Vegetation Health Classification in the Kurdistan Region Using AI and Sentinel-2A Satellite Data
DOI:
https://doi.org/10.31530/cjnst.2026.2.1.3Keywords:
Deep Learning, Environmental Monitoring, Feature Engineering, Machine Learning, NDVI, Remote Sensing, Sentinel-2A, SVM, Vegetation Health ClassificationAbstract
Background: Monitoring climate science is imperative, as it provides the scientific information needed on current changes in the Earth's climate system and their environmental and ecological consequences. However, precisely and scalably classifying vegetation health in heterogeneous semi-arid regions is challenging, particularly with limited labeled data and computational resources.
Aims: This study develops an effective vegetation health classification model using Sentinel-2 imagery and machine learning. This paper examines vegetation health assessment as an ecological indicator of environmental stress, rather than climate change prediction perse.
Methodology: A Support Vector Machine (SVM) classifier was trained using vector histogram features extracted from the Sentinel-2A spectral bands (B2, B3, and B4), and its hyperparameters were tuned via Bayesian optimization to achieve optimal performance. The model was trained and evaluated on 379 images from the Kurdistan Region, belonging to three vegetation health classes (moderately healthy, unhealthy, and dead).
Results: The proposed method achieved 90.79% test accuracy, outperforming Convolutional Neural Network (CNN), Random Forest (RF), and K-Nearest Neighbors (KNN) baselines while maintaining low computational complexity.
Conclusion: These findings highlight the effectiveness of compact spectral representations for accurate and efficient vegetation health classification, supporting scalable monitoring in resource-constrained semi-arid regions.
References
[1] S. Usman, J. O. Jayeoba, and A. Kundiri, “Climate change as a global concept: Impacts and adaptation measures,” International Journal of Environment and Cli-mate Change, vol. 14, no. 6, pp. 445–459, 2024. doi:10.9734/ijecc/2024/v14i64242. DOI: https://doi.org/10.9734/ijecc/2024/v14i64242
[2] A. Bagherzadeh, A. V. Hoseini, and L. H. Totmaj, “The effects of climate change on normalized difference vegeta-tion index (NDVI) in the northeast of Iran,” Modeling Earth Systems and Environment, vol. 6, no. 2, pp. 671–683, 2020. doi:10.1007/s40808-020-00724-x. DOI: https://doi.org/10.1007/s40808-020-00724-x
[3] C. Abel et al., “Vegetation resistance to increasing aridity when crossing thresholds depends on local environmental conditions in global drylands,” Communications Earth & Environment, vol. 5, no. 1, 2024. doi:10.1038/s43247-024-01546-w. DOI: https://doi.org/10.1038/s43247-024-01546-w
[4] V. Mazzia, A. Khaliq, and M. Chiaberge, “Improvement in land cover and crop classification based on temporal features learning from Sentinel-2 data using recurrent convolutional neural networks,” Applied Sciences, vol. 10, no. 1, 2020. doi:10.3390/app10010238. DOI: https://doi.org/10.3390/app10010238
[5] D. A. Rudd, M. Karami, and R. Fensholt, “Towards high-resolution land-cover classification of Greenland: A case study covering Kobbefjord, Disko and Zackenberg,” Re-mote Sensing, vol. 13, no. 18, 2021. doi:10.3390/rs13183559. DOI: https://doi.org/10.3390/rs13183559
[6] A. Netsianda, P. Mhangara, and E. Gidey, “Grassland biome fragmentation analysis using Sentinel-2 images and a support vector machine learning model in South Afri-ca,” Discover Sustainability, vol. 5, no. 1, 2024. doi:10.1007/s43621-024-00723-3. DOI: https://doi.org/10.1007/s43621-024-00723-3
[7] A. A. Abdullah, N. S. Mohammed, M. Khanzadi, S. M. Asaad, Z. K. Abdul, and H. S. Maghdid, “In-depth analy-sis of machine learning approaches: Techniques, applica-tions, and trends,” ARO – The Scientific Journal of Koya University, vol. 13, no. 1, pp. 190–202, 2025. doi:10.14500/aro.12038. DOI: https://doi.org/10.14500/aro.12038
[8] P. Mohammadpour, D. X. Viegas, and C. Viegas, “Vege-tation mapping with random forest using Sentinel-2 and GLCM texture features: A case study for the Lousã re-gion, Portugal,” Remote Sensing, vol. 14, no. 18, 2022. doi:10.3390/rs14184585. DOI: https://doi.org/10.3390/rs14184585
[9] L. Costanza et al., “Predicting olive tree chlorophyll fluo-rescence using explainable AI with Sentinel-2 imagery in Mediterranean environments,” Applied Sciences, vol. 15, no. 5, 2025. doi:10.3390/app15052746. DOI: https://doi.org/10.3390/app15052746
[10] A. E. Maxwell, T. A. Warner, and F. Fang, “Implementa-tion of machine-learning classification in remote sensing: An applied review,” International Journal of Remote Sens-ing, vol. 39, no. 9, pp. 2784–2817, 2018. doi:10.1080/01431161.2018.1433343. DOI: https://doi.org/10.1080/01431161.2018.1433343
[11] S. K. Abebaw Alem, “Machine learning approaches for land classification,” in Proceedings of the IEEE 8th Inter-national Conference on Reliability, Infocom Technologies and Optimization (ICRITO), 2020. doi:10.1109/ICRITO48877.2020.9197824. DOI: https://doi.org/10.1109/ICRITO48877.2020.9197824
[12] W. Utama, R. F. Indriani, M. Hermana, I. M. Anjasmara, S. A. Garini, and D. P. N. Putra, “Towards improving sus-tainable water management in geothermal fields: SVM and RF land use monitoring,” Journal of Human, Earth, and Future, vol. 5, no. 2, pp. 216–242, 2024. doi:10.28991/HEF-2024-05-02-06. DOI: https://doi.org/10.28991/HEF-2024-05-02-06
[13] R. K. Singh et al., “A machine learning-based classifica-tion of Landsat images to map land use and land cover of India,” Remote Sensing Applications: Society and Environ-ment, vol. 24, 2021. doi:10.1016/j.rsase.2021.100624. DOI: https://doi.org/10.1016/j.rsase.2021.100624
[14] M. Wakulińska and A. Marcinkowska-Ochtyra, “Multi-temporal Sentinel-2 data in classification of mountain vegetation,” Remote Sensing, vol. 12, no. 17, 2020. doi:10.3390/rs12172696. DOI: https://doi.org/10.3390/rs12172696
[15] S. Rina et al., “Application of machine learning to tree species classification using active and passive remote sensing: A case study of the Duraer forestry zone,” Re-mote Sensing, vol. 15, no. 10, 2023. doi:10.3390/rs15102596. DOI: https://doi.org/10.3390/rs15102596
[16] T. Lees, G. Tseng, C. Atzberger, S. Reece, and S. Dadson, “Deep learning for vegetation health forecasting: A case study in Kenya,” Remote Sensing, vol. 14, no. 3, 2022. doi:10.3390/rs14030698. DOI: https://doi.org/10.3390/rs14030698
[17] M. Zhang, D. Li, G. Li, and D. Lu, “Vegetation classifica-tion in a subtropical region with Sentinel-2 time series da-ta and deep learning,” Geo-Spatial Information Science, vol. 28, no. 1, pp. 145–163, 2025. doi:10.1080/10095020.2024.2336604. DOI: https://doi.org/10.1080/10095020.2024.2336604
[18] G. Aziz et al., “Remote sensing-based forest cover classi-fication using machine learning,” Scientific Reports, vol. 14, no. 1, 2024. doi:10.1038/s41598-023-50863-1. DOI: https://doi.org/10.1038/s41598-023-50863-1
[19] K. L. Du, B. Jiang, J. Lu, J. Hua, and M. N. S. Swamy, “Eploring kernel machines and support vector machines: Princples, techniques, and future directions,” Mathematics, vol. 12, no. 24, 2024. doi:10.3390/math12243935. DOI: https://doi.org/10.3390/math12243935
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Charmo Journal of Natural Sciences and Technologies
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
