Summary:

Grid-based numerical models are capable of modeling ground-water flow at either regional and local scales.  If a range of scales it to be represented, however, the models become inefficient and inaccurage, because the computation mesh must be "telescoped" down when ever local hydraulic information is needed in a regional setting.  A numerical modeling method that avoids these problems is the Analytic Element Method. (AEM) With this approach, wells, lakes, rivers, and hydrogeologic inhomogeneities are represented using analytic equations.  These equations, called analytic elements, are summed to represent the entire hydrogeologic system.  The result is a flow field representation that is exactly correct at any scale.

This tool provides an opportunity to model ground-water flow at the regional scale, and allows the result to be used to model local problems.  Our approach is to use a Geographic Information System object-oriented database (ArcInfo 8) to organize, prioritize, and parse the data to the AEM solver.   Yhis allows geologic maps, field data, remotely sensed data, and hydraulic information to be combined in an intellegent manner, even when the data are sparse and subject to large error (as is usually the case in regional modeling problems).   The computational system is designed such that solutions can be parallelized on high-performance computers, and calibration can be carried out efficiently using both analytic and numerical approaches.

The project is taking advantange of the UB Center for Computation Research (CCR)   The computing facilities housed in the Center are extensive and include a 64 processor Silicon Graphics Origin 2000 supercomputer, a 58 processor IBM RS/6000 SP supercomputer, a 64 processor Network of Workstations supercomputer (Sun Ultra5 workstations).  The computer visualization laboratory features a Pyramid Systems ImmersaDesk and several high-end computer graphics workstations.  Taken together, the supercomputers are capable of carrying out more than 90 billion operations per second (90 Gflops).   According to the Top500 Supercomputer List, this center is now one of the top 10 academic supercomputing sites in the country.

The tools resulting from this project will be tested and applied to the Niagara Falls area, which has a long history of environmental problems and can benefit from a regional flow model.  Because a finite-difference model for the site has already been built by the US Geological Survey, it will also provide an opportunity to compare modeling approaches.  Other sites will also be used as a testing ground for this project, including the Olean Creek Watershed.