- Copyright © 2001 Society of Exploration Geophysicists
Modern visualization and image processing techniques are revolutionizing the art of seismic interpretation. Emerging technologies allow us to interpret more data with higher accuracy in less time. The trend is shifting from horizon-based toward volume-based. New insights are gained by studying objects of various geologic origins and their spatial interrelationships. The standard way of highlighting objects is through seismic attribute analysis. Various attributes are tested in a trial-and-error mode, and one is selected as the optimal representation of the desired object. The selected attribute, which may be a mathematical composite of several attributes, is not sensitive to a particular geologic object but highlights any seismic position with similar attribute response.
We set out to develop a seismic-object detection method that in our opinion produces more accurate results and does not require expert knowledge. The method recombines multiple attributes into a new attribute that gives the optimal view of the targeted object. Including specific spatial knowledge about the targeted object allows us to separate objects of different geologic origin with similar attribute characteristics. The method comprises an iterative processing workflow using directive seismic attributes (i.e., attributes steered in a user-driven, or data-driven, direction), a neural network, and image processing techniques (Meldahl et al., 1999). Figure 1 is a generalized workflow of the object detection method, which has a worldwide, patent-pending status. Objects that can be detected by the method include faults, reflectors, seismic chimneys, time-lapse differences, stratigraphic features, and direct hydrocarbon indicators. The first products from simple application of the method are named The Chimney Cube and The Fault Cube.
This paper presents the basic concepts of the technology. Examples of the method to detect faults and chimneys are shown. Special emphasis is given to seismic chimneys …