Moisture‐Budget Drivers of Global Projections of Meteorological Drought From Multiple GCM Large Ensembles

Future projections of global meteorological drought are evaluated in the Multi-Model Large Ensemble Archive, including an evaluation of the atmospheric moisture budget, conditioned on drought years. Drought is defined as 5-year running-mean annual precipitation below some threshold, for example, 10th percentile. Drought increases in frequency over the subtropics, in addition to certain tropical regions, consistent with previous studies. The moisture-budget decomposition allows drought to be defined as mean-flow, eddy, or feedback droughts, depending on which term in the equation contributes the largest negative interannual anomaly. In the historical climate, mean-flow droughts constitute most droughts at low latitudes; eddy droughts are equally common at higher latitudes; feedback droughts (i.e., droughts exacerbated by land-atmosphere feedbacks) constitute almost all droughts in water-limited subtropical/Mediterranean regions. The future drought increases are predominantly due to increases in feedback droughts in regions where these droughts are common historically but also over the Amazon. However, over most Mediterranean-type regions mean-flow droughts are also large contributors, resulting from dynamics. Eddy droughts also contribute to future increases along the equatorward flanks of historical eddy-driven jets, likely reflecting poleward shifts therein. Model uncertainty is particularly large over the Amazon and Australia, a reflection of model diversity in processes associated with land-atmosphere interaction. Based on these results, an availability of 3-D atmospheric data from a wider swath of global climate model large ensembles could help constrain global drought projections based on the representation of drought mechanisms in the historical climate. Plain Language Summary Around the globe, drought is projected to intensify and increase in frequency in the future. However, the mechanisms by which drought increases have not been shown from a global perspective. Here we show that drought, defined as a 5-year period of extremely low precipitation, may fall under three categories. One is where the monthly average conditions in the atmosphere are conducive to drought; one is where there is an absence of individual storms over some region that depends on isolated events for its annual precipitation; one is where drought is made worse by a drying of the surface, which in turn further dries the atmosphere. In the future, climate models project an increase of drought mostly over regions where historically the latter category is prevalent. However, there are large disagreements between climate models over these regions, particularly over the Amazon and Australia. This reflects disagreements between the models in the interaction between the atmosphere and the Earth's surface. The results of this study help us to understand how climate models agree/disagree on drought intensification and why