Phase matched filters to remove surface wave noise from converted P-SV data
David G. Schieck
One of the greatest difficulties in obtaining and processing good quality shear wave sections is overcoming the effects of surface waves (Corbin et al., 1983). Generally, in land seismic exploration, surface waves (ground roll) should be attenuated by geophone arrays. Due to varying physical surface conditions, the desire to record non-normal incidence signal and the expense of three-component geophone arrays, the procedures in the field may be inadequate for eliminating this unwanted noise from the seismic records. Processing techniques such as frequency filters, muhichannel filters and polarization analysis is often required. Conventional frequency and multichannel filters often effect the vertical resolution of seismic data due to the overlap frequencies of ground roll and reflection data (Saatilan and Canitez, 1988).
Rayleigh waves are the dominant surface waves recorded on vertical and radial components found in exploration seismology. These waves are nondispersive in the classical case of propagation along the surface of an isotropic homogeneous half space. However for the real earth case of layered medium Rayleigh waves are dispersive and multiple modes are possible (Ren et al., 1986). An example of these modes for the vertical component of walk away data is shown in Figure 1.
It is well established that P-wave, S-wave, Rayleigh waves and Love waves have distinct polarization patterns. (Aki and Richards, 1980) Three dimensional polarization analysis of the vertical, radial and horizontal components can separate the elliptically polarized Rayleigh waves from the radial component seismic data (Vidale, 1986, Samson and Olson, 1981).
A more recent method of ground roll attenuation makes use of the dispersive character of the Rayleigh wave (Bereford-Smith and Rango, 1988, Saatqilan and Canitez, 1988). These ideas were originally applied to earthquake seismology to identify and enhance Rayleigh wave signals (Herrin and Goforth, 1977). By applying one dimensional, linear frequency modulated matched filters promising results have been demonstrated for compressional wave data as shown in figure 2. The vertical resolution is not compromised. Further, the uniform weighted stacking assumed in multichannel filters (such as a velocity filter or radon transforms) is not made.