A new postprocessing method for reservoir stochastic modeling: A solution based on information degree

A postprocessing method based on information degree is developed to solve the small-scale variation (noise) in reservoir stochastic modeling. Considering that different modeling results have different probabilities and credits, the new method uses the information degree calculated by the probabilities as weights to process the noise. Compared with the traditional postprocessing methods, this method is geologically more reasonable in that it considers both the information provided by the conditional data and the uncertainties associated with random sampling during simulation. The computation of information degree is objective, which avoids the subjective assignments of weight values in the traditional methods. Comparative studies using both conceptual and real reservoir models show that the new method effectively processes the noise in realizations. Thus, it is a prospective approach to the postprocessing family in stochastic modeling.