Integrating Radar and Multispectral Data from Sentinel Satellites for Accurate Urban Vegetation Classification using Deep Learning Methods
https://doi.org/10.32603/1993-8985-2026-29-3-112-120
Аннотация
Introduction. The Sentinel-1 (equipped with synthetic aperture radar) and Sentinel-2 (equipped with multispectral cameras) satellites are valuable tools for environmental monitoring, particularly for assessing vegetation cover and soil erosion. The data obtained by these systems can be processed using machine learning techniques to generate accurate vegetation classification maps.
Aim. To develop and evaluate machine learning models capable of efficiently fusing Sentinel satellite data to produce more accurate and detailed maps of urban vegetation, essential for urban planning, environmental monitoring, and climate change mitigation. The integration of Sentinel-1 and Sentinel-2 satellite data is intended to overcome the limitations associated with using each data type in isolation, particularly in complex urban environments where spectral signatures can be ambiguous and radar provides only structural information.
Materials and methods. A classification map of urban vegetation on Kotlin Island (St Petersburg) was generated by integrating data from Sentinel-1 (synthetic aperture radar) and Sentinel-2 (multispectral imagery) using the Random Forest algorithm, Tensorflow and Sklearn Python libraries. Conventional urban vegetation mapping often relies on a single data source, leading to limited accuracy and inability to differentiate subtle vegetation types. The vegetation classes considered in this research were coniferous forest, deciduous forest, wetland vegetation, and coastal meadows.
Results. The integrated analysis of radar and multispectral data enabled more accurate identification of erosion-prone zones in non-vegetated areas and more reliable estimation of plant moisture content. Such a fusion approach showed significantly improved classification accuracy and reduced error rate compared to techniques relying on individual indices.
Conclusion. The fusion method demonstrated superior performance in classifying vegetation types, which confirms its potential for applications in remote sensing and environmental monitoring. Future research will focus on integrating radar and multispectral data from Sentinel satellites for urban vegetation classification using the Support Vector Machine algorithm.
Об авторе
J. B. Ngoua Ndong Avele
Omar Bongo University ; Saint Petersburg Electrotechnical University
Габон
Габон
Jacques B. Ngoua Ndong Avele, Bachelor in Omar Bongo University, Libreville, Gabon, Master's degree in Quantum and Optical Electronics (2024, Saint Petersburg Electrotechnical University). Postgraduate student of the Department of Radio Engineering Systems in Radio Navigation and Radar
5 F, Professor Popov St., St Petersburg 197022
Список литературы
1. Mohammadpour P., Viergas D. X., Viegas C. Vegetation Mapping with Random Forest Using Sentinel 2 and GLCM Texture Feature–A Case Study for Lousa Region, Portugal. Remote Sensing. 2022, vol. 14, iss. 18, art. no. 4585. doi: 10.3390/rs14184585
2. Niculescu S., Billey A., Ou Ali H. T. Random Forest Classification using Sentinel-1 and Sentinel-2 series for vegetation monitoring in the Pays de Brest (France). Proc. of SPIE Remote Sensing, Berlin, Germany, 9 Oct. 2018. Vol. 10783. doi: 10.1117/12.2325546
3. Kovarnik R., Janova J. Validation of Sentinel 2 Based Machine Learning Models for Czech National Forest Inventory. Ecological Inormatics. 2025, vol. 87, art. no. 103133. doi: 10.1016/j.ecoinf.2025.103133
4. Zhang X., Shen H., Huang T., Wu Y., Guo B., Liu Zh., Luo H., Tang J., Zhou H., Wang L., Xu W., Ou G. Improved Random Forest Algorithms for Increasing the Accuracy of Forest Aboveground Biomass Estimation Using Sentinel-2 Imagery. Ecological Indicators. 2024, vol. 159, art. no. 111752. doi: 10.1016/j.ecolind.2024.111752
5. Wang T. Improved Random Forest Classification Model Combined with C5.0 Algorithm for Vegetation Feature Analysis in Non-Agricultural Environments. Scientific Reports. 2024, vol. 14, art. no. 10367. doi: 10.1038/s41598-024-60066-x
6. Majikumna K. U., Zineddine M., Mustapha M., Ahmed E. H. A. Random Forest and Sentinel-2 Based Approach for Evaluating Climate Change on Olive Trees. 4th Intern. Conf. on Innovative Research in Applied Science, Engineering and Technology, Fes, Morocco, 16–17 May 2024. IEEE, 2024, pp. 1–5. doi: 10.1109/IRASET60544.2024.10548349
7. Rynkiewicz A., Hoscito A., Aune-Lundberg L., Nilsen A. B., Lewandowska A. Detection and Quantification of Vegetation Losses with Sentinel-2 Images Using BiTemporal Analysis of Spectral Indices and Transferable Random Forest Model. Remote Sensing. 2025, vol. 17, iss. 6, art. no. 979. doi: 10.3390/rs1706097
8. Kallel A., Mura M. D, Fakhfakh S., Romdhane N. B. Physics-Based Fusion of Sentinel-2 and Sentinel-3 for Higher Resolution Vegetation Monitoring. IEEE Transactions on Geosciences and Remote Sensing. 2023, vol. 61, art. no. 5403317. doi: 10.1109/TGRS.2023.3257219
9. Roßberg T., Schmitt M. Dense NDVI Time Series by Fusion of Optical and SAR-Derived Data. IEEE J. of Selected Topics in Applied Earth Observations and Remote Sensing. 2024, vol. 17, pp. 7748–7758. doi: 10.1109/JSTARS.2024.3379838
10. Liang Y., Li Y., Li J., Li Y., Li F., Zhang Sh. Fusion of Optical and SAR Imags Via NDVI-Like Images to Reconstruct NDVI Time Series in Cloud-Prone Regions. IEEE International Geoscience and Remote Sensing Symp., Athens, Greece, 07–12 July 2024. IEEE, 2024, pp. 9307–9310. doi: 10.1109/IGARSS53475.2024.10641085
11. Vuolo F., Żółtak M., Pipitone C., Zappa L., Wenng H., Immitzer M., Weiss M., Baret F., Atzberger C. Data Service Platform for Sentinel-2 Surface Reflectance and Value-Added Products: System Use and Examples. Remote Sensing. 2016, vol. 8, iss. 11, art. no. 938. doi: 10.3390/rs8110938
12. Gašparović M., Medak D., Pilaš I., Jurjević L., Balenović I. Fusion of Sentinel-2 and Planetscope Imagery for Vegetation Detection and Monitoring. The Intern. Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. 2018, vol. XLII-1, pp. 155–160. doi: 10.5194/isprs-archives-XLII-1-155-2018
13. Cardona-Mesa A. A, Vasquez-Salazar R. D., Gomez L., Travieso-González C. M., Garavito-González A. F., Vásquez-Cano E., Díaz-Paz J. P. Dataset of Sentinel-1 SAR and Sentinel-2 RGB-NDVI Imagery. 2024, vol. 57, art. no. 111160. doi: 10.1016/j.dib.2024.111160
14. El Moussaoui E. H., Moumni A., Lahrouni A. Assessing the Influence of Different Synthetic Aperture Radar Parameters and Digital Elevation Model Layers Combined with Optical Data on the Identification of Argan Forest in Essaouira Region, Morocco. iForest - Biogeosciences and Forestry. 2024, vol. 17, iss. 2, pp. 100–108. doi: 10.3832/ifor4183-016
15. Jia P., Mohd Shafri H. Z., Yu S., Zheng Z., You Sh., Shariff A. R. M. Radar-Optical Fusion of Sentinel-1/2 for High-Resolution NDVI Reconstruction and LandscapeDriven Carbon Flux Assessment in Kuala Selangor, Malaysia (2020–2024). Intern. J. of Applied Earth Observation and Geoinformation. 2025, vol. 145, art. no. 104966. doi: 10.1016/j.jag.2025.104966
Рецензия
Для цитирования:
Ngoua Ndong Avele J. Integrating Radar and Multispectral Data from Sentinel Satellites for Accurate Urban Vegetation Classification using Deep Learning Methods. Известия высших учебных заведений России. Радиоэлектроника. 2026;29(3):112-120. https://doi.org/10.32603/1993-8985-2026-29-3-112-120
For citation:
Ngoua Ndong Avele J. Integrating Radar and Multispectral Data from Sentinel Satellites for Accurate Urban Vegetation Classification using Deep Learning Methods. Journal of the Russian Universities. Radioelectronics. 2026;29(3):112-120. https://doi.org/10.32603/1993-8985-2026-29-3-112-120
Просмотров:
131
JATS XML
Послать статью по эл. почте 