. A key advantage of Bayesian inversion is its ability to assess the confidence of inversion results, which is critical given the non-uniqueness of the inverse problem under these conditions.

Through numerical simulations and field examples, the proposed method demonstrates its ability to accurately recover both resistivity and Cole-Cole parameters, particularly in cases involving conductive and highly chargeable targets. However, the method struggles with resistive targets, where the inversion results exhibit lower accuracy and confidence. Despite this, the method is able to reliably resolve conductive regions, even when resistive regions are less accurately recovered, ensuring that model parameters are precisely estimated in conductive areas.

A field test conducted at Keyou Qianqi in Inner Mongolia confirmed the method’s effectiveness, successfully locating and validating a conductive, high-polarization silver-lead-zinc ore body. The study highlights the coupling of electromagnetic induction and IP effects, which generate time-decaying electromagnetic fields that are difficult to separate. By introducing dispersive resistivity, we can simulate these coupled responses and analyze the data’s ability to resolve dispersive resistivity model parameters. The analysis reveals that while considering IP increases the inversion’s complexity, it also enhances the resolution for resistivity and chargeability, although the resolution for time constant and frequency dependency is lower.

Our results show that multi-parameter inversion of IP-affected TEM data can effectively extract resistivity, chargeability, time constant, and frequency dependency parameters, particularly in conductive, high-polarization targets. For resistive, high-polarization targets, the inversion results are less accurate, highlighting the limitations of TEM data resolution for these structures. Nevertheless, the method still provides accurate parameter estimates for conductive zones, with high confidence in the target areas, making it a powerful tool for target identification and resource exploration.