Modeling regional crop yield and irrigation amount for corn and soybean in Central U.S.

Crop yields are greatly impacted by the climate forcing and irrigation has been conducted to improve crop yield by relieving soil water stress under drought condition. On the other hand, the agriculture expansion and irrigation practice have also altered the surface moisture and energy balance as well as the biogeochemical cycle in the agricultural ecosystem. Given their significance to the environment and food security, there is a growing trend to incorporate dynamic crop growth and irrigation into land surface models (LSMs) and it is desirable to utilize the model to provide reasonable crop yield estimate and irrigation amount. Previous LSMs with dynamic crop and irrigation module are developed at field-scale and site-specific parameters are calibrated locally. In this work, we attempted to propagate these field-scale efforts to regional scale on modeling the crop yield, for corn and soybean, and the irrigation amount in Central U.S., using Noah-MP LSM. The site-specific parameters are constrained by best-available spatial datasets, including cropping calendar, crop and irrigation fractions. The model results are evaluated against the county-level crop yield and irrigation water withdrawal report. The results found that, in the rainfed region, the root-mean-square-error (RMSE) for crop yield are 22% and 27% for corn and soybean, respectively. However, in the irrigated region, the irrigation has significantly reduced the RMSE for corn from 36% to 24%, demonstrating the critical impacts of irrigation to improve crop yield. On the other hand, the irrigation impacts for soybean are not as strong as for corn; the RMSE are 27% and 25%, with and without irrigation. A strong positive correlation between crop yield and total water use is identified and the irrigation could increase the crop yield around 35 ~ 60%. Finally, we identified several uncertainties stemmed from model parameters, including the plant photosynthesis and phenology as well as the coupling between carbon and moisture cycle. These findings have great implication for developing crop models as well as obtaining reliable crop yield estimates in the future.