Zhou, Hua-Wei2018-02-152018-02-15December 22015-12December 2http://hdl.handle.net/10657/2104Accurate mapping of subsurface structure through seismic techniques is essential in oil and gas exploration. With the development of computational power, there has been an increased focus on data-fitting related seismic-inversion techniques for producing high fidelity seismic velocity models and images, such as full-waveform inversion and least-squares migration. However, more advanced methods, such as data-fitting techniques, are generally formulated in least-squares optimization, and can be less robust and expensive in terms of computational cost. The nonlinearity of inversion problems also pose another issue for successful mapping of subsurface structure. Recently, various techniques to optimize data-fitting seismic-inversion problems have been implemented for the industrial need to better efficiency. The primary objective of this study is to optimize least-squares techniques for seismic-velocity model building and imaging. This work can be divided into three equally important parts. The first part of this work is developing a new multi-level temporal integration to make full-waveform inversion (FWI) more robust than its classic implementation. The second contribution is to maximize the capability of the least-squares migration through numerical optimization and Hessian preconditioning. The third part is to account for the large amplitude differences between field and modeled data. A new local normalization scheme is proposed for better performance of the least-squares migration. The field examples demonstrate the effectiveness of the proposed methods in generating high quality images and improving the inversion efficiency.application/pdfengThe author of this work is the copyright owner. UH Libraries and the Texas Digital Library have their permission to store and provide access to this work. Further transmission, reproduction, or presentation of this work is prohibited except with permission of the author(s).Least-squaresLeast-squares reverse-time migrationFull-waveform inversionStochastic conjugate gradient2018-02-15Thesisborn digital