Anisotropy is a valuable proxy for certain ground material properties such as rock type characterisation, fracture direction, fracture intensity identification and ground water flow direction inferred from fracture infonllation. TIns thesis describes a new method for anisotropic seismic inversion and includes extensive synthetic tests, where survey geometry and model heterogeneity levels are varied, wInch help to test under which conditions the method works. The MGMC (Multigrid Monte Carlo) method was chosen because of the hypothesis that it may find the absolute minimunl of the multi-parameter functional without needing penalty tenus or other constraints. The fit to which the functional can be lnininnsed is of very close accuracy to results of the gradient descent methods. Even with such comparable fit, some paranleters were not fully resolved, showing that even at such high levels of accuracy certain trade offs are still possible between parameters. Extensive synthetic tests show that the method works well, as long as certain conditions of angular coverage and model heterogeneity levels are accounted for. Also, the results indicate that residual maps are not at all indicative of correctly resolved anisotropy with large residuals sometimes accompanying better resolved images than relatively small residuals. Analyses of field data show encouraging results becau'se the images correspond well to previous inversion work therefore showing that the MGMC approach is a viable alternative to previous methods. Comparing the analyses, using the new MGMC inversion developed here on the one hand and the previous analysis method, which involves comparing results derived from using various combinations of penalty tenus, it was found very useful not have to compare results derived from using various combinations of penalty terms since tIns saves time and may result in a more accurate representation of the true anisotropy structure. The agreement of the results with pre-inversion apparent velocity analysis and core analysis, the independent APPSM (apparent penalty parameter search mechanism), the extremely low traveltime residual map and the inclusion of steep, near vertical raypaths all suggest that the results are reliable with high likelihood, thereby showing tIlat tIle MGMC method not only operates well for synthetic data but for real field data as well.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:485043 |
| Date | January 2008 |
| Creators | Mewes, Armin |
| Publisher | Queen's University Belfast |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
Page generated in 0.0023 seconds