On recovering distributed IP information from inductive source time domain electromagnetic data

Abstract

We develop a procedure to invert time domain induced polarization (IP) data for inductive sources. Our approach is based upon the inversion methodology in conventional electrical IP (EIP), which uses a sensitivity function that is independent of time. However, significant modifications are required for inductive source IP (ISIP) because electric fields in the ground do not achieve a steady state. The time-history for these fields needs to be evaluated and then used to define approximate IP currents. The resultant data, either a magnetic field or its derivative, are evaluated through the Biot-Savart law. This forms the desired linear relationship between data and pseudo-chargeability. Our inversion procedure has three steps: (1) Obtain a 3-D background conductivity model. We advocate, where possible, that this be obtained by inverting early-time data that do not suffer significantly from IP effects. (2) Decouple IP responses embedded in the observations by forward modelling the TEM data due to a background conductivity and subtracting these from the observations. (3) Use the linearized sensitivity function to invert data at each time channel and recover pseudo-chargeability. Post-interpretation of the recovered pseudo-chargeabilities at multiple times allows recovery of intrinsic Cole-Cole parameters such as time constant and chargeability. The procedure is applicable to all inductive source survey geometries but we focus upon airborne time domain EM (ATEM) data with a coincident-loop configuration because of the distinctive negative IP signal that is observed over a chargeable body. Several assumptions are adopted to generate our linearized modelling but we systematically test the capability and accuracy of the linearization for ISIP responses arising from different conductivity structures. On test examples we show: (1) our decoupling procedure enhances the ability to extract information about existence and location of chargeable targets directly from the data maps; (2) the horizontal location of a target body can be well recovered through inversion; (3) the overall geometry of a target body might be recovered but for ATEM data a depth weighting is required in the inversion; (4) we can recover estimates of intrinsic τ and η that may be useful for distinguishing between two chargeable targets.

3D electromagnetic modelling and inversion: a case for open source

“…SimPEG is an open-source framework and set of tools for simulation and gradient-based parameter estimation in geophysics. It includes finite volume simulations and inversion routines for a variety of geophysical applications including potential fields, vadose zone flow, DC resistivity, induced polarization, self-potential and electromagnetics (Cockett et al, 2015;Kang and Oldenburg, 2016;Heagy et al, 2017;Miller et al, 2017;Cockett et al, 2018;Miller et al, 2018;Witter et al, 2018). Simulations may be performed on several different mesh types, including cylindrically symmetric meshes, 3D tensor meshes and OcTree meshes.…”

Inversion of airborne geophysics over the DO-27/DO-18 kimberlites — Part 3: Induced polarization

“…This clearly shows a natural separation of EM and IP responses in time: EM induction is usually dominant at early times, but IP effects can be dominant at late times when EM effects have significantly decayed. This separation occurs because the buildup of polarization charges, which occurs when an electric field is applied, takes time (Smith and West, 1989;Macnae, 2015;Kang and Oldenburg, 2016). The separation of TEM signals into EM-dominant, IP-dominant, and intermediate zones will be a foundation of the following TEM-IP inversion workflow that we use to extract the conductivity and chargeability information from the VTEM data.…”

“…To extract conductivity and IP information from the VTEM data, we follow the TEM-IP inversion workflow developed by Kang and Oldenburg (2016). This workflow includes four steps: (1) invert TEM data, and recover an estimated conductivity model σ est .…”

“…Because w e ðtÞ is a dimensionless property, the unit of the pseudochargeability will be s −1 . The computations of the sensitivity matrix and the time history of electric field for an airborne TEM problem are somewhat involved, but the detailed theoretical background for these computations is explained in Kang and Oldenburg (2016). The sensitivity matrix J is generated using σ est .…”

“…Here, in part 3, we concentrate upon extracting information from the IP signal in the VTEM data set. We apply our most recent TEM-IP inversion workflow (Kang and Oldenburg, 2016) and attempt to characterize the various kimberlite units based on their chargeability properties. Distinguishing between the three kimberlite units is crucial to get the most representative view of the TKC kimberlite pipes.…”