Computer-based spatial models, representing the past and present disposition and configuration of sets of rock bodies, are a potentially important segment of a geological knowledge base. A spatial model is an interpretation that should be consistent with available data and with expectations based on knowledge of the processes which formed and deformed the rock bodies. The expectations refer not to the structure of the processes, but to their effects, which can be expressed as regional patterns, descriptive statistics, spatial relationships, material budgets, and balances. By using measures of these effects to control and modify interpolation, they can be taken into account in the modeling of surfaces and lines for display, measurement, analysis, and prediction. Processes which create features too small to be located from the data, contribute to an uncertainty envelope which can be defined around the interpolated surfaces.