Diverging climate response of corn yield and carbon use efficiency across the U.S.

Abstract In this paper, we developed an open-source package to analyze the overall trend and responses of both carbon use efficiency (CUE) corn yield climate factors for contiguous United States. Our algorithm enables automatic retrieval remote sensing data through Google Earth Engine (GEE) U.S. Department Agriculture (USDA) agricultural production at county level application programming interface (API). Firstly, integrated satellite products net primary productivity gross based on Moderate Resolution Imaging Spectroradiometer (MODIS) sensor, climatic variables from European Centre Medium-Range Weather Forecasts. Secondly, calculated CUE commonly used metrics. Thirdly, investigated spatial heterogeneity these variables. We applied a random forest identify key drivers crop yield, estimated variability using moving window regression across results show that growing degree days (GDD) has highest predictive power while extreme (EDD) is least important explanatory variable. Moreover, observed in most areas U.S., increases or stays same with higher GDD precipitation. However, decreases north shows more mixed fragmented interactions south. Notably, there are some exceptions where negatively correlated precipitation Missouri Mississippi River Valleys. As global warming continues, anticipate decrease throughout vast northern part country, despite possibility remaining stable increasing.