Iowa City's water treatment plant is now betting on a NASA-funded AI to predict nitrate spikes that the EPA no longer has the science to regulate.
Sam Fosse, a senior treatment plant operator, watched the Iowa River climb above 10 milligrams per liter in recent months after a wet December. That number is the law. It is the maximum contaminant level the EPA set in 1991 to prevent blue baby syndrome in infants. It has not been updated since. And the federal office that was supposed to re-examine whether that number protects against cancer was eliminated by the Trump administration earlier this year.
Fosse's plant now blends water from multiple sources to stay under the threshold. Soon, he will rely on an algorithm built by the University of Iowa and funded by NASA to tell him, seven days out, when the next spike is coming. The regulator stopped asking whether the standard is safe. A space agency is now telling utilities when they will violate it.
The review that vanished
The EPA announced in 2017 that it would re-assess the cancer risks of nitrates and nitrites in drinking water. In 2023, the agency accepted public comments. It has not provided an update since. Then, in 2026, the Trump administration eliminated the Office of Research and Development, the institutional home of the Integrated Risk Information System program that would have conducted the review.
The science is simply not coming.
Tyler Lobdell, a senior staff attorney with Food & Water Watch, called the decision alarming. "The assessment was intended to do the hard science to support appropriate human health protections," he told The New Lede. "It's alarming that the Trump administration has decided to undermine that process."
The consequence is not abstract. The 10 mg/L standard is the only federal backstop for water utilities that cannot afford the expensive treatment plants required to remove nitrates. Those utilities now have no federal science to tell them whether the water they are delivering, at levels that meet the 1991 rule, is actually safe.
A standard frozen in time
The current nitrate limit was built to stop methemoglobinemia, a condition that starves infant blood of oxygen. It was not designed to address cancer. Since 1991, epidemiological studies have linked chronic nitrate exposure to colorectal, thyroid, and bladder cancers. The IRIS program was the mechanism for translating that research into a revised safety threshold. That mechanism is gone.
This is not a funding gap. It is a structural void. There is no institutional capacity left at the federal level to ask the question that the EPA itself raised in 2017. The standard is frozen. The science that could change it is dead.
How NASA got into the water business
Jesus Gomez-Velez, an associate professor of civil and environmental engineering at the University of Iowa, is leading a three-year project that began in June 2026 to build a Nitrate Forecasting and Early Warning System. The partners are the cities of Cedar Rapids and Iowa City, along with Des Moines Water Works.
The system ingests NASA satellite imagery to detect crop cover and fertilizer application across the landscape. It pulls real-time nitrate measurements from 60 water quality monitoring stations managed by IIHR-Hydroscience and Engineering. An AI model learns the relationship between what the satellites see, what the ground monitors measure, and what the weather does next. The output is a seven-day nitrate forecast, integrated into the Iowa Water Quality Information System web application.
"Having a prediction beforehand of how the nitrate will look in the next seven days will allow them to actually plan better about which sources to use, but also situations where they are unlikely to meet the water quality standard," Gomez-Velez told the Iowa Capital Dispatch.
The tool does not replace the safety standard. It tells utilities when they are about to violate it. That distinction matters. The federal government set a limit in 1991 and stopped asking whether the limit is correct. A state university and a space agency are now building the early warning system that the regulator never provided.
Amy Kahler, CEO and general manager of Des Moines Water Works, described the operational reality. "Because river conditions are influenced by both the weather and activities across the landscape, it can be difficult to predict what water quality challenges our treatment plants will face on any given day," she said in a statement reported by the Iowa Capital Dispatch. Her utility issued temporary lawn watering bans in the summers of 2025 and 2026 because nitrate concentrations in the Des Moines and Raccoon rivers spiked beyond manageable levels.
The cascade
Here is what the NFEWS model will surface, and what happens next.
The model will reveal that nitrate levels in the Iowa River routinely exceed 10 mg/L for longer durations than previously understood. This is not speculative. Iowa City already measured concentrations above the threshold in 2026, as Fosse confirmed to KCRG. Des Moines Water Works has already restricted water use two summers in a row. The model makes these spikes predictable, which means utilities can no longer claim they were surprised.
First-order consequence: at least one major Iowa water utility, likely Des Moines or Cedar Rapids, will face a binary choice. Build a nitrate removal plant costing north of $50 million, or declare an emergency supply shortage when the forecast shows an extended spike that blending cannot manage. The forecast does not solve the problem. It forces the decision.
Second-order consequence: attribution becomes legally actionable. The AI model correlates upstream land use, detected via satellite, with downstream nitrate concentrations. When a utility faces a $50 million capital expenditure or an emergency declaration, it will use that data to identify the sources. Lawsuits against concentrated animal feeding operations and fertilizer applicators will follow. The model is not just a forecasting tool. It is evidence.
Third-order consequence: the EPA has no institutional capacity to respond. The IRIS program is gone. The Office of Research and Development is eliminated. When utilities start suing each other and upstream polluters, the federal agency that set the standard will have no science to offer a court about whether that standard is adequate. The next administration will face a choice: restore the office in a budget environment hostile to new spending, or watch a cascade of local lawsuits and emergency declarations unfold with no federal scientific authority to intervene.
The winners are clear. AI infrastructure vendors and the University of Iowa's hydrology program will become the de facto national authority on nitrate forecasting. The losers are the Iowans whose water utilities cannot afford the forecast tool or the treatment plant.
Within 12 to 24 months, the NFEWS model will force at least one major Iowa water utility to either build a nitrate removal plant or declare an emergency supply shortage. The EPA, with its core research arm gutted, will have no answer. This is a falsifiable prediction. Check back in 2028.
What operators should do now
For water utility operators, city managers, and public health officials in agricultural states, the federal safety net is gone. If you rely on the EPA to tell you whether your water is safe, you are betting on a dead office.
The NFEWS model is a stopgap. It does not remove nitrates. It warns you when they are coming. You need to either invest in your own monitoring and forecasting capacity, or advocate for state-level nitrate standards that are stricter than the federal 10 mg/L. Several states, including Minnesota and Wisconsin, have already begun this process through their own health departments. Iowa has not.
The alternative is to wait for the lawsuit cascade and hope you are not the defendant. Invest now in the infrastructure that makes the forecast actionable. If you cannot afford a nitrate removal plant, build the political coalition to force upstream polluters to pay for it—starting with mandatory nutrient reduction targets tied to discharge permits. The science that would have protected you from litigation by updating the standard no longer exists. You are on your own.
The 10 mg/L standard
Iowa City's treatment plant operator is staring at a NASA forecast instead of an EPA standard because the federal government set a safety limit in 1991 and then stopped asking whether it was still safe. A university and a space agency are now doing the asking.
The 10 mg/L standard is still the law. But the science that could change it is dead. And the people who drink the water are the ones who will pay the price.
