Reuters reported that the U.S. Department of Agriculture (USDA) plans to use an internally built artificial intelligence tool to screen its Nov. 10 crop report after a September corn-data correction angered farmers. Under Secretary Scott Hutchins told Reuters that USDA had concluded the tool would have flagged the problem and would serve as an added check for anomalies.
The National Agricultural Statistics Service (NASS) said the error came from how observations from irrigated and non-irrigated plots in Kansas and Nebraska were weighted in objective-yield calculations. It affected plant-population and ear-count figures in those states, as well as corn yield forecasts for Kansas, Nebraska and the United States and the national production forecast.
NASS corrected its September yield forecasts to 118 bushels per acre in Kansas, down three; 182 in Nebraska, up five; and 178.8 nationally, up 0.3. Its review found October yield and production forecasts unchanged. NASS said future reports will include added checks of acreage weights and objective-yield calculations before publication.
Reuters reported that the September correction angered farmers. It attributed eroded trust among farmers and traders to a broader series of USDA data missteps.
USDA’s September plan proposed testing remote-sensing, mapping and crop-modeling tools alongside information from producers when estimating acreage and yields, while also examining responsible applications of artificial intelligence and machine learning.
NASA and USDA announced an agreement to examine satellite data’s role in crop statistics. NASS’s release calendar schedules the next Crop Production report for Tuesday, Nov. 10, at 12 p.m. ET (17:00 UTC).
