1 Introduction

The tendency for convection to organize from isolated convective updrafts into spatially coherent clusters has long been recognized in numerical model simulations (Bretherton et al., 2005; Held et al., 1993; Tompkins, 2001). This aggregation of convection occurs even in the presence of uniform boundary conditions (self-organization) and not only impacts the intensity of precipitation events (Bao & Sherwood, 2019), but also modulates the larger-scale thermodynamic and radiative properties of the tropics (e.g., Bony et al. (2020)). Increased aggregation is associated with increased spatial variance of moisture; dry regions become drier and moist regions become moister (Dai & Soden, 2020). The increased spatial variance of moisture directly impacts both the intensity of precipitating systems and the larger-scale radiative fluxes (Bony et al., 2020; Bony et al., 2016; Bretherton et al., 2005; Wing et al., 2020).
The net effect of more aggregated convection is to dry the tropical free troposphere, particularly in cloud-free regions, resulting in a net loss of longwave radiation to space (Bretherton et al., 2005). This large-scale influence of aggregation has been proposed as a potential thermostat that may regulate the sensitivity of the tropics to radiative forcing (Mauritsen & Stevens, 2015). However, the interaction with radiation involves feedbacks. The amplification and expansion of dry regions is believed to play a key role in triggering aggregation, and radiative feedbacks involving both clouds and water vapor are essential for maintaining the aggregation in idealized models (Wing et al., 2017). Additionally, the relative importance of clear versus cloudy sky radiative processes is not well understood, as is the contributions of shortwave (SW) versus longwave (LW) radiation.
There is a long history of observational studies of convective organization (see Holloway et al. (2017) for a recent review). Most recently, Bony et al. (2020) found that increased aggregation in the tropics is associated with a reduction in high cloud cover, a drier free troposphere in the non-convective environment and increased emission of infrared radiation to space. These features lead to a net radiative cooling of the tropics.
Precipitation extremes are largely controlled by the amount of moisture present in the atmosphere (Allen & Ingram, 2002; Pall et al., 2007; Trenberth, 1999) and also atmospheric convergence (Liu et al., 2020), whereas the global-mean precipitation is constrained by the global-mean radiative cooling. From an energy balance perspective, latent heating into the atmosphere must be balanced by atmospheric radiative cooling given that heat capacity of the atmosphere is negligible (Allen & Ingram, 2002). In model simulations, the change in extreme precipitation in response to increases in greenhouse gases is found to depend on the magnitude of warming (Pendergrass et al., 2015) and the change in convective organization (Muller, 2013; Pendergrass et al., 2016). In observations, the degree of aggregation at regional scales has also been connected with extreme rainfall events (Dai & Soden, 2020).
Climate model projections indicate that the thermodynamic constraint based on the Clausius-Clapeyron relation is a good predictor for extreme precipitation changes in a warmer climate for regions where changes in the circulation are small (Pall et al., 2007). However, this may not be the case for regions with large changes in the atmospheric circulation, such as the tropics (Emori & Brown, 2005; Vecchi & Soden, 2007) or for tropical cyclones (Knutson et al., 2013; Liu et al., 2020). Indeed, several studies suggest that the sensitivity of tropical precipitation extremes is substantially larger than that predicted from the Clausius-Clapeyron relationship (Allan & Soden, 2008; Norris et al., 2019; O’gorman, 2015; Westra et al., 2013).
Among different factors contributing to radiative cooling of the atmosphere, cloud radiative effects (CRE) have been received considerable attentions. The Clouds On-Off Klimate Intercomparison Experiment (COOKIE) is designed to investigate the role of CRE in the climate system (Stevens et al., 2012). This project compares simulations with clouds that are transparent to radiation (“clouds-off”) and those including CRE (“clouds-on”). Using output from the COOKIE project, Fermepin and Bony (2014) showed that low cloud radiative effects increase tropical precipitation and strengthen winds near ocean surface. However, Li et al. (2015) found that CRE can decrease precipitation in the tropics but increase it at middle-to-high latitudes. Recently, Medeiros et al. (2021) reported that extreme precipitation over tropical ocean is strengthened by CRE. While mean precipitation and large-scale circulation can be affected by removing CRE, it is unclear how precipitation, especially extreme precipitation, will respond if radiative-convective interactions are disabled.
In idealized models, studies found that interactive radiation is key to convective aggregation over a limited domain (Muller & Bony, 2015; Muller & Held, 2012; Wing & Emanuel, 2014; Yang, 2018). In simulations of radiative convective equilibrium without rotation, Bao and Sherwood (2019) found that extreme daily precipitation gets stronger when convection is more aggregated with fully interactive radiation. Most research on the coupling of clouds and radiation on smaller time/space scales has focused on the impact of radiative feedbacks on convective organization. These studies typically have been performed using high resolution, radiative equilibrium simulations under very idealized settings. This makes it difficult to ascertain the importance of cloud-circulation feedbacks under more realistic situations (e.g., in the presence of the SST gradients or wind shears that are present in our planet). It also makes it more challenging to evaluate the model simulations with observations.
In this study, we examine the impact of radiative interactions on the spatial organization of convection and extreme precipitation events under realistic boundary conditions. Instead of completely removing the CRE, which dramatically changes the mean circulation and rainfall in a model (e.g., as is done in COOKIE), synoptic-scale radiative interactions are suppressed by prescribing radiative cooling rates using their monthly climatological values. This enables us to examine the behavior of two versions of a model with nearly identical large-scale circulations but with differing degrees of convective organization. Our simulations highlight the role of synoptic-scale radiative coupling in enhancing convective aggregation and extreme precipitation by increasing the horizontal gradient of radiative cooling which provides an upgradient transport of energy from dry to moist regions.