Computation of curvature in seismic data

Abstract

Digital analysis of seismic images is a key component in the automation of seismic data interpretation. Other authors have explored using curvature as an attribute in this kind of analysis, but only on data which has been manually interpreted. In this project the subject is the use of curvature as a feature for automatic highlighting of saltdomes in seismic images. Saltdomes are of particular interest in seismic exploration, as they are often linked to hydrocarbon finds. By using the so-called dip to extract discrete curve segments representing the seismic horizons in the image, the local curvature may be estimated. This in turn requires approxi- mation of first and second order derivatives. Derivative approximation is done by locally fitting a parabola to each data point, and approximating the derivatives in that point by the derivatives of the parabola. The presented method is applied to a set of inline seismic saltdome images. The results show the saltdomes clearly highlighted from their surroundings, with distinct areas of positive and negative curvature values.