Tropical temperature distributions over a range of climates: theory and idealised model simulations

Understanding future changes in temperature variability and extremes is an important scientific challenge. Here, the response of daily near-surface temperature distributions to warming is explored using an idealised global climate model.  Simulations of a wide range of climate states are performed with a slab-ocean aquaplanet configuration and with a simple land continent using a bucket-style model for hydrology. In the tropics, the responses of temperature extremes (i.e., high percentiles of daily near-surface temperature) to climate change contrast strongly over land and ocean. Over land, warming is amplified for hot days relative to the average day. But over ocean, warming is suppressed for hot days, implying a narrowing of the temperature distribution. 

Previous studies have developed theories based on convective coupling to interpret changes in temperature extremes over land. Building on this work, here the contrasting temperature distribution responses over land and ocean are investigated using a new theory based on strict convective equilibrium, which assumes moist adiabatic lapse rates. The theory highlights four physical mechanisms with the potential to drive differential warming across the temperature distribution: hot-get-hotter mechanism, drier-get-hotter mechanism, relative humidity change mechanism, and the free tropospheric temperature change mechanism.  Hot days are relatively dry over land due to limited moisture availability, which drives the drier-get-hotter mechanism and  amplified warming of the warm tail of the distribution. This mechanism is the dominant factor explaining the contrasting responses of hot days over land and ocean to climate change. An extended version of the theory, which relaxes the strict convective equilibrium assumption, is introduced and applied to the simulations to understand the influence of convective available potential energy (CAPE) on changes in the temperature distribution.