Summary
Curvature attribute has formed an integral part of seismic interpretation projects because of its efficient utility in detecting seismic discontinuity, seismic curve, especially in predicting fracture reservoir. However, due to getting curvature is a process of calculating the second derivative. It is sensitive to any noise. Therefore, we need to preprocess the horizon data. This study based on the actual data, after the horizon data filtered with 2D Gaussian iteration filter, we extract 13 kinds of curvature attributes for the target layer. We analyze the 13 kinds of curvature attribute and optimize them into 2 kinds. Finally, we give a picture of fracture prediction.
Introduction
Detecting faults and fractures from three-dimensional (3D) seismic data is one of the most significant tasks in subsurface exploration. Seismic attribute is an effective means. Seismic attributes which are derived from seismic data are some special measurements about kinematics, dynamics, and statistical properties. At present, we can extract more than 40 kinds of seismic attributes by using common seismic interpretation software. In recent years, curvature attribute which is a new kind of seismic attribute is developing rapidly because of its unique advantages in describing complex faults, fracture and curved structure.it already become an important technology of seismic interpretation after coherence cube technique. Roberts (2001)proposed the calculation method of curvature based on horizon time. Marfurt (2006) developed this horizon curvature to 3D volume curvature which avoided acquiring horizon time and some human factors. Deng (2013) proposed the gradient curvature attribute and it proved to be prior in small fault identification. Jing hua Gao (2014) gave another method of calculating 3D volume curvature based on dip scan with eccentric window. Marcus Cahoj and Kurt J Marfurt (2014)modeled the curvature characteristics of three types of transfer zones. Jie Qi (2014) used curvature attribute to describe the faultcontrolled karst.
Curvature attributes can be divided into two basic types: 1D curvature attributes and 2D curvature attributes and 2D curvature attributes are widely used in seismic data interpretation. It's even better than coherence slices in some respects. This study based on the actual data extracts 13 kinds of curvature attributes for the target layer and gives a result of fracture prediction.
