Can Large‐Scale Clustering of Tropical Precipitation Be Used to Constrain Climate Sensitivity?

Abstract The spatial organization of deep convection in tropical regions is posited to play an important role in determining characteristics of the tropical climate such as the humidity distribution and cloudiness and may therefore be an important control on climate feedbacks. This study analyzes one aspect of convective organization, the clustering of heavy precipitation on large scales, in both interannual variability and under warming in future climate projections. Clustering is quantified by the total areal coverage and mean size of precipitation features exceeding a climate‐dependent threshold. Both observations and global climate models (GCMs) suggest increased clustering under El Niño conditions, but the relationship of this clustering to the tropical humidity and cloud distribution differs in different GCMs. Under future warming, all models considered project an increase in clustering associated with a narrowing of the intertropical convergence zone and with a model‐spread partially explained by differences in projections of the Pacific SST gradient ( r = −0.46). However, unlike in interannual variability, a general spatial clustering with warming do not explain the climatological humidity or low‐cloud responses, limiting a simple observational constraint on feedbacks. Notably, a few model outliers drive considerable spread in tropical‐mean mid‐tropospheric drying, which appears to be partially linked to clustering of heavy rainfall meridionally toward the equator. These results suggest possible future targets for model evaluation.