Waveform inversion with exponential damping using a deconvolution-based objective function
KAUST DepartmentEarth Science and Engineering Program
Physical Science and Engineering (PSE) Division
Seismic Wave Analysis Group
Online Publication Date2016-09
Print Publication Date2016-09
Permanent link to this recordhttp://hdl.handle.net/10754/625275
MetadataShow full item record
AbstractThe lack of low frequency components in seismic data usually leads full waveform inversion into the local minima of its objective function. An exponential damping of the data, on the other hand, generates artificial low frequencies, which can be used to admit long wavelength updates for waveform inversion. Another feature of exponential damping is that the energy of each trace also exponentially decreases with source-receiver offset, where the leastsquare misfit function does not work well. Thus, we propose a deconvolution-based objective function for waveform inversion with an exponential damping. Since the deconvolution filter includes a division process, it can properly address the unbalanced energy levels of the individual traces of the damped wavefield. Numerical examples demonstrate that our proposed FWI based on the deconvolution filter can generate a convergent long wavelength structure from the artificial low frequency components coming from an exponential damping.
CitationChoi Y, Alkhalifah T (2016) Waveform inversion with exponential damping using a deconvolution-based objective function. SEG Technical Program Expanded Abstracts 2016. Available: http://dx.doi.org/10.1190/segam2016-13818075.1.
SponsorsWe are grateful to King Abdullah University of Science and Technology for financial support.
PublisherSociety of Exploration Geophysicists
Conference/Event nameSEG International Exposition and 86th Annual Meeting, SEG 2016