Impacts of Falling Ice Radiative Effects on Projections of Southern Ocean Sea Ice Change under Global Warming

AbstractThe falling ice (snow) radiative effects (FIREs) have previously been shown to contribute substantially to reduced discrepancies in simulation of present-day climatology of radiation, skin temperatures and sea ice concentration and thickness over the Southern Ocean. This study extends the earlier effort to examine the potential impacts of FIREs on simulation of sea ice changes under a scenario of gradual increase of atmospheric CO2 concentration. We perform a pair of sensitivity experiments including (SoN) and excluding (NoS) FIREs using Community Earth System Model version 1. The differences in the annual and seasonal means between the initial and warmer periods are examined. Relative to SoN, NoS simulates more surface reflected shortwave and less downward longwave radiative warming, as well as colder surface temperature, resulting in larger annual-mean sea ice extent and thickness and slower seasonal and long-term sea ice melting and thinning. Over the Southern Ocean of SoN, reduced downwelling longwave radiation in austral winter (JJA) leads to sea-ice growth with colder surface temperature while reduced net radiation resulting from increased shortwave reflection in austral summer reduces the melting of sea ice with little change in skin temperature. NoS shows seasonal and long-term trends similar to those in CMIP5 models that exclude FIREs, hinting slower future warming-driven changes and larger amplitude of the annual cycle in sea ice concentration and thickness. The ice-free Southern Ocean in peak melting season is simulated at approximately year 130 for NoS but year 100 for SoN, about 30 years later than that of the Arctic.  1. IntroductionSouthern Ocean sea ice change is critically important for Earth’s global energy balance and atmosphere-ocean heat transport [IPCC AR5, 2013; Lefebvre and Goosse, 2007; Mahlstein et al., 2013; Maksym et al., 2012; Meijers, 2012; 2014; Stammerjohn et al., 2008]. Forty years of satellite observations showed a gradual, decades-long overall increase in Antarctic sea ice extents reversed in 2014, with subsequent rates of decrease in 2014–2017 far exceeding the more widely publicized deceased rates experienced in the Arctic [Parkinson, 2019]. Extents for 2017 and 2018 were the lowest on record for both austral winter maximum and summer minimum. In 2019, both the minimum and maximum extents fell below the 1981–2010 average, but neither was a record low for that time of year [Scott, 2019]. Understanding the processes behind these changes is vital to improve estimates of albedo feedbacks, i.e., the change in Earth’s net heat absorption from reflection changes in response to temperature change. In contrast to the small observed sea ice increases before 2014, most coupled global climate models (CGCMs) simulate decreased Antarctic sea ice extent over the past 30 years [e.g., Maksym et al., 2012; Li et al., 2017]. Unlike the Arctic ocean, the strength of the Antarctic Circumpolar Current (ACC), its surrounding belt of westerlies and upwelling cool water mean that its forced response to climate change is weak relative to internal variability [Armour et al., 2016]. Processes contributing to the sea ice change, in particular the earlier growth, include stronger cyclonic flow over West Antarctica, dynamical changes in Southern Annular Mode [SAM, Turner and Overland., 2009] and strengthening of the cold, southerly winds blowing northward from the Ross Ice Shelf, which could increase vertical mixing and cooling of waters off the continental shelf [Comiso, 2010]. Although the observed Ross Sea sector extent increased, it decreased in the Bellingshausen-Amundsen sector [Comiso et al., 2011]. Sea ice near West Antarctica was also found to be influenced by the Amundsen Sea Low [ASL, Raphael et al., 2016], which is being driven by ocean circulation changes. Therefore, the aforementioned evidence for large internal variability limits our ability to extract the forced response from observations. Many present-day CGCMs, including those in the Coupled Model Intercomparison Project, phase 5 (CMIP5), exhibit large spreads with nontrivial biases in Southern Ocean sea ice extent [Li et al., 2017; Taylor et al., 2012], with some simulating less than one third of the observed annual mean extent [Turner et al., 2013; Bracegirdle et al., 2008, 2015] and much larger seasonal changes [van den Broeke, 2004; Simmonds, 2015; Li et al., 2017], suggesting that sea-ice melting rates are unrealistically high in some CGCMs. Thus, there is low confidence in Antarctic sea ice projections [Maksym et al., 2012; Turner et al., 2013; Zunz et al., 2013; Li et al., 2017; Turner and Overland, 2009; Smith et al. 2014; Hosking et al., 2016]. Reported IPCC CMIP5 simulations project decreased sea ice extent between 1986–2005 and 2081–2100, with a mean decrease of 16—67 % in February (minimum sea ice area) and 8—30 % in September (maximum sea ice area), depending on the amount of global warming in the designated scenarios [IPCC AR5, 2013]. About 75% of CMIP5 models reach a nearly ice-free state in February before 2100 under the most extreme forcing scenario, the Representative Concentration Pathway 8.5 (RCP8.5). Only small portions of the Weddell and Ross Seas stay ice-covered in February during 2081–2100. The Antarctic and Southern Ocean climate systems are the result of complex interactions between external forcing, large-scale nonlinear climate dynamics and regional feedbacks. Given the wide spread of CMIP5 simulations and the importance of the Southern Ocean for climate feedbacks [Armour et al., 2013], it is important to assess the physical processes that can bias model projections. Identifying these bias sources and reducing them should contribute to reduce uncertainties in climate change projections. A physical understanding of the link from simulated historical sea ice changes to projected changes is also important because an accurate representation of observed sea ice extent is a necessary condition for producing realistic projections [Bracegirdle et al. 2015; Li et al., 2017]. A number of physical processes have been shown to contribute to differences in CGCM representations of the energy budget and sea ice in the Southern Ocean, including the abundance and brightness of clouds [Trenberth et al., 2010], the lack of supercooled liquid clouds [Cesana et al., 2012; McCoy et al., 2015; Kay et al., 2016a, b] and the importance of regional topography and bathymetry [Nghiem et al., 2016]. The representation of the cloud effects can impact local radiation and thus contribute to sea ice changes [Kay et al., 2016a], but it is a challenge for CGCMs to have the correct radiative representation of cloud and precipitating hydrometeors such as falling ice (snow). Because the atmospheric moisture holding capacity will increase in a warming climate, falling ice (snow) is expected to grow into the future [Medley et al. 2018], which may increase the radiative effect of snow that further impacts the sea ice extent and thickness. In majority of CMIP5 models, falling ice (snow) radiative effects (FIREs) are excluded, and in this study, we attempt to quantify the impact of FIREs to simulated Antarctic sea ice changes under global warming. An earlier study has shown that inclusion of FIRE reduces model-observation discrepancy of the present-day Southern Ocean sea-ice concentration [Li et al., 2017]. That study used controlled simulations with a climate model in which FIRE was enabled or not. With FIREs, the longwave radiation warming restricted wintertime sea ice growth, resulting in a lower summertime albedo, and lower sea ice extent continued throughout the ice melting season compared with a simulation that excluded FIREs. The inclusion of FIREs resulted in a reduced sea ice extent bias relative to observations by (.17 × 106 km2, 39%) in summer and (2.12 × 106 km2, 55%) in winter (JJA). In this study, this FIRE mechanism and its importance to the Antarctic and Southern Ocean sea ice projection will be extensively examined in the context of progressive global warming, in particular, assessing how FIRE may affect projected temperature and sea ice changes over the Southern Ocean, compared to the Arctic sea ice projection presented in Li et al. [2020]. They found that both surface temperature and net radiation changes are equally important to sea ice change in the Arctic. The model sensitivity experiments and analysis methods are described in section 2. In section 3, we explore the difference introduced in the model due to FIRE and we discuss and conclude major findings in section 4.