Truncation Map Estimation for the Truncated Bigaussian Model Based on Bivariate Unit-lag Probabilities

摘要

The truncated plurigaussian model is often used to simulate the spatial distribution of random categorical variables such as geological facies. The problems addressed in this paper are the estimation of parameters of the truncation map for the truncated plurigaussian model, and improvements in the method for conditioning the latent Gaussian random variables to observations of categorical variables. Unlike standard truncation maps, in this paper a colored Voronoi tessellation with number of nodes, locations of nodes, and category associated with each node all treated as unknowns in the optimization. Parameters were adjusted to match categorical bivariate unit-lag probabilities, which were obtained from a larger pattern joint distribution estimates from the Bayesian maximum-entropy approach conditioned to the unit-lag probabilities. The distribution of categorical variables generated from the estimated truncation map was close to the target unit-lag bivariate probabilities. The conditioning of the latent Gaussian fields to log-data is also generalized for the case when the truncated bigaussian model is governed by a colored Vorono"i tessellation of the truncation map. Compared to the standard Gibbs sampler, the new approach gives better mixing properties for large amount of closely correlated data observations.