Sustainable Agriculture and Climate Change: A Deep Learning Approach to Remote Sensing for Food Security Monitoring

Remote sensing technology has revolutionized agricultural monitoring, providing crucial data for managing crops and predicting yields. Traditional models, while effective, often fail to capture the complex spatial and temporal dynamics inherent in agricultural landscapes. This study introduces a sophisticated Hybrid CNN-Transformer model tailored for agricultural monitoring via remote sensing data. Unlike conventional models, this advanced approach seamlessly integrates spatial features extracted by CNNs with the broader contextual insights offered by Transformers. The model is meticulously evaluated against nine existing methodologies including standard CNNs, RNNs, LSTMs, GANs, SVMs, Decision Trees (DT), Random Forests (RF), Gradient Boosting Machines (GBM), and Logistic Regression (LR). The proposed model demonstrates superior performance across multiple metrics, achieving an unprecedented accuracy of 98.88%. Precision, recall, F1-score, and ROC-AUC scores also surpass existing models, highlighting its effectiveness in classifying and predicting diverse agricultural conditions from remote sensing data. This paper provides comprehensive comparisons across these dimensions, underscoring the advantages of integrating advanced neural network architectures for enhanced agricultural insights and decision-making.