DEVELOPMENT AND IMPLEMENTATION OF SOURCE INDEPENDENT FULL WAVEFORM INVERSION (SI-FWI): A SYNTHETIC CASE STUDY

• Main Author: Diva Pratama, Muhamad
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/77767
• Institution: Institut Teknologi Bandung

Source Independent Full Waveform Inversion (SI-FWI) is a seismic imaging method that uses the entire seismic waveform contained within the seismogram. It addresses the challenge of dependency on wavelet sources during the velocity model reconstruction process. In SI-FWI, inaccuracies or errors in wavelet source information are mitigated by modifying the calculation of misfit and adjoint sources used in the adjoint modeling process. This modification involves convolving the seismic trace between the observed and modeled seismogram data, without altering the fundamental FWI algorithm. Consequently, the SI-FWI method becomes less reliant on wavelet sources. In this study, SI-FWI was implemented and tested on a synthetic acoustic and isotropic overthrust model. The iteration process was conducted through optimization using the conjugate gradient method and a line search algorithm, which facilitated faster and more effective convergence. The program was developed in Python and executed parallel and distributed processing to expedite program execution. Two scenarios were considered in this study, each utilizing different wavelet sources. The results obtained in both cases have reduced the misfit value significantly. In both instances, the resulting models closely matched the actual model. This was further confirmed by the 1D velocity profile analysis at different locations, which closely approximated the velocity profile of the actual model. Based on these results, this study demonstrates the capability of the SI-FWI program to eliminate the effect of wavelet sources, removing its dependence on the suitability of wavelet sources within the SI-FWI framework.