New Advances in Seismic Imaging for Exploration and Solid Earth Geophysics II
Presiding: M Bostock, University of British Columbia; K Matson, BP America
S32A-01 10:30h
How Forgetful are Seismic Waves ?
3D surface seismic and vertical seismic profiling (VSP) techniques can be employed to image crustal structures in complex geological settings. The effects of heterogeneities on seismic wave propagation can be described in terms of different propagation regimes (Wu, 1989): quasi-homogeneous for heterogeneities too small to be seen by seismic waves, Rayleigh scattering, Mie scattering and small-angle scattering. These scattering regimes cause characteristic amplitude, phase and travel time fluctuation, which can be used to obtain estimates of scale length. Horizontal resolution of exploration seismic data is often discussed in terms of Fresnel zone. For surface and VSP data, the Fresnel radius increases with increasing depth of investigation. In addition, the lateral resolution is limited by the effective frequency content of the seismic signal. Based on strong contrast in petrophysical data, crustal exploration targets (such as gas-hydrates, permafrost or massive sulfide ores) should make strong P-wave, S-wave and converted wave reflectors against most background velocity models. In the context of realistic geological models, 3D numerical simulations are required to better assess elastic wave interactions with high acoustic impedance targets. In addition, it is important to study the influence of composition and shape of high acoustic impedance targets on the full scattered wavefield through a series of numerical modeling experiments based on the 3D elastic finite-difference (FD) method. Massive sulfide ores consisting of the end-member sulfide minerals pyrite, sphalerite, and galena, which span the full range of observed P- and S- wave velocities and densities in ore rocks, as well as gabbro inclusions, are investigated for different shapes which represent the complex morphologies often observed for ore deposits. 3D FD modeling reveals that large ore deposits lead to a strong and complex scattering response that is often dominated by shear-wave events (Bohlen et al., 2003). For example, the analysis of FD snapshots of forward and backward scattered compressional waves for a simple low velocity sphere (massive sulfide) shows scattering coda generated by the target and significant travel time delays for the direct (forward scattered) wavefield. At late travel times, however the direct wave "heals". The integration of petrophysical data and 3D elastic modeling studies demonstrate that directional scattering and wavefront healing place important constraints on how to handle static corrections and azimuth binning in 3D seismic datasets. References Bohlen, T., C. Muller and B. Milkereit, Elastic seismic wave scattering from massive sulfide orebodies: On the role of composition and shape, in: Hardrock Seismic Exploration, SEG, Tulsa, 86-102, 2003. Wu, R.J., Seismic wave scattering, in: Solid Earth Geophysics, Editor: D.E. James, 1166-1187, 1989.
S32A-02 10:45h
Coherent-State Solutions of the Wave Equation
The coherent-state transform is used in obtaining a global, uniform asymptotic solution of the wave equation. This solution approximates high-frequency propagation of acoustic, optical and seismic waves in generally heterogeneous media. An understanding of the forward problem gives insight to how these type of propagators will perform as migration (imaging) operators. This insight can examine both the kinematics (travel time) and dynamic (amplitude) characteristics of the propagator. The coherent-state approximation has an advantage over more traditional methods (direct and Fourier transform), because it leads to a well-defined approximation independent of the complexity of the caustics. Two similarities of the coherent-state approximation and the Gaussian summation method are that phase functions are complex and both have a parametric dependence. A difference is that the Gaussian beam method is heuristic, whereas the coherent-state method has a theoretical basis. The ability of the coherent-state approximation to accurately model higher wave phenomena (i.e. diffractions, head waves, critical points, etc.) depends on this parameter. This parameter determines the shape of the coherent-state and controls the interpolation between ray theoretical phase and amplitude and Fourier domain (Maslov) phase and amplitude. As an illustration the edge diffraction problem is discussed. The diffracted wave has the correct asymptotic form but the accuracy depends on a parameter that controls the coherent-state. Also, an early test of a coherent-state migration is shown.
S32A-03 11:00h
Approximate hybrid reconstruction of a stochastic medium from its seismic response using scale length analysis and partial spatial phase information
Stochastic description of the Earth is an alternate starting point for analysing broadband multichannel seismic reflection data. This may be preferable if there are indications for stochastic fabric of (upper and) lower crust. Seismic imaging of such media with relatively new statistical processing methods, however, needs careful attention. Using autocorrelation techniques similar to those in earlier researches, we estimate stochastic spatial distributions (scale lengths) from the wavefield backscattered from a synthetic stochastic medium. In addition to this we attempt to retrieve spatial phase (positional) information for the low wavenumber part of the spectrum only from the data as well. These parameters (scale length and partial phase) enable an approximate reconstruction of the original stochastic medium. The estimated scale lengths are less than their true values, something which has also been observed by earlier researchers, and must thus be corrected. The phase information of the low wavenumber part of the synthetic seismic wavefield has a moderate to high correlation with the equivalent spatial phase in the original stochastic medium. The partial spatial phase information in the seismic data could thus be a useful parameter in reconstructing the medium. Used together with the corrected scale length, it enables a partial reconstruction of the medium which includes a part of the low wavenumber structure embedded within a stochastic framework. The positions of shorter scale structures are of course uncertain and cannot be resolved from the seismic data with these methods. We plan to test these ideas on the DOBRE 2000 reflection dataset, measured across the Dniepr-Donets-Pripyat Basin in Ukraine.
S32A-04 INVITED 11:15h
Green's Functions, Source Signatures, and Teleseismic Body Wavefields
The separation of Earth's impulse response or Green's function from earthquake source signature is a canonical problem in seismology and an important prerequisite to the exploitation of scattered waves in structural studies. Receiver functions represent a leading order estimate of the S contribution to the teleseismic P Green's function and have proven exceptionally useful in characterizing discontinuous shear structure in studies of the lithosphere and upper mantle. They provide no information, however, on P-to-P scattering or, accordingly, compressional properties of the Earth's subsurface. An improved estimate of the teleseismic P Green's function can be achieved through consideration of its spectral characteristics. Under conditions typical of the real Earth the P-component can be shown to be minimum phase; thus it may be recovered soley from knowledge of its power spectrum. The minimum phase property can be exploited using both auto- and cross-spectra of multichannel measurements to effect a comprehensive source-impulse response deconvolution that affords an improved estimate of the teleseismic P-Green's function and better characterization of discontinuties in compressional properties.
S32A-05 11:30h
Free Surface Effect and Separation of Teleseismic Waves: Transmission & Reflection, Primary vs Multiple
We examine the free surface effect on teleseismic waves recorded by passive seismic arrays on the surface, and develop a new way to separate the forward and free surface back scattered waves from the total teleseismic wavefield. Using one-way wavefield reciprocity the reflection responses (of seismic reflection geometry) can be reconstructed from the transmission responses (of teleseismic geometry). The free surface effect (back scattered waves) of the total teleseismic wavefield can be removed by using an inverse scattering series for the teleseismic transmission geometry. The removed free surface back scattered waves can be obtained by taking the difference between the teleseismic wavefield before and after free surface effect removal. This gives the equivalent response to that produced from reconstructed reflection waves using the reciprocity relation. Free surface multiples in the reflection waves are removed by the inverse scattering series of reflection geometry. This produces a direct separation of transmission and reflection waves of the total teleseismic wavefield without knowledge of earth models. Numerical tests of acoustic stratified layer lithospheric earth models demonstrate the effectiveness of the algorithm. Noise test shows that the algorithm can work with signal-to-noise ratio as low as 5. Reconstruction demands higher signal-to-noise ratio than the removal procedure. Test with elastic synthetic data indicates that the acoustic algorithm is still effective for small incidence angle that characterize teleseismic P waves at distance greater than 30 degrees. For shallow angles of incidence, either P and SV wave separation or an elastic algorithm will probably be necessary.
S32A-06 11:45h
Suppressing Near-Array Scattered Noise in Controlled and Passive-Source Seismic Data
A common problem in controlled and passive-source seismic imaging is the presence of secondary surface waves in the data. We focus here on surface waves that are excited as upcoming body waves hit near-array heterogeneity or nearby surface topography. These surface waves can interfere with events of interest, thereby limiting the maximum obtainable resolution in the final image. We present a method to image and suppress near-array scattered waves which was originally developed for the controlled-source (exploration) seismic case. The method revolves around the following deterministic idea: If one knows the distribution of near-surface scatterers, it is possible to predict and subsequently subtract the secondary surface waves from the data. Hence, we require an estimate of the near-surface distribution of scatterers, which we get from inversion of scattered surface waves excited by one or a few particular P-wave event(s). The prediction-and-subtraction method effectively suppresses scattered noise on data from high resolution 2D exploration seismic lines acquired in areas with strong near-surface heterogeneity and/or surface topography. This leads to increased trace-to-trace coherency and continuity of reflection events. Field data examples will be presented and used to address issues relating to assumptions and limitations inherent to the method. The availability of teleseismic data recorded at dense short-period arrays has opened the door for formulating this algorithm for teleseismic settings as well. In this context, we expect to suppress secondary surface waves in the scattered P-wave field that would otherwise be imaged with the wrong operators, giving rise to artifacts in the final image. We discuss the implications and applicability of the algorithm in a teleseismic framework, focusing in particular on data sampling requirements and preconditioning of the data before the inversion step in order to obtain the near-surface scattering distribution.