A computer model developed at the University of Nebraska in Lincoln uses current and historic weather information, in addition to other data, to more accurately estimate corn yields and how they change as the growing season progresses.
“They’re looking at this as a way to get real-time estimates of crop status that’s really location dependent and fairly accurate,” said Jeff Coulter, University of Minnesota Extension corn specialist and a collaborator on the project.
Nebraska researchers developed and used the Hybrid-Maize crop simulator, expanded its use to five other states in 2014 and are adding Minnesota, South Dakota, Indiana and Missouri this year. It simulates daily corn growth and development — and final grain yield — at 45 locations under both dryland and irrigated conditions.
“The main aim of the Yield Forecasting Center is that every player, from farmers to industry to transportation, have as much information as possible for their management decisions and for their marketing decisions,” said Francisco Morell Soler, one of the researchers on the Nebraska team.
The computer model, he said, complements work by the U.S. Department of Agriculture, which uses field researchers and surveys to estimate yields less often. But the computer can estimate yields twice a week as recent weather data is added, said Morell Soler, in turn providing more accurate and up-to-date information about whether the bushels of corn per acre are likely to be above or below typical amounts at harvest.
The system relies on collaborators like Coulter across the 10-state Corn Belt. They supply weather and other data. Minnesota information will come from two University of Minnesota research and outreach centers in Waseca and Lamberton.
The data include information about historical and recent weather, planting conditions, hybrid maturity, plant population density and other factors. Nebraska will release corn yield forecasts every other week for each location, starting about July 20 and lasting until the end of the season.
Coulter said improved corn yield projections could benefit a wide variety of people.
“That’s important from a planning standpoint so that farmers can have a better idea of what their expected yield is, and how much revenue they could have, and how much grain they’re going to have to handle,” he said. “And from a commodity trading standpoint, it could be really huge.”
Morell Soler said larger food and grain trading corporations already invest in sophisticated systems to estimate crop yields and how they evolve during the growing season. The Nebraska model will provide similar real-time information for the general public, he said.
“All of this information will determine prices and trends of corn,” Morell Soler said. It may, for example, help an individual farmer trying to decide whether to sell part of his corn crop in advance on the futures market, or to wait to sell until harvest or later.
The computer model has some caveats. Although it accounts for water stress and high temperatures during much of a corn plant’s growth, it likely won’t be as accurate for two relatively uncommon situations, researchers said. It does not account for severe heat or water stress for about a week around corn “silking and pollen shed” time, and it does not estimate very well the effects of very early killing frost that may kill the crop before “grain filling” is completed as its kernels mature.
The computer’s yield estimates also are based on optimal conditions, when the corn is not affected by lack of nutrients, disease, insect pressure or weed competition. For that reason, the projections may be somewhat higher than actual yields on farms with those problems.
Morell Soler said the computer simulation is being refined and improved each year, and has been validated by experts in the field. He said the technique might be eventually expanded to other major crops, especially soybeans, but there are no immediate plans to do so.
“It has the potential to have huge impact and use, and it could be more accurate than the USDA approach,” Coulter said. “But as with any prediction thing, it’s not going to be 100 percent perfect.”