17:15 〜 18:45
[ACG37-P04] データ同化手法による過去120年間の我が国のコメ収量の復元
キーワード：プロセスベースモデル、パラメータ最適化、品種改良、気候変動
We are facing the challenge of increasing agricultural production with limited resources to meet food demand increase caused by world population growth (Charles et al, 2010). On the other hand, cereal production is severely vulnerable to climate conditions, raising concerns that it would be affected in the future by ongoing climate change (IPCC, 2023). Given these perspectives, it is important for future food security to investigate the relationship between climate change and cereal production. Some studies have investigated past climate change impacts on cereal production using FAO statistical data (Lobell et al, 2011). These previous studies, however, couldn’t provide detailed insights into the climate change-cereal production relationship for the following reasons: (1) reliance on country-level data, ignoring regional climate variations within each country; (2) analysis periods typically covering only the last few years, making it difficult to detect long-term historical climate change trends since 1900s and their impacts on agriculture; and (3) failure to consider genetic improvements in crops resulting from crop breeding. Analyzing the effects of this crop evolution on yield increase to date is crucial for constructing future breeding strategies.
Process-based crop models are useful when considering crop breeding. MATCRO (Masutomi et al, 2016), a process-based crop model, simulates rice yield by calculating rice growth from meteorological and other environmental input data. This model contains parameters related to rice growth characteristics and enables us to make a comprehensive investigation of the connection between climate conditions, rice production, and rice characteristics. Moreover, Japan leaves a historical dataset of rice yield since the 1880s at prefectural levels, with which it becomes possible to analyze rice yield in high resolution within Japan. Therefore, this study aims to reconstruct rice yield in Japan at prefectural levels over the past 120 years, from 1896 to 2015, with MATCRO and historical yield datasets, and to quantify the contribution rate of individual environmental and genetic factors to rice yield increase to date.
At first, we optimized seven parameters related to nitrogen sensitivity, morphology, resistance to cold and hot damage, and phenology in MATCRO. Optimization was performed for each prefecture except for Okinawa, at point scale. The optimization window is 20 years, starting from 1996-2015 and regressing in 10-year intervals to 1896-1915. After optimization, we ran MATCRO with different conditions, fixing environmental and genetic parameters at values of 1896 to quantify the contribution rate of each fixed parameter to rice yield.
Results of the optimization shows a generally good reconstruction of observation data. Original parameters are lower accuracy in reconstructing yields during the periods of little nitrogen fertilization. This is because original parameters are validated with FAOSTAT yield data in recent years when there is affluent nitrogen input. For quantification, CO2 fertilization effect was the most significant, being twice the effect of nitrogen fertilization. This outcome is likely attributed to the constrained rice growth resulting from limited photosynthesis. We need to analyze the factors contributing to yield increase in more detail.
Process-based crop models are useful when considering crop breeding. MATCRO (Masutomi et al, 2016), a process-based crop model, simulates rice yield by calculating rice growth from meteorological and other environmental input data. This model contains parameters related to rice growth characteristics and enables us to make a comprehensive investigation of the connection between climate conditions, rice production, and rice characteristics. Moreover, Japan leaves a historical dataset of rice yield since the 1880s at prefectural levels, with which it becomes possible to analyze rice yield in high resolution within Japan. Therefore, this study aims to reconstruct rice yield in Japan at prefectural levels over the past 120 years, from 1896 to 2015, with MATCRO and historical yield datasets, and to quantify the contribution rate of individual environmental and genetic factors to rice yield increase to date.
At first, we optimized seven parameters related to nitrogen sensitivity, morphology, resistance to cold and hot damage, and phenology in MATCRO. Optimization was performed for each prefecture except for Okinawa, at point scale. The optimization window is 20 years, starting from 1996-2015 and regressing in 10-year intervals to 1896-1915. After optimization, we ran MATCRO with different conditions, fixing environmental and genetic parameters at values of 1896 to quantify the contribution rate of each fixed parameter to rice yield.
Results of the optimization shows a generally good reconstruction of observation data. Original parameters are lower accuracy in reconstructing yields during the periods of little nitrogen fertilization. This is because original parameters are validated with FAOSTAT yield data in recent years when there is affluent nitrogen input. For quantification, CO2 fertilization effect was the most significant, being twice the effect of nitrogen fertilization. This outcome is likely attributed to the constrained rice growth resulting from limited photosynthesis. We need to analyze the factors contributing to yield increase in more detail.