Layer | Fill | Outline |
---|
Map layers
Theme | Visible | Selectable | Appearance | Zoom Range (now: 0) |
---|
Fill | Stroke |
---|---|
Collaborating Authors
Results
Nonstationary signal decomposition via improved complete ensemble empirical mode decomposition and its application in ground roll noise attenuation
Chen, Wei (Yangtze University) | Chen, Yangkang (University of Texas–Austin) | Liu, Wei (China University of Petroleum-Beijing) | Cheng, Zixiang (University of Southern California)
It has been used to attenuate et al., 2015; Jiao et al., 2015; Chen et al., 2015b). Because both random noise and coherent ground roll by removing the of the apparent difference between the useful reflections and first one or two decomposed components in each frequency the coherent steeply dipping linear noise, one of the most effective slice, which acts as a dip filter to separate different wavenumber ways to remove the linear noise is by applying a dip components. The mode-mixing problem is the biggest filter which rejects passing the spatially non-stationary components drawback of this decomposition technique, which refers to (linear noise) through the filter. FK-based filter is one the phenomenon that each decomposed component is related of the simplest dip filters, in which the 2D fast Fourier transform with multiple oscillating frequencies. Noise-assisted variations (FFT) is utilized. Milton et al. (2009) proposed a SVD of EMDs, like ensemble empirical mode decomposition based dip filter to attenuate the ground roll noise, which is not (EEMD) and complete ensemble empirical mode decomposition an adaptive method and thus requires some parameters tuning (CEEMD), can solve the mode-mixing problem to some efforts. Bekara and van der Baan (2009) initially utilized the extent but will cause other problems, such as strong residual EMD based dip filter to attenuate the ground roll.
Improving the S-transform for high-resolution reservoir prediction and paleochannels delineation
Cheng, Zixiang (University of Southern California) | Chen, Wei (Yangtze University) | Chen, Yangkang (University of Texas–Austin) | Liu, Ying (China University of Petroleum) | Liu, Wei (China University of Petroleum) | Li, Huijian (SINOPEC Exploration and Production Research Institute)
ABSTRACT Time frequency analysis of seismic data is a flexible and robust way for characterizing the subsurface properties with high-resolution. There have existed a lot of time-frequency decomposition algorithms in the literature, among which the S transform based approaches still serves as one of the most widely used ways for time-frequency analysis because of its simple implementation, strong robustness, and high-fidelity delineation performance. It combines the separate strengths of the STFT and wavelet transforms with scale dependent resolution by using Gaussian windows, scaled inversely with frequency. One problem with the use of traditional symmetric Gaussian window is degradation of time resolution in the time-frequency spectrum due to the long front taper. In this abstract, we study the performance of an improved S transform with a bi-Gaussian window used to construct asymmetry bi-Gaussian windows. The asymmetry bi-Gaussian can obtain an increased time resolution in the front direction. This increased time resolution will result in a high-resolution event picking and a significantly improved time-frequency characterization for oil&gas traps prediction. We applied the slightly modified bi-Gaussian S transform to a synthetic trace, a 2D seismic section, and a 3D seismic cube to show the superior performance of the bi-Gaussian S transform in analyzing nonstationary signal components, hydrocarbon reservoir predictions, and paleo-channels delineations with an obviously higher resolution. Presentation Date: Tuesday, October 18, 2016 Start Time: 1:00:00 PM Location: 170/172 Presentation Type: ORAL
- Geophysics > Seismic Surveying > Surface Seismic Acquisition (1.00)
- Geophysics > Seismic Surveying > Seismic Processing (1.00)
- North America > United States > Texas > Fort Worth Basin > Boonsville Field (0.99)
- Asia > China > Xinjiang Uyghur Autonomous Region > Junggar Basin (0.99)
ABSTRACT In frequency-space domain, mixed-grid finite-difference (FD) schemes are widely used and effectively improve the forward modeling accuracy. In this article, by introducing mixed-grid FD schemes from frequency-space domain to time-space domain, we propose a new kind of Mixed-grid 2M+N (MG2M+N) FD methods, and derive the method to calculate the FD coefficients based on time-space domain dispersion relationship. Then we carry out dispersion analysis. Dispersion analysis results show that MG2M+N FD methods can suppress numerical dispersion more effectively, but larger N value with increasing computational amount can hardly improve the dispersion characteristic, so MG2M+1 (N=1) should be the first choice. Comparing to traditional high-order (T2M) FD methods and time-space domain high-order (TS2M) FD methods, with almost the same computational amount, MG2M+1 FD methods have the lowest numerical dispersion and the highest modeling accuracy. In the end, we implement numerical modeling experiments on a homogeneous model and a layer model, which further verify the inclusions drawn from dispersion analysis and demonstrate the validity and applicability of MG2M+N FD methods. Presentation Date: Wednesday, October 19, 2016 Start Time: 1:55:00 PM Location: 161 Presentation Type: ORAL