GP13A-1267
Multi-dimensional Magnetotelluric Modeling of General Anisotropy and Its Implication for Structural Interpretation

Monday, 14 December 2015
Poster Hall (Moscone South)
Zeqiu Guo1,2, Wenbo Wei3 and Gary D Egbert2, (1)China University of Geosciences Beijing, Beijing, China, (2)Oregon State University, Corvallis, OR, United States, (3)China Univ. of Geosciences, Beijing, China
Abstract:
Although electrical anisotropy is likely at various scales in the Earth, present 3D inversion codes only allow for isotropic models. In fact, any effects of anisotropy present in any real data can always be accommodated by (possibly fine scale) isotropic structures. This suggests that some complex structures found in 3D inverse solutions (e.g., alternating elongate conductive and resistive "streaks" of Meqbel et al. (2014)), may actually represent anisotropic layers. As a step towards better understanding how anisotropy is manifest in 3D inverse models, and to better incorporate anisotropy in 3D MT interpretations, we have implemented new 1D, 2D AND 3D forward modeling codes which allow for general anisotropy and are implemented in matlab using an object oriented (OO) approach. The 1D code is used primarily to provide boundary conditions (BCs). For the 2D case we have used the OO approach to quickly develop and compare several variants including different formulations (three coupled electric field components, one electric and one magnetic component coupled) and different discretizations (staggered and fixed grids). The 3D case is implemented in integral form on a staggered grid, using either 1D or 2D BC. Iterative solvers, including divergence correction, allow solution for large model grids. As an initial application of these codes we are conducting synthetic inversion tests. We construct test models by replacing streaky conductivity layers, as found at the top of the mantle in the EarthScope models of Meqbel et al. (2014), with simpler smoothly varying anisotropic layers. The modeling process is iterated to obtain a reasonable match to actual data. Synthetic data generated from these 3D anisotropic models can then be inverted with a 3D code (ModEM) and compared to the inversions obtained with actual data. Results will be assessed, taking into account the diffusive nature of EM imaging, to better understand how actual anisotropy is mapped to structure by 3D isotropic inversion.