Technology advances, such as WAZ/FAZ acquisition, reverse time migration (RTM) and iterative salt imaging in model building, have greatly improved subsalt image quality. However subsalt imaging under complex salt overburden is still challenging. The fundamental issue lies with the imperfect illumination from limited acquisition. In the poor illumination zones, the migration artifacts and coherent noises, such as residual multiples and converted waves, become prevalent and can severely contaminate the images and mislead the interpretation. Migration vector offset output provides a domain in which the true imaging energy and noise can be separated as the imaging contributions focus in different vector offset tiles according to the subsurface structure orientation (azimuth and dip). Thus selective stacking the migration vector offset outputs based on the structural interpretation can achieve migration noise reduction and improve subsalt images.
This paper presents a novel approach for estimating thin bed thickness by using gradient of spectral amplitudes resulting from spectral decomposition. The gradient is calculated from the differences between amplitudes spectrum at two adjacent frequencies. The technique is based on the concept that seismic reflections from a thin bed have characteristic signatures in the frequency domain. Thicker beds have shorter period of spectral notches than thinner beds have. Consequently, steeper gradients of spectral amplitude representing thicker beds and lower gradients representing thinner beds. In this paper, the Discrete Fourier Transform (DFT) method is used for spectral decomposition. The application of the technique for mapping thin bed thickness distribution is presented. The gradient of spectral amplitude has been found to have better correlation to thickness than the conventional spectral amplitude to thickness.
Today''s advanced borehole acoustic analysis not only enhances the rock physic characteristics of shale gas plays but it also provides new insights in geomechanics applications and fracture identification. Current borehole acoustic tools can detect the azimuthal and transverse shear wave intrinsic rock anisotropy. They also help identify stress-sensitive formations and help in the evaluation of natural fractures, those that intersect the borehole as well as those that do not.
The Marcellus shale has become an active area for gas exploration, and has fracture and lithology characteristics that make it a good candidate for multicomponent data. It is well established that similar vertical wavelength ranges must be preserved in multicomponent data and that wavelengths of P- and S-waves must match in order to sample reflectivity in an equivalent manner. Although registration functions align corresponding stratigraphic events of the P-wave and P- to S-wave (C-wave) reflections, they distort the seismic wavelet because global average velocity properties are independent of local interval properties that define wavelength. In this study we apply a velocity-based wavelet correction method to C-waves, which are expressed as a function of interval and average
In this paper we present a new trace interpolation and denoising method in the frequency-space (
Shang, Yongsheng (BGP, CNPC) | Wang, Changhui (BGP, CNPC) | Zhang, Mugang (BGP, CNPC) | Zhou, Xuefeng (China University of Petroleum) | Dong, Lieqian (China University of Petroleum) | Zhenchun, Li (China University of Petroleum) | Fenglei, Li (China University of Petroleum)
Slip sweep recording offers the potential to dramatically improve vibroseis production rates and to reduce acquisition costs. However, slip sweep data is contaminated by harmonic noise, because one vibrator begins sweeping before the previous sweep has terminated. We present a new method for eliminating harmonic distortion in slip sweep data. The method is derived using only the recorded ground force and the start times for the sweeps. It is simple, robust, and fast and can be easily implemented. It has been applied to synthetic and real data and gives satisfactory practical results.
Compressive inversion is a technique used to reduce the dimensionality of the data space to improve the efficiency of inversion. We develop a compressive inversion method for inverting large-scale multichannel geophysical datasets. We then develop the framework to apply this technique to airborne electromagnetic (AEM) surveys to allow for the rapid inversion of airborne time-domain data.
Based on neglecting the anomalous electromagnetic (EM) fields under Green''s integral operator, Born Approximation is an important approach to obtain the fast solution of the EM integral equations (IE). In the half-space conductivity, Born Approximation could precisely reflect the truly information of the anomalous conductivities embedded in the half-space, which the data is collected on the ground. Comparing to the method of finite difference (FD) and finite element (FE), IE could solve the problem in comparative low time-consuming and high precise, while the “anomalous induction number” does not get too large. Take advantage of the method of IE, the abstract computes the EM field induced by anomalous conductivity in half-space region. Born inversion is also considered basing on the work has been done previously. It demonstrates that Born method, no matter of the forward modeling or inversion, is a high useful and efficient way in geophysical data processing.
Seismic trace interpolation can be formulated as an underdetermined least squares inverse problem. In order to force interpolation to follow the directions of seismic events, the weighting function is designed to build the interpolated energy along the desired direction. Usually, such a weighting function is built in the Fourier domain. However, when Fourier spectra are aliased, especially in the case of up-sampling seismic traces (creating more output than input), the design of the weighting function faces problems. In this paper, we discuss these problems and a proposed solution using the f-p domain.
Seismic image flattening plays an important role in both interpretation and processing of seismic data. We introduce a new algorithm for seismic image flattening. The algorithm transfers seismic data to a new coordinate system and consists of two steps. First, local slopes of seismic events are estimated by plane-wave-destruction , then information is spread along the seismic data and horizons are picked by predictive painting algorithm. These picked horizons can be considered as level sets of the first axes of the new coordinate system. Next, an upwind finite-difference scheme is used to find the other axes which are perpendicular to the first axis by solving the relevant gradient equation. This method may find different applications in interpretation, velocity analysis, and seismic imaging. We demonstrate the performance of our method using synthetic and real data examples.