Haibin Di

Date of Graduation


Document Type


Degree Type



Eberly College of Arts and Sciences


Geology and Geography

Committee Chair

Dengliang Gao

Committee Co-Chair

Timothy Carr

Committee Member

Rong Luo

Committee Member

Ryan Shackleton

Committee Member

Thomas Wilson


In 3D subsurface exploration, detection of faults and fractures from 3D seismic data is vital to robust structural and stratigraphic analysis in the subsurface, and great efforts have been made in the development and application of various seismic attributes (e.g. coherence, semblance, curvature, and flexure). However, the existing algorithms and workflows are not accurate and efficient enough for robust fracture detection, especially in naturally fractured reservoirs with complicated structural geometry and fracture network. My Ph.D. research is proposing the following scopes of work to enhance our capability and to help improve the resolution on fracture characterization and prediction.;For discontinuity attribute, previous methods have difficulty highlighting subtle discontinuities from seismic data in cases where the local amplitude variation is non-zero mean. This study proposes implementing a gray-level transformation and the Canny edge detector for improved imaging of discontinuities. Specifically, the new process transforms seismic signals to be zero mean and helps amplify subtle discontinuities, leading to an enhanced visualization for structural and stratigraphic details. Applications to various 3D seismic datasets demonstrate that the new algorithm is superior to previous discontinuity-detection methods. Integrating both discontinuity magnitude and discontinuity azimuth helps better define channels, faults and fractures, than the traditional similarity, amplitude gradient and semblance attributes.;For flexure attribute, the existing algorithm is computationally intensive and limited by the lateral resolution for steeply-dipping formations. This study proposes a new and robust volume-based algorithm that evaluate flexure attribute more accurately and effectively. The algorithms first volumetrically fit a cubic surface by using a diamond 13-node grid cell to seismic data, and then compute flexure using the spatial derivatives of the built surface. To avoid introducing interpreter bias, this study introduces a new workflow for automatically building surfaces that best represent the geometry of seismic reflections. A dip-steering approach based on 3D complex seismic trace analysis is implemented to enhance the accuracy of surface construction and to reduce computational time. Applications to two 3D seismic surveys demonstrate the accuracy and efficiency of the new flexure algorithm for characterizing faults and fractures in fractured reservoirs.;For robust fracture detection, this study presents a new methodology to compute both magnitude and directions of most extreme flexure attribute. The new method first computes azimuthal flexure; and then implements a discrete azimuth-scanning approach to finding the magnitude and azimuth of most extreme flexure. Specially, a set of flexure values is estimated and compared by substituting all possible azimuths between 0 degree (Inline) and 180 degree (Crossline) into the newly-developed equation for computing azimuthal flexure. The added value of the new algorithm is demonstrated through applications to the seismic data set from Teapot Dome of Wyoming. The results indicate that most extreme flexure and its associated azimuthal directions help reveal structural complexities that are not discernible from conventional coherence or geometric attributes.;Given that the azimuth-scanning approach for computing maximum/minimum flexure is time-consuming, this study proposes fracture detection using most positive/negative flexures; since for gently-dipping structures, most positive is similar to maximum flexure while most negative flexure to minimum flexure. After setting the first reflection derivatives (or apparent dips) to be zero, the localized reflection is rotated to be horizontal and thereby the equation for computing azimuthal flexure is significantly simplified, from which a new analytical approach is proposed for computing most positive/negative flexures. Comparisons demonstrate that positive/negative flexures can provide quantitative fracture characterization similar to most extreme flexure, but the computation is 8 times faster than the azimuth-scanning approach.;Due to the overestimate by using most positive/negative flexure for fracture characterization, 3D surface rotation is then introduced for flexure extraction in the presence of structural dip, which makes it possible for solving the problem in an analytical manner. The improved computational efficiency and accuracy is demonstrated by both synthetic testing and applications to real 3D seismic datasets, compared to the existing discrete azimuth-scanning approach.;Last but not the least, strain analysis is also important for understanding structural deformation, predicting natural fracture system, and planning well bores. Physically, open fractures are most likely to develop in extensional domains whereas closed fractures in compressional ones. The beam model has been proposed for describing the strain distribution within a geological formation with a certain thickness, in which, however, the extensional zone cannot be distinguished from the compression one with the aid of traditional geometric attributes, including discontinuity, dip, and curvature. To resolve this problem, this study proposes a new algorithm for strain reconstruction using apparent dips at each sample location within a seismic cube.