Page 365 - Acquisition and Processing of Marine Seismic Data
P. 365

356                                  6. DECONVOLUTION


























           FIG. 6.42  An example application of designature to a shot gather. (A) Input shot, and (B) after designature. Blue and red
           insets show corresponding zoom-in areas.


                6.9 SURFACE CONSISTENT                  consistent convolutional model for the recorded
                     DECONVOLUTION                      seismogram x(t), source wavelet w(t), and
                                                        reflectivity at source-receiver midpoint r(t)in a
              Surface consistent deconvolution is an effec-  noise-free environment is given by
           tive resolution improvement method generally              ∗       ∗        ∗       (6.24)
                                                                               ð
           used in land seismic data processing. The seismic  xtðÞ ¼ s i tðÞ h i jð  Þ=2 tðÞ r i + jÞ=2 tðÞ e j tðÞ
           trace is decomposed into the convolution effects  where s i (t) is the effect of the source at location i,
           of source, receiver, reflectivity series, and offset  e j (t) is the effect of the receiver at location j, and
           by surface consistent deconvolution, which   h(t) is the effect of the offset. Taking the Fourier
           accounts for the weathering effects just below  transform of Eq. (6.24), we get
           the source and/or receiver locations, and the
           effect of offset. The deconvolution is then applied  X ωðÞ ¼ S ωðÞH ωðÞR ωðÞE ωðÞ  (6.25)
           as an inverse filter after decomposition. Surface
                                                        and taking the natural logarithm of the ampli-
           consistent deconvolution is used to correct ampli-
                                                        tude spectra for linearity, we obtain
           tude, phase and frequency variations caused by
           geophone responses, poor coupling, and near-
                                                           j ½
                                                                     j ½
                                                                                             j ½
                                                                                     j ½
                                                                             j ½
                                                         ln X ωðÞjŠ ¼ ln S ωðÞjŠ ln H ωðÞjŠ ln R ωðÞjŠ ln E ωðÞjŠ
           surface seismic transmission properties for land
                                                                                              (6.26)
           seismic data. The most important advantage of
           this method is its stability and effectiveness on  ln[jS(ω)j] and ln[jE(ω)j] for source and receivers
           data with poor quality and low S/N ratio in the  can be solved by a least-squares iterative
           presence of high noise amplitudes.           approach, finally to obtain the inverse filter
              In practice, it is assumed that the shape of  operator as (Yılmaz, 1987)
           the wavelet embedded in the data depends on
                                                                 1          1  t              (6.27)
           the source and receiver locations. A surface        s i  t ðÞs i tðÞ ¼ e j ðÞe j tðÞ ¼ δ tðÞ
   360   361   362   363   364   365   366   367   368   369   370