Drinking water contamination is an ongoing issue across the United States. However, tracking water quality violations and notifying residents about them is challenging, and there is no systematic approach for prioritizing assistance once a violation is detected. Using a dataset intended to assess bottled water marketing trends, Maura Allaire, an assistant professor in Urban Planning and Public Policy at UC Irvine, and her collaborators are tackling these challenges and gaining a better understanding of how communities deal with contaminated water.

In a new paper published in the Proceedings of the National Academy of Sciences, the researchers explain how tracking bottled water sales might be help improve the process of identifying drinking water violations, notifying residents, and providing assistance. People reduce their exposure to contaminated water in many ways including boiling or filtering, but because bottled water sales can be traced, they may be a helpful new indicator of how communities react to water quality violations.

In the U.S., drinking water violations are split into two categories. The first are short-term, with acute effects that require immediate public notification. Allaire and colleagues found that for these kinds of violations, which are often due to pathogens like E. Coli, bottled water sales increased 14%. For the second type of violations, those related to long-term exposure risk and the category most violations fall into, bottled water sales increase on average 4.9%.

This means that, generally, bottled water sales are most responsive to violations that could cause acute health effects and require public notification within 24 hours. However, the actions that people take to deal with water contamination vary across communities. For example, low-income rural communities do not show a significant response to nitrate violations, which pose the highest health risks to infants under six months old.

“It was surprising for us to find that low-income rural communities did not respond to nitrate violations, since these populations are the ones facing the brunt of nitrate contamination. It is possible that these communities take other actions, such as installing nitrate filters, but we'd need to investigate further,” says Allaire.  

The researchers also expected to find that communities might make a more permanent switch to bottled water after experiencing a violation. However, Allaire says, “the effect we found was very small. After an acute violation occurred for the first time, bottled water sales are slightly higher at around 2%, but this declines as time goes on. For locations with many violations, sales of bottled water drop back to normal levels after violations end.”

Analyzing all those bottled water purchases was no small feat. “This was a two-year effort by a team of four researchers. At first, it was a challenge to work with such a large dataset of weekly bottled water sales. The raw data contained over 5,000 Universal Product Codes (UPCs) of bottled water purchased at over 25,400 stores. Sorting through terabytes of bottled water purchase data took quite a bit of time and processing before it could be used in the analysis. For some of the students, this was their first introduction to working with big data,” says Allaire.

There is yet more work to do. One outstanding challenge when it comes to drinking water violations is that regulations that guide public notification have not been updated in two decades. The guidelines recommend notifying people via radio, television, and notices posted in public places, and the researchers note that more modern forms of communication could improve public awareness and ability to respond.

In addition, tracking consumer purchases could detect emerging water quality concerns, especially for contaminants that are sampled infrequently and effect water taste or appearance. Additional monitoring could improve response of state and local agencies on issues that otherwise go undetected or unreported.

Allaire says that “it was exciting to use this extensive dataset to look at people's behavior. While this dataset was originally intended to assess marketing trends, it can shed light on how communities are affected by contaminated water, and help improve both notifications and assistance.” 

This article was first published in The Confluence blog.