Skip to main content
ARS Home » Midwest Area » Columbia, Missouri » Cropping Systems and Water Quality Research » Research » Publications at this Location » Publication #353412
Title: Improving corn nitrogen rate recommendations through tool fusion

Author
item RANSOM, C - University Of Missouri
item Kitchen, Newell
item CAMBERATO, J - Purdue University
item CARTER, P - Dupont Pioneer Hi-Bred
item FERGUSON, F - University Of Nebraska
item FERNANDEZ, F - University Of Minnesota
item FRANZEN, D - North Dakota State University
item LABOSKI, C - University Of Wisconsin
item NAFZIGER, E - University Of Illinois
item SAWYER, J - Iowa State University
item SHANAHAN, J - Fortigen

Submitted to: International Conference on Precision Agriculture Abstracts & Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 6/5/2018
Publication Date: 6/24/2018
Citation: Ransom, C.J., Kitchen, N.R., Camberato, J.J., Carter, P.R., Ferguson, F.G., Fernandez, F.G., Franzen, D.W., Laboski, C.A., Nafziger, E.D., Sawyer, J.E., Shanahan, J.F. 2018. Improving corn nitrogen rate recommendations through tool fusion. 14th International Conference on Precision Agriculture, June 24-27, 2018, Montreal, Quebec, Canada. Paper No. Paper No. 5106.

Interpretive Summary: Improving corn nitrogen (N) fertilizer recommendations can improve farmer’s profits and help reduce excessive applications and therefore N pollution. One way to improve N recommendations may be to employ multiple decision making tools for making recommendations. This could be thought of as “tool fusion”. The objective of this research was to improve corn N management by combining N recommendation tools used for N fertilizer applications. We found that any combination of two or three N recommendation tools together improved performance compared to using any one tool alone. The best fused tool from this research included a yield goal approach, the Iowa Late-Spring Nitrate Test, and canopy reflectance sensing. When these tools were combined, there was a 75% increase over when the best tool was used alone. However even when these tools were combined, only about 45% in the variation in economic optimal N rate was predicted, indicating there is opportunity for additional tool improvement. This analysis demonstrated that combining tools is a valid way to improve N recommendations, and thus could aid farmers in better managing N fertilizer than using a single tool by itself. Better N management will increase farmer profits and help reduce over-application of fertilizer that leads to environmental problems.

Technical Abstract: Improving corn (Zea mays L,) nitrogen (N) fertilizer rate recommendation tools can improve farmer’s profits and help mitigate N pollution. One way to improve N recommendation methods is to not rely on a single tool, but to employ two or more tools. This could be thought of as “tool fusion”. The objective of this analysis was to improve N management by combining N recommendation tools used for guiding rates for an in-season N application. This evaluation was conducted on 49 N response trials that spanned eight states and three growing seasons. An economical optimal N rate (EONR) was calculated for N treatments receiving 45 kg /ha applied at planting and the remaining fertilizer N applied at the V9 corn developmental stage. A yield goal approach, the Iowa Late-Spring Nitrate Test (IA LSNT), and canopy reflectance sensing were the three recommendation tools used to evaluate the tool fusion concept. Tools were fused using either an elastic net or decision tree approach. Using the elastic net approach, tools were fused with all combinations of main and two- or three-way interaction terms regressed against EONR. The decision tree was developed using only the main effects compared against EONR. Regardless of the method used to combine tools, any combination of two or three N recommendation tools together improved performance compared to using any one tool alone. The best elastic net based tool fusion occurred when all three recommendation tools and all possible interactions were included in the model which helped explain 42% of the variation around EONR, a 75% increase over the best tool alone. Additionally, the root-mean-square error (RMSE) improved from 68 kg N/ha (best tool used alone) to 55 kg N/ha. However, the best combination occurred when using the three N recommendation tools in a decision tree. The decision tree method explained 45% of the variation in EONR and had a RMSE value equal to 53 kg N/ ha. This analysis demonstrated that combining tools is a valid way to improve N recommendations, and thus could aid farmers in better managing N than using a single tool by itself.