Variability of Parameters of ORYZA (v3) for Rice under Different Water and Nitrogen Treatments and the Cross Treatments Validation

Abstract

To reveal the variability in ORYZA (v3) model parameters among different water and nitrogen treatments, and illustrate the model performance and uncertainty in rice biomass simulation with different treatment specific calibrated parameters, ORYZA (v3) model were calibrated and validated treatment specifically based on data from four different water and nitrogen treatments. Generally, the treatment specific calibrated ORYZA (v3) model is accurate in modelling rice biomass accumulation in the exact specific treatment, and it performed a little better under potential condition than water or nitrogen limited condition. Variability of treatment specific calibrated parameters for ORYZA (v3) model was low, with the maximum CV of 6.43% for biomass partitioning factor to leaf. The cross treatment validation indicated the errors in simulated total above ground biomass (WAGT) and panicles biomass (WSO) for a specific treatment were enlarged in a certain degree, when it was simulated by parameters calibrated from others. For WAGT simulation, the FF (flooding irrigation + farmers' fertilization practice) treatment calibrated parameters performed slightly better than others. For WSO simulation, all the four group calibrated parameters underestimated it in the early panicle growth phrase under water or nitrogen limited conditions, and FF and NFS (non-flooding controlled irrigation + site specific nitrogen management) treatment parameters performed slightly better than FS (flooding irrigation + site specific nitrogen management) treatment and NFF (non-flooding controlled irrigation + farmers' fertilization practice treatment). There would be a certain degree of uncertainty when rice biomass production simulated by ORYZA (v3) model with different treatment specific calibrated parameters. The range and standard deviation of simulated WAGT or WSO is low at beginning and increased along with crop development, with the values in FF and FS treatment are generally lower than that in NFF and NFS.

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

“…The ORYZA_V3 model can accurately describe the N transformation processes and the effects of N fertilizer on paddy rice growth (Bouman et al., 2001; Li et al., 2017) and has a strong capability to simulate paddy rice growth (Hameed et al., 2019; Li et al., 2017; Xu et al., 2018; Yuan et al., 2017). In the present study, this crop model produced accurate predictions of the yield, AGB, FNLVE, and FNLVM under different input N fertilizer rates.…”

“…As the latest in the ORYZA series (Li et al., 2017), the ORYZA_V3 model is considered the most comprehensive representation of available knowledge on paddy rice growth (IRRI, 2017). It has been widely used to simulate paddy rice growth under different agricultural systems and in various rice‐growing areas (Hameed et al., 2019; Li et al., 2017; Radanielson et al., 2018; Wang et al., 2018; Xu et al., 2018; Yuan et al., 2017). Crop model predictions are inherently uncertain due to the uncertainty in the model structure, parameters, and input data (Rotter et al., 2012; Tao et al., 2018).…”

“…For the ORYZA_V3 model, Xu et al. (2018) found that a low variability in the treatment‐specific calibrated parameters could result in uncertainty in rice biomass predictions under the same water–N field condition. In ORYZA_V3, the growth of biomass and yield are dependent upon the in‐season growth variables, such as the leaf area index (LAI) and leaf N concentration (FNLV) (Bouman et al., 2001; Li et al., 2017).…”

“…Parametric uncertainty can lead to high levels of variability in crop growth predictions (Dzotsi et al, 2013(Dzotsi et al, , 2015Gao et al, 2020;Iizumi et al, 2009;Sexton et al, 2016). For the ORYZA_V3 model, Xu et al (2018) found that a low variability in the treatment-specific calibrated parameters could result in uncer-…”

ORYZA (v3) rice crop growth modeling for MR269 under nitrogen treatments: Assessment of cross-validation on parameter variability

“…Extensive evaluation of rice crop models such as CERES-Rice, ORYZA2000, and ORYZA (v3) for rice varieties have been conducted in China (Yuan et al 2017, Xu et al 2018, Lu et al 2020, India (Halder et al 2020, Jha et al 2020, Iran (Amiri et al 2014, Tari et al 2017, and the Philippines (Li et al 2016, Radanielson et al 2018. Fewer evaluations of rice crop models have been reported for Thailand (Wikarmpapraharn and Kositsakulchai 2010, Babel et al 2011, Boling et al 2011, Indonesia (Agustiani et al 2018), and Vietnam (Tan Yen et al 2019).…”