Development of ground-water vulnerability database for the U.S. Environmental protection agency's hazard ranking system using a geographic information system

Geographic information system (GIS) methods were applied to the U.S. Environmental Protection Agency's (EPA) hazard ranking system (HRS) to evaluate the vulnerability of ground water to contamination from actual or potential releases of hazardous materials from waste-disposal sites. Computerized maps of four factors influencing ground-water vulnerability - hydraulic conductivity, sorptive capacity, depth to water, and net precipitation - were derived for the Southeastern United States from digitized copies of published maps and from computerized databases, including the U.S. Geological Survey's (USGS) national water information system. To test the accuracy of the derived data coverages used to assess ground-water vulnerability, GIS-derived values for hydraulic conductivity, depth to water, and net precipitation were compared to corresponding values assigned by EPA's field investigation teams (FIT) at 28 hazardous waste sites. For each factor, site data were divided into three physiographic groupings: (1) Coastal Plain, (2) Valley and Ridge-Interior Low Plateaus, and (3) Piedmont-Blue Ridge. The best correlation between the paired data sets was for the net precipitation factor, where most GIS-derived values were within 0 to 40% of the FIT data, and 79% were within the same HRS scoring range. For the hydraulic conductivity factor, the best correlation between GIS and FIT data was for values derived from a published surficial deposits map, where most of the values were within one order of magnitude of the FIT data, and on the average were within 1.24 orders of magnitude of the FIT data. For this map, the best match between data sets was in the Coastal Plain province, where the difference in order to magnitude averaged 0.92. For the depth-to-water factor, most of the GIS derived values were within 51 to 100% of the FIT data, and only 44 to 50% of the sites were within a common scoring range. The best correlation for depth to water was in the Coastal Plain where GIS derived values were within 8 to 100% of the FIT data.