Enhanced cross-well seismic inversion by fuzzy c-means clustering
|dc.contributor.author||Jafarzadeh Rastin, Sepideh|
|dc.identifier.citation||Kieu, D. and Jafarzadeh Rastin, S. and Kepic, A. 2019. Enhanced cross-well seismic inversion by fuzzy c-means clustering.|
© 2019 24th European Meeting of Environmental and Engineering Geophysics. All rights reserved. A cross-well tomography with a geostatistical constraint via fuzzy C-means clustering (FCM) is proposed to extract geologically realistic velocity models for acquisition during mineral resource definition drilling. Constraining a damped least squares inversion process with the additional requirement to keep velocity values in clusters counters the smearing of geological boundaries by the addition of smoothing constraints, which are typically used . The FCM center values were either set as priori known values from borehole measurements or estimated from the data during the cross-well inversion process. We have applied our FCM constrained inversion to three synthetic models and then compared these results to a conventional smoothness constrained inversion result. The three models approximate three simple-to-challenging possible scenarios within a polymetallic deposit embedded in meta-sedimentary rocks. A significant improvement resulted using FCM constraints in the recovery of the true structure, particularly at the boundaries, and various artifacts were better damped. The additional accuracy can provide considerable benefits in improve the resource models in the mine planning stage, and optimize ground-support design and blasting parameter estimation during mining.
|dc.title||Enhanced cross-well seismic inversion by fuzzy c-means clustering|
|dcterms.source.title||24th European Meeting of Environmental and Engineering Geophysics|
|dcterms.source.series||24th European Meeting of Environmental and Engineering Geophysics|
|curtin.department||WASM: Minerals, Energy and Chemical Engineering (WASM-MECE)|
|curtin.accessStatus||Fulltext not available|
Files in this item
There are no files associated with this item.