Geostatistical joint inversion of frequency-domain electromagnetic data and direct current resistivity data for near-surface modelling

Geophysical methods, such as electrical resistivity tomography (ERT) and frequency-domain electromagnetic (FDEM) induction, have been widely used for imaging and modelling the first meters of the Earth, in fields like agriculture, urban development, resources exploration. These methods are sensitive to subsurface electrical conductivity (EC), which can be estimated through data inversion. However, the different spatial resolutions of both methods, along with the inherent nonlinearity of these geophysical inverse problems, make the joint inversion challenging and, therefore, the individual inversion of each data type remains the standard. This study presents an iterative geostatistical joint inversion approach that integrates FDEM and ERT data to increase the accuracy in modelling small-scale spatial heterogeneities that typically characterizes near-surface environments. The method uses geostatistical simulation and co-simulation as stochastic model perturbation and update techniques to couple the data domains in a consistent spatial model. The inversion is guided by the simultaneous reduction of misfit between predicted and observed FDEM and ERT data. We validate the method in a synthetic data set that illustrates a complex and highly heterogeneous near-surface environment, and apply it to real field data from a heterogeneous and high-conductivity site. The proposed joint inversion method improves the resolution of small-scale heterogeneities and reduces uncertainty at depth, outperforming individual inversion methods in both synthetic and real case applications.