Arsenic concentrations, related environmental factors, and the predicted probability of elevated arsenic in groundwater in Pennsylvania

Abstract

Analytical results for arsenic in water samples from 5,023 wells obtained during 1969–2007 across Pennsylvania were compiled and related to other associated groundwater-quality and environmental factors and used to predict the probability of elevated arsenic concentrations, defined as greater than or equal to 4.0 micrograms per liter (µg/L), in groundwater. Arsenic concentrations of 4.0 µg/L or greater (elevated concentrations) were detected in 18 percent of samples across Pennsylvania; 8 percent of samples had concentrations that equaled or exceeded the U.S. Environmental Protection Agency’s drinking-water maximum contaminant level of 10.0 µg/L. The highest arsenic concentration was 490.0 µg/L. Comparison of arsenic concentrations in Pennsylvania groundwater by physiographic province indicates that the Central Lowland physiographic province had the highest median arsenic concentration (4.5 µg/L) and the highest percentage of sample records with arsenic concentrations greater than or equal to 4.0 µg/L (59 percent) and greater than or equal to 10.0 µg/L (43 percent). Evaluation of four major aquifer types (carbonate, crystalline, siliciclastic, and surficial) in Pennsylvania showed that all types had median arsenic concentrations less than 4.0 µg/L, and the highest arsenic concentration (490.0 µg/L) was in a siliciclastic aquifer. The siliciclastic and surficial aquifers had the highest percentage of sample records with arsenic concentrations greater than or equal to 4.0 µg/L and 10.0 µg/L. Elevated arsenic concentrations were associated with low pH (less than or equal to 4.0), high pH (greater than or equal to 8.0), or reducing conditions. For waters classified as anoxic (405 samples), 20 percent of sampled wells contained water with elevated concentrations of arsenic; for waters classified as oxic (1,530 samples) only 10 percent of sampled wells contained water with elevated arsenic concentrations. Nevertheless, regardless of the reduction-oxidation classification, 54 percent of samples with low pH (13 of 24 samples) and 25 percent of samples with high pH (57 of 230 samples) had elevated arsenic concentrations. Arsenic concentrations in groundwater in Pennsylvania were correlated with concentrations of several chemical constituents or properties, including (1) constituents associated with redox processes, (2) constituents that may have a similar origin or be mobilized under similar chemical conditions as arsenic, and (3) anions or oxyanions that have similar sorption behavior or compete for sorption sites on iron oxides. Logistic regression models were created to predict and map the probability of elevated arsenic concentrations in groundwater statewide in Pennsylvania and in three intrastate regions to further improve predictions for those three regions (glacial aquifer system, Gettysburg Basin, Newark Basin). Although the Pennsylvania and regional predictive models retained some different variables, they have common characteristics that can be grouped by (1) geologic and soils variables describing arsenic sources and mobilizers, (2) geochemical variables describing the geochemical environment of the groundwater, and (3) locally specific variables that are unique to each of the three regions studied and not applicable to statewide analysis. Maps of Pennsylvania and the three intrastate regions were produced that illustrate that areas most at risk are those with geology and soils capable of functioning as an arsenic source or mobilizer and geochemical groundwater conditions able to facilitate redox reactions. The models have limitations because they may not characterize areas that have localized controls on arsenic mobility. The probability maps associated with this report are intended for regional-scale use and may not be accurate for use at the field scale or when considering individual wells.