2006 Philadelphia Annual Meeting (22–25 October 2006)

Paper No. 3
Presentation Time: 8:40 AM


LIN, Yu-Feng1, BAJCSY, Peter2, WANG, Jihua3, VALOCCHI, Albert3 and HUNT, Randall4, (1)Center for Groundwater Science, Illinois State Water Survey, 2204 Griffith Drive, Champaign, IL 61820, (2)National Center for Supercomputing Applications, University of Illinois at Urbana-Champaign, Urbana, IL 61801, (3)Department of Civil and Environmental Engineering, University of Illinois at Urbana-Champaign, Urbana, IL 61801, (4)Wisconsin Water Science Center, U.S. Geological Survey, Middleton, WI 53562, yflin@uiuc.edu

Groundwater recharge and discharge (R/D) rates and patterns result from a set of complex, uncertain processes that generally are difficult to characterize. There is currently no single method capable of estimating R/D rates and patterns for all practical applications. Therefore, cross analyzing results from various estimation methods and related field information will likely be superior than using only a single estimation method.

Using numerical methods and image processing algorithms, we have developed a Pattern Recognition Utility (PRU) that can help hydrogeologists to estimate R/D in a more efficient way than conventional methods. Our Geographic Information System (GIS) software package is capable of processing and visualizing shallow R/D patterns and rates. In order to ensure the accuracy of the numerical R/D estimation from the PRU, the point-to-zone pattern learning (P2Z) software will be used to cross analyze results from the PRU with field information, such as land coverage, soil type, topographic maps and previous estimates. The learning process of P2Z will cross examine each initial recognized R/D pattern with the weighted segment in the reference spatial dataset, and then calculate a quantifiable reliability index. This quantifiable reliability index will provide an indication for the optimum R/D pattern. P2Z with graphic user interface will enable hydrogeologists to recognize R/D patterns objectively and compare R/D estimations from various approaches. Moreover, the reliability indices from P2Z will provide significant information to aid research in groundwater R/D estimation because they provide objectively quantifiable conceptual bases for further stochastic and uncertainty analysis.

Both the PRU and P2Z are designed to be universal utilities for pattern recognition and learning from point data to zonation delineation. The PRU has been tested against an intensively studied field site in Wisconsin. Currently, the PRU is being used for several active projects in Illinois and Wisconsin, and the P2Z is being developed based upon the image spatial data analysis algorithms from the National Center for Supercomputing Applications (NCSA). The final product of the PRU and P2Z will be applied immediately to several groundwater studies in northeastern Illinois.