Reducing aliasing artifacts in seismic data processing using sharpened (TAU-P-Q) transforms
AssigneeBaker Hughes Incorporated
InventorsRonald E. Chambers, Necati Gulunay
Life of this patent
Estimating the term…
Abstract
Marine seismic data sets are generally under-sampled spatially because of the relatively long listening times required in deep water. It is customary to use very long spreads in the field thereby enhancing aliasing and interference from coherent noise. A seismic-signal data processing method is proposed that applies a combination of a forward parabolic Radon transform and a linear Radon transform to the data, followed by a further transform to a three-dimensional frequency domain. In this domain, a deterministic operator is applied to the data to sharpen the Radon-domain response thereof. The data are then scavenged of noise in the Radon domain and inversely transformed back into the time-space domain.
Claims5
A method comprising:
- providing a data set of consecutive CMP seismic-data-signal gathers having a predetermined geometry, the signals being discretely recorded in t-x space and corrected for hyperbolic moveout;
- executing a parabolic Radon transformation, embracing a predetermined moveout window, of said data set in the x direction and executing a linear Radon transformation encompassing a predetermined dip window in the y direction to form a forward-transformed Radon gather, D(.tau.,p,q);
- transforming D(.tau.,p,q) from the (.tau.-p-q) domain to the frequency domain to define the spectral data response, D(f,k.sub.p,k.sub.q);
- defining the spectral geometry response, G(f,k.sub.p,k.sub.q) for a zero-dip, zero-moveout model built with the geometry characteristic of the data set;
- multiplying D(f,k.sub.p,k.sub.q) by .vertline.1/G(f,k.sub.p,k.sub.q).vertline. and inversely transforming the product from the frequency domain to the (.tau.-p-q) domain as a sharpened Radon gathers (.tau.,p,q);
- scavenging undesired noise from s(.tau.,p,q) and inverse-transforming the residue to the t-x-y domain to recover a substantially noise-free, unaliased data set S(t,x,y).
A method for processing seismic data sets that have been sparsely sampled in t-x-y space, comprising:
- providing a data set of consecutive CMP seismic-data-signal gathers having a predetermined geometry and distributed across t-y space, the signals being discretely recorded in t-x space within each said CMP and corrected for hyperbolic normal moveout;
- executing a parabolic Radon transformation, embracing a predetermined moveout window, of said data set in the x direction and executing a linear Radon transformation encompassing a predetermined dip window in the y direction to form a forward-transformed Radon gather, D(.tau.,p,q);
- transforming D(.tau.,p,q) from the tau-p-q domain to the three-dimensional frequency domain to define the spectral data response function, D(f,k.sub.p,k.sub.q);
- defining a zero-dip normalizing function, .vertline.G(f,k.sub.p,k.sub.q).vertline. for a flat-lying symmetric model built with the geometry characteristic of the data set;
- establishing a threshold value, .epsilon., for the zero-dip normalizing function;
- multiplying .vertline.D(f,k.sub.p,k.sub.q).vertline. by 1/.epsilon. when .vertline.G(f,k.sub.p,k.sub.q).vertline.<.epsilon. but by .vertline.1/G(f,k.sub.p,k.sub.q).vertline. otherwise, and inversely transforming the product from the frequency (f-k.sub.p -k.sub.q) domain to the (.tau.-p-q) domain as an improved-resolution Radon gather s(.tau.,p,q) gather;
- scavenging undesired noise from s(.tau.,p,q) and inversely transforming the residue to the t-x-y domain to recover a substantially noise free, unaliased data set S(t,x,y);
- and
- processing the noise-free, unaliased data set to provide a model of subsurface earth layers.
Dependencies are inferred from the claim text.