Surface geometry inversion of potential field and electromagnetic geophysical data
Tuesday, 15 December 2020: 16:16
Virtual
Xushan Lu, Memorial University of Newfoundland, St John's, Canada, Christopher Galley, Memorial University of Newfoundland, St John's, NL, Canada, Peter G Lelievre, Mount Allison University, Department of Mathematics and Computer Science, Sackville, NB, Canada and Colin Farquharson, Memorial University of Newfoundland, Department of Earth Sciences, St John's, Canada
Surface geometry inversion (SGI) methods directly invert for the contact surface between different rock units, making it consistent with geologists’ typical interpretations of the Earth. SGI methods parameterize the subsurface Earth model in terms of wireframe surfaces, representing the interfaces between rock units, and solve for the coordinates of the facet vertices of these surfaces. To reduce the non-uniqueness of the inversion, physical properties of each unit are held at a priori values during the inversion. Our objective function consists of a single data misfit term, although extra regularization terms can be added to reduce the number of acceptable models. The number of data is easily larger than the number of parameters in our SGI method, making the inverse problem overdetermined. The objective function is minimized using the genetic algorithm (GA) global optimization method. After the minimum solution is found, a Markov chain Monte Carlo (MCMC) stochastic sampling is utilized to provide the mean and standard deviations of the surface vertex positions.
We applied our SGI methods to both potential field data and EM data inversions. Gravity and magnetic field data collected at the seafloor were jointly inverted to develop wireframe models of seafloor massive sulphide deposits. The gravity data provided a means to model the higher density massive sulphide ore, whereas the magnetic data allowed the intense hydrothermal alteration around the deposit to be resolved.
Synthetic examples based on thin graphitic fault conductors in the Athabasca Basin, Canada, were used to test our SGI methods on transient electromagnetic (TEM) data. We used a deformed wireframe surface 3D model as the initial model for the conductor and the inversion manipulated the coordinates of the surface vertices. The vertices were then duplicated with a fixed distance along the normal direction of the surface. An intermediate tetrahedral mesh was used to calculate the TEM responses. Compared to the conventional minimum-structure inversion, our SGI methods are able to provide models that better resemble realistic geological structures.
