Joint full-waveform inversion of time-lapse seismic datasets with model-difference regularization is a promising new technique for identifying production-induced changes within hydrocarbon reservoirs and in the overburden. The proposed technique achieves robust inversion of velocity changes in the presence of challenging survey repeatability issues.

Check out the abstract of my upcoming presentation at the annual meeting of the Society of Exploration Geophysicists in Denver, 26-31 October, 2014. Visit the Technology section for a poster presentation explaining how cross-updating achieves artifact reduction.

Joint full-waveform inversion of time-0lapse seismic data sets

Link to the AGU 2013 presentation on a new, computationally efficient technique for numerical extrapolation of elastic wave-fields. The proposed method can be applied to a variety of problems, from global to exploration seismology.

Is static stress transfer a significant contributor to earthquake triggering or the phenomenon is mostly dynamic? The referenced paper presents some results of inverting for spatially-heterogeneous source event slip from the distribution of presumed triggered events. Inverted slips matching known data may confirm static stress triggering as a plausible triggering mechanism (see the referenced paper)....display full article text

http://arxiv.org/abs/1304.7821

Functional determinants arise naturally in normalization of functional integrals. When efficient numeric or analytic methods for computing functional determinants are available, functional integration (i.e. Feynman path integral) may offer a computationally viable alternative to perturbative and asymptotic methods.

The referenced paper reviews methods for the efficient computation of functional determinants for 1D and radially-symmetric (Dunne, 2008) potentials, and discusses extension for arbitrary operators....display full article text

http://arxiv.org/abs/1304.0042

Check out this extended abstract under the Technology section (click on the thumbnail on the left)