Future Global Convective Environments in CMIP6 Models

The response of severe convective storms to a warming climate is poorly understood outside of a few well studied regions. Here, projections from seven global climate models from the CMIP6 archive, for both historical and future scenarios, are used to explore the global response in variables that describe favorability of conditions for the development of severe storms. The variables include convective available potential energy (CAPE), convection inhibition (CIN), 0–6 km vertical wind shear (S06), storm relative helicity (SRH), and covariate indices (i.e., severe weather proxies) that combine them. To better quantify uncertainty, understand variable sensitivity to increasing temperature, and present results independent from a specific scenario, we consider changes in convective variables as a function of global average temperature increase across each ensemble member. Increases to favorable convective environments show an overall frequency increases on the order of 5%–20% per °C of global temperature increase, but are not regionally uniform, with higher latitudes, particularly in the Northern Hemisphere, showing much larger relative changes. The driving mechanism of these changes is a strong increase in CAPE that is not offset by factors that either resist convection (CIN), or modify the likelihood of storm organization (S06, SRH). Severe weather proxies are not the same as severe weather events. Hence, their projected increases will not necessarily translate to severe weather occurrences, but they allow us to quantify how increases in global temperature will affect the occurrence of conditions favorable to severe weather. Plain Language Summary Severe weather can occur when some combination of atmospheric ingredients are present. These ingredients are called “convective environments” and refer to quantities that measure, for example, atmospheric instability and wind shear. By combining these convective environments into so‐called severe weather proxies, modelers can measure the favorability of occurrence of severe convective storms. Moreover, they can address a recurrent challenge in severe weather modeling: to find a way to robustly analyze phenomena (hail storms, tornadoes, straight‐line winds) that are highly intermittent and not resolved in coarse numerical models. CMIP6 models, for example, cannot resolve directly these phenomena because of both temporal and spatial resolution limitations. Therefore, we computed the convective environments for a subse