Global agriculture faces the challenge of increasing productivity while maintaining environmental sustainability. Agricultural system models support climate-smart land management; however, accurately predicting soil organic carbon (SOC) density changes and indexing soil health remains difficult because of the complex interactions among soils, climate, land use, and management practices. Emerging technologies such as Artificial Intelligence (AI), including remote sensing, drones, and spectroscopy, provide opportunities for real-time monitoring and decision-making despite having several challenges.
This study advances the MODASYS (Model for Decision-Making in Agri-Environmental Systems) framework, facilitating the assessment of nutrient and land-management strategies aligned with the Climate Action Plan. By integrating high-resolution georeferenced soil datasets with the Land Parcel Identification System (LPIS). For this, SoilGrids (Global Soil Map) and the National Soil Database of Ireland were harmonised using ArcGIS and Multiple Imputation by Chained Equations (MICE), supported by advanced statistical and Machine Learning approaches, to generate a robust and spatially consistent soil dataset and map. The resulting Soil Health Indices (SHIs) benchmark observed (O) SOC against Typical (T) values, which remain limited for organo-mineral and organic soils, revealing substantial spatial variability across land uses and soil types in Ireland.
By integrating LPIS-mapped datasets with the MODASYS API, the framework enables high-resolution simulations of soil and climate-driven variables to predict SHIs and quantify carbon footprints, facilitating the precise delineation of optimal carbon-management zones. Simulations from an Irish dairy farm case study using nine land-use scenarios showed that, although the highest SHI (O/T SOC = 3.8) was predicted in grasslands, mixed farming systems incorporating agroforestry and reduced nitrogen inputs achieved high SHI values (0.94) while simultaneously lowering carbon footprints and maintaining profitability through reduced reliance on synthetic fertilisers. In contrast, a New Zealand case study with twelve crop-field sequences predicted declining O/T SOC ratios (from 1.20 to 1.12 over five years) despite fields remaining within a “High” SHI category, highlighting the limitations of relative indices in detecting absolute carbon depletion.
The MODASYS framework demonstrates the power of integrated soil-health modelling for strategic agricultural planning, carbon farming, and climate resilience. Future iterations will incorporate AI-driven digital twin methodologies, evolving the model into a dynamic platform for predictive soil management and sustainable intensification in the face of escalating climate risks.