Assessing models of arsenic occurrence in drinking water from bedrock aquifers in New Hampshire

Journal of Contemporary Water Research and Education
By: , and 

Links

Abstract

Three existing multivariate logistic regression models were assessed using new data to evaluate the capacity of the models to correctly predict the probability of groundwater arsenic concentrations exceeding the threshold values of 1, 5, and 10 micrograms per liter (µg/L) in New Hampshire, USA. A recently released testing dataset includes arsenic concentrations from groundwater samples collected in 2004–2005 from a mix of 367 public-supply and private domestic wells. The use of this dataset to test three existing logistic regression models demonstrated enhanced overall predictive accuracy for the 5 and 10 μg/L models. Overall accuracies of 54.8, 76.3, and 86.4 percent were reported for the 1, 5, and 10 μg/L models, respectively. The state was divided by counties into northwest and southeast regions. Regional differences in accuracy were identified; models had an average accuracy of 83.1 percent for the counties in the northwest and 63.7 percent in the southeast. This is most likely due to high model specificity in the northwest and regional differences in arsenic occurrence. Though these models have limitations, they allow for arsenic hazard assessment across the region. The introduction of well-type (public or private), well depth, and casing length as explanatory variables may be appropriate measures to improve model performance. Our findings indicate that the original models generalize to the testing dataset, and should continue to serve as an important vehicle of preventative public health that may be applied to other groundwater contaminants in New Hampshire.

Study Area

Publication type Article
Publication Subtype Journal Article
Title Assessing models of arsenic occurrence in drinking water from bedrock aquifers in New Hampshire
Series title Journal of Contemporary Water Research and Education
DOI 10.1111/j.1936-704X.2017.03238.x
Volume 160
Issue 1
Year Published 2017
Language English
Publisher Wiley
Contributing office(s) New England Water Science Center
Description 17 p.
First page 25
Last page 41
Country United States
State New Hampshire
Google Analytic Metrics Metrics page
Additional publication details