PENENTUAN MODEL KECEPATA UNTUK PRE-STACK DEPTH MIGRATION MENGGUNAKAN METODE FULL WAVEFORM INVERSION (FWI)

• Main Author: Ardana, Andriana
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/84800
• Institution: Institut Teknologi Bandung

Seismic migration is an important step in the seismic data processing workflow to correct reflectors into their actual position. Based on seismic cross section result, seismic migration is divided into two domains namely Time-Migration and Depth-Migration. Conventional depth-migration methods generally use interval velocity models derived from RMS or stacking velocity using the Dix transformation. This interval velocity model often contains poor velocities resulting inaccurate depth migration results, especially in complex geological models. This study uses velocity model resulting from Full-Waveform Inversion (FWI). FWI can produce high resolution velocity models, that can be used as input for depth migration. In addition, the depth-migration stage can also evaluate the accuracy of the FWI velocity model based on the seismic cross-section result. FWI is a seismic imaging method that reconstructs velocity of the medium by minimizing misfit between observed and simulated seismograms. FWI is carried out in several stages. Firstly, forward modelling of wave propagation using an initial velocity model from a source to a receiver. Secondly, estimating residual waveform by comparing modelled and recorded data. Thirdly, carrying out an iterative process of minimizing residual data by updating the initial model with adjoint state and optimization methods. The FWI method is implemented using the Python programming language. The FWI workflow is applied to the Overthrust synthetic velocity model. This reasearch conduct Reverse Time Migration (RTM) as depth-migration algorithm. The results shows that RTM using input good FWI velocity model is more accurate for imaging complex geological structures than conventional model. Furthermore result show that smooth velocity model input provide more accurate RTM image.