Assessing parametric and nitrogen fertilizer input uncertainties in the ORYZA_V3 model predictions

The ORYZA_V3 model is widely used for simulating paddy rice (Oryza sativa L.) growth. However, few studies have assessed the uncertainties in the ORYZA_V3 model predictions. This study examined the effects of uncertainties in crop parameters and input nitrogen (N) fertilizer rates on four growth variable predictions using the Markov Chain Monte Carlo and Monte Carlo methods. The results showed: (a) This model generated accurate predictions of the yield and aboveground biomass (AGB) at harvesting and leaf N concentration (FNLV) at the early (FNLVE) and middle (FNLVM) development stages but poor predictions of the FNLV at the late development stage (FNLVL) and leaf area index (LAI) at tillering, stem elongation, and panicle initiation development stages. (b) Parametric uncertainty had a significant effect on FNLVE and FNLVM predictions but had a negligible effect on LAI predictions at the three development stages. (c) Different input N fertilizer rates produced different degrees of parametric uncertainty in yield, AGB, FNLVM, and FNLVL predictions. (d) Incorporating the uncertainty in the input N rates increased the predictive variance of the variables. The average magnitude of increase was 69.45% for yield; 9.73% for AGB; 0.43% for LAI; and 0, 101.71, and 146.91% for FNLVE, FNLVM, and FNLVL, respectively. However, the mean values of the prediction distributions varied slightly after incorporating the uncertain input N rates. This study demonstrated the importance of the input N rates for simulating the yield, AGB, FNLVM, and FNLVL. The ORYZA_V3 model can be used to determine the impacts of N on rice growth.