This is a collaborative multi-disciplinary three-year project that addresses the fundamental problem of determining (or "imaging") the location of subsurface geologic materials and the spatial distributions of their physical properties that control movement of groundwater and contamination. These spatial variations occur in complex patterns and at all size scales. Subsurface engineering applications that require accurate imaging of these variations include reliable environmental monitoring, predictive modeling, and efficient groundwater remediation. The project will develop the next generation of subsurface imaging tools to significantly improve estimates of formation and property distributions, and to improve quantification of the corresponding predictive uncertainty to provide a sound basis for management or policy decisions. A team of scientists and engineers with overlapping expertise in mathematics, statistics, modeling, and hydrogeology has been assembled from Stanford, Rice, Utah, and Boise State universities. Theoretical and modeling developments will be combined with controlled experiments at a field-scale test facility (Boise Hydrogeophysical Research Site, or BHRS) with three known scales of sedimentary structure and property variation, including layers and lenses with both high-contrast and gradational boundaries. In particular, the research team will: (i) develop a firm mathematical foundation for the analysis of inverse problems (or imaging) under realistic assumptions about the completeness of measurements, including improved methods for representing complex systems; (ii) employ novel statistical tools that exploit recent advances and trends in computation; (iii) develop new analytical approaches for stochastic (or statistically uncertain) systems with realistic variability; (iv) combine these developments with experimental studies and independent evaluation of model performance against archive data sets available from BHRS; and (v) advance an emerging field method (hydraulic tomography) to acquire data sets for modeling 3D hydraulic conductivity distributions in aquifers. Students and a post-doctoral scientist will work with senior researchers and will participate in all aspects of this project to gain cross-disciplinary knowledge and experience. In addition to dissemination through peer-reviewed literature and professional meetings, the team will develop web-based tutorials and training sets with data and models from the project, and a short course on field and modeling methods from the project. This project has broad impacts for society and for scientific and engineering infrastructure. Most available freshwater is stored in the subsurface. Groundwater is the primary source of water for over 50 percent of Americans, and for roughly 95 percent in rural areas. In the world, many of the most important aquifers are being gradually depleted. In coastal areas, where world population is growing the fastest, seawater intrudes into aquifers as groundwater levels drop and/or sea levels rise. This research will lead to better methods for management of this important resource by developing the next generation of subsurface imaging capabilities based on advancements in the mathematics of inverse modeling, stochastic differential equations, multi-scale simulations, and new field methods such as hydraulic tomography.
