Premium
Practical aspects of wavefield separation of two‐component surface seismic data based on polarization and slowness estimates
Author(s) -
Richwalski Sandra,
RoyChowdhury Kabir,
Mondt Jaap C.
Publication year - 2000
Publication title -
geophysical prospecting
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.735
H-Index - 79
eISSN - 1365-2478
pISSN - 0016-8025
DOI - 10.1046/j.1365-2478.2000.00204.x
Subject(s) - slowness , amplitude , smoothing , geology , synthetic data , polarization (electrochemistry) , a priori and a posteriori , seismic wave , surface wave , data processing , seismology , algorithm , optics , mathematics , computer science , physics , statistics , philosophy , chemistry , epistemology , operating system
The processing of multicomponent seismic data, carried out individually on the different wavetypes (P‐, S‐ and converted waves), should result in an improved image of the subsurface. We examine the wavefield‐separation method proposed by Cho and Spencer. We discuss practical aspects related to the separation of interfering waves on two‐component surface seismic data and illustrate these using synthetic data. A sliding spatial window is used for analysis. The choice of its width represents a trade‐off between stabilizing the method in the presence of random noise and ensuring a good spatial resolution. No a priori knowledge of the subsurface is required, but locally, the characteristic parameters of the waves, i.e. horizontal slowness and polarization, are assumed to be constant inside the analysis window. These parameters are estimated at each frequency, but a statistical analysis provides a more robust estimate, especially in the presence of random noise. This approach also solves the problem of eigenvalue sharing and switching. Additional smoothing of the estimates according to a model may further improve the results. The width of the analysis window may be decreased if the waves inside the data window differ significantly in amplitude. The dominant wave in each case is separated from the lower‐amplitude waves and subtracted from the data. This novel iterative approach thereby allows for the isolation of low‐amplitude events.