Statistical modelling and validation of correlation in turbidites : an example from the Tertiary Piedmont Basin (Castagnola Fm., Northern Italy)

A reservoir-scale analysis, which integrated outcrop and geostatistical studies, was applied to a turbidite sandbody (Late Oligocene-Miocene) cropping out in the eastern part of the Tertiary Piedmont Basin (NW Italy). The geometry, stratigraphic relationships, facies distribution, and palaeocurrent directions all indicate that the sandbody was deposited in a confined basin, principally controlled by synsedimentary tectonic activity. This geological model forms a template for subsequent geostatistical simulations that provide equi-probable representations of small-scale geological heterogeneities. The method used to obtain the conditional simulation of facies (Heresim methodology) is based on a truncated Gaussian random function. The application of such a methodology includes five main steps: (i) database compilation, (ii) geometrical reconstruction of the external surfaces, (iii) statistical analysis of facies distribution, (iv) variogram computation, and (v) lithofacies simulation.Conditional simulations lead to a better characterization of the depositional environment in terms of facies variation that are consistent with the observed sedimentological framework. In addition to their qualitative interest, the geostatistical simulations enabled the quantification of geological parameters, the testing of different geological hypotheses and to appreciate the uncertainties of the results when less geological information is available to constrain the model. The integration of outcrop data and geostatistical analysis allows a quantitative sedimentological model to be constructed, which helps to improve the efficiency of outcrop observation as a predictive tool for estimating facies heterogeneity in subsurface reservoirs. © 2003 Elsevier Ltd. All rights reserved.