PEMODELAN INVERSI SATU-DIMENSI DATA

• Main Author: Akmam
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/4566
• Institution: Institut Teknologi Bandung

<b>Abstract</b><p align="justify"> <br /> Interpretation of earth electricity fields can often be assisted by inversion methods. In the magnetotelluric case the method used to find physical parameters of the earth, (the physical parameters are the resistivities layer and the thickness of layer). The inversion of magnetotelluric data does not yield a unique solusion, but inevitably a single model to interpret the observations is sought. Another problem is the small change ini the observation data can cause the large change in the result of inversion parameters and cause a large number of iteration. In other word, the parameters result of inversion depends on the value of estimate parameters. For that, the inversion methods which the parameters of result does not depend on estimate parameters are necessary.<p align="justify"> <br /> To solve the problem, the iterarive inversion method is developed through analysis problem based on lokal linearization. Iteration steps are optimalized through _ parameterized problems to the thickness of layer, and the non-linear forward problems is linearized in the estimate model of the Occam's inversion. In this case, the truncation singuler value of the Marquardt iteration method is used. In addition, the coresponding Jacobian matrix is corrected using the Occam correction method. In the practice is assumed that Gaussian distribution noise.<p align="justify"> <br /> The porformance of the proposed inversion method is tested using synthetic data and also noisy data. The test shows that; First, this inversion method can predict a well conductive layers without depend on estimate parameters. Second, there is not <br /> ' significant different between the synthetic data and the results of inversion, if the apparent "Fesistivities did not add by noise.<p align="justify"> <br /> Disadvantage of this method is when the initial istimate is too far from the real parameters that need a large number of iteration and senstive to ploting point error when runnin programe take place.