Assessment of Rice Yield Forecasting Accuracy Using MODIS LAI Assimilation in DSSAT in Ayodhya District, India

Abstract

Accurate estimation of rice yield is essential for food security and effective agricultural management under changing climatic conditions. This study evaluated the performance of the CERES-Rice crop simulation model integrated with MODIS-LAI data for regional-scale rice-yield prediction and verification during 2023–24 and 2024–25. Three rice varieties, namely Sarjoo-52, NDR-359, and NDR-370133, were evaluated across the different blocks of the study area. Among them, NDR-370133 consistently recorded the highest productivity, with predicted yields ranging from 5.08 to 6.10 t ha⁻¹ during 2023–24 and from 4.91 to 5.96 t ha⁻¹ during 2024–25. The highest average productivity was observed in Milkipur block, whereas comparatively lower productivity was recorded in Tarun and Amaniganj blocks. Validation indicated varying agreement between predicted and observed yields, with deviations ranging from −7.94% to 24.91% during 2023–24 and from −6.29% to 20.83% during 2024–25. Lower deviations in several blocks indicated better model performance in those locations. The integrated MODIS-LAI and CERES-Rice framework represented crop-growth dynamics and regional yield variability. The findings indicate the potential of integrating remote sensing data with crop simulation models for large-scale crop monitoring, yield forecasting, and agricultural decision support.