Introduction: 

Accurate simulation of rice growth and yield using process-based crop models depends on the proper calibration of genetic coefficients for specific cultivars and production systems. In Brazil, where rice is cultivated under both upland (rainfed) and lowland (irrigated) conditions, the lack of calibrated parameters for major national cultivars limits the application of crop models for management analyses and climate risk assessments. The objective of this study was to calibrate and evaluate the CSM-CERES-Rice model for two representative Brazilian rice cultivars grown under contrasting production systems: upland rice (BRS Primavera) and lowland rice (IRGA 424 RI).

Methods: 

Model calibration and evaluation were conducted using experimental data from field trials in central and southern Brazil, including observations of phenology, leaf area index, aboveground biomass, and grain yield. Model performance was assessed using independent evaluation treatments when available, leave-one-out cross-validation, and standard statistical metrics.

Results: 

The calibrated model accurately reproduced anthesis and physiological maturity in both systems and showed good agreement with observed canopy development, total aboveground biomass, and grain yield. Although some uncertainties were observed in biomass partitioning among vegetative organs, these did not affect the model’s ability to simulate whole-canopy growth dynamics or yield formation.

Discussion: 

In summary, the calibrated CSM-CERES-Rice model provided a robust representation of rice development and productivity under Brazilian conditions. The genetic coefficients generated in this study enhance the applicability of CSM-CERES-Rice for crop management analyses, climate risk assessments, and scenario-based simulations involving upland and lowland rice systems in Brazil.