AI in agriculture: hybrid machine learning models for nitrogen simulation

农业模拟模型用于指导氮素管理以减少氮素损失及其对环境的影响，但它们是使用受限的数据集开发的，这限制了它们只能进行特定地点或区域的模拟。该项目采用了一种新的方法，通过应用基于机器学习的数据分析来解决这些问题。该项目将完善氮素损失与其主要驱动因素之间的联系，并通过数据分配、参数优化和模块增强来改进现有的农业生态系统模型。该项目的成果将导致对农业氮素损失的准确预测、农业生态系统模型的进步及其对全球背景的适应性。

Agricultural simulation models are used to guide nitrogen management to reduce nitrogen loss and its environmental impact, but they were developed using constrained datasets, which restricts them to site- or regional-specific simulations. This project adopts a novel approach to addressing these problems by applying machine learning-based data analytics. The project will refine the linkages between nitrogen losses and their key drivers, and improve the existing agroecosystem models through data imputation, parameter optimisation and module enhancement. The outcomes of this project will lead to an accurate prediction of nitrogen losses from agriculture, advancement in agroecosystem models and their adaptability to a global context.