Showing papers in "Seg Technical Program Expanded Abstracts in 2015"

Adjustive full waveform inversion

Abstract: One fundamental challenge of full waveform inversion(FWI) is the local minimum issue caused by the cycle skipping between the predicted and acquired data. To overcome this challenge and achieve a successful inversion, a good initial model becomes a necessary ingredient of FWI. In others words, the background model must be accurate enough to start FWI. Our proposed adjustive FWI (AdFWI) is designed to build the relation between travel time shift and model error in a different and novel way, so that FWI can be used to correct the erroneous background model, and therefore mitigate cycle-skipping issues and improve the robustness of FWI with inaccurate initial models.

Damped Multichannel Singular Spectrum Analysis for 3D Random Noise Attenuation

Abstract: Multichannel singular spectrum analysis (MSSA) is an effective algorithm for random noise attenuation in seismic data, which decomposes the vector space of the Hankel matrix of the noisy signal into a signal subspace and a noise subspace by the truncated singular value decomposition (TSVD). However, this signal subspace actually still contains residual noise. We derived a new formula of low-rank reduction, which is more powerful in distinguishing between signal and noise compared with the traditional TSVD. By introducing a damping factor into the traditional MSSA for damping the singular values, we proposed a new algorithm for random noise attenuation. The proposed modified MSSA is named as the damped MSSA. The denoising performance is controlled by the damping factor and the proposed approach reverts to the traditional MSSA approach when the damping factor is sufficiently large. Application of the damped MSSA algorithm on synthetic and field seismic data demonstrates a superior performance compared with the conventional MSSA algorithm.

Acoustic Multi-Parameter FWI for the Reconstruction of P-Wave Velocity, Density and Attenuation: Preconditioned Truncated Newton Approach

Abstract: Multi-parameter full waveform inversion is a challenging problem, mainly because of the existence of trade-offs between subsurface parameters. Mitigating these trade-offs requires to account accurately for the inverse Hessian operator during the minimization process. In this study, we investigate the efficiency of the truncated Newton minimization approach in the 2D acoustic frequency-domain approximation for the simultaneous reconstruction of P-wave velocity, density and attenuation factor. To further enhance the performances of the truncated Newton algorithm, a preconditioner adapted to this multi-parameter problem is designed, based on a 3× 3 block approximation of the Hessian operator where each block is diagonal. Numerical results on the synthetic Valhall model illustrate the interest of this strategy.

Supervised Learning to Detect Salt Body

Abstract: Summary In this paper we are presenting a novel workflow to detect salt body base on seismic attributes and supervised learning. The machine learning algorithm Extremely Random Trees Ensemble is used to train and automatically identifying and classify salt regions. We have used a complex synthetic seismic dataset from phase I model of the SEG Advanced Modeling Corporation (SEAM), that represents deepwater regions of Gulf of Mexico. This dataset has very low frequency and there are sediments locations with similar amplitude value than salt body. After a first step of our proposal, where machine learning is applied directly to the seismic data, we obtained accuracy values of around 80%. A second (post-processing) step brings up accuracy to around 95%. We conclude that machine learning is a promise mechanism to identify salt bodies on seismic data when the selected model exhibits enough capacity to model the complex decision boundaries needed during class discrimination.

Investigating the Possibility of Locating Microseismic Sources using Distributed Sensor Networks

Abstract: Distributed sensor networks are designed to provide computation in-situ and in real-time. The conventional time-reversal imaging approach for microseismic event location may not be optimal for such an environment. To address this challenge, we develop a methodology of locating multiple microseismic events with unknown start times based on the cross-correlation imaging condition borrowed from active-source seismic imaging. The imaging principle states that a true microseismic source must correspond to the location where all the backwardpropagated events coincide in both space and time. Instead of simply stacking the backward-propagated seismic wavefields, as suggested by time-reversal imaging, we perform multiplication reduction to compute a high-resolution microseismicity map. The map has an extra dimension of time, indicating the start times of different events. Combined with a distributed sensor network, our method is designed for monitoring microseismic activities and mapping fracture development during hydraulic fracturing in-situ and in real-time. We use numerical examples to test the ability of the proposed technique to produce high-resolution images of microseismic locations.

EMD-Seislet Transform

Abstract: The seislet transform uses a prediction operator that is connected to the local slope or frequency of seismic events. We have combined the 1D nonstationary seislet transform with empirical-mode decomposition (EMD) in the f-x domain. We used EMD to decompose data into smoothly variable frequency components for the following 1D seislet transform. The resultant representation showed remarkable sparsity. We developed a detailed algorithm and used a field example to demonstrate the application of the new seislet transform for sparsity-promoting seismic data processing.

An Angle-Domain Wavenumber Filter for Multi-Scale Full-Waveform Inversion

Abstract: The scale length information of velocity perturbations carried by scattered wave is not only dependent on wave frequency but also related to the scattering angle. We introduce a angle-domain wavenumber filter in fullwaveform inversion. Both source and receiver side waves are decomposed into local plane waves, followed by sorting scattering events according to their incidence and scattering angles. The small-angle scatterings are more responsible for large-scale velocity perturbations, while the large scattering angles are related to small-scale perturbations. By controlling scattering angles, we can perform multiscale inversion. Numerical examples reveal, when initial model has large errors, the new inversion method can significantly improve the convergence. The angle-domain wavenumber filter is highly localized, flexible and efficient, and can be combined to most FWI methods.

Detection of Salt-dome Boundary Surfaces in Migrated Seismic Volumes Using Gradient of Textures

Abstract: SUMMARY Salt domes, an important geological structure, are closely related to the formation of petroleum reservoirs. In many cases, no explicit strong reflector exists between a salt dome and neighboring geological structures. Therefore, interpreters commonly delineate the boundaries of salt domes by observing a change in texture content. To stimulate the visual interpretation process, we propose a novel seismic attribute, the gradient of textures, which can quantify texture variations in three-dimensional (3D) space. On the basis of the attribute volume, we apply a global threshold to highlight regions containing salt-dome boundaries. In addition, with region growing and morphological operations, we can remove noisy boundaries and detect the boundary surfaces of salt domes effectively and efficiently. Experimental results show that by utilizing the strong coherence between neighboring seismic sections, the proposed method can delineate the surfaces of saltdome boundaries more accurately than the state-of-the-art detection methods that label salt-dome boundaries only in twodimensional (2D) seismic sections.

Up-Down Separation Using Hilbert Transformed Source for Causal Imaging Condition

Abstract: We show an explicit scheme that separates seismic source and receiver wave-fields individually into upand downgoing components. The main challenge of the work is generating wavefields that correspond to positive or negative temporal frequencies in space-time 4D volume. This difficulty arises because the seismic propagators we typically employ store wave-fields with slowest dimension in time but the Fourier transform operates most efficiently on data that are stored contiguously. We solve this issue by a temporal Hilbert transform of the source term of the wave-equation followed by the conventional propagations. The pair of wave-fields, namely, the wave-field propagated without a Hilbert transformed source and the wave-field generated by the Hilbert transformed source, constitute the desired positive or negative temporal components in real and imaginary parts, respectively and separately. The down-going and up-going wave components can then be conveniently obtained by applying 1D Fourier filters in depth. We pursue a causal imaging condition that correlates the down-going source component with the up-going receiver component for subsurface imaging. We demonstrate that by applying the causal imaging condition certain strong near-salt imaging artifacts are removed.

Microseismic event estimation in noisy data via full waveform inversion

Abstract: Full waveform inversion accurately estimates the full spatial and temporal description of a microseismic source which includes not only the location and origin time of the source but also the waveform itself. Assuming two-dimensional acoustic wave propagation, the gradient is computed via the adjointstate method for both the spatial radiation pattern and the temporal waveform of the source. Neither of these gradients requires storing the forward solution of the wave equation as is required by the imaging condition for velocity inversion. This approach identifies multiple sources, handles extremely low signal-to-noise ratio data, and produces accurate results in the absence of a good initial estimate.

3D Seismic Image Processing for Faults

Abstract: Numerous methods have been proposed to automatically extract fault surfaces from 3D seismic images, and those surfaces are often represented by meshes of triangles or quadrilaterals. However, extraction of intersecting faults is still a difficult problem that is not well addressed. Moreover, mesh data structures are more complex than the arrays used to represent seismic images, and they are more complex than necessary for subsequent processing tasks, such as that of automatically estimating fault slip vectors. We have represented a fault surface using a simpler linked data structure, in which each sample of a fault corresponded to exactly one seismic image sample, and the fault samples were linked above and below in the fault dip directions, and left and right in the fault strike directions. This linked data structure was easy to exchange between computers and facilitated subsequent image processing for faults. We then developed a method to construct complete fault surfaces without holes using thi...

Choice of mother wavelets in CWT spectral decomposition

Abstract: Summary Spectral decomposition carried out with the use of the continuous wavelet transform requires the choice of a mother wavelet, which in turn is used to derive a family of wavelet functions. These wavelet functions are scaled and shifted to ‘fit’ them to the input seismic data traces. Unlike the fixedlength discrete Fourier transform method, the continuous wavelet transform (CWT) window varies with frequency, resulting in better temporal resolution at high frequencies and better frequency resolution. We evaluate the relative value and use of Morlet, Mexican Hat, Derivative of Gaussian (DOG), and the Shannon wavelets in the analysis of a fluvial-deltaic system. Spectral decomposition carried out on two seismic data volumes shows that the Morlet wavelet is more robust and yields better results than the others. While we do not suggest that this conclusion be generalized, we do recommend that this exercise be carried out on a test volume to select the best mother wavelet to be used in the spectral decomposition.

Application of Robust Principal Component Analysis (RPCA) to Suppress Erratic Noise in Seismic Records

Abstract: Seismic data are always contaminated with noise. Therefore, signal-to-noise ratio enhancement plays an important role in seismic data processing. This paper illustrates a robust principal component analysis (RPCA) method to suppress erratic noise that contaminates seismic data. The method operates in the frequency-space domain and relies on a robust low-rank approximation of the seismic data volume. We adopt a nuclear norm constraint that yields the low-rank approximation of the desired data while using an `1 norm constraint to properly estimate the erratic (sparse) noise. The problem is then tackled via the first-order gradient iteration method with two steps of softthresholding. We illustrate the effectiveness of this method via synthetic examples.

Preconditioning least-squares RTM in viscoacoustic media by Q-compensated RTM

Abstract: Reverse-time de-migration (RTDM) is formulated as the adjoint operator of reverse-time migration (RTM). In acoustic medium, RTM provides a good approximation to the inverse of RTDM, and can be used to iteratively invert for the reflectivity image in least-squares RTM (LSRTM). In viscoelastic medium, however, the adjoint of the RTDM operator is far from its inverse because of amplitude attenuation during both forward and backward wave propagation. As a result, LSRTM in attenuating medium may suffer from a slow convergence rate due to the ill-conditioned wave-equation Hessian. To improve the convergence rate, we propose preconditioning LSRTM by replacing the original RTM operator with a better approximate inverse to the RTDM operator, namely the Q-compensated RTM (Q-RTM). Since the inverted matrix is numerically non-Hermitian, we use the Generalized Minimum Residual (GMRES) method instead of the Conjugate Gradient (CG) method as the iterative method. Numerical tests demonstrate that the proposed Q-LSRTM approach converges significantly faster than LSRTM, and is capable of producing high-quality attenuation-compensated images within the first few iterations.

VSP Field Trials of Distributed Acoustic Sensing in Trinidad and Gulf of Mexico

Abstract: Distributed acoustic sensing (DAS) is a novel technology that can take almost any fiber-optic installation and turn the fiberoptic cable into a large seismic array, which can provide enhanced imaging and monitor fluids and pressures in the reservoir. Two key marine borehole seismic field trials using fiberoptic DAS technology were recently executed. The first field trial was conducted by BP in a Trinidad Mahogany production well using standard fiber to pressure/temperature (P/T) sensors. The second field trial was conducted by Shell and was acquired simultaneously in two injector wells at the Mars Field, located in the deepwater Gulf of Mexico, with BP as a co-owner. Successful imaging results from the two trials demonstrate many potential applications of DAS technology.

Using Inverse Transmission Matrices for Marchenko Redatuming in Highly Complex Media

Abstract: The goal of Marchenko redatuming is to reconstruct, from single-sided reflection data, wavefields at virtual subsurface locations containing transmitted and reflected primaries and internal multiples, while relying on limited or no knowledge of discontinuties in subsurface properties. Here, we address the limitations of the current Marchenko scheme in retrieving waves in highly heterogeneous media, such as subsalt or subbasalt. We focus on the initial focusing function that plays a key role in the iterative scheme, and propose an alternative focusing function that uses an estimate of the inverse transmission operator from a reference model that contains sharp contrasts (e.g., salt boundaries). Using a physics-driven estimate of the inverse transmission operator, we demonstrate that the new approach retrieves improved subsurface wavefields, including enhanced amplitudes and internal multiples, in a subsalt environment.

Reverse-Time Migration for Microseismic Sources Using the Geometric Mean as an Imaging Condition

Abstract: Time reversal is a powerful tool used to image directly the location and mechanism of passive seismic sources. This technique assumes seismic velocities in the medium and propagates time-reversed observations of ground motion at each receiver location. Assuming an accurate velocity model and adequate array aperture, the waves will focus at the source location. Because we do not know the location and the origin time a priori, we need to scan the entire 4D image (3D in space and 1D in time) to localize the source, which makes time-reversal imaging computationally demanding. We have developed a new approach of time-reversal imaging that reduces the computational cost and the scanning dimensions from 4D to 3D (no time) and increases the spatial resolution of the source image. We first individually extrapolate wavefields at each receiver, and then we crosscorrelate these wavefields (the product in the frequency domain: geometric mean). This crosscorrelation creates another imaging condition, and focusi...

Sparse Least-Squares Reverse Time Migration using Seislets

Abstract: We propose sparse least-squares reverse time migration (LSRTM) using seislets as a basis for the reflectivity distribution. This basis is used along with a dip-constrained preconditioner that emphasizes image updates only along prominent dips during the iterations. These dips can be estimated from the standard migration image or from the gradient using plane-wave destruction filters or structural tensors. Numerical tests on synthetic datasets demonstrate the benefits of this method for mitigation of aliasing artifacts and crosstalk noise in multisource least-squares migration.

Numerical Temporal Dispersion Corrections For Broadband Temporal Simulation, RTM and FWI

Abstract: A time-domain seismic simulator used to compute either the forward simulation of a source or the adjoint simulation of a recorded wave field is often implemented using a time stepping algorithm based upon a selected explicit or implicit finite difference approximation to either a first or a second time derivative. Any finite-order approximation provides a solution that suffers from some degree of temporal dispersion, particularly at higher frequencies. The temporal dispersion errors can be corrected via a resampling operation in the frequency domain when using this type of simulator for forward simulation, Reverse Time Depth Migration (RTM), and Full Waveform Inversion (FWI) gradients. This correction can impact how well broad-band RTM or FWI inversion results tie to well logs.

Full Waveform Inversion using Preserved Amplitude Reverse Time Migration

Abstract: A great deal of effort has been expended to improve the amplitude reliability of migration. The similarity of reverse time migration (RTM) to the gradient of full waveform inversion (FWI) indicates that preserved amplitude RTM can help improve FWI. We develop the theoretical derivation of an improved gradient for FWI based on common shot preserved amplitude RTM. We validate our approach on the Marmousi II model and the Chevron SEG 2014 dataset, showing that it significantly improves the convergence rate of FWI.

Practical Challenges in Adaptive Marchenko Imaging

Abstract: Through the multidimensional Marchenko equation, seismic redatuming can be expressed as a series. Unlike in conventional redatuming, where only the first term of these series is evaluated, not only primary reflections, but all orders of internal multiples are taken into account by this approach. By crosscorrelation of the redatumed wavefields with their corresponding (direct) source wavefields, as computed in a macro velocity model, a seismic image can be obtained without artifacts from internal multiples. Unfortunately, this approach requires accurate knowledge of the source signature, which is not always available. Moreover, the method is sensitive for source / receiver ghosts, coupling effects, attenuation and other noise. In practice, the individual terms in the series can also be added adaptively. This procedure is successful if internal multiples don’t interfere with primary reflections, but has its limitations in more complex media. In this contribution, we analyze the feasibility of adaptive addition, using only the first two terms in the series. The result appears useful for internal multiple suppression, as we illustrate on synthetic data with severe event interfererence and on field data.

Image-Based Q Tomography using Reverse Time Q Migration

Abstract: This paper first developed a technique to tomographically estimate a Q model from the migrated image using two-way wavefield continuation method. When compared with the previous work with a one-way downward-continuation method, this technique exhibits better capability of handling the steep structure, e.g. the salt flank. Numerical results on a complex model with a salt body demonstrate the effectiveness of this two-way method on resolving the overturned wave propagation introduced by the steep structures.

Full Waveform Inversion via Matched Source Extension

Abstract: Matched Source Waveform Inversion introduces additional degrees of freedom into waveform modeling in the form of trace-dependent source modification, allowing close data fit at all stages of the inversion process. Penalizing source modification leads to an optimization problem with the same global minimum as Full Waveform Inversion, but with less tendency to develop local minima caused by cycle-skipping. We implement and analyze a simple variant of this technique, using constant density acoustics and a transmission configuration, and explain the very close relation of this method to traveltime tomography for mildly heterogeneous velocity models. In common with other data domain waveform tomography methods, the Matched Source objective function may develop multiple local minima, despite avoiding cycle-skip in data residual, if multiple ray paths connect sources and receivers.

Estimation of Elastic Constants in HTI Media Using Gauss-Newton and Full-Newton Multi-Parameter Full Waveform Inversion

Abstract: We investigate the estimation of elastic constants for a fractured medium, using multi-parameter FWI when considering the naturally fractured reservoirs as an equivalent anisotropic medium. Multiparameter FWI remains exposed to a range of challenges, one of which being the cross-talk problem resulting from overlap of Fréchet derivative wavefields. Cross-talk is strongly influenced by the scattering patterns of different physical parameters, which govern the amplitude variations with varying scattering angle. In the numerical section, we illustrate the analytic and numerical scattering patterns of different elastic constants in HTI media for cross-talk analysis. We also analyze the role of multi-parameter approximate Hessian in suppressing cross-talk. The gradient vectors are also contaminated by the doubly-scattered energy in the data residuals. The second-order term in the Hessian, which we construct using the adjoint-state technique, can suppress the multi-parameter second-order scattering effects in the gradient. We apply Gauss-Newton and Full-Newton multiparameter FWI on several numerical examples to verify the role of multi-parameter Hessian in suppressing cross-talk and second-order scattering effects.