Summary

Fluid-saturated rocks are generally expected to have frequency-dependent velocities, and it is attractive to try to use this property to discriminate different fluids with seismic data. Previous work has demonstrated how to combine spectral decomposition techniques with AVO inversion to obtain direct estimates of dispersion from prestack data. In this paper, we present the application of frequency-dependent AVO inversion to real seismic data. The Wigner-Ville distribution based method is used for spectral decomposition in order to achieve high resolution. Numerical studies and a real example illustrate the potential of this method for detection of seismic dispersion due to fluid saturation. We may also avoid stacking related “frequency-shadows” as spectral decomposition is performed on pre-stack data.

Introduction

Frequency dependent seismic attributes are of fundamental interest because they are believed to be directly related to the scale length of heterogeneities, rock permeability and saturating fluid. Particular interest has focused on frequency dependence of the time delay between split shear-waves and azimuthal variations in compressional wave attenuation. Recently, laboratory studies show that seismic velocities are frequency-dependent at a relative high frequencies regime and appeared to be caused by fluid mobility (Batzle et al., 2006), which is defined as the ratio of rock permeability to fluid viscosity. Theoretical investigation (eg. Jakobsen and Chapman, 2009) is also conducted to understand this frequency-dependence. Being able to measure frequency-dependence of velocity from reflection data would greatly assist fluid discrimination efforts. Chapman et al. (2005) performed a theoretical study of reflections from layers which exhibit fluid-related dispersion and attenuation, and showed that in such cases the AVO response was frequency-dependent. Application of spectral decomposition techniques allows the behavior to be detected on synthetic seismograms. Wilson et al. (2009) extended this analysis, introducing a frequency-dependent AVO inversion concept aimed at allowing a direct measure of dispersion to be derived from pre-stack data. In this paper, we combine this frequency-dependent AVO inversion with a high-resolution Smoothed Pseudo Wigner- Ville distribution. Numerical studies and field data application from North Sea show that this method has the potential to be useful for fluid detection.

Smoothed Pseudo Wigner-Ville Distribution

We use Wigner-Ville Distribution (WVD) based method for spectral decomposition. WVD is well-recognized as an effective method for time-frequency analysis of nonstationary signals (Cohen, 1995).

Numerical Example

We consider a two-layer Class III AVO model presented by Chapman et al. (2005), where the top elastic layer had Pand S-wave velocities of 2743ms-1 and 1394ms-1. For the dispersive model, the lower layer is defined as a material under water-saturation then substituted with gas by changing the fluid bulk modulus from 2GPa to 0.2GPa. For the elastic model, the P- and S-wave velocities of lower layer were calculated from elastic tensor for the dispersive model at low frequency. Eleven traces for each model are generated using 40Hz Ricker wavelet as the source. The trace space is 100m. Figure 1 displays the synthetic gathers of the elastic and dispersive models at the interface respectively, both of which the amplitudes increase with the offset gradually.

This content is only available via PDF.
You can access this article if you purchase or spend a download.