A Simulation Study On Rice Productivity Under Future Climate Change

Rice (Oryza sativa L.) is one of the staple food in China. More than 65% of the population are feed on rice, Therefore, stable growth of rice productivity plays a key role in ensurance of national food security. So far, climate change is one of the most important and severe global environmental problems facing mankind, and has attracted the attention of governments and people all over the world. Crop growth model provides new methods and tools to simulate the impacts of climate change on rice growth and development. This study combined the method and rice field experiments of Agricultural Model Intercomparison and Improvement Project (AgMIP) rice model team with existing experimental data to improve, calibrate and validate the biomass partitioning and CO2 impact in rice growth model RiceGrow. Then, a sensitivity analysis on the effects of temperature, CO2 and nitrogen levels on rice phenology and yield was studied by three different models (ORYZA2000, RiceGrow and CERES-Rice) at major rice producing areas in China including representative varieties of 8 single season and 4 double season rice sites. Finally, we analyzed the trends of climate variables during growing season of 30 years’historical weather data in 12 typical rice production sites, then the effects of future climate on rice phenology and yield of these 12 sites were predicted by the three models. The results would support the growth and development, management practices and production decisions of rice under future climatic conditions.Based on the original model-based framework, The submodel of aboveground biomass partitioning and processes related to ambient air CO2 concentration were improved by several experiments involved with different sites, years, and species, the above ground biomass partitioning improved model parameters used partitioning index based on physiological development time (PDT) and algorithms of CO2 concentration used the factor method that can balance rationality and practicality. Then a set of independent experiment data was used for calibration and validation. The results indicated that the stability and regularity of partition index based on physiological development time (PDT) and algorithms of CO2 concentration were relatively good.Based on the regionalization of rice cropping in China and the dataset of rice planting site, we selected 12 representative sites (single season:Xinghua, Hefei, Zhongxiang, Hanyuan Xinyang, Xuzhou, Wuchang and Guangyuan; double season:Gaoyao, Nanchang, Hengyang and Wugang) in the main rice producing provinces of China, and we selected a representative variety for each site. Then we used the three models to predict and compare rice growth, development and yield based on these selected varieties, and did sensitivity analysis on temperature, CO2 and nitrogen levels of the three models. After calibration and validation, simulation results of three models on phenology are relatively good, the RMSE of ORYZA2000, RiceGrow and CERES-Rice were between 1.94-3.05 days, with an average of 2.66 days. NRMSE of three models were between 1.48-2.33%. The predicted yield among three models of "blind simulation" ranged from 3070.8-15036.6 kg ha-1. After calibration and validation, the results showed a marked improvement, the range reduced to 4028.3-10736.0 kg ha-1. Except minority sites such as the 2003 in Hefei of ORYZA2000 and RiceGrow and the 2008 in Hanyuan of CERES-Rice, most of the results after calibration and validation were within-10%-10% error line, and some were within-20%-20%, which indicated a good fit of simulated and measured values.The phenology predicted by the two models (CERES-Rice, ORYZA2000) were not sensitive to the changes in CO2 concentration except for RiceGrow. In generally, rice growth duration shortened as temperature increased from baseline temperature for all three models. Overall, predicted yields decreased with increasing temperature and increased with increasing CO2 and nitrogen application rate. And there was no significant difference of sensitivity to CO2 among three models. Predicted average yields varied a lot among sites, yield of all sites decreased with increasing temperature. What’s more, predicted average yield increased as the latitude decreased at most sites. And from the results of CO2 concentration sensitivity analysis, elevation of CO2 concentration could partially compensate for the negative effect of temperature increased. The more the temperature increased, the more yield decreased.Compared of future climate and historical climate, except for only a few sites, the average temperature of the rest sites showed an increasing trend, the average total sunshine hours and total rainfall also showed significant increase. The simulated growth duration under future climate scenarios spanned a lot, however, they all showed a shorter trend comparing to baseline climate (1980-2010). The predicted average yield of single season rice declined, and the yield declining of 80s (2070-2100) was greater than 50s (2040-2070), the elevated CO2 concentration partially compensated for the negative effects caused by temperature increasing. The average predicted yield of double season rice in 50s showed an increasing trend. Except for early rice at Gaoyao, early rice at Nanchang and early rice at Hengyang decreased, the rest all experienced an increasing trend in 80s.