Page 488 - Acquisition and Processing of Marine Seismic Data
P. 488

10.4 SPECIFIC STACKING METHODS                      479

































           FIG. 10.21  Comparison of mean and median stack. (A) An NMO corrected synthetic CDP gather with three genuine
           reflections at 150, 500, and 850 ms along with high-amplitude noise bursts. (B) Mean stack, and (C) median stack traces
           of (A).






           schematic illustration in Fig. 10.20, the amplitude  in astack sectionwithmuchhigher S/N ratio
           value at trace number 3, indicated with a blue bar,  with  no  high-amplitude  noise  bursts
           is the median.                               (Fig. 10.22B).
              Fig. 10.21 compares mean stack and median    Applications show that the median stack may
           stack on an NMO corrected synthetic CDP      be successful in the case of the noisy seismic
           gather. The input CDP contains three reflec-  data, especially for higher amplitude spike-like
           tions at 150, 500, and 850 ms along with     noise bursts. It is especially useful in the case
           high-amplitude noise bursts. The mean stack  of extreme strong noise amplitudes in low fold
           trace in Fig. 10.21B suffers from noise ampli-  seismic data, and it produces nice results when
           tudes, whereas the median stack trace is     the noise amplitudes appear at less than half of
           almost free of noise bursts. Fig. 10.22 com-  the total number of traces involved in the stack-
           pares mean stack and median stack on a real  ing. However, since it reserves only the median
           marine data example. There are extreme spike  value for the stacking and ignores all other
           noise bursts distributed randomly on the     amplitude values, the median stack may reduce
           traces of the stack section in Fig. 10.22A. The  the quality of the stack sections in the case of
           median stack of the same seismic data results  low-amplitude noise.
   483   484   485   486   487   488   489   490   491   492   493