STOCHASTIC CHARACTERIZATION OF THE ABERJONA VALLEY AQUIFER NEAR WOBURN, MASSACHUSETTS

Indicator geostatistics are being used to characterize proportion, geometry, and patterns of different facies in Aberjona Valley aquifer near Woburn, Massachusetts. Flow and transport in the Aberjona Valley aquifer has been previously studied by Metheny (1998) and Bair (2000) with a deterministic approach. As a complement to these prior studies, stochastic models are being developed from characterizations of the facies within this aquifer system. To accomplish this, the indicator geostatistics formalism was employed. First, an indicator data set was created defining the lithofacies category (high permeability or low permeability) present at all points of known lithology. These data were first used to estimate the proportions of the lithofacies using a declustering algorithm to remove the bias due to the non-uniform density of sampling points. It was found that the valley is composed of 37% low permeability lithofacies and 63% high permeability lithofacies. An iterative procedure was employed to explore and define regions in which the proportions of lithofacies are stationary. After finding the proportion of the lithofacies in each partition, the boundaries were adjusted until the proportion of any lithofacies was maximized in high proportion regions and minimized in low proportion regions. Four regions along the axis of the valley were defined; these have a systematic decrease in the proportions of low-permeability material down the valley. Perpendicular to the valley axis, however, proportions are stationary. Statistics characterizing the geometry and the pattern of the facies in the regions are being computed and modeled with transition probabilities.