Undiscovered petroleum accumulation mapping using model-based stochastic simulation

Stochastic simulation has been proven to be a useful tool for revealing uncertainties in petroleum exploration and exploitation. The application to petroleum resource assessment would result in predicted potential accumulations with geographic locations, a desirable feature for improving both resource management and exploration efficiency. The associated uncertainties with the prediction provide information useful for exploration risk analysis. This attempt has been encumbered by two typical technical difficulties: biased observation data and lack of information with respect to the undiscovered accumulation locations. In this paper we propose a model-based simulation approach, in which models are used to perform unbiased parameter estimation from biased data and to facilitate the location of undiscovered petroleum accumulations based on reasoning of available geological and geophysical observations. The Fourier transform algorithm is chosen for the simulation because the spatial correlation and location-specific features can be studied separately from different data sources and integrated in the simulation in a frequency domain. The proposed approach is illustrated by an example from the Rainbow petroleum play in the West Canadian Sedimentary Basin. In the application example, a pre-1994 exploration history data set was used as input, and the predictions are then checked against the locations of post-1993 exploratory drilling results. The comparison of the predictions from the proposed approach and the traditional conditional simulation shows that the model-based approach captures the essentials of geological controls on the spatial distribution of petroleum accumulation, thus improving the projections of undiscovered petroleum accumulations.