A fundamental quantity in tropical dynamics is the ‘convective mass flux’, which measures the upwards mass flux in tropical convection. Changes in convective mass flux under warming are an important, and uncertain, aspect of climate change. Here we build off recent work linking changes in mass flux to the clear-sky energy budget to show that convective mass fluxes decrease along isotherms at around 3-5 % K-1 under warming. We show that this constraint holds throughout the free-troposphere and across a hierarchy of models; from idealized radiative-convective equilibrium simulations to CMIP6 models. This decrease in convective mass flux with warming is driven by a stabilization of the lapse rate and can be captured with a simple analytical model. We also revisit previous work by Held and Soden (2006), who proposed a scaling for changes in the convective mass flux at cloud-base with warming. We show that the Held and Soden scaling does not capture inter-model spread in cloud-base mass flux changes under warming in cloud-resolving models. Altogether, this work provides a quantitative constraint on changes in convective mass flux throughout the troposphere which can be derived from first principles, and which is verified across a hierarchy of models.

The climatological atmospheric circulation is key to establishing the tropical ‘pattern effect’, whereby cloud feedbacks induced by sea surface temperature (SST) warming depend on the spatial structure of that warming. But how patterned warming-induced circulation changes affect cloud responses is less clear. Here we use idealized simulations with prescribed SST perturbations to understand the contributions to changes in tropical-mean cloud radiative effects (CRE) from different circulation regimes. We develop a novel framework based on moist static energy to understand the circulation response, targeting in particular the bulk circulation metric of ascent fraction. Warming concentrated in regions of ascent leads to a strong ‘upped-ante’ effect and spatial contraction of the ascending region. Our framework reveals substantial contributions to tropical-mean CRE changes not only from traditional ‘pattern effect’ regimes, but also from the intensification of convection in ascent regions as well as a smaller contribution from cloud changes in convective margins.