- 274 views
- 1280 downloads
Full waveform inversion using simultaneous encoded sources based on first- and second-order optimization methods
-
- Author / Creator
- Anagaw, Amsalu Y.
-
Full waveform inversion (FWI) is an emerging seismic technology engine for estimating subsurface model parameters such as velocity, density and attenuation through local minimization of an objective function. The minimization problem is often performed iteratively via gradient-based method that is based on the first-order derivatives of the objective function. One of the major obstacles of FWI that hinders its practical applications for estimating subsurface model parameters for large scale problems is its huge computational demand for performing many forward modeling for multiple sources. One way to reduce the overall cost of waveform inversion is by adopting a simultaneous source strategy. In other words, multiple sources are simultaneously fired to simulate super-shot gathers and thereby reduce the number of seismic modeling simulations that are performed during the inversion. However, the use of simultaneous sources introduces cross-talk artifacts that arise from the interference among the sources that constitute a super-shot.
In this thesis, we study and extend the practical applications of the second-order optimization methods in the framework of simultaneous sources. First, we examined the effect of model parameterizations on velocity model building using Newton-based optimization methods. Three model parameterizations for the acoustic FWI were investigated. These include velocity, slowness and slowness squared. We then analyze the influence of different simultaneous multi-frequency selection strategies on cross-talk artifacts. Our analysis focuses on a multiscale frequency-domain FWI algorithm that is implemented with simultaneous sources that are randomly encoded with a source-encoding function. In the multiscale conventional FWI inversion strategies, a finite set of discrete frequencies will be selected and the inversion is carried out sequentially from low to high frequency data components. We examine six frequency selection strategies and test the performance of the algorithm with encoded datasets. Numerical tests show that high fidelity results can be attained by inverting partially overlapped groups of temporal frequencies. In order to mitigate cross-talk artifacts during the numerical inversion, a new encoding is generated at every iteration. We also found that high resolution images can be obtained by re-sampling new source positions and new encoding functions at every iteration of the FWI algorithm. Finally, we present a full Newton FWI algorithm and its application in the presence of high amplitude multi-reflected waves that are generated by strong velocity contrast and/or from free surface reflections.
-
- Subjects / Keywords
-
- Graduation date
- Fall 2014
-
- Type of Item
- Thesis
-
- Degree
- Doctor of Philosophy
-
- License
- This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.