9628299 Ritzi Contaminant problems have created the need to understand how hydraulic conductivity (K) is distributed in aquifer systems. Extreme-value pathways exert a dominant physical control on contaminant transport, and thus the boundary between high-K and low-K materials (hydrofacies) must be quantified. Although the exact geometry of the hydrofacies can only be determined by a prohibitive amount of subsurface information, stochastic methods such as conditional indicator simulation provide a way to compute a number of realizations of hydrofacies distributions that honor available data at their locations, with specified functions for the principal directions of spatial correlation of the hydrofacies, the scale of correlation in these directions, and the proportions. Running transport simulations using a large number of these realizations and examining the central tendency and variance of simulated contaminant travel times, or the simulated contaminant pathways, can give insight into how contaminants will move through the real system. These simulations can also be incorporated into a formal decision-analysis framework. However, the field data at a specific site commonly are insufficient for the purpose of defining the spatial correlation and proportions of the hydrofacies directly. Therefore, analog sites that are data rich need to be studied in order to find models for spatial correlation and proportions of hydrofacies that, on the basis of similar geology, can be transferred to data-poor sites. Indeed, one of the research needs identified by the National Research Council in Alternatives for Ground Water Cleanup, is the development of statistical models representing common geologic environments, which can be used to assess the variability of aquifer properties and thereby improve the success of aquifer restoration (NRC, 1994, p. 73). The proposed research will develop statistical models for aquifer properties in buried-valley aquifers, which occur across the central a nd western glaciated plains of North America and are vital to the water resources of the region. In such glacially derived aquifers, it is common to have low-K facies (e.g., till or lacustrine clay) juxtaposed with high-K facies (e.g., sand and gravel outwash). This is a renewal proposal. In the prior phase, several locations were studied in a buried-valley aquifer system located in Ohio, with the successful development of models for the hydrofacies distribution, proportions, principal directions of spatial correlation, and correlation scale in these directions. These statistical models were linked to the deposition processes that created the aquifer system, and therefore are likely to be applicable to other locations in the aquifer, and to other aquifer systems where similar depositional processes have operated. However, hydrofacies models are most useful when based on the features of numerous well-studied examples. The generality and usefulness of quantitative geostatistical models will likewise increase with their numbers, if more data-rich analogs are examined. One goal of this renewal proposal is to develop new geostatistical models for a buried-valley aquifer system in Indiana. However, the main focus of this work will be to critically evaluate methods of linking spatial correction of hydrologic parameters and geologic structure. The transition probability has been shown to be an even more interpretable representation of spatial correlation than the variography used in the prior phase. Thus the transition probability may provide a new tool for linking spatial correlation and deposition processes. The innovative aspect of this work is that it will provide a way to test the hypothesis that generic hydrofacies models can be developed for a specific class of aquifers. This will be the first comparison of spatial correlation of hydrogeologic parameters among aquifers, and is part of the longer-term goal of developing a catalogue of quantitative hydrofacies. ??
