A Recommendation Engine to Optimize Crop Yields Utilizing the Dynamic Relationship between Nitrogen, Irrigation, and Precipitation

Abstract

This thesis builds a recommendation engine that can be used to help growers optimize their crop yields by evaluating and utilizing the dynamic relationship between nitrogen, irrigation, and precipitation. In this project, optimizing crops relates to weighing the environmental impact with maximizing the crops harvest weight and profit. In order to assess this, a model was built using the Decision Support System for Agrotechnology Transfer (DSSAT). The model creates a randomized weather file and a soil file that can be manipulated for different irrigation and nitrogen fertilization amounts. The manipulated files are then inputted to DSSAT which outputs relevant information about the crop throughout the growing period. The outputs are used to evaluate the effects nitrogen and water have on both the crop’s harvest weight and the overall environment. The model creates a number of these trials to generate a distribution. The distribution can be evaluated to get a recommendation for the amount of irrigation and nitrogen fertilization based on optimizing the crop. The model was tested by building assumptions to evaluate optimizing maize growth in Des Moines, Iowa. The model showed the need for sufficient amounts of water in order to fully utilize the nitrogen used and that an increase in the amount of nitrogen and irrigation gave the grower greater control. The increased control decreased the disparity of harvest weights across trials of the same conditions but increased the negative environmental impact. The recommendation based on the model is to use 50 𝑘𝑔ℎ𝑎 of nitrogen and 400 𝑚𝑚ℎ𝑎 of irrigation for Des Moines.