ABSTRACT

Carbonate aquifers are an important source of water in the United States; however, these aquifers can be particularly susceptible to contamination from the land surface. The U.S. Geological Survey National Water-Quality Assessment (NAWQA) Program collected samples from wells and springs in 12 carbonate aquifers across the country during 1993–2005; water-quality results for 1,042 samples were available to assess the factors affecting ground-water quality. These aquifers represent a wide range of climate, land-use types, degrees of confinement, and other characteristics that were compared and evaluated to assess the effect of those factors on water quality. Differences and similarities among the aquifers were also identified. Samples were analyzed for major ions, radon, nutrients, 47 pesticides, and 54 volatile organic compounds (VOCs).

Geochemical analysis helped to identify dominant processes that may contribute to the differences in aquifer susceptibility to anthropogenic contamination. Differences in concentrations of dissolved oxygen and dissolved organic carbon and in ground-water age were directly related to the occurrence of anthropogenic contaminants. Other geochemical indicators, such as mineral saturation indexes and calcium-magnesium molar ratio, were used to infer residence time, an indirect indicator of potential for anthropogenic contamination. Radon exceeded the U.S. Environmental Protection Agency proposed Maximum Contaminant Level (MCL) of 300 picocuries per liter in 423 of 735 wells sampled, of which 309 were drinking-water wells.

In general, land use, oxidation-reduction (redox) status, and degree of aquifer confinement were the most important factors affecting the occurrence of anthropogenic contaminants. Although none of these factors individually accounts for all the variation in water quality among the aquifers, a combination of these characteristics accounts for the majority of the variation. Unconfined carbonate aquifers that had high percentages of urban or agricultural land, or a combination of both, had higher concentrations and higher frequency of detections for most of the anthropogenic contaminants than areas with other combinations of land use and degree of aquifer confinement. Redox status is an indicator of more recently recharged water and affects the fate of some contaminants.

Median concentrations of nitrate were highest in the Valley and Ridge and Piedmont aquifers and lowest in the Biscayne and Silurian-Devonian/Upper carbonate aquifers. Nitrate concentrations were significantly higher in unconfined aquifers than in confined aquifers and semiconfined/mixed confined aquifers (wells in aquifers with breached confining layers or wells open to both a confined and an unconfined aquifer). Water recharged after 1953 had significantly higher concentrations of nitrate than water recharged prior to 1953. Redox status was also a key factor affecting nitrate concentrations; in recently recharged waters, samples in oxic waters had significantly higher concentrations of nitrate than anoxic waters, regardless of land use in the area around the well. Samples from 54 wells (5 percent) exceeded the U.S. Environmental Protection Agency MCL of 10 mg/L for nitrate in drinking water. Most of the samples exceeding the drinking-water standard (52 samples, or 5 percent) were in domestic supply wells in agricultural areas. The Piedmont and Valley and Ridge aquifers had the largest number of samples (45) exceeding the MCL; in the remaining aquifers only 9 samples had concentrations of nitrate that exceeded the MCL (about 1 percent). None of the water recharged prior to 1953 and only a single sample from a confined aquifer had nitrate concentrations that exceeded 10 mg/L as N.

Wells were sampled for a minimum of 47 pesticides. Detection frequencies and comparisons varied depending on the assessment level used. At least 1 of the 47 pesticides was detected at 510 (50 percent) of the 1,027 sites where pesticide data were available using the ‘all detections’ assessment level—that is, including any quantified detection as well as any estimated values where the compound was definitively detected. Multiple pesticides were frequently detected in a sample of water from a site; 34 percent of the samples had two to five pesticides detected in the same sample, and 4 percent of the samples had six or more pesticides detected. Dieldrin was detected at 20 sites, 9 of which were from either domestic or public supply wells, at a concentration above the Health-Based Screening Level (HBSL) of 0.002 µg/L. Diazinon was detected at a concentration greater than the HBSL of 1 µg/L at a single site, which was also a domestic supply well. These are the only samples where a pesticide exceeded a human-health benchmark.

The most frequently occurring pesticide compounds were four herbicides—atrazine, simazine, metolachlor, and prometon—and deethylatrazine, a degradate of atrazine. These pesticides typically were detected at concentrations that were less than 10 percent of a human-health benchmark. Of the four frequently occurring pesticides, only samples for atrazine (3 percent) and simazine (0.1 percent) had concentrations that exceeded 10 percent of the human-health benchmark; most of these cases were in agricultural areas. It is important to note, however, that the most frequently occurring pesticide degradate compound—deethylatrazine—has no human-health benchmark. Using a common assessment level of 0.01 µg/L, four of the aquifers—Biscayne, Mississippian, Piedmont, and Valley and Ridge—had at least one of these five compounds detected in more than 30 percent of the wells sampled. These four aquifers, along with the Ordovician, Ozark Plateaus, and Prairie du Chien aquifers were the aquifers or aquifer systems that had concentrations of pesticides that exceeded 10 percent of a human-health benchmark. Water recharged after 1953 had a significantly higher percentage of detections of pesticides than water recharged before 1953, and water from unconfined aquifers had a significantly higher percentage of detections of pesticides than water from confined or semiconfined/mixed confined aquifers. Water from sites in unconfined aquifers, where land use was agricultural or urban, accounted for the vast majority of detections of pesticides. Dissolved oxygen concentration was positively related to pesticide occurrence, which likely reflects the positive association between dissolved oxygen concentration and recently recharged water.

Water samples were collected for analysis of VOCs at 793 sites—154 samples were analyzed for 54 VOCs from 1993 through 1995 and 639 samples were analyzed for 86 VOCs from 1996 through 2005. Twenty percent of samples contained one or more VOCs at concentrations greater than or equal to 0.2 µg/L (159 of 793 samples). The aquifers with the highest percentage of samples containing one or more VOCs were the Castle Hayne (about 41 percent of samples) and Biscayne aquifers (34 percent). The most frequently detected VOCs were chloroform, tetrahydrofuran, tetrachloroethene (PCE), toluene, acetone, ethylmethylketone, methyl tert-butyl ether (MTBE), and trichloroethene (TCE). Low-level concentrations of VOCs occurred in a much larger percentage of a subset of the data (the 639 samples analyzed using a low-level analytical method). In these samples, 69 percent of the 639 samples contained 1 or more VOCs, indicating the vulnerability of the carbonate aquifers to low-level VOC contamination. Four VOCs were detected at concentrations exceeding their respective MCLs in five samples, all of which were from drinking-water wells. Vinyl chloride concentrations exceeded the MCL of 2 µg/L in two samples from urban areas in the unconfined Biscayne aquifer. PCE, TCE, and 1,2-dichloropropane each had one sample with a concentration greater than their MCLs of 5 µg/L; these samples were from agricultural and urban areas in the unconfined Mississippian aquifer.

Water quality in the 12 carbonate aquifers was highly variable. Most of the samples met drinking-water standards. The occurrence of anthropogenic contaminants was related to contaminant sources but also was affected by degree of aquifer confinement, ground-water age, and redox status. Areas with higher amounts of agricultural or urban land in unconfined aquifers were the most likely to have elevated concentrations of anthropogenic contaminants.

First posted June 22, 2009