Decreasing dynamic predictability of global agricultural drought with warming climate

Abstract

Droughts have been occurring frequently worldwide in a warming climate with an adverse impact on water–food–energy–ecology security, which raises substantial challenges for drought predictability. However, little is known about the future changes in dynamic predictability of agricultural drought over the globe and the dominant confounders causing this change. Here we leveraged Bayesian model averaging ensemble vine copula model to reveal changes in agricultural drought predictability globally in warm seasons at three projected global warming levels. We found that the projected dynamic predictability of agricultural drought would significantly decrease over 70% of the global land areas in +2 °C and +3 °C worlds, especially over North America, Amazonia, Europe, eastern and southern Asia and Australia. This was primarily attributed to the weakening soil moisture memory, background aridity and weakening land–atmosphere coupling. Our findings highlight that stakeholders should employ dynamic climate adaptations to cope with the decreasing drought predictability in a warmer climate.