Dr. Mrinal K. Sen
 Institute for Geophysics
John A.& Katherine G. Jackson School of Geosciences
The University of Texas at Austin

ABSTRACT
Seismic inversion involves estimating elastic properties or some attributes of the subsurface rocks by iterative fitting of observed seismograms with the theoretical seismograms. For many applications, optimal fitting of data with model may become computationally intractable due to nonlinearity, ill-posedness and expensive forward modeling. I will discuss several efficient approaches to address these issues. The non-linear optimization problem is solved very efficiently using a method called simulated annealing (SA) that does not usually get trapped in a local minimum of the fitness function that measures the similarities or differences between observed and synthetic data. The SA approach can be applied to post-stack inversion for impedances, wavelet estimation, and pre-stack waveform inversion. It can also be used for estimation of uncertainties in the derived results. The efficiency can be increased further by a combination of SA with a local optimization. I will show several field data examples to demonstrate the applicability and usefulness of our algorithm.