This use cases addresses challenges for process-based crop model applications to optimize nitrogen use.

Background

Minimizing nitrogen losses to the environment from arable crop production has benefits for biodiversity preservation, water quality, climate change mitigation and resource use efficiency. With improved knowledge of the spatial and temporal heterogeneity in crop-soil nitrogen pathways, simulation models driven by inter- and intraseasonal weather projections can evaluate the probability of optimizing tradeoffs between yield levels and nitrogen losses for various crop rotations and nitrogen management regimes. These models can also inform about nitrogen and carbon interactions to better understand options for carbon sequestration.

Despite this potential, the use of process-based models to optimize nitrogen use has been limited due to a number of data-related challenges. First, model simulations are often highly uncertain due to the lack of experimental data for model development and parameterization.

While many experiments have quantified the response of soil organic matter and nitrogen to soil types and agricultural management (e.g., Denef et al. (2007); Pulleman et al. (2005), the data are often not findable or available in standardised form for model use. Furthermore, these studies rarely report sufficient information or data required for simulations, such as soil, weather, management, market or farm economic data. A second key challenge relates to the scaling of input and simulation data to consistent levels (pedon to field to farm to region). The results of aggregation can be heavily influenced by the choice of land use mask, years considered in production area weighting and the aggregation method (Porwollik et al., 2017). Overcoming these data-related challenges results in more robust projections with reduced uncertainties to inform optimal nitrogen management for sustainable crop and soil (carbon) management.

Objectives

The first objective is to facilitate the translation of published research results into FAIR datasets for use in model calibration to improve the quality of crop-soil nitrogen model simulations. The second objective is to reduce model simulation uncertainties associated with data aggregation and scaling.

Progress & next steps

UC Update: Bridging Literature and Models: Creating Agricultural Datasets with AI

This new contribution by Chen et al. (2025) presents an innovative workflow that leverages artificial intelligence to extract, structure, and harmonize data from scientific literature for use in agricultural models. The approach bridges the gap between published research and model applications, paving the way for more data-rich and interoperable simulations in crop and climate modelling.