De-noising Receiver function data using the unsupervised deep learning approach

Summary The converted wave data (P-to-s or S-to-p), traditionally termed as receiver functions, are often contaminated with noise of different origin that may lead to the erroneous identification of phases and thus influence the interpretations. Here we utilize an unsupervised deep learning approach called Patchunet to de-noise the converted wave data. We divide the input data into several patches, which are input to the encoder and decoder network to extract some meaningful features. The method de-noises an image patch-by-patch and utilizes the redundant information on similar patches to obtain the final de-noised results. The method is first tested on a suite of synthetic data contaminated with various amount of Gaussian and realistic noise and then on the observed data from three permanent seismic stations: HYB (Hyderabad, India), LBTB (Lobatse, Botswana, South Africa), COR (Corvallis, Oregon, USA). The method works very well even when the signal-to-noise ratio is poor or with the presence of spike noise and deconvolution artifacts. The field data demonstrate the effectiveness of the method for attenuating the random noise especially for the mantle phases, which show significant improvements over conventional receiver function based images.

Approximations for traveltime, slope, curvature, and geometrical spreading of elastic waves in layered transversely isotropic media

Each seismic body wave, including quasi compressional, shear, and converted wave modes, carries useful subsurface information. For processing, imaging, amplitude analysis, and forward modeling of each wave mode, we need approximate equations of traveltime, slope (ray-parameter), and curvature as a function of offset. Considering the large offset coverage of modern seismic acquisitions, we propose new approximations designed to be accurate at zero and infinitely large offsets over layered transversely isotropic media with vertical symmetry axis (VTI). The proposed approximation for traveltime is a modified version of the extended generalized moveout approximation that comprises six parameters. The proposed direct approximations for ray-parameter and curvature use new, algebraically simple, equations with three parameters. We define these parameters for each wave mode without ray tracing so that we have similar approximate equations for all wave modes that only change based on the parameter definitions. However, our approximations are unable to reproduce S-wave triplications that may occur in some strongly anisotropic models. Using our direct approximation of traveltime derivatives, we also obtain a new expression for the relative geometrical spreading. We demonstrate the high accuracy of our approximations using numerical tests on a set of randomly generated multilayer models. Using synthetic data, we present simple applications of our approximations for normal moveout correction and relative geometrical spreading compensation of different wave modes.

Feasibility study of fracture interpretation using multicomponent seismic data: SEAM II Barrett model

Several P-wave azimuthal anisotropy studies have been conducted for the SEAM II Barrett model data. However, these analyses provide fracture property estimation that is inconsistent with the actual model properties. Therefore, we perform a feasibility study to understand the influence of the overburden and reservoir properties, and the processing and inversion steps, which together determine the success of the fracture interpretation from seismic data. 1D model properties (orthorhombic for both overburden and reservoir) are first extracted from the actual Barrett model properties at two locations. Anisotropic prestack reflectivity modeling exposes the true orthorhombic response of the 1D medium in the form of Common Offset and Common Azimuth (COCA) gathers. The true anisotropic response is obscured in the Barrett data (generated by finite element modeling) due to the mild lateral velocity variations and orthorhombic anisotropy in the overburden. We then expose the reservoir anisotropic response by using an isotropic overburden in the reflectivity modeling. This shows that the P-wave VVAZ responses generated by the reservoir itself are weak, which leads to an unstable VVAZ inversion to estimate the interval NMO velocity anisotropy. The reservoir thickness (125m or 65ms TWT) or NMO velocity anisotropy (6-7%) needs to be at least doubled to obtain a stable VVAZ inversion. Anisotropic geometrical-spreading correction improves the amplitude-versus-azimuth (AVAZ) inversion results when reflectivity modeling models orthorhombic overburden. The converted wave ( C-wave) has a stronger VVAZ response compared to the P-wave. We suggest that the C-wave data could be useful to constrain fracture interpretation in the Barrett model. We conclude that the results of previous studies are due to the combination of the residual influence of overburden after processing and imaging, and the weak anisotropy responses from the reservoir.

Converted-wave model building and imaging based on common-focus-point methodology

Seismic images can be viewed as photographs for underground rocks. These images can be generated from different reflections of elastic waves with different rock properties. Although the dominant seismic data processing is still based on the acoustic wave assumption, elastic wave processing and imaging have become increasingly popular in recent years. A major challenge in elastic wave processing is shear-wave (S-wave) velocity model building. For this reason, we have developed a sequence of procedures for estimating seismic S-wave velocities and the subsequent generation of seismic images using converted waves. We have two main essential new supporting techniques. The first technique is the decoupling of the S-wave information by generating common-focus-point gathers via application of the compressional-wave (P-wave) velocity on the converted seismic data. The second technique is to assume one common VP/ VS ratio to approximate two types of ratios, namely, the ratio of the average earth layer velocity and the ratio of the stacking velocity. The benefit is that we reduce two unknown ratios into one, so it can be easily scanned and picked in practice. The PS-wave images produced by this technology could be aligned with the PP-wave images such that both can be produced in the same coordinate system. The registration between the PP and PS images provides cross-validation of the migrated structures and a better estimation of underground rock and fluid properties. The S-wave velocity, computed from the picked optimal ratio, can be used not only for generating the PS-wave images, but also to ensure well registration between the converted-wave and P-wave images.