Global Impact Assessment of Internal Climate Variability on Maize Yield Under Climate Change

Abstract

Internal climate variability (ICV) is well-known to mask forced climate change patterns and is thus expected to also impact crop yield trends. To date, a global picture of ICV effect on crop yield projection remains unclear, which inhibits effective adaptation and risk management under climate change. By combining initial condition large ensembles from multiple climate models with machine-learning based crop model emulators, an ensemble of 2002 global maize yield simulations are conducted. The ICV effect is quantified for by the middle and end of 21st century under the business-as-usual scenario. ICV is shown to have significant influence on both the magnitude and sign of future yield change, with relatively higher impact in the top producing countries. The results imply that future yield projections considering relatively limited samples of ICV can be highly misleading as they may, by chance, indicate low yield loss risk in areas which will, instead, be at high risk (or vice versa). Further analysis reveals that the ICV effect is 2.30 ± 0.02 and 1.25 ± 0.03 times larger for yield projections than temperature and precipitation projections, respectively, suggesting an amplification of ICV effect from climate system to agricultural system. This study highlights that crop yield projections are substantially more uncertain than climate projections under the influence of ICV.