Gravity and Magnetic Modeling using Monte Carlo Method.

Nacim Foudil Bey and Jean-Jacques Royer and Cheng Li Zhen.. ( 2009 )
in: Proc. 29th Gocad Meeting, Nancy

Abstract

Tetrahedral grids are very successful to describe subsurface features in geosciences modeling. In particular, they provide unprecedented accuracy tool for modeling heterogeneous complex geological structures; in addition, this approach requires fewer elements compared to conventional prismatic or Cartesian grids. In this work, tetrahedral grids are used to describe 3D geological structure and to calculate potential field responses. The gravity field is calculated assuming that each elementary cell mass is reduced to a single point at its center, while the elementary tetrahedral cell is reduced to a magnetic dipole to estimate the magnetic field components. Consequently, these approximations are valid and can be used as a first approximation when the distance between the observed point at the surface and the tetrahedron center is much larger than the cell dimensions. In order to obtain more accurate results, the near surface coarse cells are then further subdivided recursively into 8 smaller tetrahedral cells. This process is performed automatically and recursively for each tetrahedron as long as the previous distance condition is not respected. This technique is not only applied to near surface coarse cells, but also to anomalous zones such as faults, favorable mineralization zones, etc., where there is possibility of sudden variations in the physical property and thus of the structure. The investigated inverse method implemented in a Gocad plug-in is based on the Monte Carlo sampling (random sampling) approach which simulates 3D densities and/or magnetic susceptibility of the model distributions. This method combines to the prior information of the magnetic susceptibility and/or densities masses distributions with measurement data is used in order to determine the posterior probability of the model. The Metropolis algorithm consisting of sampling close to the previous model is also implemented for simulations.

Download / Links

BibTeX Reference

@inproceedings{FoudilBeyGM2009,
 abstract = { Tetrahedral grids are very successful to describe subsurface features in geosciences modeling. In particular, they provide unprecedented accuracy tool for modeling heterogeneous complex geological structures; in addition, this approach requires fewer elements compared to conventional prismatic or Cartesian grids. In this work, tetrahedral grids are used to describe 3D geological structure and to calculate potential field responses. The gravity field is calculated assuming that each elementary cell mass is reduced to a single point at its center, while the elementary tetrahedral cell is reduced to a magnetic dipole to estimate the magnetic field components. Consequently, these approximations are valid and can be used as a first approximation when the distance between the observed point at the surface and the tetrahedron center is much larger than the cell dimensions. In order to obtain more accurate results, the near surface coarse cells are then further subdivided recursively into 8 smaller tetrahedral cells. This process is performed automatically and recursively for each tetrahedron as long as the previous distance condition is not respected. This technique is not only applied to near surface coarse cells, but also to anomalous zones such as faults, favorable mineralization zones, etc., where there is possibility of sudden variations in the physical property and thus of the structure. The investigated inverse method implemented in a Gocad plug-in is based on the Monte Carlo sampling (random sampling) approach which simulates 3D densities and/or magnetic susceptibility of the model distributions. This method combines to the prior information of the magnetic susceptibility and/or densities masses distributions with measurement data is used in order to determine the posterior probability of the model. The Metropolis algorithm consisting of sampling close to the previous model is also implemented for simulations. },
 author = { Foudil Bey, Nacim AND Royer, Jean-Jacques AND Zhen., Cheng Li },
 booktitle = { Proc. 29th Gocad Meeting, Nancy },
 title = { Gravity and Magnetic Modeling using Monte Carlo Method. },
 year = { 2009 }
}