Simultaneous Autoregressive Models for Fluvial Channel-Belt Facies and Paleomorphodynamics of the Lower Wasatch Formation, Uinta Basin, Utah, USA
Stochastic reservoir simulations of fluvial channel belts require knowledge of both facies proportions within the channel belt, as well as how facies are spatially distributed throughout the reservoir. Facies within fluvial channel belts are characterized by different reservoir properties such as (1) sorting, (2) rounding, (3) grain size, and (4) degree of bioturbation. In order to address this problem, six different reservoir properties were spatially documented within outcrops of five channel belts within the lower Wasatch Formation of the Uinta basin in eastern Utah. Using a lattice approach, spatial dependence of reservoir properties was documented both globally using Moran's I and locally using local indicators of spatial autocorrelation. Maximum grain size within all five channel belts shows significant evidence for spatial dependence when observed Moran's I values are compared to Monte Carlo simulations of randomly distributed maximum grain size. To model the maximum grain size using other reservoir properties, a simultaneous autoregressive (SAR) model was used with row-standardized weights. Using the spatial position of the bedsets, as well as the average sorting of the bedsets, valid models were constructed for four of the five channel belts. The normally distributed and spatially independent residuals of the models validate these SAR models. Furthermore, depositional processes and temporal persistence in flow conditions within the channel belts are documented by spatial autocorrelation of average flow velocity within the channel belt via a paleomorphodynamics workflow.
AAPG Datapages/Search and Discovery Article #90217 © 2015 International Conference & Exhibition, Melbourne, Australia, September 13-16, 2015