Journal of Agronomy and Field Innovations

Agriculture and the agrifood domain are major anthropogenic sources of GHG emissions, but agriculture can also help climate mitigation, as it can provide significant opportunities for SOC sequestering. Artificial intelligence in this regard, along with geospatial technologies, IoAT and digital twins, is increasingly regarded as a technological foundation of carbon-smart farming which aims to marry productivity, economic benefits, and carbon management.
This review aims to summarize the existing information published mostly between 2019 and 2026 regarding AI-enabled carbon-smart farming, so as to consolidate the existing proof on the use of ML, DL, remote sensing, GIS, IoAT and digital twins technologies in carbon monitoring, GHG emissions reduction and agroecosystems climate friendly management.
Framework: The reviewed works are grouped into four connected themes: (i) AI-driven predictive modelling of SOC and GHG emissions; (ii) carbon measurement through remote sensing and GIS; (iii) IoAT and digital twin systems for real-time field monitoring; and (iv) carbon accounting, measurements, reporting and verification that convert modelling and monitoring findings into verifiable carbon credits.
General findings: The materials reviewed testify to the fact that tree-based ensemble algorithms (random forest, gradient boosting, XGBoost) as well as deep neural architectures are mostly used for SOC and GHG emissions modelling (the coefficients of determination of the developed models often exceed 0.80 thanks to the application of remote sensing, radar and soil proximal sensing methods). Digital twin and IoAT systems are still being introduced and face lots of obstacles due to variability of data, cost of sensors and challenges related to interoperability. Moreover, MRV procedures have some inconsistencies in their methodology in terms of sampling depth and accounting method, which does not allow for adequate comparison of carbon credits from different programmes. The majority of studies highlight regenerative techniques (like conservation tillage, cover cropping and crop rotation) used for optimisation of the process in both spatial and temporal terms.
Scientific significance and conclusion The convergence of AI, geospatial monitoring and carbon accounting infrastructure represents a scientifically credible pathway toward scalable and verifiable soil carbon sequestration, but the literature consistently identifies data standardisation, model transferability across agroecological zones, and transparent MRV governance as prerequisites for translating research advances into climate-neutral agricultural systems at scale.