Improving Simulations of Rice in Response to Temperature and CO₂

Abstract

Crop models are frequently used to assess the impact of climate change responses. Evaluation of model performance against empirical data is crucial to establish confidence, particularly for rice (Oryza sativa L.), one of the world’s important cereal crops. Data from soil-plant-atmosphere-research (SPAR) chambers and field plots were used to assess three versions of the ORYZA model to a range of climate conditions. The three versions were: V1–the original, V2–V1 plus a revised heat stress component, and V3–V2 plus a coupled leaf-level gas exchange algorithm. Comparison against SPAR datasets, which covered a range of temperatures at two CO2 levels, indicated successive improvement in yield predictions with the model version. Root Mean Square Error (RMSE) decreased by 520 and 647 kg ha−1 for V2 and V3, respectively, and Wilmott’s index of agreement improved by 10 and 12% compared with V1 when averaged across 20 treatments and three cultivars. Similar improvements were observed from 17 field dataset simulations with two additional varieties. These results indicated the importance of improving heat sterility functions and carbon assimilation methodologies that incorporate direct responses to air temperature and CO2 concentration in rice models. Accounting for cultivar differences in thermal sensitivity is also an important consideration for climate assessments.

Keywords

• crop model
• photosynthesis
• high temperature
• CO₂
• spikelet fertility
• heat stress