ALLUVSIM: A program for event-based stochastic modeling of fluvial depositional systems

This paper presents an algorithm for the construction of event-based fluvial models. The event-based approach may be applied to construct stochastic pseudo-process-based fluvial models for a variety of fluvial styles with conditioning to sparse well data (1–5 wells) and areal and vertical trends. The initial models are generated by placing large-scale features, such as channels and crevasse splays, into the model as geometric objects. These large-scale features are controlled by geometric input parameters provided by the user and are placed into the model to roughly honor well data through a rejection and updating method. Yet, some model to well data mismatch may still occur due to inconsistency in the size and positioning of complicated features relative to the well data. An image processing algorithm is used to post-process realizations to exactly honor all well data. The final, cell-based models, have no data mismatch and contain geologically realistic fluvial features that would be difficult to obtain with other pixel-based methods and precise conditioning that is difficult to obtain with object-based methods.