Modelling rice performance in Telangana: CERES-rice calibration and validation for post-monsoon cultivar adaptation

Rice, the staple food for more than 60 % of the global population, faces increasing production challenges due to climate variability and resource constraints. Accurate crop growth modelling and yield prediction are therefore essential to strengthen food security and optimise management. This study aimed to calibrate and validate the CERES-rice model within the decision support system for agrotechnology transfer (DSSAT v4.8.5.0) for three cultivars, RNR-15048, KNM-1638 and JGL-24423, across multiple sowing dates under Rabi conditions in Telangana, India. Field experiments were conducted during the rabi seasons of 2023 and 2024 using a split-plot design with five sowing windows and three cultivars. Site-specific weather, soil, crop management and genotype datasets were compiled and genetic coefficients were estimated through iterative calibration. Model performance was evaluated using root mean square error (RMSE), normalised RMSE (nRMSE) and the index of agreement (d). Calibration results showed excellent accuracy, with phenology simulated within ±1 day (nRMSE 1.2–3.7 %; d >0.83) and grain yield deviations <2.5 % (nRMSE 3.4–3.6 %; d >0.85). Biomass and straw yields were reproduced with good agreement (nRMSE <10.5 %). Validation confirmed robust model performance, with phenology and grain yield consistently accurate across cultivars and biomass predictions genotype-dependent but acceptable. Straw yield was more variable, particularly for KNM-1638 (nRMSE 11.2 %), though overall trends were reliable. The calibrated genetic coefficients were physiologically realistic and consistent with Indica rice, confirming the model’s suitability for yield forecasting, sowing window assessment and management optimisation under Rabi conditions. Overall, these findings highlight the potential of the CERES-rice model to support climate-smart rice production strategies in South Asia.