Rice Yield Estimation Using Vegetation Indexes (NDVI and EVI) Derived from Sentinel-2 Imagery for Sustainable Agriculture

Organic farming has emerged as an alternative agricultural production practice, accompanied by a growing market for organic products. Advanced remote sensing technology, exemplified by Sentinel-2 imagery as an optical satellite, can efficiently determine the rice planting season and project total rice production. It also enables the monitoring of plant health, irrigation, and fertiliser management, and environmental quality in organic cultivation areas. This study explored two vegetation indices, namely NDVI (Normalized Difference Vegetation Index) and EVI (Enhanced Vegetation Index). The study was conducted on 30 agricultural parcels in Candipuro District, Lumajang Regency, East Java, Indonesia. NDVI and EVI values were derived from Sentinel-2 imagery corresponding to each harvesting date; therefore, two harvesting periods were used in this study. Statistical indices—including the Spearman-rho test, correlation analysis, linear regression, coefficient of determination, and RMSE (Root Mean Square Error)—were used to compare rice yield estimates from satellite imagery with actual measurement results. The findings conclude that the EVI index provided the most accurate prediction, with R² ≈ 0.7 and RMSE = 0.03 for the second period. Both NDVI and EVI demonstrated promising outcomes for pre-harvest rice yield estimation, with EVI showing slightly superior performance, although data support remains limited.