COMPARISON OF SINGULAR VALUE DECOMPOSITION, GAUSS-JORDAN, AND REGULAR INVERS METHOD IN 1D INVERSION MODELING MAGNETOTELLURIC METHOD

• Main Author: Hasbi Ashshiddiqi, Ighfar
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/60910
• Institution: Institut Teknologi Bandung

The magnetotelluric (MT) method is a passive electromagnetic method that utilizes the natural electric field E and magnetic fields B that induce the material beneath the earth's surface. This method is used to obtain information on the subsurface resistivity structure of the earth with a depth up to hundreds of kilometers. The MT method has a higher sensitivity in detecting conductive coatings. In the MT data inversion process, it is necessary to have a perturbation in the calculation data and the perturbation must be a function that is minimized. This perturbation contains a Jacobian matrices and the Jacobian matrices must have an inverse value. There are several methods of determining the inverse of a matrices which are partly used in this final project, namely ordinary inverse, singular value decomposition, and Gauss-Jordan. Things will be done in the three methods is how effective it is in calculating the inverse matrices with variations in the earth model and variations in the initial guess value. The benchmark for effectiveness is the graph between the root mean square (RMS) value and the resulting iteration. The fewer iterations produced to get to a certain convergent value, the more effective it is.