An Intelligent Data-Driven Framework for Optimizing Wheat Management toward High Yields with Reduced Nitrogen Inputs

Abstract

Achieving a balance between high and stable wheat yields and reduced nitrogen inputs remains a major challenge for sustainable agriculture. To address the limited generalizability of current nitrogen management models across environment and management gradients, this study develops an AI- and data-driven decision framework for wheat based on a harmonized global field trial database. The framework is built on datasets compiled from published field experiments and integrates two modeling solutions (AutoGluon and TabPFN) to predict and optimize wheat yield and nitrogen uptake across different climate zones. The models exhibit strong performance and robustness, with an R2 of 0.88 and an RMSE of 26.3kg N{$\cdot$}ha{$^-$}1 for predicted nitrogen uptake and an R2 of 0.82 and an RMSE of 865.9kg DM{$\cdot$}ha{$^-$}1 for predicted yield. Scenario analysis shows that nitrogen application can be reduced by 13% while maintaining yield, and that improved management can increase average yield by 19.6% while reducing nitrogen application by 18.2%, demonstrating the potential for multi-factor synergistic optimization. Analysis of management contributions using sensitivity analysis and Shapley decomposition revealed that improvements in nitrogen application methods and forms, as well as multiple applications, had the greatest impact on yield enhancement. Shapley decomposition further validated these conclusions and quantified the relative contributions of each management factor under multi-factor conditions. Overall, this study provides quantitative evidence that AI-driven management can identify agronomic interventions that increase yields while reducing nitrogen inputs, thereby informing scalable, environment-specific decision frameworks for wheat production.

Prof. Dr. Kang Yu

Professor of Precision Agriculture

My research interests include precision crop farming, hyperspectral remote sensing, and AI in agriculture.