Use this URL to cite or link to this record in EThOS:
Title: Practical solutions for seismic free-surface and internal multiple attenuation based on inversion
Author: Zhang, Nan
ISNI:       0000 0004 2732 4849
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2013
Availability of Full Text:
Access from EThOS:
Access from Institution:
Multiple prediction through inversion (MPI) is an effective method for seismic multiple attenuation. The research in this thesis aims to make the MPI method more practical for both free-surface and internal multiple attenuation. For free-surface multiple attenuation, the MPI scheme requires the input data to be dense and regularly sampled, and with one shot at each receiver position. In order to meet these requirements, I use a multilevel B-spline method for seismic data reconstruction. This method can perform regularisation and interpolation on seismic data without any prior-knowledge of models. For free-surface multiple attenuation on marine data, MPI can generate superior results compared to SRME (surface-related multiple attenuation). However, MPI is more computationally expensive due to the large amount of matrix operations involved. The conventional implementation addresses this by approximating the multiple model prediction operator as a pentadiagonal or a tridiagonal matrix. Tackle this problem by solving the full prediction operator using a Graphic Processing Unit (GPU), this accelerates the processing and improve the multiple attenuation results, especially for far-offset traces. As extensions of SRME for internal multiple attenuation, both the CFP (common-focus-point) technique and correlation method have problems. The results can be improved using the MPI method with GPU acceleration. The correlation method is preferred as the initial step for MPI because it can be implemented as a fully data-driven pre-stack domain approach in either forward data space or inverse data space. In all cases, the MPI scheme generates internal multiple models with improved kinematic and dynamic accuracy.
Supervisor: Rao, Ying ; Wang, Yanghua Sponsor: Imperial College London
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral