Seismic interpretation using predictive painting
MetadataShow full item record
Seismic interpretation plays a crucial role in extracting geologic information from seismic images in order to provide a better understanding of the earth's subsurface. Although there are different methods introduced in structural interpretation and stratigraphic interpretation to evaluate and predict reservoir properties, the challenge of predicting lithological and petrophysical properties of reservoir, revealing features that appear more subtle in conventional seismic data, and automating common interpretation tasks still remain hot topics among geophysicists. Developments in interpretation algorithms and attributes help interpreters to achieve a better understanding of features and properties of interest, leading to a better interpretation of seismic data which can dramatically reduce the possibility of dry wells in oil and gas exploration. Therefore, there is a room for improving the accuracy of current methods and algorithms and prompting novel ideas in the field of interpretation and attributes. This dissertation consists of six main parts. In the first part, I review the predictive painting method, which plays a fundamental role in the proposed methods. I employ predictive painting to generate horizon cubes and to spread information in 3D seismic volumes by following the local structure of seismic events. Next, I review nonlinear structure-enhancing filtering which is applied along seismic events in order to improve lateral continuity and remove random noise. Next, I introduce a new coordinate system and framework for seismic interpretation and processing. The stratigraphic coordinates are aligned with horizons, and the vertical direction in stratigraphic coordinates corresponds to the direction normal to the major reflection boundaries. In the presence of dipping layers any data processing and interpretation tasks in which the vertical direction is commonly assumed to be normal to reflection boundaries may yield biased and inaccurate results. In contrast, the stratigraphic coordinate system offers a local reference frame naturally oriented to sample the unbiased seismic waveform and, hence, promises to yield more accurate results. Next, I develop a novel attribute for highlighting faults and other discontinuities. The conventional coherence measures operate on a spatial window of neighboring traces and a temporal analysis window of samples above and below the analysis point and can hardly cope with non-stationarity in fault information. In contrast, the proposed method involves neither temporal nor spatial windows in coherence computation which honors non-stationary changes of fault information and achieves high resolution in both vertical and lateral directions. After that, I introduce a novel approach to computing volumetric curvature. The key idea is to transfer seismic image into a coordinate frame in which geometry follows the natural shape of each reflector, therefore assigning a horizon to each point in the seismic data volume. The proposed approach also enables multispectral curvature computation. Finally, I introduce a new method for integration of well-log and seismic data. Di fferent seismic attributes can be used to interpolate rock properties between well-logs but often seismic reflectors play no role in the essential task of guiding this process. In contrast, my proposed method honors both seismic image structures and the relationship of seismic amplitudes or other attributes to well-log properties which provides useful clues in the interpolation process.