Reproducibility of Crop Yield Simulated by iGAEZ Model with High-resolution GCM Output

It is widely accepted that uncertainties in calculated crop yields through the global-scale agricultural model are induced by the type and spatial resolution of the input data. In general, a spatial resolution of 0.5° is employed in crop simulation on a global scale. Datasets with higher spatial resolution are not available over the long term due to limited availability and uncertainty in climate conditions. The purpose of this research is to simulate the crop yield with an improved Global Agro-Ecological Zones (iGAEZ) model on a global scale for 1979-1999 by using outputs from the Climate Research Units Global 0.5° Monthly Time Series (CRU-TS2.1, 0.5° grid) and the Meteorological Research Institute Global Climate Model with a 20-km mesh horizontal resolution (MRI-GCM20, 0.1875° grid) as climate input datasets and to evaluate the availability of super-high-resolution information as forcing data on reproducibility of crop yield and harvested area in large and small regions or countries. Results on crop model simulation indicate that (i) when using MRI-GCM20 outputs, crop yields and harvested areas correlate strongly with results of CRU-TS2.1 output in large regional areas or large croplands for long-term statistics. Therefore, there is little merit in increasing the spatial resolution, particularly for regions and countries with large croplands. (ii) However, there may be large potential and advantages in simulating crop yields by using MRI-GCM20 at a country level and in small cultivated areas and low-latitude zones. Simulations using the GCM output are effective in impact studies and may contribute to the adaptation research on the influence of crops due to future climate changes.