PREDICTING THE PROBABILITY OF ARSENIC OCCURRENCE IN GROUNDWATER FROM BEDROCK WELLS IN NEW HAMPSHIRE FOR PUBLIC HEALTH STUDIES

Inorganic arsenic in groundwater used for drinking water is a worldwide concern due to adverse impacts on human health related to ingestion. Arsenic is prevalent in groundwater from bedrock aquifers in New England and concentrations in bedrock well water are generally low to moderate (< 10 to 50 µg/L). However, parts of the region, particularly in New Hampshire and Maine, have occasional wells with arsenic concentrations exceeding 1,000 µg/L. Models predicting occurrence and/or concentrations of arsenic in wells are being used by environmental and health scientists in the northeast.

The USGS is using a regression modeling approach to predict the probability of occurrence of elevated arsenic concentrations in groundwater in relation to regional geological, geochemical, hydrologic, physiographic, and land-use features. Inorganic arsenic concentrations in sampled bedrock wells vary by bedrock geologic formations and by dominant lithology types. In one regionally extensive belt of variably calcareous metasedimentary rocks, about 30 percent of wells have groundwater arsenic concentrations exceeding the U.S. Environmental Protection Agency standard of 10 µg/L. A previous New England–wide model predicting groundwater arsenic concentrations exceeding 5 µg/L indicated that geologic information was among the most significant predictors, in addition to hydrologic and geochemical variables (Ayotte and others, 2006).

For the New Hampshire models that are currently under development, 364 of 1,716 arsenic observations (dependent variable) had concentrations ≥ 10 µg/L and these data were censored at multiple reporting levels. Models predicting the probability of arsenic in groundwater at thresholds of 1, 5, and 10 µg/L were evaluated. Preliminary explanatory variables for the New Hampshire models are related to a variety of regional and local-scale geologic features (Robinson and Ayotte, 2006). Other variables such as rainfall, recharge, elevation, glacial marine inundation, stream alkalinity, stream-sediment geochemistry, land use, population density, and waste-site information were predictive. Maps indicate that high probabilities of arsenic ≥ 1 µg/L are widespread in New Hampshire whereas probabilities of arsenic ≥ 10 µg/L occur mostly in the southeastern and central part of the state. Modeled probabilities will be stored in the New Hampshire Environmental Public Health Tracking database. Using a similar modeling approach, Canadian health researchers are interested in joint collaboration to develop a large-area groundwater-arsenic model for New England and Eastern Canada.

Ayotte, J.D., Nolan, B.T., Nuckols, J.R., Cantor, K.P., Robinson, G.R., Baris, D., Hayes, L., Karagas, M., Bress, W., Silverman, D.T., and Lubin, J.H., 2006, Modeling the Probability of Arsenic in Groundwater in New England as a Tool for Exposure Assessment: Environ. Sci. Technol., v. 40, no. 11, p. 3578-3585.

Robinson, J., Gilpin R., and Ayotte, J.D., 2006, The influence of geology and land use on arsenic in stream sediments and ground waters in New England, USA: Applied Geochemistry, v. 21, no. 9, p. 1482-1497.