MAPPING INTENTIONALLY UNCERTAIN GROUNDWATER DATA WITH GEOGRAPHIC INFORMATION SYSTEMS

Locational uncertainty is inherent in any map of groundwater contamination. Concentration values are known at wells as represented by points, but values are interpolated or extrapolated elsewhere. Recently, many well locations have intentionally become more uncertain to address issues of national security. On eastern Long Island, New York, the Suffolk County Water Authority (SCWA) reports monitoring well locations based on street name and geographic zones within the district. This study maps groundwater concentrations based on this information, assigning values to street line segments. This is done to determine if the resulting map patterns are significant with the addition of this intentional uncertainty.

Geographic Information System (GIS) is used in this study to make groundwater concentration maps from spreadsheets of data provided as part of the 2003 SWCA Annual Drinking Water Quality Water Report. A GIS layer is created which contains all streets referenced in the spreadsheet, and is intersected with outlines of the district's zone. A GIS procedure called geocoding is then used to assign 223 wells from the spreadsheet to the appropriate line segment. The results are not represented as points at the location of the well (as this is unknown), but rather as the street segment on which the well occurs. The spreadsheet also contains information on which of the three aquifers, the Upper Glacial, the Magothy and the Lloyd, was sampled. The resulting GIS layer allows any of the 78 groundwater contaminants or parameters to be mapped for the aquifer(s) in which they occur.

Results show detailed patterns of groundwater concentrations. This study looks at patterns of dissolved iron, a regional, naturally occurring element, and methy tertiary-butyl ether (MTBE), a local, synthetic compound. Dissolved iron had been mapped in previous studies, but this study shows more detailed patterns of its distribution. The map of MTBE concentrations identified several polluted areas. Using ancillary data, known tank failures, spills, tank test failures or MTBE spills could be identified as candidates for nearly two thirds of these contaminated areas.