Search

In this article we propose a workflow for creating geologically realistic depth to basement maps and apply it to the undercover area of Cloncurry, located in the Mount Isa province, Queensland. A magnetotelluric (MT) survey consisting of 457 soundings was inverted using a 1D trans‐dimensional Bayesian algorithm designed to be robust to non‐1D effects present in the data. Automated change‐point analysis was then used to derive probability distributions on basement depth for each site independently. These were laterally combined, along with drill hole data and a structural model derived from aeromagnetic and geological data, using a Bayesian estimate fusion algorithm to create a region wide depth to basement probabilistic map. Combining many different constraints in this way lead to a significant reduction in posterior uncertainty. Individual MT site change‐point posteriors were highly multi‐modal in certain areas, with multiple resistivity transitions that could correspond to the cover‐basement interface. The estimate fusion process correlates these uncertainties and the combined posterior was thus much less multi‐modal. Our results show that the sedimentary cover gradually thickens toward the north, while toward the east its thickening is controlled by a two‐steps fault system. This workflow highlights the value gained from integrating different types of geoscientific data but also shows the capability of the MT method used within a probabilistic workflow to accurately image depth to basement even using limited constraints and assumptions. Plain Language Summary: Within regions covered by sediments, defining the basement geometry underneath is essential to understand the distribution of mineral and groundwater resources. This study introduces a methodology and applies it to characterize the buried basement geometry in the Cloncurry region (90 km × 60 km area) of Queensland, which is a well‐known metallogenic province in Australia. The methodology we present uses independent estimates of depth to basement derived using data from different sources (electromagnetic geophysics and drill holes) and geological constraints on it, all providing complementary information, at many locations over the entire area. This produces a map of depth to basement, alongside a quantitative estimation of its reliability, which is based on uncertainty analysis of the various data inputs and their combination. This workflow produced a geologically realistic probability image which highlights the value of our workflow to integrate different types of data in a flexible and robust manner. Key Points: Magnetotelluric data used within a probabilistic workflow is a reliable tool to derive a realistic depth to basement probabilistic mapThe uncertainty of the probabilistic map derived can be significantly reduced by combining many different probabilistic constraintsThe semi‐automated probabilistic workflow we present is mainly data‐driven and relies on limited expert assumptions [ABSTRACT FROM AUTHOR]

Bayesian inversion of magnetotelluric (MT) data is a powerful but computationally expensive approach to estimate the subsurface electrical conductivity distribution and associated uncertainty. Approximating the Earth subsurface with 1-D physics considerably speeds-up calculation of the forward problem, making the Bayesian approach tractable, but can lead to biased results when the assumption is violated. We propose a methodology to quantitatively compensate for the bias caused by the 1-D Earth assumption within a 1-D trans-dimensional Markov chain Monte Carlo sampler. Our approach determines site-specific likelihood functions which are calculated using a dimensionality discrepancy error model derived by a machine learning algorithm trained on a set of synthetic 3-D conductivity training images. This is achieved by exploiting known geometrical dimensional properties of the MT phase tensor. A complex synthetic model which mimics a sedimentary basin environment is used to illustrate the ability of our workflow to reliably estimate uncertainty in the inversion results, even in presence of strong 2-D and 3-D effects. Using this dimensionality discrepancy error model we demonstrate that on this synthetic data set the use of our workflow performs better in 80 per cent of the cases compared to the existing practice of using constant errors. Finally, our workflow is benchmarked against real data acquired in Queensland, Australia, and shows its ability to detect the depth to basement accurately. [ABSTRACT FROM AUTHOR]