Challenges with starting models, correlation and convergence for inversion of IP-affected TEM data: examples from African countries

Transient electromagnetics (TEM) is a geophysical method well suited for hydrogeological investigations, offering extensive spatial coverage and the ability to resolve hydrogeological units in the order of tens of meters across diverse terrains. However, in certain environments – such as sulfide mineral deposits, clay-rich materials or permafrost - the TEM signal can be affected by Induced Polarization (IP), a phenomenon that introduces perturbations and limits the reliability of traditional inversion workflows. TEM data affected by Induced Polarization (IP), collected using a towed TEM system across four Sub-Saharan African countries, illustrate the challenges associated with IP effects. Inversion of the TEM data using the Maximum Phase Angle (MPA) re-parameterization reveals low convergence rates during inversion, with strong dependence on the choice of starting model. The resulting models show limited sensitivity to MPA parameters (ϕ_max,τ_φ,C), in contrast to resistivity, which exhibits comparatively lower uncertainty. Integration of borehole or other ground-truthing data proved essential for reducing ambiguity in IP-affected TEM data interpretation. These findings underscore the limitations of current inversion approaches under IP conditions and highlight the need for improved parameterizations that yield reliable constraints on IP behavior.