Zhang Guangzhi

Dispersion-dependent attribute and application in hydrocarbon detection

Previous work has demonstrated that seismic velocity dispersion is often related to a hydrocarbon-saturated reservoir with a high value of attenuation. The velocity dispersion properties can be used as new hydrocarbon indicators if they can be estimated quantitatively from seismic data. In this context, based on the assumption that the reflection coefficient of the interface between dispersive media varies with frequency, we propose an approximation of the reflection coefficient which involves the dispersion-dependent attributes indicating the level and gradient of P-wave velocity dispersion. We develop a new generalization of an AVO inversion scheme to extract the dispersion-dependent attributes from multi-frequency prestack seismic data obtained by the Morlet wavelet decomposition method. The method test and real application show that the dispersion-dependent attributes have the potential to be useful in the discrimination of hydrocarbon accumulation. However, the hydrocarbon detection quality of the dispersion-dependent attributes can be degraded by the inverse Q-filtering procedure which compensates the high-frequency amplitudes.

Application of wavelet analysis in distinguishing seismic sequence

Wavelets are a relatively recent development in applied mathematics. With the local time-frequency of wavelet transforms, we can deal much more with the signal reflected by thin-mutual strata. Therefore, we have put forward a method to distinguish the seismic sequence using the wavelet analysis theory. Through the seismic sequence analysis using wavelet transforms, we can obtain vertical and horizontal phase profiles from real seismic data. We will distinguish deposition cycle and material components in the vertical phase profile, and compare and track deposition sequence and structure interface horizontally in the horizontal phase profile.

Noise suppressing by wavelet transform in seismic data processing

This paper gives a new approach to suppress noise in seismic data. According to the time-frequency local characteristics of wavelet transform and the optimal filtering characteristics of KL transform, we have advanced this approach. On the basis of the fact that the noise energy is different from reflection energy in each frequency range, prestack seismic data is divided into different frequency band by wavelet transform first, and then it is filtered by KL transform. Synthetic and real data processing results show that the method can desirably eliminate noise.

Application of support vector machine in prediction of reservoir parameters

The conventional method is not performing well in reservoir parameters prediction because of lacking learning samples. The support vector machine method could help us in this situation. We repeat an experiment to verify the excellent generalization ability of SVM. Four applications of real data processing were done by us, and they were all working very well. The result shows that this method would bring us to a nice place.

Estimation of porosity and saturation by Markov Chain Monte Carlo prestack seismic inversion

Porosity and saturation are important parameters to describe the reservoir properties. Here, we develop a method to inverse porosity and saturation within a Bayesian framework using Markov Chain Monte Carlo (MCMC) method. First, we discuss the changes of P-wave velocity, S-wave velocity, density due to changes in porosity and saturation. Then, we build a three-layer model to do the inversion making use of the full Knott-Zoeppritz equation to get the seismic AVO data. The advantages of MCMC methods are: (1) the accuracy of solution is improved with the prior information joined in, and the uncertainty is greatly reduced; (2) solving the nonlinear problem directly, so the process of inversion is to do the seismic forward in each iteration; (3) the dependence of results and initial values is not strong, because the algorithm are not use the objective function with a single optimal solution; (4) optimization process can jump out of local optimum, and ultimately obtain the global optimal solution. By testing a layered model, shows the MCMC method based on Metropolis-Hastings algorithm is available and can obtain good results by random searching solution space.

Using Wavelet Transform to Extract Seismic Attributes

Seismic attributes have been widely used since their introduction and have already become a valid analytical tool for lithology prediction and reservoir characterization; wavelet transform based on the matching pursuit analysis approach using a specific type of Gabor-atoms, it can extract useful seismic attributes that reflect reservoir property, and can be used to divide favorable reservoir area; model and real data application shows that the proposed method is simple and effective and can be easily used

3D F-T analysis of seismic profile

On the basis of frequency time (F-T) analysis for a single seismic trace, this paper studies the 3D F-T analysis of a seismic profile using the wavelet transform. Once the seismic profile is transformed into a 3D X-T-F data cube, seismic sequence analysis can be done on the 3D cube. Distinguishing the seismic sequence cycle and determining the sequence interface location can be realized in the F-T slice. Tracing continuity characteristics of the sequence interface and picking geological body boundaries and tectonic faults can be implemented in the X-T slice. It becomes the basis for 4D F-T analysis of the 3D seismic data cube and for quantitative interpretation on 4D X-Y-T-F cubes.