Precision Agriculture | April 20, 2023
Researchers from the University of Nebraska conducted a study to improve nitrogen (N) fertilizer management in winter crops like wheat and barley. By applying the economic optimal nitrogen rate (EONR), farmers can enhance N fertilizer efficiency, increase profits, and reduce environmental impacts. The study utilized on-farm precision experimentation (OFPE) to collect extensive data for estimating the EONR.
Machine learning techniques, specifically generalized additive models (GAM) and random forest (RF), were employed to predict crop yields and EONR. The models analyzed various factors such as soil conditions, terrain characteristics, and remote-sensed variables.
The results showed that both GAM and RF models accurately predicted crop yields with an average error of 13.7%. However, the estimated EONR values differed significantly between the two models. Soil phosphorus availability and organic matter content emerged as influential factors, but their impact varied across fields.
The study highlights the importance of site-specific considerations when determining the EONR, as different fields may require tailored nitrogen recommendations. Further research is needed to refine machine learning methods and ensure reliable and automated N fertilizer recommendations as agriculture transitions into the digital era.
Location of experimental sites (A, B) and an example of the Data-Intensive Farm Management nitrogen trial (DIFM N-trial) (C, D) for Field_ID 1