MAGNETOTELLURIC DATA PROCESSING BASED ON MESTIMATION AND BOUNDED INFLUENCE

• Main Author: Rusdiana, Randi
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/67367
• Institution: Institut Teknologi Bandung

Data processing is an important part of the Magnetotelluric method. The data processing process is crucial in determining the quality level of the resulting subsurface structure model. MT data measured in the field is affected by noise which causes low data quality and less accuracy. Getting good data quality requires a long stage. In addition, MT is a relatively new method compared to other geophysical methods. Therefore, the data processing device is still inadequate and being developed. In this study, MT data processing was done using Python starting from time series data to obtain apparent resistivity and phase curves. The coherence values can know the improvement of MT data quality of electric and magnetic fields. However, the coherence value is no longer a reference at low frequencies. In some data, the increase in the coherence value is not proportional to the change in the apparent resistivity curve and the smoother phase at low frequencies. Therefore, in this study, statistical weighting using the MEstimation and Bounded Influence methods is proposed to improve the quality of MT data. The results showed an increase in data quality ranging from 15.35% to 25%, with an average of 19.95% between raw data and robust processing. The apparent resistivity and phase curves with weighted M-Estimation and Bounded Influence are better than Robust Ordinary Least Square, especially at low frequencies. The M-Estimation weighting tends to show the trend of the change in the resistivity value variation curve at low frequencies, which is smoother than the Bounded Influence and Ordinary Least Square weighting. In the MT Deadband zone around 0.1 Hz, the Bounded Influence method shows a smoother curve than the M-Estimation.