A test of analog-based tools for quantitative prediction of large-scale fluvial architecture

Outcrop analogs are routinely used to constrain models of subsurface fluvial sedimentary architecture built through stochastic modeling or interwell sand-body correlations. Cone lability models are analog-based quantitative templates for guiding the well-to-well correlation of sand bodies, whereas indicator variograms used as input to reservoir models can be parameterized from data collected from analogs, using existing empirical relationships. This study tests the value and limitations of adopting analog-informed correlability models and indicator variogram models and assesses the effect and significance of analog choice in subsurface workflows for characterizing fluvial reservoirs. A 3.2-km (2-mi)-long architectural panel based on a virtual outcrop from the Cretaceous Blackhawk Formation (Wasatch Plateau, Utah) has been used to test the methodologies. Vertical dummy wells have been constructed across the panel, and the intervening fluvial architecture has been predicted using correlability models and sequential indicator simulations. The correlability and indicator variogram models employed to predict the outcrop architecture have been compiled using information drawn from an architectural database. These models relate to (1) analogs that partially match with the Blackhawk Formation in terms of depositional setting and (2) empirical relationships relating statistics on depositional element geometries and spatial relations to net-to-gross ratio, based on data from multiple fluvial systems of a variety of forms. The forecasting methods are assessed by quantifying the mismatch between predicted architecture and outcrop observations in terms of the correlability of channel complexes and static connectivity of channel deposits. Results highlight the effectiveness of correlability models as a check for the geologic realism of correlation panels and the value of analog-informed indicator variograms as a valid alternative to variogram model parameterization through geostatistical analysis of well data. This work has application in the definition of best-practice use of analogs in subsurface workflows; it provides insight into the typical degree of realism of analog-based predictions of reservoir architecture, as well as the effect of analog choice, and draws attention to associated pitfalls.