STOCHASTIC INVERSION OF 1-D MAGNETOTELLURIC DATA USING THE GLOBAL OPTIMIZATION METHOD, FLOWER POLLINATION ALGORITHM (FPA)

• Main Author: Raya Fadhillah, Muhammad
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/77960
• Institution: Institut Teknologi Bandung

The Magnetotelluric Method (MT) is one of the ill-posed geophysical methods. The stochastic inversion approach using the Flower Pollination Algorithm (FPA) is expected to be used to find solutions for MT inversion problems. FPA inversion was tested with synthetic models, synthetic models with noise, field data, and compared with the Levenberg-Marquardt algorithm (WinGLink by Schlumberger) and the Least Squares method (ZondMT1D by Kaminsky). The field data used were wide period bandwidth sounding data (0.18Hz-8192Hz). Inversion using synthetic data without noise (maximum error of 3.9%) and with noise resulted in good fitness. The results of the FPA inversion compared to applications show fairly good values (maximum error of 4.9%) and exhibit a model trend similar to the inversion results in the application. FPA inversion does not require a priori information to perform calculations. However, with the addition of such information, the program will run faster