Scientists from the University of Vermont and Utah State University have helped develop a new tool that could detect low water quality, enabling communities to issue better warnings and more efficiently provide a clean water supply.

The tool uses artificial intelligence to predict when runoff into lakes and rivers would carry harmful substances that affect the water quality.

The scientists combined the existing National Water Model, which forecasts streamflow, with AI and sensor data, to predict water quality, according to a Thursday press release from the University of Vermont.  

“This new tool can be implemented across the country and broadly utilized by folks that could use water quality forecasts in any number of applications,” UVM’s Andrew Schroth said in the release. “With the first ever application of the National Water Model to predict water quality, we’ve opened a new window that can really benefit the country as a whole moving forward.”

A key determinant of water quality is turbidity, or the concentration of particles, such as sediment.

At Esopus Creek, a major supplier of drinking water for New York City where researchers tested the new technology, results showed that high turbidity caused poor water quality because of how high particle concentrations interfere with treatment processes. The researchers concluded that forecasting high turbidity threats is essential to maintaining efficient water supply, according to the release.

“When too much sediment comes into the reservoir during or after big storms, New York City has to limit supply and modify their operations,” Schroth said in the release.

Most tap water in the United States comes from lakes, rivers, reservoirs or groundwater, according to the U.S. Centers for Disease Control and Prevention. More than half of Vermonters get their water from public water systems, with the rest getting it from private sources, like wells and springs, according to the Vermont Department of Health. Municipalities like Burlington, Montpelier and Brattleboro use bodies of water such as Lake Champlain, Berlin Pond and Pleasant Valley Reservoir, respectively, to supply residential customers.

The scientists are working toward a demonstration of the tool’s ability to forecast the runoff of phosphorus and chloride to the Lake Champlain Basin, after flooding or high water forecast. This prediction method could promote the maintenance of clean water by alerting farmers of harmful substances in water sources due to their use of road salt or fertilizer, according to Schroth and Utah State scientist John Kemper, his research partner.

With the ability to see the weather’s impact on water quality in real time, many communities could benefit from the use of this new tool, the duo said. Schroth, Kemper and their team of researchers are working to develop a version of the tool for widespread use.

Schroth and Kemper are working to apply the tool to watersheds in the Lake Champlain area, they said via email, which “will allow Vermonters to get near real-time forecasts of water quality in the streams draining to the lake.” They said they plan to have that work done later this year.

The duo said they are pursuing funding to create a product that will operate on a national scale, which would be a larger undertaking. With continued investment in their research, they hope to have a national version operational roughly by 2027.