Time-Lapse Full Waveform Inversion Methods

  • Author / Creator
    Alemie, Wubshet M
  • In the time-lapse seismic method, often referred to as 4D seismic, a series of seismic data sets are acquired at different times to study the temporal variation of a target subsurface reservoir. This technique is being used as a subsurface monitoring tool in the oil and gas industry for fluid driven production enhancement as well as for environmental monitoring purposes in CO2 sequestration. Assuming that the geological structures are time-invariant, the expected time-lapse changes are due to temporal variations in the reservoir. These changes, for instance, occur due to injection or depletion of fluid into or from the reservoir that creates a dynamic process. However, the time-lapse signature that can be obtained from time-lapse seismic data does not always reflect lithological changes. Time-lapse data are also liable to contamination by acquisition and processing artifacts. A major contributor to artifacts in the time-lapse signature is the non-repeatability of the seismic experiment. The qualitative interpretation that can be made by analyzing amplitude changes and time shifts on post-stack time-lapse seismic data are often not sufficient for a complete understanding of detailed reservoir conditions. Thus, to quantitatively characterize the dynamic of a reservoir, one should adopt pre-stack imaging and inversion algorithms that yield high-resolution images of physical parameters. Traditionally, time-lapse seismic inversion is carried out by processing baseline and monitor data-sets independently. Then, the difference between baseline and monitor surveys are computed. Artifacts that accumulate as a result of acquisition and data processing errors may often have a strength that is comparable to the actual time-lapse signal of interest. Consequently, artifacts can have negative impacts on the interpretation of time-lapse signals. Therefore, developing seismic inversion methods that are capable of suppressing the aforementioned artifacts is a problem of practical interest. In this thesis, the current full waveform inversion algorithms are reviewed. The performance of four optimization methods (gradient descent, approximate diagonal Hessian, Gauss-Newton, and Quasi-Newton) are assessed with a variety of numerical examples. The primary goal is to extend the traditional full waveform inversion algorithm into a time-lapse inversion algorithm. In particular, a new time-lapse full waveform inversion algorithm, namely, Joint Reparameterized Time-Lapse Full Waveform Inversion (JRTL FWI) is proposed in comparison with the traditional algorithms that include independent and double difference inversions. This algorithm allows passing a prior information from baseline to the monitor velocity models. Therefore, it plays a role of regularization. To demonstrate the JRTL FWI with synthetic examples, the seismic data are modeled via a 2D frequency domain acoustic wave equation with Perfectly Matched Layer (PML) boundary conditions. The inversion algorithm is derived using a least squares data misfit that is minimized via the Limited-memory Broyden-Fletcher-Goldfarb-Shanno (L-BFGS) optimization method. Two synthetic velocity models are considered in various time-lapse seismic monitoring settings. The examples show that when models have high complexity and the seismic experiments are highly non-repeatable, the proposed algorithm improves the quality of time-lapse velocity signature as compared to the traditional ones.

  • Subjects / Keywords
  • Graduation date
  • Type of Item
  • Degree
    Doctor of Philosophy
  • DOI
  • 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.
  • Language
  • Institution
    University of Alberta
  • Degree level
  • Department
  • Specialization
    • Geophysics
  • Supervisor / co-supervisor and their department(s)
  • Examining committee members and their departments
    • Van der Baan, Mirko (Physics)
    • Heimpel, Moritz (Physics)
    • Gu, Yu Jeffrey (Physics)
    • Innanen, Kristopher (Geoscience, University of Calgary)
    • Han, Bin (Mathematical and Statistical Sciences)