CROP MODELLING IN THE ERA OF SMART FARMING: TRENDS, CHALLENGES AND OPPORTUNITIES

Abstract

Modern smart farming leverages artificial intelligence, Internet of Things (IoT), and digital twin technologies to transform crop modelling from static, parameter-heavy systems into adaptive, real-time predictive frameworks. The paper illustrates how the integration of process-based models with machine learning, IoT enabled sensing platforms, and digital twin technologies contributes to a new paradigm in predictive agriculture. The reviews investigate the design principles, comparative strengths, and applications of traditional tools such as DSSAT and APSIM alongside emerging approaches that include IoT–ML hybrids, generative models, and explainable AI frameworks. Furthermore, research highlights the novel role of digital twin systems as dynamic virtual counterparts of crops, emphasizing their ability to synchronize field data with computational simulations for enhanced monitoring and optimization. The study also identifies critical challenges including heterogeneous datasets, limited model generalization across agro-ecological contexts, cybersecurity vulnerabilities, and the heavy computational demands of AI-intensive infrastructures. The review further discusses hybrid frameworks that integrate explainable AI with mechanistic crop models, explicitly linking genotype × environment × management interactions to yield outcomes and strengthening interpretability. The work demonstrates a conceptual and comparative novelty by framing future research needs in multi-scale modelling, multi-modal data fusion, and edge–cloud deployments, thereby providing an integrative roadmap for sustainable, resilient, and transparent agricultural systems.