Satellite-Based Crop Monitoring Using Hyperspectral Imagery and Deep Learning Approaches

Agricultural crop monitoring plays a crucial role in ensuring food quality and sustainable development, yet traditional field surveys are labor-intensive, time-consuming, and often error-prone. These limitations hinder timely detection of crop stress over large areas. In contrast, satellite-based hyperspectral imagery (capturing hundreds of contiguous spectral bands) offers broad coverage and rich spectral detail, revealing subtle indicators of vegetation health across expansive regions. We propose a novel two-phase methodology combining statistical computer vision techniques with deep learning to improve crop monitoring accuracy and efficiency. In Phase 1, hyperspectral images preprocessed with ENVI software (radiometric calibration and atmospheric correction) undergo histogram analysis and red channel intensity distribution evaluation. The red band (corresponding to near-infrared in false-color composites) is especially indicative of vegetation vitality, as stressed crops exhibit diminished reflectance in this range. Binning red channel intensities into discrete ranges enabled clear differentiation between healthy and unhealthy vegetation. Phase 2 employs a YOLOv11 deep learning model for crop classification. The model was trained on labeled hyperspectral images from multiple Indian states (including Punjab and Gujarat) encompassing major crops such as wheat, rice, and cotton. It achieved ~82% classification accuracy in distinguishing crop types. Integrating these complementary approaches leverages both spectral feature insights and data-driven modeling, enabling accurate crop-type mapping alongside early detection of crop stress. The results demonstrate that this efficient and scalable satellite-driven framework can reliably assess crop condition and species, providing a valuable tool for precision agriculture. In real-world applications, it supports sustainable crop management and informs decision support systems for optimized agricultural practices.