Factors affecting occurrence and distribution of selected contaminants in ground water from selected areas in the Piedmont Aquifer System, Eastern United States, 1993-2003

Abstract

Results of ground-water sampling from 255 wells and 19 springs in 11 studies done by the U.S. Geological Survey National Water-Quality Assessment (NAWQA) Program within the Piedmont Aquifer System (PAS) were analyzed to determine the factors affecting occurrence and distribution of selected contaminants. The contaminants, which were selected on the basis of potential human-health effects, included nitrate, pesticides, volatile organic compounds (VOCs), and radon. The PAS was subdivided on the basis of the general rock type of the aquifers into three areas for the study—crystalline, carbonate, and siliciclastic. The 11 studies were designed to areally represent an individual aquifer rock type and overall are representative of the PAS in their distribution; 7 studies are in the crystalline-rock aquifers, 3 studies are in the siliciclasticrock aquifers, and 1 study is in the carbonate-rock aquifers. Four of the studies were focused on land use, 1 in an agricultural area and 3 in urban areas. The remaining studies had wells representing a range of land-use types. Analysis of results of nitrate sampling indicated that in 8 of the 10 areas where nitrate concentrations were measured, median concentrations of nitrate were below 3 mg/L (milligrams per liter); 2 of the 10 areas had statistically significant higher median concentrations when compared to the other 8 areas. The agricultural land-use study in the carbonate-rock aquifer in the Lower Susquehanna River Basin had the highest median nitrate concentration (11 mg/L), and 60 percent of the wells sampled exceeded the U.S. Environmental Protection Agency (USEPA) Maximum Contaminant Level (MCL) of 10 mg/L. The major aquifer study in the crystalline-rock aquifer of the Lower Susquehanna River Basin Study Unit had the second-highest median nitrate concentration. Nitrate concentrations were positively correlated to the percentage of agricultural land use around the well, the total input of nitrogen from all sources, dissolved oxygen concentration, lithology, depth to water, and soil-matrix characteristics. A linear regression model was used to determine that increases in the percentage of agricultural land use, the input of nitrogen from all sources, and dissolved oxygen were the most significant variables affecting increased concentration of nitrate. A logistic regression model was used to determine that those same factors were the most significant variables affecting whether or not the nitrate concentration would exceed 4 mg/L. Of the analysis of samples from 253 wells and 19 springs for 47 pesticides, no sample had a pesticide concentration that exceeded any USEPA MCL. The most frequently detected pesticide was desethyl atrazine, a degradation product of atrazine; the detection frequency was 47 percent. Other frequently detected pesticides included atrazine, metolachlor, simazine, alachlor, prometon, and dieldrin. Detection frequency was affected by the analytical reporting limits; the frequency of detection was somewhat lower when all pesticides were censored to the highest common detection limit. Source factors such as agricultural land use (for agricultural herbicides), urban land use (for insecticides), and the application rate were found to have positive statistical correlations with pesticide concentration. Transport factors such as depth to water and percentage of well-drained soils, sand, or silt typically were positively correlated with higher pesticide concentrations. Sampling for VOCs was conducted in 187 wells and 19 springs that were sampled for 59 VOCs. There were 137 detections of VOCs above the common censoring limit of 0.2 µg/L. The most frequently detected VOCs were chloroform, a trihalomethane, and methyl-tert butyl ether (MTBE), a fuel oxygenate. Seventy-nine wells had at least one VOC detected. The detections were related to land use and well depth. Kendall’s tau correlations indicated a significant positive correlation between chloroform concentration and urban land use, leaking underground storage tanks, population density, and well depth. MTBE concentrations also were positively correlated to urban land use, leaking underground storage tanks, population density, and well depth. Radon was sampled at 205 sites. The subdivisions used for analysis of other contaminants were not adequate for analysis of radon because radon varies on the basis of variations in mineralogy that are not reflected by the general lithologic categories used for the rest of the studies. Concentrations of radon were highest in areas where the crystalline-rock aquifers had felsic mineralogy, and the lowest concentrations of radon were in areas where the crystalline-rocks aquifer had mafic mineralogy. Water from wells in siliciclastic-rock aquifers had concentrations of radon lower than that in the felsic crystalline-rock aquifers. More than 90 percent of the wells sampled for radon exceeded the proposed MCL of 300 pCi/L (picoCuries per liter); however, only 13 percent of those wells had concentrations in water that exceeded the alternative maximum contaminant level (AMCL), a higher level that can be used by municipalities addressing other sources of radon exposure. Overall, concentrations of constituents were related to land-use factors for nitrate, pesticides, VOCs, and to aquifer lithology for radon. None of the 47 pesticides or 59 VOCs analyzed exceeded the MCLs where those constituents were sampled. Concentrations exceeded the MCL for nitrate in 11 percent of the wells sampled. Nearly 91 percent of the wells sampled exceeded the proposed MCL for radon. Additional sampling in selected areas would improve overall understanding of the PAS and increase the possibility of creating predictive models of ground-water quality in this area.