Development of impact functions on regional paddy rice yield in Japan for integrated impact assessment models

For this study, we developed and validated two types of impact functions on regional paddy rice yield in six regions of Japan. One is simple, and the other is slightly detailed. An impact function has the following features: (1) it is a look-up table that shows the quantity of interest (regional paddy rice yield in this study) in response to various forcing conditions; (2) it is provided by a mechanistic but heavy computational-load model; and (3) it is implemented in integrated impact assessment models to reduce the simulation time. The developed impact functions will be submitted for use in integrated impact assessment models and will contribute to examining the target level of stabilization on greenhouse gas emissions. The impact functions were compared with the observation and the conventional crop model simulation (full-simulation). The slightly detailed (simple) impact function agreed with the observation with a mean-absolute-percentage (MAP) error of 7.8% (7.7%). Under warming conditions, the slightly detailed (simple) one was comparable to the full-simulation with a MAP difference of 4.5% (4.6%). The majority of the difference arose from the temperature fluctuation within a short time period of less than two months. These results indicate that the application of the impact functions is permissible for assessing the impact of the average climate change on regional yield. However, the impact functions tend not to be a very reliable tool for assessing the impacts of extreme climate events and interannual climate variation on regional yield.