Volumetric curvature attributes for fault/fracture characterization

Abstract

Seismic attributes have proliferated in the last three decades at a rapid rate and have helped interpreters in making accurate predictions in hydrocarbon exploration and development. Attributes sensitive to amplitude, such as impedance inversion and AVO, are widely used for lithological and petrophysical prediction of reservoir properties. Other attributes, such as coherence and curvature, are particularly useful in mapping the structure and shape of geological features of interest. It is these latter types of attributes that are of interest for fault/fracture characterization, and in this paper we discuss the applications of volumetric curvature attributes for this purpose. Horizon-based curvature attributes (Roberts, 2001) have been used in seismic data interpretation for predicting fractures ever since Lisle (1994) demonstrated the correlation of curvature values to fractures measured on an outcrop. Different measures of curvature (Gaussian, strike, dip, etc.) have been shown by different workers to be highly correlated with fractures (Hart, 2002; Ericsson et al., 1988; Sigismondi and Soldo, 2003; Massaferro et al., 2003); many more applications can be found in Chopra and Marfurt (2007a, 2007b). As the name implies, horizon-based curvature is computed directly from a picked seismic horizon which in general requires that the data quality be good and that the horizon of interest corresponds to a prominent impedance contrast. Horizons picked on noisy surface seismic data or when picked through regions where no continuous surface exists can produce misleading curvature measures. A common means of addressing such problems is to spatially filter the horizon picks, with the goal of removing the noise and retaining features of geologic interest (Bergbauer et al., 2003; Chopra et al. 2006). Once picked and filtered, a mathematical quadratic surface is fitted to the picked data within a user-defined aperture. The different measures of curvature are then computed analytically from the coefficients of the quadratic surface. Roberts (2001) demonstrated the application of different curvature attributes including minimum and maximum curvatures, mean curvature, dip curvature, strike curvature, mostpositive and most-negative curvature, and shape index. Of this list, we find the most-positive and most-negative curvature measures to be the easiest to directly relate to commonly encountered geologic structural and stratigraphic features.