The potential impacts of the spatial resolution of sedimentary structure data on solute dispersion in heterogeneous porous formations are investigated with Lagrangian-based transport models. The models rely on a covariance function that represent a hierarchical organization of sedimentary facies types by using facies physical properties, such as volume proportions and mean length as well as auto- and cross-transition probabilities and their log-permeability covariance. A detailed sedimentary architecture data (i.e. indicator data) provides better representation of the spatial correlation structures of the global covariances through capturing its underlying structure defined by transition probabilities more accurately. However, the extent to which such data affect time-dependent transport parameters (i.e. dispersivity) is unclear. In this study, we parameterize transport models using detailed collocated sedimentary architecture and permeability data from an outcrop in Española Basin, NM. In addition, we perform global sensitivity analysis based on Polynomials Chaos Expansion to understand the significance of parameters in the transport models. The results show that dispersivity and particle displacement variance are under-estimated if less resolved facies (i.e. indicator) data are used even if the global covariance structure is well captured. Dispersivity is sensitive to the correlation scale that is directly calculated from sedimentary architecture data, and the mean log-permeability. However, anisotropy ratio and mean log-permeability are the most sensitive parameters for the transverse dispersivity.
Read full abstract