Multi-sensor integration and cloud-native AI for climate-smart agricultural monitoring: A systems framework
Computers and Electronics in Agriculture, cilt.253, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Derleme
- Cilt numarası: 253
- Basım Tarihi: 2026
- Doi Numarası: 10.1016/j.compag.2026.112182
- Dergi Adı: Computers and Electronics in Agriculture
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Aerospace Database, Applied Science & Technology Source, BIOSIS, Compendex, Environment Index, Geobase, INSPEC, Academic Search Ultimate (EBSCO), Engineering Source (EBSCO), Technology Collection (ProQuest)
- Anahtar Kelimeler: Agricultural monitoring, Cloud-native computing, Google earth engine, Machine learning, Multi-sensor data fusion
- Karadeniz Teknik Üniversitesi Adresli: Evet
Özet
Climate change intensifies drought, heat stress, and production instability, requiring scalable monitoring systems to support climate-smart agriculture. Remote sensing has evolved into operational decision-support infrastructure enabled by open-access satellite archives, multi-sensor integration, cloud-native computing, and machine learning. This review synthesizes peer-reviewed literature (2000–2025) and proposes a unified five-layer computational framework for agricultural climate adaptation. The framework integrates data acquisition, cloud processing, machine learning analytics, data fusion, and decision-support outputs into end-to-end workflows. The synthesis is based on a structured review of 120 core peer-reviewed studies, complemented by additional foundational and methodological references to ensure technical depth. Comparative evidence indicates that multi-sensor fusion enhances robustness relative to single-source approaches. Optical-radar integration using Sentinel-1 and Sentinel-2 has achieved classification accuracies of up to 89 % in cloud-prone regions, compared to approximately 76 % using optical data alone. ML-based yield prediction models frequently improve performance by 5–20 % in R2 relative to empirical regression under data-rich conditions, while multi-temporal ML approaches enable detection of crop stress 2–4 weeks before visible symptoms occur. However, ML superiority over process-based models is conditional on calibration quality, data availability, and cross-regional transferability. Cloud-based platforms have reduced computational barriers and enabled large-scale time-series analysis, supporting operational monitoring from field to regional scales. Despite these advances, reproducibility limitations, ground-truth data scarcity, and reduced transferability across agroecological zones remain significant constraints, particularly in smallholder-dominated systems. Advancing climate-smart agricultural monitoring therefore depends not only on improved sensing technologies, but on interoperable, reproducible, and transferable computational frameworks capable of delivering reliable decision support across heterogeneous agricultural landscapes.