Layer | Fill | Outline |
---|
Map layers
Theme | Visible | Selectable | Appearance | Zoom Range (now: 0) |
---|
Fill | Stroke |
---|---|
Collaborating Authors
Results
High-fidelity full-waveform inversion with an initial velocity model from multiple wells interpolation
Chen, Yangkang (University of TexasโAustin) | Xiang, Kui (China University of PetroleumโBeijing) | Chen, Hanming (China University of PetroleumโBeijing) | Chen, Xiaohong (China University of PetroleumโBeijing)
ABSTRACT Full waveform inversion (FWI) is a promising technique for inverting a high-resolution subsurface velocity model. The success of FWI highly depends on a fairly well initial velocity model. We propose a method for building a remarkable initial velocity model that can be put into the FWI framework for inverting nearly perfect velocity structure. We use a well log interpolated velocity model as a high-fidelity initial model for the subsequent FWI. The interpolation problem is solved via a least-squares method with a structural regularization. In order to obtain the geological structure of subsurface reflectors, an initial reverse time migration (RTM) with a fairly realistic initial velocity model is used to roughly calculate the local slope of subsurface structure. The well log interpolated initial velocity model can be very close to the true velocity while having small velocity anomaly or over-smoothing caused by the imperfect velocity interpolation, which however can be compensated during the subsequent FWI iterations. Regarding the field deployment, we suggest that future drilling should be seismic-oriented, which can help fully utilize the well logs for building initial subsurface velocity model and will facilitate a wide application of the proposed methodology. Presentation Date: Wednesday, October 19, 2016 Start Time: 4:25:00 PM Location: 162/164 Presentation Type: ORAL
Summary Observed seismic data is always irregular and seismic data interpolation is an essential procedure to provide accurate complete data for Surface Related Multiple Elimination (SRME) and wave equation based migration and inversion. While most interpolation methods belong to iterative method, how to define reconstruction error reasonably for terminating iterations duly is important for efficient seismic data interpolation. In this abstract, Projection Onto Convex Sets (POCS) method is achieved through the view of Iterative Hard Threshold (IHT) method and a novel reconstruction error definition is proposed with the information related with the observed seismic data. Tests on synthetic and real datasets demonstrate the validity of the proposed method. Introduction Acquired seismic data is always irregular sampled in spatial coordinates because of the presence of obstacles, forbidden areas, feathering and dead traces. Since multi-channel processing techniques, such as Surface Related Multiple Elimination (SRME), wave-equation based migration and inversion, require complete seismic data, seismic data interpolation technique, which can provide accurate complete seismic data for these multi-channel processing methods, is becoming an essential stage. Interpolation methods can be divided into four categories (Gao et al., 2012; Wang et al., 2014): mathematical transform-based methods, prediction filtering-based methods, wave equation-based methods and rank reduction -based methods. While most interpolation methods belong to the category of iterative methods, the reconstruction error definition becomes essential for efficient seismic data interpolation. Gao et al. (2012) gave two reconstruction error definitions to monitor the convergence of the iterative interpolation methods: the first one uses the original complete seismic data and can only be used in theoretical research; the second one uses the adjacent iterative solutions and may trap in local minimum which leads to unsatisfactory interpolation results. In this abstract, firstly, Projection Onto Convex Sets (POCS) method is achieved in the view of Iterative Hard Threshold (IHT) method; secondly, a novel reconstruction error definition is proposed with the information related with the observed seismic data. Tests on synthetic and real data prove the validity of the proposed method.
Summary Projection Onto Convex Sets (POCS) method is an efficient iterative method for seismic data interpolation. In each iteration, observed seismic data is inserted into the updated solution, therefore it has difficulty for interpolation in noisy situations. Weighted POCS method can weaken the noise effects because it uses a weight factor to scale the observed seismic data, then fewer noisy data is inserted into the updated solution, but it still inserts some random noise. In this abstract, a novel method is proposed by combining the advantages of the weighted POCS method and the Iterative Hard Threshold (IHT) method: the weighted POCS method used for interpolation and the IHT method used for random noise elimination. The novel method can be used for simultaneous interpolation and random noise removal of seismic data, and its validity is demonstrated on synthetic and real datasets. Introduction Spatial irregularity and random noise observed in seismic data can affect the performance of Surface-Related Multiple Elimination (SRME), wave-equation based migration and inversion. Therefore, interpolation and random noise elimination is pre-requisite for multi-channel processing techniques. Interpolation methods can be divided into four categories (Gao et al., 2012; Wang et al., 2014): mathematical transform based methods, prediction filters based methods, wave-equation based methods and rank-reduction based methods. Among these methods, mathematical transform based methods are easy to implement and have drawn much attention. While the random noise in observed seismic data can affect the interpolation performance and the irregularity of observed data can also affect the results of random noise elimination. Therefore, simultaneous interpolation and random noise attenuation is developed (Naghizadeh, 2012; Oropeza and Sacchi, 2011), while it is suitable for linear or quasi-linear events and should be handled window by window for curved events.
Seismic Data Reconstruction via Fast Projection onto Convex sets in the Seislet Transform Domain
Gan, Shuwei (China University of Petroleum, Beijing) | Wang, Shoudong (China University of Petroleum, Beijing) | Chen, Yangkang (The University of Texas at Austin) | Chen, Xiaohong (China University of Petroleum, Beijing)
Summary According to the compressive sensing (CS) theory in the signal-processing field, we proposed a new seismic data reconstruction approach based on a fast projection onto convex sets (POCS) algorithm with sparsity constraint in the seislet transform domain. The FPOCS can obtain much faster convergence than conventional POCS (about two thirds of conventional iterations can be saved). The seislet transform based reconstruction approach can achieve obviously better data recovery results than f โk transform based scenarios, considering both signal-to-noise ratio (SNR) and visual observation, because of a much sparser structure in the seislet transform domain. Both synthetic and field data examples demonstrate the performance of the proposed approach. Introduction Due to different reasons, seismic data may have missing traces. Seismic data reconstruction is such a procedure to remove sampling artifacts, and to improve amplitude analysis, which is very important for subsequent processing steps including highresolution processing, wave-equation migration, multiple suppression, amplitude-versus-offset (AVO) or amplitude-versusazimuth (AVAZ) analysis, and time-lapse studies (Trad et al., 2002; Liu and Sacchi, 2004; Abma and Kabir, 2005, 2006; Wang et al., 2010; Naghizadeh and Sacchi, 2010; Li et al., 2012, 2013; Chen et al., 2014a). In recent years, because of the popularity of compressive sensing (CS) based applications (Cand`es et al., 2006b), there exists a new paradigm for seismic data acquisition that can potentially reduce the survey time and increase the data resolution (Herrmann, 2010). Compressive sensing (CS) is a relatively new paradigm in signal processing that has recently received a lot of attention. The theory indicates that the signal which is sparse under some basis may still be recovered even though the number of measurements is deemed insufficient by Shannonโs criterion. The principle of CS involves solving a least-square minimization problem with a L1 norm penalty term of the reconstructed model, which requires compromising a least-square data-misfit constraint and a sparsity constraint over the reconstructed model. The iterative shrinkage thresholding (IST) and the projection onto convex sets (POCS) are two common approaches used to solve the minimization problem in the exploration geophysics field.
ABSTRACT The sparse Radon transform (RT) represents seismic data by the superposition of a few constant amplitude events, and thus it has trouble dealing with amplitude-versus-offset (AVO) variations. We integrated the gradient and curvature parameters of AVO into the RT. With these additional properties, the lateral continuity of the eventsโ amplitude was modeled in the transformation and it could be fitted with orthogonal polynomials. This resulted in a higher order RT, which included AVO terms. The high-order RT is a highly underdetermined problem, which was solved by extracting the major model parameters from energy distribution in a high-order Radon domain and by decreasing the number of inversion parameters. Thus, a high-order sparse RT was achieved. The proposed method can be used for data interpolation as well as extrapolation. The AVO-preservation performance of the proposed algorithm in data reconstruction was illustrated using both synthetic and field data examples, and the results showed the feasibility of the method.
ABSTRACT Rank reduction strategies can be employed to attenuate noise and for prestack seismic data regularization. We present a fast version of Cadzow reduced-rank reconstruction method. Cadzow reconstruction is implemented by embedding 4D spatial data into a level-four block Toeplitz matrix. Rank reduction of this matrix via the Lanczos bidiagonalization algorithm is used to recover missing observations and to attenuate random noise. The computational cost of the Lanczos bidiagonalization is dominated by the cost of multiplying a level-four block Toeplitz matrix by a vector. This is efficiently implemented via the 4D fast Fourier transform. The proposed algorithm significantly decreases the computational cost of rank-reduction methods for multidimensional seismic data denoising and reconstruction. Synthetic and field prestack data examples are used to examine the effectiveness of the proposed method.
- North America > Canada > Saskatchewan > Western Canada Sedimentary Basin > Alberta Basin (0.99)
- North America > Canada > Northwest Territories > Western Canada Sedimentary Basin > Alberta Basin (0.99)
- North America > Canada > Manitoba > Western Canada Sedimentary Basin > Alberta Basin (0.99)
- (2 more...)
We have developed a novel method for random noise attenuation in seismic data by applying regularized nonstationary autoregression (RNA) in the frequency-space () domain. The method adaptively predicts the signal with spatial changes in dip or amplitude using RNA. The key idea is to overcome the assumption of linearity and stationarity of the signal in conventional domain prediction technique. The conventional domain prediction technique uses short temporal and spatial analysis windows to cope with the nonstationary of the seismic data. The new method does not require windowing strategies in spatial direction. We implement the algorithm by an iterated scheme using the conjugate-gradient method. We constrain the coefficients of nonstationary autoregression (NA) to be smooth along space and frequency in the domain. The shaping regularization in least-square inversion controls the smoothness of the coefficients of RNA. There are two key parameters in the proposed method: filter length and radius of shaping operator. Tests on synthetic and field data examples showed that, compared with domain and time-space domain prediction methods, RNA can be more effective in suppressing random noise and preserving the signals, especially for complex geological structure.
ABSTRACT We propose a novel method for random noise attenuation in seismic data by applying nonstationary autoregression (NAR) in frequency-space (f, -x, ) domain. The method adaptively predicts the signal with special changes in dip or amplitude using f, -x, NAR. The key idea is to overcome the assumption of linearity and stationarity of the signal in conventional f, -x, deconvolution technique. The conventional f, -x, deconvolution uses short temporal and spatial analysis windows to cope with the nonstationary of the seismic record. The proposed method does not require windowing strategies in spatial direction. We implement the algorithm by iterated scheme using conjugate gradient method. We constrain the coefficients to be smooth along space and frequency in f, -x, domain. The shaping regularization in least square inversion controls the smoothness of the coefficients of f, -x, NAR. There are two key parameters in the proposed method: filter length and radius of shaping operator. Synthetic and field data examples demonstrate that, compared with f, -x, deconvolution, f, -x, NAR can be more effective in suppressing random noise and preserving the signals, especially for complex geological structure.
ABSTRACT Rank reduction strategies can be employed to attenuate noise and as a basic template for pre-stack regularization of seismic data. We propose to utilize the rank reduction method of Multichannel Singular Spectrum Analysis (MSSA) to implement a fast 5D seismic data reconstruction method by embedding 4D spatial data into a block Toeplitz matrix and rank reduce this matrix via the Lanczos bidiagonalization algorithm, rather than using Singular Value Decomposition (SVD). The computational cost of the Lanczos bidiagonalization is dominated by the cost of multiplying a block Toeplitz matrix by a vector. The latter can be efficiently implemented via multidimensional Fast Fourier Transforms. The proposed algorithm significantly decreases the computational cost of the rank-reduction stage needed for de-noising and reconstruction with respect to algorithms that utilize the Singular Value Decomposition (SVD). In essence, our algorithm exploits the structure of block Toeplitz matrices to accelerate the rank-reduction step of de-noising and reconstruction strategies used by Multichannel Singular Spectrum Analysis (MSSA) or Cadzow matrix completion methods. Synthetic data examples and a field data test were used to examine the proposed algorithm.