Vytautas Valinskas | 2026.09.03
https://www.technology.org/2026/09/02/usda-satellites-ai-crop-estimates-pilot/
A Technology Pilot for Federal Crop Data
The US Department of Agriculture is starting a pilot project that leans harder on technology to improve its crop acreage and yield estimates, Secretary Brooke Rollins said on Tuesday. The announcement follows sustained criticism from farmers and traders over whether government agricultural data can still be trusted.
Key Takeaways
- The pilot will pair improved satellite imagery with geospatial tools, crop models and farmer polls, with USDA also exploring artificial intelligence and machine learning.
- Rollins said the department will work with NASA and other federal agencies to make the reporting more accurate.
- Grain prices fell more than 5% after USDA’s January final estimates, at a point when growers were already struggling to make money.
What the Pilot Will Try
Rollins laid out the data modernisation plan at the Farm Progress Show in Boone, Iowa, after a month of listening sessions with farmers and agricultural groups.
“We will conduct a pilot to evaluate the use of improved satellite imagery, working with NASA and other federal government agencies across the country to figure out how we can make this reporting more accurate,” Rollins said during the show.
The project combines geospatial tools and crop models with the existing farmer polls. The department also plans to look at machine learning. Remote sensing is not new to agriculture, and precision farming tools have used satellite and drone imagery commercially for years. The change here is the federal statistical system adopting them.
A second goal is reducing the work asked of farmers. Scott Hutchins, USDA’s under secretary for research, education, and economics, said the department wants to make sharing information easier, including putting polls online and pre-populating fields with data USDA already holds, which should lift response rates.
USDA has attempted data modernisation before, and some of those efforts took years to arrive. Hutchins gave no detailed timeline but promised faster movement. “We’re gonna have results by the end of the term, for sure,” he said.
Why Farmers Stopped Trusting the Numbers
Producers have criticised deep staff cuts at USDA, arguing that collection and accuracy suffered. Their sharpest questions concern the National Agricultural Statistics Service, whose reports move commodity markets, shape the prices growers receive and feed into federal farm programmes.
The January final estimates hit hard. Already low grain prices sank more than 5%, hurting growers who had little margin left.
At the Farm Progress Show, several farmers said the traditional methods were overdue for replacement. “Some of these methods have not changed since the 1970s,” said Jeff Mundorf, an Iowa farmer visiting the show. “These are big projects for one administration and there’s only so much they can do, but I’m optimistic.”
The complaints intensified after unusually large gaps opened between initial and final estimates of 2025 corn acreage, a swing that grain traders and economists noticed alongside farmers.
Politics in the Background
Rollins’ Iowa trip came after the non-partisan Cook Political Report shifted the state’s US Senate race from “lean Republican” to “toss up.” Iowa is an agriculture-dominated Republican stronghold where farmers have been squeezed by tariffs and by rising diesel and fertiliser costs following the Iran war. Democrats need to hold all current seats and gain four more to take the Senate.
USDA started its listening sessions in Minnesota on 4 August, and Rollins plans more through the autumn, department spokesman Harry Fones said on Tuesday.
Those sessions have covered more than statistics. Farmers have raised staffing cuts and other administration policies affecting the farm economy. Rollins has acknowledged the loss of confidence and said USDA needs to rebuild farmers’ trust in its data. The meetings have also become venues for broader grievances. When Rollins visited the Iowa State Fair last month, cattle producers criticised the president’s announcement that the US would increase beef imports, a move meant to lower consumer prices that producers feared would depress cattle prices.
What Rides on Getting This Right
Federal agricultural statistics are unusual among government data products because they carry direct financial consequences the moment they publish. The agency’s reporting calendar covers crop production, acreage, grain stocks and livestock inventory, and traders position around release times. When an estimate moves several percent between preliminary and final, that is not an academic revision. It lands on balance sheets. The agency already publishes geospatial products, including gridded crop condition layers built from confidential county data, so the satellite component is less of a departure than the headline suggests.
The harder question is whether better instruments fix a problem caused partly by fewer people. Satellite imagery can estimate planted area and biomass well. It cannot ask a farmer what he actually harvested, or why a field came in below what the model expected. Ground truth still comes from surveys, and surveys need staff and respondents. Coverage of the announcement noted USDA also intends to publish more about its methodologies, response rates and data limitations, which may matter as much as the imagery. Transparency about uncertainty is cheaper than new satellites, and it addresses the actual complaint, which is that nobody outside the agency could tell how much confidence the numbers deserved. Farms already generating their own sensor and equipment data may end up being the richest input the department is not yet using.