EAGE/SEG Research Workshop 1990

One of the most commonly used tool for obtaining stacking velocities is the velocity spectrum. An expert system for aid in the interpretation of velocity spectra on a profile is presented. The raw velocity spectra computed by conventional seismic data processing software and a simplified description of the geological and geophysical context of the survey are input data for the system. In the expert system, the reasoning of the geophysicist picking a velocity spectrum is expressed in the form of rules using the objects of the velocity spectrum (spectrum extrema) and the information about the survey context. The generating of the search space and its filtering are carried out with heuristics specific to geophysical criteria and adapted to the velocity analysis problem.

In mid-1989, Shell Canada acquired a 3D seismic survey, Ram 3D, in the foreland basin of the Canadian Rocky Mountains. The surveyed area includes steeply dipping layers from surface outcrop to Precambrian basement. The azimuths of these layers vary greatly throughout.

For some time the seismic processing industry has possessed the algorithmic capability to accurately migrate seismic data from areas of structural complexity. What has stopped this ability being used in practice is the fact that the required depth migrations are themselves very sensitive to the supplied velocity depth model.

In this paper a macro model estimation technique is presented based on non-recursive wave field extrapolation. By shot record redatuming, using an initial macro model, true Common Depth Point (CDP-) gathers are generated at grid points along one or more lateral positions (such as potential boreholes).

Time domain processing has been extensively developed during the eighties to the stage that, with DMO, good quality images can be formed for most situations. However, in structurally complex areas, converting the time image to a correct depth image has been a serious problem. Recently, prestack depth migration has attracted more and more attention because of its capability of placing the image at the correct location, laterally and vertically.

There are basically two approaches to the dipping layers problem: the combination of normal move-out (NMO) and dip move-out (DMO) followed by post-stack migration and depth conversion, and full pre-stack depth migration.

Several methods for dip moveout (DMO) and prestack Imaging (PSI) have reduced the operations to invariant convolutlons by resampling the data.

Summary not available

In order to obtain from seismic data a good image of complex geologic structures we must meet several demands. First among these is a correct velocity model whose detail is related to the complexity of the structures and which is geologically plausible. In order to limit our search for the velocity model to part of the model space we should incorporate in our model construction a priori geophysical and geological constraints on the model. My presentation deals with these limits on the velocity model.

A velocity analysis process, SIVA, is described which combines the information in CMP gathered trace data with interpreted seismic data (eg, time maps) to create a model of structure and P wave interval velocity.