SEMINAR: Physics Seminar

http://tinyurl.com/ydqk7nc http://tinyurl.com/yl79g3g

The uncertainty and risk associated with subsurface geology is an important input to business decisions in the petroleum and mining industry. Combination of various remote sensing measurements (seismic, E&M, magnetic, gravity, etc.), each with their own error, into an integrated assessment of the subsurface geologic model with uncertainty is needed. Starting with forward models that predict the measurement response given the geologic model, Bayes' theorem is used to commute the conditional probability for the probability of the geologic model given the measurement. Metropolis methods are used to estimate this posterior probability distribution (since the forward models are not analytically invertable). Geologic concepts are incorporated through the prior model probability. Methods have been developed and applied to real data for both seismic and marine E&M data. Results will be shown. Future directions will be presented for this research based on computer simulations of geologic processes and a path integral based approach to the statistical data assimilation.