Tage Røsten

Rough-sea deghosting of streamer seismic data using pressure gradient approximations

A new method is presented for seismic deghosting of towed streamer data acquired in rough seas. The deghosting scheme combines pressure recordings along one or several cables with an estimate of the vertical pressure gradient (or the vertical component of the particle velocity). The estimation of the vertical pressure gradient requires continuous elevation measurements of the wave height directly above the receivers. The vertical pressure gradient estimate is obtained by spatially weighting the pressure field. Each spatial weight generally is the product of two weight functions. The first is a function of partial derivatives acting solely along the horizontal Cartesian coordinates. It can be implemented by finite-difference or Fourier derivative operations. The second is a function of the vertical Cartesian coordinate and accounts for the varying sea state. This weight can be changed from one receiver to the next, making the deghosting a local process. Integrated with the measured pressure field, the esti...

A 2.5D finite-element-modeling difference method for marine CSEM modeling in stratified anisotropic media

We have developed a 2.5D finite-element modeling (FEM) method for marine controlled-source electromagnetic (CSEM) applications in stratified anisotropic media. The main feature of the method is that delta sources are used to solve the governing partial differential equations for cases with and without a resistive target and to obtain the difference of these two solutions as the scattered field from the target. The total field is then the sum of the analytical background field calculated with a 1D modeling method and the difference or scattered field mentioned above. Compared with a conventional direct solution (using delta sources directly in a 2.5D formulation), the new method has smaller near-field error as a result of the source singularity and smaller boundary reflections. The new method does not require a dense mesh in the source region, which thereby reduces the total number of variables to be solved. In this way, the modeling time can be kept within a few minutes for some cases. We show that the ma...

Integration of multiple electromagnetic imaging and inversion techniques for prospect evaluation

The use of controlled source electromagnetics (CSEM) in the marine environment has grown rapidly in the past few years from a simple anomaly fluid-hunting technique used in geologically simple environments to a modeling and inversion based technique applied in structurally and lithologically complex environments (Carazzone et al., 2005). The tool set most commonly available to interpreters includes one-, twoand three-dimensional forward and inverse modeling codes. All previous examples, reported in the literature, of inversion codes applied to marine CSEM data have been cell-based regularized techniques designed to produce the smoothest possible isotropic conductivity model (in twoor three-dimensions) which fits the observed data. We report on the development of a new technique, anisotropic sharp-boundary inversion in which the model is parameterized by two-dimensional interfaces. In this approach anisotropic conductivity can have sharp contrasts across interfaces. Regularization is applied to the smoothness of the interface and the lateral variations of conductivity between interfaces. We demonstrate a work flow that progresses from forward modeling through fast depth migration to smooth cell based inversion, concluding with sharp boundary inversion for the final interpreted conductivity image.

Filter bank decomposition of seismic data with application to compression and denoising

The use of discrete wavelet based analysis, feature extraction, denoising, and compression methods have led to extremely interesting developments in the field of seismic data processing. Notwithstanding, discrete wavelets belong to a wider class of filter banks. The use of more general filter banks allows the design of filter coefficients matching the signal’s properties. Consequently, general filter banks bring forth the performance of discrete wavelet based seismic data processing techniques. In this paper, we discuss basics of general filter bank theory, and its applications to seismic data compression and denoising. We show that properly designed filter banks are able to outperform discrete wavelets in both instances.

On the relationships between depth migration of controlled-source electromagnetic and seismic data

Remote sensing techniques record variations in subsurface petrophysical parameters, such as seismic, electromagnetic, and potential-field properties. Seismic is by far the most common technique, using acoustic and elastic waves to map boundaries between buried strata with contrasting pressure (P) and shear (S) wave velocities. High-resolution data provide information about geologic structures and potential hydrocarbon traps and can allow inferences to be made about the composition of reservoir pore fluids (oil, gas, brine, or combinations) and their relative saturations. Nevertheless, hydrocarbon-filled sandstone reservoirs are often better detected by electrical resistivity rather than seismic parameters (e.g., P-wave velocities are less affected by variations in brine saturation than electrical resistivity).

On the optimality of filter banks in subband compression of seismic stack sections

The performance of seismic data subband coders depends on the proper choice of filter banks and frequency partitioning in the subband domain. We propose a variable-rate seismic subband coder system based on a non-adaptive filter bank and subband block classification as a common framework. However, the nonadaptive filter bank is a combination of a uniform parallel and an octave-band tree-structured filter banks: An 8 x 8 uniform nonunitary analysis and synthesis filter bank is used to decompose and reconstruct the original seismic signal, and a 3-stage octave-band nonunitary analysis and synthesis filter bank is used to analyze and synthesize the lowpass-lowpass subband seismic signal. Favorable rate-distortion results are obtained with this method compared to results reported in the literature. The low complexity and high rate-distortion performance make the proposed seismic coder useful for storage and transmission applications.

Imaging salt bodies using explicit migration operators offshore Norway

We have tested the performance of 3D shot-profile depth migration using explicit migration operators on a real 3D marine data set. The data were acquired offshore Norway in an area with a complex subsurface containing large salt bodies. We compared shot-profile migration using explicit migration operators with conventional Kirchhoff migration, split-step Fourier migration, and common-azimuth by generalized screen propagator (GSP) migration in terms of quality and computational cost. Image quality produced by the explicit migration operator approach is slightly better than with split-step Fourier migration and clearly better than in common-azimuth by GSP and Kirchhoff migrations. The main differences are fewer artifacts and better-suppressed noise within the salt bodies. Kirchhoff migration shows considerable artifacts (migration smiles) within and close to the salt bodies, which are not present in images produced by the other three wave-equation methods. Expressions for computational cost were developed for all four migration algorithms in terms of frequency content and acquisition parameters. For comparable frequency content, migration cost using explicit operators is four times the cost of the split-step Fourier method, up to 260 times the cost of common-azimuth by GSP migration, and 25 times the cost of Kirchhoff migration. Our results show that in terms of image quality, shot-profile migration using explicit migration operators is well suited for imaging in areas with complex geology and significant velocity changes. However, computational cost of the method is high and makes it less attractive in terms of efficiency.

Compression of Seismic Stack Sections Using Singular Value Decomposition

Singular value decomposition (SVD) is an adaptive transform with excellent energy compaction properties for separable data. Hence, applying SVD on seismic stack sections seems to be a good choice. In order to achieve a good SVD compression system, bit-efficient representations must be found for both the transform (singular vectors) and the transform coefficients (singular values). In this paper we present an SVD coding system which exploits the orthonormality of the singular vectors to find a minimum number of parameter representation for the singular vectors. These parameters and the singular values are quantized using pdf optimized quantizers. The SVD coder is designed to give full control of the mse within each image block. The proposed method has the ability to compress seismic stack sections at a ratio of 30:1 with no loss of visual information. Regarding the novelty of this scheme this must be considered to be a good result, although it is outperformed by a standard JPEG image coder in terms of signal to noise ratio.

Wave equation versus Kirchhoff prestack depth migration of OBC data

In areas with complex velocity models, wave equation depth migration has been shown to give significantly better results than the industry standard Kirchhoff depth migration. For very simple velocity models, on the other hand, Kirchhoff depth migration is expected to give similar results as wave equation algorithms. In this paper, we compare Kirchhoff and wave equation prestack depth migration using an ocean bottom cable (OBC) data set from a North Sea area with a velocity model of intermediate complexity. The wave equation depth migration gives better resolution than the Kirchhoff method, but the wave equation algorithm seems not to image the steeply dipping part of the reflectors at large depths as well as the Kirchhoff depth migration. However, this can be an artificial result due to the blurring introduced by the Kirchhoff method.

Seismic data compression and its effect on the amplitudes

We investigate the impact of lossy seismic data compression on the amplitudes of a North Sea 2-D seismic data set processed conventionally. Processing includes attenuation of waterbottom multiples, true amplitude recovery, predictive deconvolution, attenuation of peg-leg multiples, prestack time migration, and stacking. Two sequences of the lossy compression method are investigated. In the first sequence the seismic data are compressed before predictive deconvolution (denoted predecon compression), while in the second sequence the seismic data are compressed after predictive deconvolution (denoted post-decon compression). Both prestack and poststack amplitude analysis are performed. For pre-decon and post-decon compression, compression ratios between 7.5:1-15:1 and 15:1-30:1, respectively, provide excellent reconstruction quality of the seismic data set. In general, migration and stacking reduce the effect of the coding noise at all compression levels, while for high compression ratios (i.e., much greater than 10:1) the coding noise has a destructive effect on the predictive deconvolution step and on the two applied multiple attenuation methods.