Full text loading...
-
Fully Non-linear Self-organizing Inversion of Cross-borehole Tomographic Data
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, Near Surface Geoscience 2015 - 21st European Meeting of Environmental and Engineering Geophysics, Sep 2015, Volume 2015, p.1 - 5
Abstract
Summary
Geophysical model reconstruction by data inversion is usually ill-posed and suffers ambiguity due to limited number and accuracy of the available observations. Here, we formulate a new concept for an almost completely data-driven fully-nonlinear inversion to achieve an ensemble of different tomographic models fitting the underlying data set equally well. Since we parameterize the model reconstruction area using regular grids with grid cell dimensions not exceeding the spatial resolution of the tomographic data set we can find models of any reasonable complexity. To avoid the finding of models characterized by numerous anomalies below the spatial resolution limit of the data set we concurrently address a data misfit objective and a model structure objective formulated in fuzzy domain. The final ensemble of models addresses the data objective equally well but the model structure objective only to different degrees and allows for the finding of model ensembles illustrating the tomographic reconstruction ambiguity reasonably.
© EAGE Publications BV