G-GRID: PROPOSED DEPLOYMENT OF MULTIDIMENSIONAL INFORMATION INFRASTRUCTURE FOR GROUNDWATER MAPPING IN THE UPPER US MIDWEST

In Minnesota, 70% of the population obtains drinking water from groundwater, but the capacity of this resource and its vulnerability to climate change, contamination, and overuse are not well known. Water resource managers are, however, increasingly committed to ensuring that groundwater usage is sustainable. New approaches to make better use of existing databases therefore are being developed for construction of the groundwater maps that are required for water resource management in this and similar regions. Vast amounts of information are available, largely in the form of databases of water well records. Unfortunately, however, the information tends not to be used at the regional scale because the infrastructure to do so has not been adequately developed.

A recently completed model for the Winnipeg region of Canada, however, demonstrated methods for creating a regional 3D geologic map through interpretation of large databases. This approach is now being coordinated with paleoclimatic research, regional characterization of subsurface material property heterogeneity using tools such as borehole geophysics, and geostatistical methods for quantitatively modeling heterogeneity of subsurface strata on a regional basis - in part through transfer of methods from the hydrocarbon industry. Large portions of these databases based on or containing hundreds of thousands of well records can already be viewed and interpreted visually with emerging visualization systems such as the GeoWall. But when appropriate interpolation and extrapolation methods are applied at the regional scale to the prediction of subsurface properties on a regular grid - a basic requirement for analysis - a multi-dimensional digital model beyond the capability of conventional computing resources can readily be generated. Our current efforts therefore are emphasizing methods for making both large groundwater databases and resultant models more available to users in a networked environment.