Efficient Regularization Technique for Full Waveform Inversion Using Selective Smoothing Coefficients
Y. Park and S. Pyun
Event name: 76th EAGE Conference and Exhibition 2014
Session: FWI - Theoretical and Numerical Advancements
Publication date: 16 June 2014
Info: Extended abstract, PDF ( 1.11Mb )
Price: € 20
A smoothing filter has a good effect on regularizing functions harmonizing adjacent values. For this reason, we use regularization when we perform full waveform inversion. However, regularization with the same smoothing coefficients blurs boundaries between layers and important structures so the result becomes unclear. In order to compensate for the blurring problem, we designed a selective smoothing filter that adopts different smoothing coefficients as the inversion process runs. At first, we establish effective maximum wavenumbers regarding to all smoothing coefficients. Second, we determine wavenumber coverage effectively with a few of frequencies. Then, the coverage and frequencies decide varying smoothing coefficients for the inversion. In order to verify the accuracy of our regularization, we perform full waveform inversion with selected smoothing coefficients on the SEG/EAGE salt model. The inverted result shows for the correct salt velocity and clear structures including sand lens, faults, and layer boundaries. It is observed that our scheme is both simple and powerful for stabilizing and sharpening inverted image.