10:45 AM - 12:15 PM
[ACG39-P06] Development of MATCRO-Maize for global estimation of maize production
Keywords:climate change, process-based model, maize
Climate change such as the rising temperature or changing precipitation patterns has been proceeding, and it would give various impacts on crop growth and yield. For example, global maize yield declined by 3.1 % from 1980 to 2008 because of the increase in global mean temperature (Lobell et al., 2011). There is a concern about the further decrease in crop yield due to future climate change. On the other hand, the increase in the global population has been increasing food demand. In 2050, global food demand for crops is predicted to reach around 3.6 billion, which is 1.7 times higher than in 2010 (Ministry of Agriculture, Forestry and Fisheries in Japan,2019). With these two reasons, projecting future yield is important for understanding future imbalance between food supply and demand.
There are two common approaches for predicting crop yield: statistical model and process-based model. Statistical models generally don’t consider physiological mechanisms to represent crop growth, therefore it is difficult to understand the details of relationships between climate change and crop growth. On the other hand, most of the process-based models can represent such kind of mechanisms, so it is suitable to study the interactions between crops and climate change. Therefore this study aims to develop a process-based crop model for maize called MATCRO-Maize, by modifying MATCRO-Rice (Masutomi et al., 2016) with adding a new function for C4 plant Maize: C4 photosynthesis.
To develop MATCRO-Maize, firstly some model parameters for rice (i.e. cardinal temperature, plant organ partitioning ratio) were changed for maize, and the function for calculating photosynthesis was changed for the C4 plant. Second, the model was validated at four validation sites: France, USA, Brazil, and Tanzania (these data have already been used for the AgMIP study, Bassu et al., 2014) by comparing simulated LAI (leaf area index), total aboveground biomass and yield with observations. Finally, MATCRO-Maize simulated global yield from 2000 to 2005 with the spatial resolution of 0.5 x 0.5 degree, then the simulated yields are compared with FAOSTAT yield data for the top 10 major countries of maize production.
Simulated results for the point scale show generally good agreement with observation data. R square and RMSE for the yield comparison were about 0.71 and 2.4 t/ha, respectively.
The simulated yield in Brazil was underestimated probably because the current MATCRO does not consider soil nitrogen dynamics. Nitrogen fertilizer wasn’t applied in Brazil because there was sufficient nitrogen in the soil due to organic matter mineralization. MATCRO-Maize didn’t consider this soil nitrogen, and this might make the yield in Brazil be underestimated.
The statistical values for the global validation show the correlation coefficient about 0.6 and RMSE 2.37 t/ha. Simulated top 10 countries’ yields were partly overestimated probably because of excess leave growth in simulation. For improvement, it is needed to represent maize growth more accurately, especially with leave growth in MATCRO-Maize.
There are two common approaches for predicting crop yield: statistical model and process-based model. Statistical models generally don’t consider physiological mechanisms to represent crop growth, therefore it is difficult to understand the details of relationships between climate change and crop growth. On the other hand, most of the process-based models can represent such kind of mechanisms, so it is suitable to study the interactions between crops and climate change. Therefore this study aims to develop a process-based crop model for maize called MATCRO-Maize, by modifying MATCRO-Rice (Masutomi et al., 2016) with adding a new function for C4 plant Maize: C4 photosynthesis.
To develop MATCRO-Maize, firstly some model parameters for rice (i.e. cardinal temperature, plant organ partitioning ratio) were changed for maize, and the function for calculating photosynthesis was changed for the C4 plant. Second, the model was validated at four validation sites: France, USA, Brazil, and Tanzania (these data have already been used for the AgMIP study, Bassu et al., 2014) by comparing simulated LAI (leaf area index), total aboveground biomass and yield with observations. Finally, MATCRO-Maize simulated global yield from 2000 to 2005 with the spatial resolution of 0.5 x 0.5 degree, then the simulated yields are compared with FAOSTAT yield data for the top 10 major countries of maize production.
Simulated results for the point scale show generally good agreement with observation data. R square and RMSE for the yield comparison were about 0.71 and 2.4 t/ha, respectively.
The simulated yield in Brazil was underestimated probably because the current MATCRO does not consider soil nitrogen dynamics. Nitrogen fertilizer wasn’t applied in Brazil because there was sufficient nitrogen in the soil due to organic matter mineralization. MATCRO-Maize didn’t consider this soil nitrogen, and this might make the yield in Brazil be underestimated.
The statistical values for the global validation show the correlation coefficient about 0.6 and RMSE 2.37 t/ha. Simulated top 10 countries’ yields were partly overestimated probably because of excess leave growth in simulation. For improvement, it is needed to represent maize growth more accurately, especially with leave growth in MATCRO-Maize.