Every day, the U.S. water supply system works hard to provide safe water. Now, scientists at the University of Vermont (UVM) have created a new AI tool to help communities predict water quality issues before they happen.
Their research shows that the National Water Model, a system used by the government to predict water flow, can be improved with AI and real-time sensor data. This upgrade allows it to forecast water quality, not just stream flow.
“This tool can be used nationwide to help people prepare for water quality changes,” said Andrew Schroth, the lead researcher. “It’s the first time the National Water Model has been used this way, and it can benefit the entire country.”
Testing in New York City
To see how well the tool works, researchers tested it on New York City’s water supply. They focused on Esopus Creek in the Catskill Mountains, which feeds the Ashokan Reservoir. This reservoir provides about 40% of NYC’s drinking water.
One big challenge for water quality is turbidity, or cloudiness caused by sediment. When sediment levels get too high, the NYC Department of Environmental Protection (DEP) must limit the reservoir’s use. This disrupts water management.
“After big storms, too much sediment enters the reservoir, forcing NYC to change its operations,” Schroth explained. The tool helps predict these events, making it easier to manage the water supply.
A Nationwide Solution
The National Water Model already predicts streamflow across the U.S., helping communities prepare for floods. This new AI-powered version can also predict water quality issues, like sediment levels and harmful algal blooms.
Researchers combined data from the National Water Model with sensor data from streams. This allowed them to create more accurate forecasts of water quality. The tool can help:
- Water treatment plants plan for storms affecting water quality
- Authorities close beaches if harmful algal blooms are expected
- Farmers adjust fertilizer use based on water conditions
“Turning a streamflow tool into a water quality forecasting tool helps communities prepare for changes,” said researcher John Kemper.
This model can be used across the U.S. to predict issues like nitrogen, phosphorus, and chloride levels in water. It’s a major step toward improving water quality management nationwide.
Source: phys
