It sounds like the stuff of science fiction — but with enough data, new agricultural models really are able to predict the future of any crop-growing season with startling accuracy, Phys.Org reports.

The approach is called digital-twin technology, and it’s been developed by teams at Texas A&M AgriLife Research who published their findings in the Computers and Electronics in Agriculture journal…

In their first trial, the team collected over 250,000 data points on a single 200-acre Texas farm in one season, studying canopy cover, crop height, and vegetation numbers. After that, they developed a way to interpret the data using artificial intelligence.

For example, the model correctly identified that one farmer’s harvest would need to be prepared earlier than usual. “The farmer said ‘no way. I usually defoliate in July,'” Landivar explained, “but field observations … confirmed the model’s accuracy.” Ultimately, the farmer lost what Landivar calculated to be approximately $70 per acre in potential profit due to not heeding the model’s recommendations.

This is an excerpt. Read the oriignal post here