MODELING MULTI-SCALE HETEROGENEITY IN BURIED-VALLEY AQUIFERS USING INDICATOR GEOSTATISTICS

Hydrogeologic investigations of glaciated regions are difficult because the stratigraphic relationships among sedimentary facies are more complex than can be determined by available borehole, geophysical, and hydrologic records. One approach is to identify and then characterize assemblages of facies, the distributions of which are more tractable than the individual facies they contain. Heterogeneity at the scale of facies must still be characterized but this can be accomplished by grouping lithologic units into a smaller number of facies with distinct permeability modes. Lithologic data from boreholes can then be coded with indicators to represent the occurrence of facies and facies assemblages. Importantly, facies assemblages must be identified as regions that are uniform within themselves but that differ from adjacent regions in the characteristics to which any model will be conditioned. These characteristics are the proportions, geometry, and spatial correlation of facies they contain. Because these features can be characterized by univariate and spatial bivariate statistics, identifying facies assemblages can be seen as a statistical analog to traditional stratigraphic interpretation.

Identifying facies assemblages requires delineating their boundaries, computing their statistical characteristics, modifying the boundaries, and re-computing statistics. Several iterations are normally required to maximize the similarity within and differences among regions. Although it may be possible to use objective methods to optimize facies assemblage boundaries, these units must ultimately fit within the framework of interpreted depositional environments. As with traditional stratigraphy, geologic insight is a valuable guide to reasonable interpretations. Realizations of facies distributions can be created by combining sequential indicator simulations of facies within facies assemblages with those of facies assemblages themselves to represent two scales of heterogeneity. We illustrate this approach with examples drawn from the central and western glaciated plains regions of North America.