Report to Landmark Graphics Corporation, University Partnership Program
MetadataShow full item record
Seismic processing requires accurate knowledge of the earth's velocity structure to properly image and interpret multi-fold seismic data. Conventional analysis methods are based on determining the best fit hyperbolas to seismic travel-times, T, as a function of source-receiver offset, X, in CMP gathers. The results of this analysis are the two-way normal-time, and the stacking velocity for each event analyzed. If the source-receiver offsets are not too large compared to the reflector depth, the stacking velocities can be equated to the RMS velocity. From knowledge of the RMS velocity and two-way normal times above and below a zone of interest, the interval velocity can be determined. Even if the earth is truly one-dimensional, i.e., velocity varies only as a function of depth, errors arise from the departure of the actual travel-times trajectories from the assumed T(X) hyperbola and the departure of the stacking velocity from the RMS velocity. These errors are in addition to the uncertainties involved in determining both the stacking velocity and the two-way normal-times from limited offset, band limited data in the presence of coherent and random noise. An alternative interval velocity analysis method can be implemented if we first perform a plane wave decomposition of the seismic data. By transforming the data to the the domain of intercept time, t, and horizontal ray parameter, p, velocity analyses can be performed exactly for a one-dimensional earth model without the need for intermediate quantities such as the stacking and RMS velocities. Workstation technology, such as the LandmarkTM, can then be used to do this velocity analysis interactively. For example, the original seismic data are plane wave decomposed on a remote computer, e.g., a Cray, and are then transferred either via ethernet or tape to the Landmark for interpretation. The interpretation is done directly in the ?-p domain by interactively 3 defining ?-p travel time curves and superimposing these curves on the ? -p data. Once reasonable agreement is achieved, the plane wave data are NMO corrected in the ?-p domain and then redisplayed. (The NMO corrections can be to two-way time or to depth.) The interpretation procedure is now repeated in the NMO domain to refine the velocity depth structure. The data can be windowed in time and ray parameter prior to analysis and the window changed during the interpretation process. The parameters determined directly by the interpreter are the interval velocity and the thickness and/or two-way normal-time of each layer. No approximations are required and all source receiver offsets are implicitly included in the analysis.