RADIOELECTRONIC AND COMPUTER SYSTEMS

This study focuses on developing a unified, reproducible cloud-to-model pipeline for parcel-scale cropland delineation from multispectral Sentinel-2 and Landsat imagery in the Kur–Araz region. The goal of this study is to produce accurate, boundary-faithful, and computationally efficient farmland maps that remain transparent, scalable, and deployable on commodity hardware for use by government agencies, water authorities, and agricultural producers. The tasks to be addressed include the following: specification of the study scope, data sources, and evaluation protocol that integrates pixel-wise and boundary-sensitive accuracy metrics; construction of a multi-index feature stack in addition to surface reflectance bands, followed by screening of features for cross-seasonal stability; design and training of a memory-efficient U-Net architecture with a hybrid loss that simultaneously balances calibration and overlap; and validation of model generalization across different cropping seasons and neighboring subregions of Kur–Araz, with complete provenance tracked from preprocessing through evaluation. The methods used in the study include cloud-native preprocessing in Google Earth Engine, consisting of cloud masking, seasonal compositing, medoid and percentile mosaicking, and stratified patch sampling with spatial blocking. The datasets were exported to TFRecords and used to train a compact U-Net encoder–decoder network with skip connections and a hybrid objective function. Training involved strong normalization, data augmentation to account for agricultural scene variability, overlap-tiling for inference, and three ablation studies to isolate the contribution of individual vegetation indices. Complete logging of assets, random seeds, and parameters supported exact reruns of the pipeline, replication, and auditing. Conclusions. The scientific novelty of the results obtained is as follows: (1) a unified, provenance-complete pipeline was developed that directly integrates Google Earth Engine preprocessing with deep learning model training and boundary-aware evaluation, ensuring full reproducibility and auditability; (2) a resource-efficient U-Net architecture with skip connections and a hybrid overlap-weighted loss was designed and validated, which preserves parcel borders while maintaining calibration, outperforming heavier transformer-based backbones under limited compute budgets; (3) the complementarity of multi-index feature stacks was systematically quantified through ablation studies, demonstrating their role in mitigating spectral saturation and soil brightness effects and improving overlap fidelity. Finally, a standardized evaluation protocol was proposed that combines pixel-wise and boundary-sensitive metrics, allowing robust assessment of farmland delineation performance across seasons and spatial subregions.

Google Earth Engine; multispectral segmentation; U-Net; spectral indices; farmland boundary delineation; Kur–Araz; reproducible pipeline

Persello, C., Wegner, J. D., Hänsch, R., Tuia, D., Ghamisi, P., Koeva, M., & Camps-Valls, G. Deep learning and earth observation to support the sustainable development goals: Current approaches, open challenges, and future opportunities. IEEE Geoscience and Remote Sensing Magazine, 2022, vol. 10, no. 2, pp. 172-200. DOI: 10.1109/MGRS.2021.3136100.

Gorelick, N., Hancher, M., Dixon, M., Ilyushchenko, S., Thau, D., & Moore, R. Google Earth Engine: planetary-scale geospatial analysis for everyone. Remote Sensing of Environment, 2017, vol. 202, pp. 18-27. DOI: 10.1016/j.rse.2017.06.031.

Kumar, L., & Mutanga, O. Google Earth Engine applications since inception: usage, trends, and potential. Remote Sensing, 2018, vol. 10, no. 10, article no. 1509. DOI: 10.3390/rs10101509.

Rouse, J. W., Haas, R. H., Schell, J. A., & Deering, D. W. Monitoring vegetation systems in the Great Plains with ERTS. Proceedings of the Third Earth Resources Technology Satellite-1 Symposium, 1974, vol. 1, pp. 309-317. Available at: https://ntrs.nasa.gov/citations/19740022614 (accessed 25.09.2025).

Huete, A. R. A soil-adjusted vegetation index (SAVI). Remote Sensing of Environment, 1988, vol. 25, no. 3, pp. 295-309. DOI: 10.1016/0034-4257(88)90106-X.

Wang, J., Chen, T., Zheng, L., Tie, J., Zhang, Y., Chen, P., Luo, Z., & Song, Q. A multi-scale remote sensing semantic segmentation model with boundary enhancement based on UNetFormer. Scientific Reports, 2025, vol. 15, article no. 14737. DOI: 10.1038/s41598-025-99663-9.

Li, Y., Guo, R., Li, R., Ji, R., Wu, M., Chen, D., Han, C., Han, R., Liu, Y., Ruan, Y., & Yang, J. An improved U-Net and attention mechanism-based model for sugar beet and weed segmentation. Frontiers in Plant Science, 2025, vol. 15, article no. 1449514. DOI: 10.3389/fpls.2024.1449514.

Jonnala, N. S., Bheemana, R. C., Prakash, K., Bansal, S., Jain, A., Pandey, V., Faruque, M. R. I., & Al-Mugren, K. S. DSIA U-Net: deep shallow interaction with attention mechanism U-Net for remote sensing satellite images. Scientific Reports, 2025, vol. 15, article no. 549. DOI: 10.1038/s41598-024-84134-4.

Wang, J., Wang, Y., Li, G., & Qi, Z. Integration of remote sensing and machine learning for precision agriculture: a comprehensive perspective on applications. Agronomy, 2024, vol. 14, no. 9, article no. 1975. DOI: 10.3390/agronomy14091975.

Ramos, L. T., & Sappa, A. D. Leveraging U-Net and selective feature extraction for land cover classification using remote sensing imagery. Scientific Reports, 2025, vol. 15, article no. 84795. DOI: 10.1038/s41598-024-84795-1.

Gayibov, A. Development of a zero-shot classification method for cross-regional crop mapping demonstrating domain transferability in Sentinel-2 imagery. Eastern-European Journal of Enterprise Technologies, 2025, vol. 4, no. 2 (136), pp. 93-101. DOI: 10.15587/1729-4061.2025.338000.

Howard, A., Sandler, M., Chen, B., Wang, W., Chen, L.-C., Tan, M., Chu, G., Vasudevan, V., Zhu, Y., Pang, R., Adam, H., & Le, Q. Searching for MobileNetV3. Proceedings of the IEEE/CVF International Conference on Computer Vision, 2019, pp. 1314-1324. DOI: 10.1109/ICCV.2019.00140.

Claverie, M., Ju, J., Masek, J. G., Dungan, J. L., Vermote, E. F., Roger, J.-C., Skakun, S. V., & Justice, C. The Harmonized Landsat and Sentinel-2 surface reflectance data set. Remote Sensing of Environment, 2018, vol. 219, pp. 145–161. DOI: 10.1016/j.rse.2018.09.002.

Huete, A. R., Didan, K., Miura, T., Rodriguez, E. P., Gao, X., & Ferreira, L. G. Overview of the radiometric and biophysical performance of the MODIS vegetation indices. Remote Sensing of Environment, 2002, vol. 83, no. 1-2, pp. 195-213. DOI: 10.1016/S0034-4257(02)00096-2.