IDENTIFIKASI STRES KEKERINGAN PADA TANAMAN PADI BERBASIS ANALISIS SPEKTRAL CITRA MULTISPEKTRAL MENGGUNAKAN U-NET SEGMENTATION
DOI:
https://doi.org/10.54314/jssr.v8i3.4190Abstract
Abstract: Global climate change has increased the frequency and intensity of droughts that threaten national food security, particularly in rice agriculture sector. This research develops a system for identifying drought stress in rice plants using spectral analysis of multispectral images based on U-Net segmentation. The research methodology includes collecting multispectral images from rice fields, radiometric preprocessing and atmospheric correction, vegetation area segmentation using U-Net architecture, spectral feature extraction through vegetation index calculation (NDVI, NDWI, SAVI), and drought stress level classification. The U-Net architecture is trained to separate rice plant areas from the background with high accuracy, then analyzes the spectral response of plants to water deficiency. Research results show that the system can identify four levels of drought stress (normal, mild, moderate, severe) with an accuracy rate of 87.5%. The output is a spatial map of drought stress distribution that can be used as a basis for decision-making in irrigation management. This system provides significant contributions to precision agriculture and climate change adaptation in supporting sustainable food security.
Â
Keyword: Drought Stress, Rice Plants, Spectral Analysis, Multispectral Imagery, U-Net Segmentation
Â
Abstrak: Perubahan iklim global telah meningkatkan frekuensi dan intensitas kekeringan yang mengancam ketahanan pangan nasional, khususnya pada sektor pertanian padi. Penelitian ini mengembangkan sistem identifikasi stres kekeringan pada tanaman padi menggunakan analisis spektral citra multispektral berbasis segmentasi U-Net. Metodologi penelitian meliputi pengumpulan citra multispektral dari ladang padi, preprocessing radiometrik dan koreksi atmosferik, segmentasi area vegetasi menggunakan arsitektur U-Net, ekstraksi fitur spektral melalui perhitungan indeks vegetasi (NDVI, NDWI, SAVI), serta klasifikasi tingkat stres kekeringan. Arsitektur U-Net dilatih untuk memisahkan area tanaman padi dari latar belakang dengan akurasi tinggi, kemudian menganalisis respon spektral tanaman terhadap kekurangan air. Hasil penelitian menunjukkan sistem mampu mengidentifikasi empat tingkat stres kekeringan (normal, ringan, sedang, berat) dengan tingkat akurasi 87,5%. Output berupa peta spasial distribusi stres kekeringan yang dapat digunakan sebagai dasar pengambilan keputusan dalam manajemen irigasi. Sistem ini memberikan kontribusi signifikan terhadap pertanian presisi dan adaptasi terhadap perubahan iklim dalam mendukung ketahanan pangan berkelanjutan.
Â
Kata kunci: Stres Kekeringan, Tanaman Padi, Analisis Spektral, Citra Multispektral, U-Net Segmentation
Downloads
References
Banjade, T. P., Zhou, C., Chen, H., Li, H., Deng, J., Zhou, F., & Adhikari, R. (2024). Seismic Random Noise Attenuation Using DARE U-Net. Remote Sensing, 16(21). doi: 10.3390/rs16214051
Fu, B., He, X., Yao, H., Liang, Y., Deng, T., He, H., Fan, D., Lan, G., & He, W. (2022). Comparison of RFE-DL and stacking ensemble learning algorithms for classifying mangrove species on UAV multispectral images. International Journal of Applied Earth Observation and Geoinformation, 112. doi: 10.1016/j.jag.2022.102890
Geng, H., Zhao, H., Shi, Z., Wu, F., Geng, L., & Ma, K. (2024a). MBFE-UNet: A Multi-Branch Feature Extraction UNet with Temporal Cross Attention for Radar Echo Extrapolation. Remote Sensing, 16(21). doi: 10.3390/rs16213956
Geng, H., Zhao, H., Shi, Z., Wu, F., Geng, L., & Ma, K. (2024b). MBFE-UNet: A Multi-Branch Feature Extraction UNet with Temporal Cross Attention for Radar Echo Extrapolation. Remote Sensing, 16(21). doi: 10.3390/rs16213956
Guo, M., Liu, H., Xu, Y., & Huang, Y. (2020). Building extraction based on U-net with an attention block and multiple losses. Remote Sensing, 12(9). doi: 10.3390/RS12091400
Konatowska, M., MÅ‚ynarczyk, A., Rutkowski, P., & Kujawa, K. (2024). Impact of Site Conditions on Quercus robur and Quercus petraea Growth and Distribution Under Global Climate Change. Remote Sensing, 16(21).
Wang, L., Peng, L., Gui, R., Hong, H., & Zhu, S. (2024). Unsupervised PolSAR Image Classification Based on Superpixel Pseudo-Labels and a Similarity-Matching Network. Remote Sensing, 16(21). doi: 10.3390/rs16214119
Win, K. S., & Sasaki, J. (2024). The Change Detection of Mangrove Forests Using Deep Learning with Medium-Resolution Satellite Imagery: A Case Study of Wunbaik Mangrove Forest in Myanmar. Remote Sensing, 16(21). doi: 10.3390/rs16214077
Zhang, K., Li, P., & Wang, J. (2024). A Review of Deep Learning-Based Remote Sensing Image Caption: Methods, Models, Comparisons and Future Directions. In Remote Sensing (Vol. 16, Issue 21). Multidisciplinary Digital Publishing Institute (MDPI). doi: 10.3390/rs16214113
Zhao, H., Wang, J., Guo, J., Hui, X., Wang, Y., Cai, D., & Yan, H. (2024). Detecting Water Stress in Winter Wheat Based on Multifeature Fusion from UAV Remote Sensing and Stacking Ensemble Learning Method. Remote Sensing, 16(21). doi: 10.3390/rs16214100
