The authors study abrupt transitions with gradual increase in sea surface temperature in an idealized tropical atmosphere. The approach the authors take is the singlecolumn modeling that incorporates the weak temperature gradient (WTG) approximation. Their numerical experiment with different SST show that abrupt transition can arise from the interplay between local deep convection and the large-scale adjustments parameterized via WTG. The authors further suggested the abrupt transition is related to evaporative cooling.

We document a feature of the tropical atmosphere that could be relevant to episodes of abrupt transitions in global climate that regularly occurred during the last ice age. Using a single-column model (SCM) incorporating the weak temperature gradient (WTG) approximation, we find that abrupt transitions occur as the sea surface temperature is steadily increased. Because these transitions arise from the interplay between local deep convection and the largescale adjustments that are required to maintain weak temperature and pressure gradients, they are only present with the WTG approximation relevant for the tropics but may be of interest as a trigger for abrupt transitions in global climate. These transitions are marked by an abrupt change in the partitioning of rainfall between convective and large-scale (microphysics) subroutines in addition to various other features of the column, including cloudiness, vertical velocity, temperature, and humidity. We conclude that the transitions are initiated by a failure of evaporative cooling in the lower free troposphere. This leads to lower-column heating and a burst of convection that heats the upper free troposphere, increasing the large-scale rainfall rate, which allows for sufficient evaporative cooling to restabilize the column.

We find that part of the uncertainty in the amplitude and pattern of the modeled precipitation response to CO2 forcing traces to tropical condensation not directly involved with parameterized convection. The fraction of tropical rainfall associated with large-scale condensation can vary from a few percent to well over half depending on model details and parameter settings. In turn, because of the coupling between condensation and tropical circulation, the different ways model assumptions affect the large-scale rainfall fraction also affect the patterns of the response within individual models. In two single-model ensembles based on the National Center for Atmospheric Research (NCAR) Community Atmosphere Model (CAM), versions 3.1 and 5.3, we find strong correlations between the fraction of tropical large-scale rain and both climatological rainfall and circulation and the response to CO2 forcing. While the effects of an increasing tropical large-scale rain fraction are opposite in some ways in the two ensembles—for example, the Hadley circulation weakens with the large-scale rainfall fraction in the CAM3.1 ensemble while strengthening in the CAM5.3 ensemble—we can nonetheless understand these different effects in terms of the relationship between latent heating and circulation, and we propose explanations for each ensemble. We compare these results with data from phase 5 of the Coupled Model Intercomparison Project (CMIP5), for which some of the same patterns hold. Given the importance of this partitioning, there is a need for constraining this source of uncertainty using observations. However, since a “large-scale rainfall fraction” is a modeling construct, it is not clear how observations may be used to test various modeling assumptions determining this fraction.

It has recently been shown that the presence of a spectator pseudoscalar field, coupled to photons through a Chern-Simons term, can amplify the primordial tensor spectrum without observationally disrupting the primordial scalar spectrum. The amplification occurs due to an instability that develops for the vector fields. We consider the extension of previous studies to account for the contribution arising from an inhomogeneous vector background which is generated prior to the onset of inflation. We find that there may be contributions in which net momentum is transferred between the inhomogeneous vector background and the gravitons, which would give rise to a signature different than in the absence of the semiclassical corrections. We discuss the properties the classical vector field form must have in order for these signatures to leave observable imprints, though we were unable to construct a model for generating such a vector field.