Multicomponent seismic data registration for subsurface characterization in the shallow Gulf of Mexico

Abstract

Abstract Using multicomponent ocean-bottom seismic technology, it is possible to obtain robust estimates of several subsurface parameters. We apply an automatic data registration (warping) algorithm to find a mapping between P-wave and converted wave migrated images. The algorithm improves the matching of the two seismic volumes obtained by previous manual interpretation. There are two main products of this process. First, it improves a combined interpretation of the gas cloud zones that are obscured in the conventional one-component (P-wave) seismic images. Second, interval Poisson ratios get extracted directly from the warping function. This extraction provides a petrophysical characterization of the subsurface at a resolution unobtainable by other methods. Introduction Multicomponent seismic exploration with ocean-bottom technology opens new possibilities for improving seismic imaging and for extracting valuable additional information about subsurface physical characteristics (Stewart et al, 2003). The benefits of using multicomponent data have been proven for imaging through gas clouds and identifying shallow gas hazards (Granli et al, 1999; Englehart et al, 2001; Knapp et al, 2001). Joint interpretation of multiple image components (P-P and P-S images) depends on our ability to identify and register reflection events from similar reflectors. This task is especially challenging in shallow sediments, where the high ratio of Pand S-velocities causes large differences in the corresponding traveltimes. DeAngelo et al (2003) describe a careful strategy of joint P-P and P-S interpretation with application to subsurface characterization in the shallow Gulf of Mexico. In this study, we extend the registration procedure with an accurate automatic algorithm for representing P-S reflection events in the corresponding P-P time. By examining the algorithm performance on simple synthetic data, we observe that the differences in the frequency content of the P-wave and converted-wave data have a major impact on the registration accuracy. We implement a non-stationary spectral balancing method to take these differences into account. Balanced images are then automatically registered (warped) to estimate a point-by-point correlation function. The time derivative of this function produces the time-variable ratio of the P and S seismic velocities. This ratio and the related Poisson ratio are major physical attributes useful for interpreting subsurface structures. Our method enables extracting them directly from time-migrated P-P and P-S images. Application of this technique to data from the Gulf of Mexico reveals the structure of sediments around shallow gas clouds with a resolution unobtainable by other methods. Theory If we denote a P-P seismic image as a function of the vertical P-P traveltime t as P(t) and the corresponding converted-wave image as a function of the vertical P-S traveltime ? as C(?), then the relationship between the two images can be expressed as Mathematical equation (1)(Available in full paper) where w(t) is the warping function establishing the correspondence of reflection events in the two images, and a(t) is the amplitude gain function compensating for the difference in reflectivity. We assume the reflection events to be correctly positioned laterally in the migrated images so that the differences can be explained by vertical transformations only.