|CAMBERATO, JAMES - Purdue University|
|CARTER, PAUL - Farmer|
|FERGUSON, RICHARD - University Of Nebraska|
|FERNANDEZ, FABIAN - University Of Minnesota|
|FRANZEN, DAVID - North Dakota State University|
|MYERS, D. BRENTON - Corteva Agriscience|
|NAFZIGER, EMERSON - University Of Illinois|
|SAYWER, JOHN - Iowa State University|
|SHANAHAN, JOHN - Agoro Carbon Alliance|
Submitted to: Soil Science Society of America Journal
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 2/20/2023
Publication Date: 3/5/2023
Citation: Ransom, C.J., Kitchen, N.R., Camberato, J.J., Carter, P.R., Ferguson, R.B., Fernandez, F.G., Franzen, D.W., Laboski, C.A., Myers, D., Nafziger, E.D., Saywer, J.E., Shanahan, J.F. 2023. Combining corn nitrogen recommendation tools for an improved economical optimal nitrogen rate estimation. Soil Science Society of America Journal. 87(4):902-917. https://doi.org/10.1002/saj2.20539
Interpretive Summary: Corn nitrogen (N) fertilization is a complex problem for both farmers and the environment. Under applying reduces profit for the farmer, in terms of reduced grain yields, while over applying results in excess N lost to the environment—resulting in increased N related pollution. Determining the correct amount of N fertilizer to apply varies within and across fields (soil driven), and by year (weather driven). To account for this, many corn N rate recommendation tools have been developed. They have all been less reliable than desired at being adaptive to variations in soil and weather. This investigation looked at combing two or three tools together to get a more accurate estimation of the optimal N rate. Using 49 site-years collected across three years and eight states, two machine learning techniques were used to determine the best method for combining tools. Tools used in various combinations included a yield goal method, soil nitrate tests, a computer simulation crop model, and canopy reflectance sensing. Both techniques proved helpful, as the R^2 based accuracy improved from = 0.24 (using one tool) to = 0.46 (using three tools combined). Using this technique could allow farmers to better manage their N fertilizer by applying close to the optimal N rate.
Technical Abstract: Improving corn (Zea mays L.) nitrogen (N) rate fertilizer recommendation tools can improve farmers’ profits and help mitigate N pollution. Numerous efforts have been given to improve these tools, but to date improvements for predicting economically optimum N rate (EONR) have been modest. This work’s objective was to use ensemble learning to improve our estimation of EONR (for a single at-planting and split N application timing) by combining multiple N recommendation tools. The evaluation was conducted using 49 N response trials that spanned eight states and three growing seasons. Elastic net and decision tree approaches regressed EONR against three unique tools for each N application timing. Tools used in various combinations included a yield goal method, soil nitrate tests, a computer simulation crop model, and canopy reflectance sensing. Any combination of two or three N recommendation tools improved or maintained performance metrics (R^2, RMSE, and number of sites close to EONR). The best results for a single at-planting recommendation occurred when the three at-planting N recommendation tools were combined (including interactions) using an elastic net regression model. This combined recommendation tool had a significant linear relationship with EONR (R^2 = 0.46), an increase of 0.27 over the best tool evaluated alone. Combining multiple tools increased the implementation cost but it did not reduce profitability and, in some cases, improved profitability. These result prove that combining tools is a valid way to improve N recommendations to better match EONR, and thus could aid farmers in better managing N than using a single tool by itself.