High-Order Data-Driven Spatial Simulation of Categorical Variables

Abstract Modern approaches for the spatial simulation of categorical variables are largely based on multi-point statistical methods, where a training image is used to derive complex relationships using relevant patterns. In these approaches, simulated realizations driven by utilized, while statistics actual sample data ignored. This paper presents data-driven, high-order approach approximation indicator moments. The expressed as functions distances that similar variogram models two-point higher-order connected with lower-orders via boundary conditions. Using an advanced recursive B-spline algorithm, reconstructed from available and subsequently construction conditional distributions Bayes’ rule. Random values all unsampled grid nodes. main advantages proposed technique its ability (a) simulate without reproduce data, (b) adapt model’s complexity information in data. practical intricacies effectiveness demonstrated through applications at two copper deposits.