#TITLE_ALTERNATIVE#

• Main Author: ENGGELIA P. S, DIAN
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/21720
• Institution: Institut Teknologi Bandung

Basis pursuit inversion (BPI) is a new inversion method that can be used to reconstruct the subsurface layer with better resolution. This method assumed that subsurface layer is comprised by a number of sparse layers with different thicknesses, within which each layer can be decomposed by a sum of odd and even pairs of dipole reflectivity. Seismic trace is therefore assumed as a convolution between a wavelet and a sequence of odd and even pairs of dipole reflectivity with different thicknesses. If the best-fit wavelets can be found, then basis pursuit inversion can be explained as process to determine the coefficient of dipole reflectivity, which can reconstruct seismic trace appropriately. <br /> <br /> <br /> <br /> <br /> <br /> <br /> BPI is accomplished by building a dictionary of functions representing a sequence of odd and even pairs of dipole reflectivity with different thicknesses, varies from minimum sample rate until maximum as thick as tuning thickness. This dictionary is therefore known as wedge dictionary. Convolution between wedge dictionary and a wavelet can produce over-complete kernel matrix (over-complete dictionary), because it consist all possibility of odd and even pairs of dipole reflectivity with different thicknesses, which is used as basis to reconstruct the seismic trace. The inverse problem of under-determined dictionary can be solved by using basis pursuit method. In this final project, BPI method is applied to do the inversion of Elastic Impedance (EI). BPI has ability to deliver better result than SSI. This is affected by the regularization parameter (lambda), which contributes as smoothing factor to get sparse layer that equally fit with well logs.