Integration of geologic information into geostatistical models

Stochastic subsurface models generally lack the geologic realism found in process based and conceptual geologic models. Stochastic algorithms are limited in their ability to integrate geologic information; they do not reflect the complicated patterns and interrelationships observed in geology.; Simulated subsurface models that integrate more geologic information: (1) will appear more realistic and command greater credibility, (2) should perform more realistically when used in flow simulation and (3) will provide an improved assessment of uncertainty for optimal decision making. Pseudo process, advanced geometric and rule based paradigms provide methods for directly integrating geologic information.; This work represents an integrated approach on this topic. This dissertation includes the development of algorithms and work flows for surface and object based simulation, the construction of static and flexible training image libraries and the development of building blocks for efficient simulation directly to unstructured grids.; This work has resulted in advancements over a broad range of topics. The algorithms developed in this dissertation greatly augment the geostatistical tool box available to the practitioner and seed further development. Practical work flows are discussed and demonstrated. Future research directions are provided throughout and are summarized in the conclusions.