Warm Clouds Biases in CMIP6 Models Linked to Indirect effects of Falling ice-Radiation Interactions over the Tropical and Subtropical Pacific

Abstract We examine the spatial distributions of CMIP6-simulated cloud liquid water path (CLWP) and content (CLWC) against MODIS and CloudSat synthesized data over the tropical and subtropical Pacific. Three subsets of models are categorized based on their treatments of frozen ice-radiative interactions. CLWP/CLWC are generally well simulated in subset with separately-calculated radiative effects of cloud ice and falling ice (SON2). Too much warm clouds above 750 hPa are produced in either subset with total frozen ice (SON1) or no radiative effective effects of falling ice (NOS) and thus CLWP/CLWC are overestimated over the open ocean including the trade-wind regions. Stratocumulus clouds off the coasts of North and South America are severely underestimated in NOS models. We attribute the overestimates of clouds above the trade-wind boundary layers to anomalous ascending motion associated with warm sea surface temperature and weaker surface wind stress linked to indirect effects of falling ice-radiation interactions. The three key points: Key point #1: CloudSat-MODIS synthesized estimates of warm cloud liquid water path/content are used to evaluate three CMIP6 model subsets Key point #2: CLWP/CLWC are simulated well with separately-calculated radiative effects of cloud ice and snow over subtropical and tropical Pacific Key point #3: Too much warm clouds above 750 hPa produced in other two subsets leading to overestimates of CLWP/CLWC over open ocean and trade-wind regions Plain Language Summary We explore on how different models simulate frozen particles and their effects on the representation of warm clouds in the subtropical and tropical Pacific with three subsets of models in CMIP6. We found that models with a separated treatment of floating and falling ice radiative effects (SON2) performed better than models without falling ice radiative effects (NOS) or combined (SON1) in terms of biases of cloud liquid water content and cloud liquid water path against satellite measurements. The study suggests that large-scale environmental biases with anomalous ascending motion over warm sea surface temperature and weaker surface wind stress, may be causing an excess of warm clouds above the boundary layer in NOS and SON1, leading to excessive cloud liquid water. 1. Introduction How to accurately represent marine low-level clouds remains a challenge for global climate models (GCMs) (Bony and Dufresne, 2005; Mueller et al., 2011; Konsta et al., 2022; Mülmenstädt et al., 2021; Nam et al., 2012; Boucher et al. 2013). These clouds reflect a significant amount of solar radiation but emit amount of infrared radiation similar to that of the ocean surface. Increase/decrease in their coverage/liquid amount can either cool or warm the ocean surface. This study pertains to quantifying model’s disparity in cloud liquid water path and content (CLWP and CLWC) simulation, which helps to increase the understanding of cloud-climate feedbacks. There has been a wide disagreement in CLWP and CLWC simulations among the Coupled Model Intercomparison Project Phase 5 (CMIP5) models (Li et al., 2008, 2011, 2018; Lin et al., 2014) and Phase 6 (CMIP6) models (Jian et al., 2021). For example, Li et al. (2018) found significant biases with overestimated CLWP by factors of 2-10 compared to observational estimates and most models showed significant systematic biases in CLWC vertical profile with overestimates occurring at all levels above 700 hPa. In CMIP5, most GCMs ignored the radiative effects of precipitating (falling) ice (Gettelman et al., 2010, 2015; Li et al., 2012, 2013, 2014a,b; Michibata et al., 2019; Waliser et al., 2009, 2011). Without falling ice radiative effects (FIREs), models tend to have excessive upper-level outgoing longwave radiative cooling that makes atmospheric columns more unstable over the intertropical convergence zone (ITCZ), South Pacific Convergence Zone (SPCZ) and tropical western Pacific (TWP). This leads directly to extra upward motion with excessive downward motion and anomalously low surface wind stress due to downdraft-induced outflows away from convective zones into the trade-wind regions, leading to warmer sea surface temperatures (SSTs) on the flanks of the ITCZ. A consequence, that is, the indirect effects of FIREs, is the increased middle- and high-level clouds with excessive precipitation biases (parts of double-ITCZ) over the trade-wind regions (Li et al., 2021). Li et al. (2021) showed that no-FIREs simulation has a cyclonic-like wind pattern anomaly in the lower-troposphere relative to with-FIREs simulation with weaker subsidence over the northeast side, leading to moisture convergence, horizontal warm and moist advections, and stronger effective ascending motion. These differences favor the occurrence of middle- and high-clouds instead of shallow cumulus clouds in the trade-wind regions where in nature SST is cooler and surface wind stress is stronger. CMIP6 has more models with FIREs included (Li et al., 2020b), which exhibit reduction in low-level outflow over ITCZ/SPCZ and increase in surface wind stress in the trade-wind regions, leading to colder SSTs on the flanks of the ITCZ and reduced cloud and precipitation biases over the trade-wind regions. Li et al. (2020c, 2022a, 2022b) further divided 23 CMIP6 models into three subsets: those without FIREs (referred to as NOS) and those with FIREs using separate (referred to as SON2) and combined (referred to as SON1) floating and falling frozen hydrometeors to compute the optical properties of frozen hydrometeors. Li et al. (2020c) found that SON mitigated part of the long-standing double-ITCZ problem in precipitation field. Li et al. (2022a) showed that SON2 significantly improved radiation fields compared to SON1 and NOS, which led to reduced positive SST biases and surface wind stress biases over the trade-wind regions (Li et al., 2022b). The difference in treating ice-cloud radiative properties between SON2 and SON1 seems to be a significant factor impacting the performance of model group although individual models differ greatly in many other aspects. This study focuses on evaluating CLWP and CLWC in the above-mentioned three subsets of CMIP6 models against CloudSat-MODIS synthetized estimates developed by Li et al. (2018), with an emphasis on the indirect effects of frozen hydrometeors-radiation interactions over the tropical and subtropical Pacific. The CloudSat-MODIS synthetized data are described in Section 2. Model subsets and liquid water hydrometeor output are explained in Section 3. Results are presented in Section 4. Summary of results and further discussions are given in Section 5.