Two current challenges in seismic imaging are to obtain more detailed images of complex structures from reflection data and to constrain the regional structure of the Earth using wide-angle data. These are complex problems for which traditional methods fail because they are based on too many simplifying assumptions. I develop a unified approach which addresses these tasks by starting with a fundamental problem formulation, leading to a practical numerical solution which converges rapidly. It is based on pre-stack depth migration and cell-stripping tomography in heterogeneous media, which allow layer-stripping and retain all the information from the data, incorporates a depth focusing technique for improved image resolution and utilizes user-interaction and geologic input to guide and constrain the imaging process.
I first describe a fast and accurate dynamic ray-tracing scheme in heterogeneous media which allows complex model definition and rapid two-point ray tracing. This ray tracing method is then used to compute Green functions in a layer-stripping pre-stack depth migration algorithm. The algorithm itself is based on a Kirchhoff integral in heterogeneous media using exact weighting factors and specialized to 2.5 D migration. I examine the migration results with a depth-focusing technique which analyzes common image panels for horizontal alignment, relating the degree of non alignment, or Migration Moveout (MMO) to corrections in the velocity model along the raypaths. Finally, I develop a cell-stripping tomography (CST) algorithm which distributes velocity residuals only to the relevant cells, allowing resolution of both horizontal and vertical discontinuities and providing starting models for migration.
Although computer-intensive, this unified approach is successful both in synthetic tests and for obtaining local and regional images of the edge of the Santa Maria basin in central California. It is more faithful to the velocity and dip information contained in the data, allows more control over the imaging process and with available computing power promises to be routinely applicable.
| Identifer | oai:union.ndltd.org:RICE/oai:scholarship.rice.edu:1911/16457 |
| Date | January 1991 |
| Creators | Lafond, Claude F. |
| Contributors | Levander, Alan |
| Source Sets | Rice University |
| Language | English |
| Detected Language | English |
| Type | Thesis, Text |
| Format | 143 p., application/pdf |
Page generated in 0.0022 seconds