<![CDATA[Sugarcane is a vital commodity in the national sugar industry, requiring accurate growth monitoring to support precision agriculture. In Indonesia, conventional monitoring methods remain limited in spatial and temporal coverage. This study aims to monitor sugarcane phenology using multitemporal Sentinel-1 Synthetic Aperture Radar and to develop a machine learning–based growth phase classification model. The study was conducted in Central Lampung during one complete growing season from November 2023 to December 2024. Time-series analysis of VV and VH backscatter coefficients and the Normalised Polarisation Ratio was applied to capture sugarcane growth dynamics. Growth phase classification models were developed using Random Forest (RF) and Support Vector Machine (SVM) and evaluated using the confusion matrix, overall accuracy, Kappa coefficient, coefficient of determination, and root mean square error. The results indicate that RF consistently outperformed SVM across all model configurations. The best performance was achieved using combined VV–VH polarization, yielding an R² of 0.929, RMSE of 0.295, and classification accuracy of 93%. In contrast, SVM models showed weak predictive performance with negative R² values and classification accuracy below 32%. These findings demonstrate that multitemporal Sentinel-1 SAR data combined with RF provide an effective approach for spatial and temporal monitoring of sugarcane phenology]]>
Copyrights © 2026
