Farmnet -XG-A Hybrid Digital Twin model of Crop Recommendation in Smart Farming

This research focuses on digital twin implementation of hybrid AI model for crop classification using NASA’s Distributed Active Archive Centers (DAAC) Virtual Environment (DAV) remote sensing data. By integrating climate, soil, and environmental parameters, we assess the efficiency gains of NASA DAV over traditional spectral analysis in precision agriculture. A novel hybrid Farmnet-XG model achieved 96.55% accuracy, outperforming Random forest, Support Vector Machines, K Nearest Neighbors, XG Boost, and Gradient Boosting. SHapley Additive exPlanations (SHAP) and Local Interpretable Model-agnostic Explanations (LIME) were used to ensure explainability, demonstrating temperature, NDVI and soil moisture as the strongest features. This implementation result shows improvement in crop recommendation performance for enhanced yield prediction, increased efficiency of resources used, improved management of land-use, and real time monitoring using digital twins.