ALBERT

Description: Multi-dimensional inversion of Transient electromagnetic data is a computationally expensive task. Only few developments and practical interpretation tools exist. Here, we present a multidimensional inversion framework for loop source time-domain electromagnetic data. The developed algorithm is a robust, efficient, and user-oriented tool for the multi-dimensional inversion of typical loop source time-domain electromagnetic configurations. A time-domain finite volume discretization and the direct solver MUMPS are utilized to solve the 3D TEM forward problem. An iterative Gauss-Newton optimization method is implemented for the inversion kernel. The code is parallelized for calculating multiple sources simultaneously to accelerate the inversion. Based on exploration tasks, different configurations exist for commonly used loop source TEM configurations and typical field scales. Synthetic examples are used to verify the effectiveness and benchmark the developed 3D algorithm. Considering that TEM data is often gathered along profiles, adjusting the model roughness along the different modeling domain directions, sufficiently constrains to allow for 2D imaging. In addition to the vertical signal components, we also included horizontal components for large scale fixed loop applications. Subsequent to synthetic validation, the inversion algorithm is further verified using ~120 dense TEM soundings collected over a clay pan site in the Atacama Desert, Chile, to provide bedrock geometry information and suitable coring sites. The 3D inversion result provided an excellent depth estimate of sedimentary infill as well as the bedrock topography and was later confirmed by deep coring. Another interesting site is the Roter Kamm impact crater in Namibia. Our preliminary results obtained from largescale multicomponent fixed loop TEM data reveal a sedimentary infill down to ~300 m depth. In conclusion, our presented 3D inversion code is capable to handle data from various exploration scenarios and provides a robust tool for advanced EM interpretation.

Description: In petrophysics, physical rock properties are typically established through laboratory measurements of individual samples. These measurements predominantly relate to the specific sample and can be challenging to associate with the rock as a whole since the physical attributes are heavily reliant on the microstructure, which can vary significantly in different areas. Thus, the obtained values have limited applicability to the entirety of the original rock mass. To examine the dependence of petrophysical measurements based on the variable microstructure, we generate sets of random microstructure representations for a sample, taking into account macroscopic parameters such as porosity and mean grain size. We show that the methodology can adequately mimic the physical behavior of real rocks, showing consistent emulation of the dependence of electrical conductivity on connected porosity according to Archie's law across different types of pore space (micro-fracture, inter-granular, and vuggy, oomoldic pore space). Furthermore, properties such as the internal surface area and its fractal dimension as well as the electrical tortuosity are accessible for the random microstructures and show reasonable behavior. Finally, the possibilities, challenges and meshing strategies for extending the methodology to 3D microstructures are discussed.

Description: This study investigates the optimization of Semi-Airborne Electromagnetic (SAEM) surveys for enhanced subsurface imaging in mineral exploration. It highlights the utility of multi-transmitter systems and explores real data utilization and the challenges of large-scale surveys. With emphasis on Data obtained from DESMEX project surveys. The use of multiple transmitters is crucial. Single transmitters can distort results and mask subsequent bodies. Employing two transmitters on both sides of the target enhances resolution and depth accuracy. results are based on finite element forward operator custEM and pyGIMLi’s inverse solver [1]. substantial advantages of combining single and multi-patch inversion data. This integration results in improved resolution, reduced artifacts, enhanced continuity of geological structures, superior anomaly detection, minimized edge effects, and improved depth penetration [2]. These findings open promising avenues for further exploration and research in geosciences, offering valuable insights into the Earth's subsurface and its intricate geological features. The next logical step involves expanding our methodology to large-scale inversion using more than three transmitters.

Description: We present a finite element software library written in Matlab for the numerical simulation and inversion of electromagnetic fields in two and three dimensions. It is designed in a modular way to easily plug together fundamental building blocks for various electromagnetic applications from DC to the inductive range in the frequency and even time domain. External modules comprise the mesh generator and the equation solver library. Through its homogeneous software concept the adoption to any field application is relatively simple and makes the code suitable to open source distribution. We introduce the key features of this library including higher-order Lagrange and Nédélec finite elements formulated on unstructured tetrahedral grids, a Gauss- Newton inversion approach using linear Raviart-Thomas elements for H1 regularization, and the ability to incorporate any geometric feature such as topography, bathymetry and internal voids like caves, tunnels and mine buildings. The library is currently being tested with large real data sets to confirm its usefulness as a tool for practical data interpretation. Therefore, case studies for the magnetotelluric, direct current resistivity, controlled source electromagnetic and induced polarization methods in the field and laboratory are briefly outlined as examples with challenging geometric features.