We develo
ped a
process-based model to
predict the
probability of arsenic exceeding 5
![](/images/entities/mgr.gif)
g/L in drinking waterwells in New England bedrock aquifers. The model is beingused for ex
posure assessment in an e
pidemiologic studyof bladder cancer. One im
portant study hy
pothesis that mayex
plain increased bladder cancer risk is elevatedconcentrations of inorganic arsenic in drinking water. Ineastern New England, 20-30% of
private wells exceed thearsenic drinking water standard of 10 micrograms
perliter. Our
predictive model significantly im
proves theunderstanding of factors associated with arseniccontamination in New England. S
pecific rock ty
pes, higharsenic concentrations in stream sediments, geochemicalfactors related to areas of Pleistocene marine inundationand
proximity to intrusive granitic
plutons, and hydrologicand landsca
pe variables relating to groundwater residencetime increase the
probability of arsenic occurrence ingroundwater. Previous studies suggest that arsenic inbedrock groundwater may be
partly from
past arsenical
pesticide use. Variables re
presenting historic agriculturalin
puts do not im
prove the model, indicating that this sourcedoes not significantly contribute to current arsenicconcentrations. Due to the com
plexity of the fracturedbedrock aquifers in the region, well de
pth and relatedvariables also are not significant
predictors.