Various geophysical methods are used to estimate subsurface geolectric properties, such as the resistivity of
subsurface layers of the earth and their geometric properties. Bayesian methods were used in this study to
estimate resistivity and thickness of subsurface horizons. Specifically, Monte Carlo Methods based in Markov
Chains (MCMC) were proposed, which are the most common bayesian methods used in parameter inversions.
First, MCMC Metropolis-Hastings (MH) were used to fix different thickness to each layer and then estimated
the resistivity of each. Then, Reversible Jump MCMC (RJ-MCMC) was applied using the modelgenerated
thicknesses and number of layers as variables. The MH model returned acceptable resistivity values of the
shallow layers, generating imprecision in the intermediate and deep layers. The RJ-MCMC model was better
suited for self-parameterizing and for faster computing time, generating uncertainty in the deeper layers. Our
results show that for studies of shallow layers RJ-MCMC correctly adjusted the three-layer model with low
deviation of real values as related to correlations between the pair of properties.
|
