Key points are not available for this paper at this time.
Abstract Anthropogenic climate warming is expected to accelerate the hydrological cycle with significant consequences for hydrological droughts. However, a systematic understanding of climate warming impacts on the global hydrological droughts and their driving mechanisms is still lacking. Here, we integrate bias‐corrected climate experiments, multiple hydrological models (HYs), and a multivariate analysis of variance (ANOVA) with a machine learning modeling framework, to examine the evolving frequency and multivariate characteristics of hydrological droughts and their mechanisms under climate warming for 6,688 catchments in the five principal Köppen‐Geiger climate zones. Results show that the total frequency of hydrological droughts is likely to stay unchanged while extreme hydrological droughts (e.g., events with a 30 yr joint return period, JRP) are projected to occur more frequently across the 21st century. The historical 30 yr JRP events assessed during the historical baseline period of 1985–2014 could become twice as frequent over ∼60% of global catchments by 2071–2100 under the middle and high emission scenarios (ESs). Climate uncertainty (i.e., from global climate models and ESs) is the major source of uncertainty over temperate and tropical catchments, versus HY uncertainty in arid catchments with locally complex runoff regimes. Our machine learning framework indicates that precipitation stress controls the development of historical droughts over ∼87% of global catchments. However, with climate warming, air temperature variations are expected to become the new primary driver of droughts in high‐latitude cold catchments. This study highlights an increasing risk of global extreme hydrological droughts with warming and suggests that rising temperatures in high latitudes may lead to more extreme hydrological droughts.
Gu et al. (Thu,) studied this question.