Parameterization, Verification, and Validation of a Physiologically Complex Age-structured Rice Simulation Model

A model of rice (Oryza sativa L.) growth and development is presented which accurately simulates the seasonal numbers and mass of roots, culms, leaf sheaths, leaf blades, panicle vegetative components, and grains for the main tillers and subtillers. The physiological detail imbedded within the model departed to the greatest degree from previously published rice simulation models in its tiller cohort andplant part age structure, and its metabolite partitioning structure. Possibly the greatest distinction between RICEPSM and previously developed rice models is the inclusion of a statistically rigorous model parameterization, verijication. and validation process. The verijicationvalidation approach used in this paper has the advantage of enabling the individual and collectiveJit of multiple response variables. The verijied model explained 88% of the variability across 13 data types, with 88 and 93% of the variability explained for main tiller and subtiller grain yield, respectively. When validated against a second year of data, 87% of the variability across the 13 data types, and 92 and 95% of main tiller and subtiller grain yield, respectively, was explained. The paper contrasts the verification and validation approach presented in this paper with that used in other simulation-basedpape: s.