Exploration Geophysics - Volume 28, Issue 3, 1997

This paper compares the seismic expression of Lower Tertiary mudstone-dominated sequences of the North Sea basin with Cretaceous siliciclastic sequences of the Eromanga Basin. The paper focuses on the development of layer-bound systems of minor extensional faults that pervasively deform these sequences in both basins. Layer-bound fault systems have only recently been recognised in the North Sea with the availability of high resolution 3D seismic data. The resolution achievable with 3D mapping has revealed that in map-view, these faults are arranged in polygonal networks reminiscent of dessication cracks. This property of polygonality is accounted for in a genetic model based on basin-wide volumetric contraction of muddy sediments during the early stages of compactional dewatering.

Deep-seated structures within the Warburton Basin have been intermittently reactivated during development of the overlying Cooper and Eromanga Basins. Four major deep structures preceding deposition of the Cooper Basin succession are described. 1) A complex deformation zone occurs on the eastern flank of the Birdsville Track Ridge near Kalladeina 1. 2) Simple reverse faults are common and widespread in the Warburton Basin strata. 3) Gentle folds occur in the Coongie area and become increasingly tight towards the Birdsville Track Ridge. 4) Imbricate thrust faults form the cores of structures along the northeast-southwest trending Gidgealpa–Merrimelia–Innamincka Ridge. Warburton Basin structural control in the overlying basins is characterised by reactivation and propagation of thrust faults along the Gidgealpa–Merrimelia–Innamincka Ridge, and by further folding and faulting of the complex deformation zone on eastern flank of the Birdsville Track Ridge. The episodic, compressive reactivation has controlled depositional and erosional patterns of the Cooper and Eromanga Basins, and hence controlled hydrocarbon occurrence in these two basins. Potential footwall traps may be formed by propagation of the thrust faults.

Improved seismic sections can be obtained by processing through to stack in the τ-ρ domain. This approach provides significant improvements over routine processing, using hyperbolic velocity filtering (HVF) to suppress the τ-ρ transform artefacts. HVF is a form of time and offset variant filtering that only allows each point in the t-x domain to contribute to a small number of p traces during a τ-ρ transform. HVF can be incorporated into a point-source τ-ρ transform, extending previous implementations that applied HVF during the slant stack. This assists gap deconvolution by providing plane-wave amplitudes in a τ-ρ gather where reverberations are exactly periodic along p traces and where S-wave reflections and long period multiples are suppressed. The τ-ρ domain also offers improved velocity analysis through the use of wide-angle reflections and because primary reflection ellipses never cross each other. The τ-ρ transform of a CMP gather can then be stacked with further advantages resulting from the elliptical moveout correction, which minimises wavelet stretch and approximates the exact reflection traveltime better than NMO. Examples from the Carnarvon and Gippsland Basins confirm that the cumulative theoretical advantages of τ-ρ domain processing are achieved in practice.

Semblance displays are commonly used to indicate the velocities required to flatten reflection hyperbolas during a velocity analysis of seismic data. “Model-based velocity analysis” is a scheme for interpreting semblance analyses that uses an elastic synthetic seismogram computed from a detailed depth model. The depth model is constructed from logs taken from an adjacent well. The semblance analysis of the synthetic and the P-wave RMS velocity function derived from the depth model are then used to guide the velocity analysis of nearby field records. Semblance analyses of additional synthetics, such as those computed without multiples, enable noise events to be more readily identified. Strong noise events remain identifiable on field data semblance analyses computed away from the well. A model-based approach thereby allows more effective extrapolation of the well-log information.

A model-based velocity analysis revealed that interbed multiples had been mistaken for primaries in conventional velocity analyses from the Gippsland Basin. High amplitudes and lack of differential moveout of the interbed multiples relative to the primaries resulted in picks that were 5-10% lower than the stacking velocities of the primaries. Restacking using the model-based velocities provided far greater continuity across a 4 km seismic line from the Gippsland Basin that had been processed through to stack in the τ-ρ domain.

The model-based approach can be applied with greater accuracy by exploiting the higher resolution of semblance peaks computed in the τ-ρ domain. The resolution improvement results from the use of wide-angle reflections and the theoretical advantages of the elliptical moveout correction. Semblance analyses of field records showed a 2× resolution improvement in the target zone after the application of a point-source τ-ρ transform. A 5× resolution improvement was achieved at the shallow Miocene carbonates where the reflection ellipses were flatter and contained a supercritical segment that was muted in the t-x domain.

The exploding reflector model has been suggested as a way to simplify calculation of wave propagation. Previous work has produced numerical solutions to a model where two media are in contact and a signal is generated at the surface of the upper layer. Scatter from the interface was modelled as an exploding reflector. The strength and shape of the signal that arrives at the base of the lower layer were calculated following the scatter from the interface. These numerical solutions gave results that were comparable to finite difference solutions of the wave equation.

The physical meaning and an extension of the exploding reflector model are discussed here. Exact analytical solutions to a two layer model separated by an incoherent scattering layer are derived by two distinct methods, one using the mathematical computer package Maple and then by simple geometrical arguments. Exact analytical solutions could be found for the case where the layers had identical thickness and sound velocities and the interface scattered the waves incoherently.

Two different laws for the reduction in intensity of the signal as a function of distance were examined being the inverse unity power law and inverse square law. Two different laws for scattering from the interface were also examined being an isotropic scatter for which scattering intensity is independent of scattering angle, and a Lambertian scatter which obeys Lambert's Cosine Law.

Analytical solutions to the model with unequal thicknesses or unequal acoustic velocities were investigated. Analytical solutions could be found in some cases but could not usually be reduced to simple expressions that could be presented here. The general solution is possible but required the solution to the general quartic equation. Solution expressions are very long and can only reasonably be given as process flows in a mathematical computing package such as Maple.

The models described here and the solutions found previously may find a use in describing the incoherent scatter from interfaces between layers.