Three-dimensional Multi-Scale and Multi-Method Inversion to Determine the Electrical Conductivity Distribution of the Subsurface Using Parallel Computing Architectures

Short title: Multi-EM

Geophysical methods are applied to investigate the Earth's interior. We obtain models of the Earth by imaging physical parameters such as density, electrical conductivity, or elastic properties using a range of techniques. Here, we consider geophysical techniques exploiting natural and impressed electromagnetic (EM) fields to reconstruct the electrical conductivity structure at depth. During the last years, these techniques have experienced rapid development for exploration purposes of the drillable subsurface. For instance, active controlled-source electromagnetic (CSEM) techniques are now frequently used together with seismic techniques to characterize resistive hydrocarbon reservoirs in offshore petroleum exploration. Deep saline aquifers exhibit high electrical conductivities and constitute one of the prime targets for electrical imaging methods, making these techniques one of the most important geophysical tools to characterize target horizons for CO₂ storage or geothermal reservoirs.

In this project, we attempt to optimize the resolution capabilities of geoelectric potential field and electromagnetic diffusion methods covering a wide range of scales from boreholes to regional or lithosphere dimensions. To reach these goals, we pursue an inter-disciplinary concept, integrating research groups from applied and numerical geophysics, information technology and numerical mathematics. The project partners come from university and non-university research institutions.

For tomographic imaging of a single physical parameter, the electrical conductivity, a number of well-proven techniques are available. In this project, we combine geoelectric methods (direct current - DC), transient electromagnetics (TEM), natural-source magnetotellurics (MT) and controlled-source magnetotellurics (CSMT). The resolution power of the individual methods depends on the experimental design, the strength, geometry, and frequency content of the source fields and the characteristics of the induced current systems. Multi-scale, multi-method inversion strategies yield complementary but usually higher sensitivities when compared with existing inversions of the individual methods. Hence, a combination of the various sensitivity patterns is expected to result in (i) a better coverage of the model space, (ii) a more complete and better resolved reconstruction of the conductivity structure, and (iii) a reduction of model ambiguities. An effective implementation of multi-method, single-parameter inversions is a prerequisite for subsequent development of multi-method, multi-parameter inversions (e.g. a joint inversion of MT and seismic).

Realistic predictions of structures and materials at depth require three-dimensional modelling. The underlying differential equations can only be solved with numerical approaches, usually using finite difference (FD), finite element (FE) or integral equation (IE) techniques. The subsurface and the physical fields are discretized on grids, which in the three-dimensional domain easily consist of millions of individual cells. The number of these grid cells and associated computations determines the requirements for memory and CPU time. However, a comprehensive image of the subsurface is only obtained with sufficiently fine discretization on a correspondingly large mesh. Solving these complex equation systems requires extremely powerful computers and highly optimized algorithms, which are often beyond the realms of individual working groups.

Considering the enormous numerical complexity of "normal" single-method three-dimensional inversions, the new algorithms will be designed from the beginning for massively parallel computing architectures. Initially, local computer clusters will serve as development environments, but integration with D-GRID or similar international structures is envisaged. The overall goal is to transfer the newly gathered knowledge and infrastructure to the wider geosciences community.

Here we show short movie presenting visualization of current research topics within Multi-EM. We reccomend to watch this movie in HD (720p) quality.

 

This project is funded by the German Ministry of Education and Research (BMBF) within the GEOTECHNOLOGIEN Programme.