**1. Introduction**

Recent reports of devastation resulting from record-breaking heavy precipitation around the world have provided strong indications that humanity is already experiencing

**Citation:** Lau, W.K.M.; Kim, K.-M.; Harrop, B.; Leung, L.R. Changing Characteristics of Tropical Extreme Precipitation–Cloud Regimes in Warmer Climates. *Atmosphere* **2023**, *14*, 995. https://doi.org/10.3390/ atmos14060995

Academic Editors: Xiaolei Zou, Guoxiong Wu and Zhemin Tan

Received: 28 April 2023 Revised: 28 May 2023 Accepted: 31 May 2023 Published: 8 June 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

the disastrous effects of increased extreme precipitation, i.e., flash floods, soil erosion, landslide, degradation of the eco-system, destruction of properties and loss of lives, attributable to anthropogenic greenhouse warming. Without a timely reduction in the emissions of greenhouse gases, the current trend in extreme precipitation will continue, and adverse impacts on the socio-economic system are likely to become worst [1]. Extreme precipitation in the tropics not only adversely affects the livelihood of more than 40% of the world population but is also a primary driver of global climate variability and change [2–5]. Hence, a better understanding of the physical processes underlying tropical extreme precipitation and its global impacts is paramount for the development and implementation of effective adaptation and mitigation strategies for global climate variability and change.

In the tropics and subtropics, climatologically strong surface heating and low-level moisture convergence lead to increased convective instability, enhancing heavy precipitation preferentially in regions with a warm surface temperature, i.e., the Inter-Tropical Convergence Zone (ITCZ), monsoon regions, and the maritime continent [6–9]. Changes in precipitation under global warming generally follow a geographic distribution pattern of "wet-gets-wetter" and "warmer-gets-wetter" [10–15]. A necessary condition for precipitation is the formation of clouds. Both precipitation and clouds, and their associated temporal and spatial distributions, are strong functions of atmospheric heating/cooling and moistening/drying processes, modulated by the surface temperature, heat and moisture fluxes, cloud microphysics, convection, and large-scale circulation [16–22]. Previous research on precipitation and clouds under climate variability and change have emphasized: (a) regional extreme precipitation events, cloud microphysics, and latent heating and forcing by a mesoscale convective system (MCS) [23–29], and (b) radiation heating feedback by various cloud types in determining global climate sensitivities [30–37]. While much knowledge has been gained and both approaches need to be continued in order to narrow down uncertainties, an emerging paradigm is that a deeper understanding of the myriad factors leading to extreme precipitation under climate change is predicated on a more comprehensive approach based on the broader context of interactions and enhanced by feedback processes involving cloud radiation, convection, and large-scale circulation [38–44].

Previous observational and climate modeling studies have shown that under global warming, the rate of increase in the top 0.1% of tropical daily precipitation has been estimated to be near 10% K−1, significantly higher than those in the extratropics, which is limited by a thermodynamic rate of 6–7% K−1, governed by the Clausius–Clapeyron relationship for atmospheric saturated moisture and temperature [2–5]. Models and observations have also shown that as Earth's surface and the atmosphere warm up under anthropogenic CO2 radiative forcing, convection becomes more vigorous, and clouds grow faster, wider, and taller, producing more extreme precipitation [45]. An increasing number of recent studies [5,46–48] have shown that extreme precipitation events attributable to GHG warming tend to occur preferentially in tropical/subtropical regions with a strong and sustained organization of deep convection embedded in extended areas of high anvil clouds associated with long-lived strong mesoscale convective systems (MCS). Even though such long-lived MCS occur in less than 5% of the tropical precipitation events in preferred climatological wet regions, they account for more than 40% of the extreme precipitation amount [49]. This could mean that extreme precipitation, which occurs on hourly/daily time scales, could have organization signals on monthly and longer time scales over specific land or oceanic regions, and even over the entire tropics.

In spite of the increasing reports on devastating and destructive impacts on populated land regions, the scientific question of whether extreme cloud–precipitation organization is (a) fundamentally different and (b) more or less intense and/or frequent over land vs. ocean on climatic time scales remains uncertain. In this paper, we focus on addressing these questions and the scientific rationales underlying them based on general circulation model (GCM) simulations. However, because of the GCM's coarse resolution (>50–100km), MCS are not explicitly resolved and not well simulated in traditional GCM cloud–precipitation parameterizations. More recently, MCS-like features have been simulated and tracked in a

moderate resolution (50 km) GCM, with improved physical cloud–precipitation parameterization that includes organization features occurring across the scales [50]. In this study, we conducted AMIP (Atmospheric Model Intercomparison Project)-type [51] simulations using the Department of Energy's Exascale Energy Earth System Model (E3SM), which includes an improved unified parameterization of clouds and precipitation types, to examine its capability in simulating MCS-like features and contributing to extreme tropical precipitation on climatic time scales. See a further discussion on the E3SM model's physics in Section 2.

Specifically, we disentangled the effects of surface warming vs. atmospheric heating and moistening by increased CO2 radiation forcing, leading to an occurrence of extreme precipitation–cloud regimes, with respect to changes in the stratiform vs. convective precipitation, precipitation efficiency, and thermodynamic vs. dynamical forcing over land and ocean. The organization of the paper is as follows. In Section 2, we describe the methodology, including the key physical parameterizations of the clouds and precipitation processes and the experimental design of the E3SM model experiments. In Section 3, we present the key results of the experiments. The conclusions and scope of continuing work are discussed in Section 4.

#### **2. Model Description and Methodology**

The U.S. Department of Energy (DOE)'s Energy Exascale Earth System Model Version 1 (E3SMv1) [52] was developed with the aim of addressing the grand challenge of actionable prediction of the Earth system's variability and changes to meet scientific and societal needs. The E3SMv1 is a fully coupled ocean–atmosphere–land–biosphere model, developed on the foundation of the Community Earth System Model version 1 (CESM1), but it includes adaptations and improvements to optimize the computational performance and science/application requirements of the DOE.

For clouds and precipitation, the E3SM atmospheric model (EAM) uses an improved version of Cloud Layers Unified by Binormals (CLUBB), which includes a third-order turbulence closure parameterization that unifies the treatment of boundary-layer clouds, shallow and deep convection, and cloud microphysics [53,54]. In the E3SM, improving the model of shallow cumulus clouds and stratocumulus clouds and precipitation was achieved by optimizing the scale dependence of the CLUBB parameterization for a diurnal cycle of precipitation over land [55]. Deep convective clouds and precipitation are based on the improved version of the Zhang and McFarlane (1995) [56] scheme, which included a recent update on the bulk parameterization of updraft processes (entrainment, detrainment, condensation, and precipitation) and downdraft processes (entrainment and evaporation of falling rain) from both liquid- and ice-phase precipitation [57]. Aerosol and cloud microphysics interactions in stratiform clouds are included in an updated version of the Modal Aerosol Module (MAM4) [58], which predicts the concentrations of major aerosol species (sulfate, black carbon, primary and secondary organic matter, mineral dust, and sea spray). The Morrison and Gettelman Version 2 [59] aerosol–cloud microphysics parameterization, coupled with CLUBB and MAM4, was used for the generation of shallow and stratiform clouds. The implementation of a convective gustiness adjustment to CLUBB significantly improved the simulation of stratiform and shallow clouds over the tropical ocean, where the climatological surface mean winds are weak [60]. Radiation–cloud– convection–circulation interaction (RC3I) processes in the EAM have also been significantly improved by better microphysics-based treatment of wet scavenging and re-suspension of evaporating precipitation, which affect the abundance and size of cloud condensation nuclei for liquid- and ice-phase precipitation, respectively [61]. In addition, this study used a variant of EAMv1 that adopted a consistent set of parameter adjustments, including sub-grid scale wind variance, resulting in better simulations of cloud properties [55].

Based on the Community Land Model (4.5) of CESM2, the land model (ELM) of the E3SM includes improvements in the representation of the water cycle processes of soil hydrology, river routing, coastal erosion, and biogeochemistry fluxes [52]. A new river routing Model for Scale Adaptive River Transport (MOSART) was implemented, with particular emphases on human activities, including the management of water availability from river flow and the mitigation of floodplain inundation [62–64]. Two-way coupling between the MOSART and ELM was implemented to estimate the amount of water available from precipitation, river run-off, and storage in reservoirs for irrigation.

A key motivation for our analytical approach was to assess the degree to which the E3SM parameterization of fast and subgrid-scale cloud microphysical processes reflect the important contribution of mesoscale convective systems (MCS) to extreme cloud– precipitation organization on climatic (monthly and longer) time scales. Key features of MCS producing heavy precipitation over the ocean and land have been well documented [24,65]. During peak MCS development, a deep core with intense convective precipitation is coupled with extensive anvil clouds in the downwind regions, where stratiform precipitation dominates (Figure 1a). In the stratiform region, condensation heating associated with increased precipitation by active ice-phase microphysics (deposition, riming, and aggregation) causes a large-scale ascent in the upper troposphere above the freezing level (0 ◦C isotherm) near 500 hPa. At the same time, evaporative cooling by falling rain results in a large-scale mean descent in the lower troposphere. For tropical extreme precipitation events, the associated MCS life cycle may consist of multiple clusters of MCS complexes at various stages of development, starting with predominant convective precipitation and evolving to an increased contribution from stratiform precipitation. The results of cloud-resolving model simulations have shown that for non-MCS (100% convective) precipitation, the heating profile has a maximum near 500 hPa, while for "pure" stratiform precipitation, the heating profile shows a dipole structure with maximum heating (cooling) in the upper (lower) troposphere (Figure 1b). As a result, the degree of MCS development is reflected in the elevation of the level of maximum condensation heating, relative to that of convective (non-MCS) precipitation. However, it is important to note that the presence of an active convective core coupled with a substantial fraction of the stratiform (anvil cloud) region is essential for the development and maintenance of an MCS. A stratiform anvil cloud–precipitation regime decoupled from its convective core lacking in a sustained supply of ice-phase condensates from the convective core region represents a decaying MCS that readily dissipates and ceases to rain [24]. Hence, while the proportion of stratiform rain in a developing and active MCS is expected to increase with increasing extreme precipitation, it is unlikely to be close to 100% over the life cycle of the development of multiple organized MCS systems in extreme precipitation [65].

**Figure 1.** (**a**) Schematic showing organization of convective and stratiform cloud precipitation associated with a mesoscale convective system (adopted from Houze et al., 1989 [24]). Formation of water droplet and precipitation ice are denoted as • and \*, respectively. (**b**) Idealized vertical profiles of latent heating as a function of stratiform precipitation fraction from simulations of cloud-resolving models (adopted from Sui et al., 2020 [66]).

For model integration, the control experiment was represented by an equilibrium solution of an AMIP-type 10-year integration of the E3SM with a 100 km latitude–longitude resolution, and 72 layers with variable thickness in the vertical direction, with a top at 60 km, under present-day sea surface temperature (SST), sea-ice conditions, and an atmospheric concentration of CO2. To disentangle the effects of surface warming vs. CO2 radiative forcing, equilibrium solutions based on two separate AMIP simulations identical to the control were conducted. First, the SST was increased by including an idealized plus-4K (P4K) anomaly uniformly across the globe. Second, an SST anomaly (SSTA) and CO2 radiative forcing were imposed based on the climatology of the last 30-year simulation of an abrupt 4 times CO2 (4xCO2) experiment with the coupled ocean–atmosphere version of the E3SM, as part of the Coupled Model Intercomparison Project phase 6 (CMIP6) [67,68]. Changes in the tropical cloud precipitation characteristics for extreme precipitation were compared among the control, P4K, and 4xCO2 simulations. Emphases were placed on a better understanding of the forcing and competing influences and feedback arising from surface warming vs. atmospheric heating and moistening processes. The realism of the model parameterization of the MCS and extreme precipitation, in terms of the changes in stratiform vs. convective precipitation, precipitation intensity, and large-scale circulation, was evaluated.
