Improving the Pattern Reproducibility of Multiple-Point-Based Prior Models Using Frequency Matching

摘要: Some multiple-point-based sampling algorithms, such as the snesim algorithm, rely on sequential simulation. The conditional probability distributions that are used for simulation based statistics of multiple-point data events obtained from a training image. During simulation, with zero in image may occur. This is handled by pruning set conditioning until an event non-zero found. resulting distribution sampled algorithms pruned mixture model. strategy leads to lacks some information provided image, which reduces reproducibility patterns outcome realizations. When models prior inverse problems, local re-simulations performed obtain perturbed Consequently, these lead additional data, further deteriorates pattern reproduction. To mitigate this problem, it here suggested combine model frequency matching realizations combined has improved degree match An efficient algorithm samples suggested. Finally, tomographic cross-borehole problem expressed (prior) demonstrate effect resolution problem.