TWO DIMENSIONAL MAGNETOTELLURIC INVERSION USING FINITE ELEMENT METHOD

Magnetotelluric method is a geophysical method that utilize natural electromagnetic source field to map electrical resistivity structure of earth subsurface. In order to obtain subsurface electrical resistivity profile from magnetotelluric data, inversion method can be used. Inversion problems in ma...

Full description

Saved in:
Bibliographic Details
Main Author: (NIM : 20215051), TIFFANY
Format: Theses
Language:Indonesia
Online Access:https://digilib.itb.ac.id/gdl/view/24530
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Institut Teknologi Bandung
Language: Indonesia
Description
Summary:Magnetotelluric method is a geophysical method that utilize natural electromagnetic source field to map electrical resistivity structure of earth subsurface. In order to obtain subsurface electrical resistivity profile from magnetotelluric data, inversion method can be used. Inversion problems in magnetotelluric are usually ill-posed, where the inversion solusions could change significantly with small change in data. This leads to incorrect and inaccurate solutions of inversion process. Singular value decomposition is a method which is able to stabilize inversion process in ill-posed problems. In this research, we perform inversion for 2D transverse electric problem with vector based finite element method modelling program. This forward modelling program is chosen because the vector based finite element method scheme could models electromagnetic fields accurately with the solutions satisfying physical concepts of electromagnetic fields. Inversion performed with truncated singular value technique, where small singular values and the ones which are close to zeros are omitted and not used to inverse Jacobian matrix. We test the inversion program using homogenous half-space model, layered earth model with two and three layers, vertical contact model and resistive and conductive anomaly model. We also test the inversion for several variation of singular values. Truncation of too large singular values also producing inaccurate result , where the corresponding data which should contribute to the model are omitted.