Conservation of lateral stochastic structure of a medium in its simulated seismic response

Abstract

[1] Mapping correlation lengths of continental lower crustal heterogeneity directly from deep reflection data can be a valuable additional interpretation tool for deep seismic imaging. In this feasibility study we investigate the conservation of lateral correlation length and power law structure in the seismic response of lower-crustal stochastic lithology. The conservation of von Karman power law heterogeneity in backscattered wavefields is evaluated in a range of challenging seismic scenarios. We pay special attention to the limitations of this conservation due to the effect of spatial band limitation, caused by limited seismic recording aperture and to the effect of complex seismic scattering conditions. The comparison of average lateral correlations in von Karman media and their modeled reflection responses reveals two important facts: spatial band limitation by limited seismic illumination severely affects the estimated power law exponent (the Hurst number) of the von Karman heterogeneity distribution, but the estimated correlation length is unaffected and scales linearly with the original value. The original type of lateral power law distribution is conserved under nearly all seismic scattering conditions investigated, albeit with an underestimated correlation length and overestimated Hurst number. Our results can be quantitatively explained which allows for stable estimations of correlation length under various seismic scattering scenarios. This study confirms the potential of utilizing estimates of lateral von Karman structure to characterize the middle and lower crust from deep seismic reflection data.Citation: Carpentier, S. F. A., and K. Roy-Chowdhury (2009), Conservation of lateral stochastic structure of a medium in its simulated seismic response,

Delineating tectonic units beneath the Donbas Fold Belt using scale lengths estimated from DOBRE 2000/2001 deep reflection data

“…Adding noise raises the average value of estimated a x (Figure 5e) and migration spatially slightly shifts some patches of correlation length (Figure 5f), but both operations leave the character of the separated blocks intact. Similar effects of migration and noise on the estimated von Karman parameters have also been observed by Pullammanappallil et al [1997], Hurich and Kocurko [2000], and Carpentier and Roy‐Chowdhury [2009].…”

“…These studies found empirical evidence indicating that the von Karman‐type distribution of causative scattering lithology is inherited in both synthetic and real near‐vertical reflection data. A recent study by Carpentier and Roy‐Chowdhury [2009] shows that this von Karman power law structure of the seismic medium is conserved in the reflected wavefield in a wide range of seismic scattering conditions. This applies especially to the horizontal (lateral) correlation length.…”

“…We wish to estimate these von Karman parameters from an arbitrary seismic reflection data set. The approach taken here follows along the lines of previous work by Holliger et al [1994], Hurich and Kocurko [2000], and Carpentier and Roy‐Chowdhury [2009].…”

“…Vertical correlation length is less easily accessible, although attempts to retrieve it have also been reported by Poppeliers and Levander [2004] and Poppeliers [2007]. Additionally, Hurich [1996], Pullammanappallil et al [1997], Hurich and Kocurko [2000], and Carpentier and Roy‐Chowdhury [2007, 2009] have found that parameters estimated from synthetic viscoelastic data scale linearly with the original input values of the heterogeneous von Karman medium. This provides an extra indication that the relative variations in the estimated values are reliable.…”

Stochastic Heterogeneity Mapping around a Mediterranean salt lens

“…Likewise, a x can be underestimated by a factor of 3-6 (Carpentier and Roy-Chowdhury, 2007). More recently, for complex scattered visco-elastic reflection data, Carpentier et al (2009a) found underestimation factors for a x of between 6 and 10 and overestimation factors of up to 10 for ν. These errors are due to the fact that, although the impedance field is highly correlated to the reflectivity field, they are not equivalent.…”