Linking Global Land Surface Temperature Projections to Radiative Effects of Hydrometeors under a Global Warming Scenario

AbstractLand skin temperature (Ts) is directly influenced by surface energy balance, in particular, radiative energy, which can be linked to model’s representation of radiative effects of hydrometeors in the atmosphere. This indirect link is inferred by examining the changes of geographical distribution and seasonal cycle of surface radiation, surface turbulent fluxes and Ts between a pair of 140-yr sensitivity experiments under 1% per year increase of atmospheric CO2. One is with radiative effects of falling ice (snow) hydrometeors on (SON) and the other off (NOS) using CESM1-CAM5 and the results are compared with CMIP5 models without these effects. For boreal winter, NOS relative to SON simulates less surface downward longwave and net flux (~10—15 W m-2), resulting in colder Ts, over mid- and high latitudes (~2—3 K colder), but more solar radiative flux, resulting in warmer Ts, over subtropical and tropical land (~1—3 K). These differences between NOS and SON are amplified as climate warms. The results from CMIP5 ensemble generally match with those of NOS. Temporal correlation analysis indicates that the indirect linkage between Ts and falling ice hydrometeor changes is through one between Ts and downward longwave and net fluxes in high latitudes, but strongly weakened by shortwave changes in low latitudes (and boreal summer) due possibly to changes in other hydrometeors. NOS and CMIP5 Ts are underestimated throughout the seasonal cycle but with smaller differences in summer from SON. The seasonal cycle of Ts is likely not solely determined by that of net radiative flux in CMIP5 models.1. IntroductionLand surface energy balance and its changes are parts of a complex land-atmosphere interactive system. They are a key factor for the climate change, which is influenced by the net surface radiative fluxes at a variety of temporal and spatial scales (Susskind 1993; Jin et al., 1997; Pitman, 2003; Knutson et al., 2013; Brooks et al., 2015). The land radiative skin (surface) temperature is primarily controlled by the atmospheric radiative fluxes at the skin interface, which acts as an upper boundary that constrains the land surface energy and hydrological cycle and then influences surface turbulent fluxes (latent and sensible) and surface precipitation (Gu et al., 2006; Koster et al., 2009; Haghighi et al., 2018; Schwingshackl et al., 2017; Seneviratne et al., 2006, 2010). Despite that land-related processes within land surface models have been advanced with the help of satellite measurements of, for example, albedo, emissivity, and leaf area index (Pitman, 2003), they still suffer from the uncertainties in representations of physical processes from the atmospheric components of coupled global climate models (GCMs) such as clouds, precipitation and their interactions with radiation (Randall et al., 2007; Stephens et al., 2005; 2012a,b; 2015; Trenberth and Fasull, 2009). Land surface properties such as temperature and moisture are changed with those of the upper boundary conditions from atmospheric radiation forcing. It is therefore critical to investigate the impacts from the upper boundary conditions including clouds, surface radiative flux on land surface properties (Lo and Famiglietti, 2013; Anderson et al., 2015; Wey et al., 2015). As discussed below, the lack of the fidelity in the present-day CMIP (Coupled Model Intercomparison Project) model land surface simulations can lower the confidence level for future projection of land surface properties (Hua et al., 2014; Li et al., 2016a,b). It has been reported that modeled land surface temperature of the present-day climate simulations in CMIP Phase 5 (CMIP5; Taylor et al., 2002) seems to be underestimated in mid- and high-latitudes in boreal winter season but biased warm in low- and mid-latitudes, especially over arid and semiarid regions in boreal summer season, attributed mainly to the discrepancies of the net shortwave radiation and surface longwave radiation (Hua et al., 2014; Li et al., 2016a,b). That is, overestimation of downward shortwave radiation at the surface (RSDS) and underestimation of downward longwave radiation (RLDS) occuring over most continental regions, in particular, during boreal winter. Li et al. (2016a) discussed the key role of treating radiative effects of both floating ice (cloud ice) and falling ice (snow) hydrometeors in simulating the seasonal cycle of land skin temperature variations with sensitivity experiments. They found that such treatments are more important for the boreal winter season than the other seasons in the present-day climate simulations. This is mainly because downward longwave radiation increases due to the greenhouse effect of snow hydrometers, which in turn restricts the cooling of land surface when there is less solar radiation during winter. Since land surface temperature and surface energy budgets are critical factors that control near surface air temperature (SAT) through surface layer turbulent fluxes, their biases in most current GCMs with treatment of radiative effect of floating ice hydrometeor only impose non-trivial effects on modeled land-atmosphere interactions, soil layer and land hydrological processes through soil moisture and temperature changes (Li et al., 2016b). These biases in the present-day climate simulations may influence the confidence level for projecting future land surface properties. Only a few CMIP5 GCMs treat radiative effects of falling ice hydrometeor, including the National Center for Atmospheric Research-Department of Energy (NCAR-DOE) CESM1-CAM5 (CESM1-CAM5; Neale et al., 2012) and HadGEM2 (Martin et al., 2011). Following Li et al. (2016a), in this study, we will use CMIP5 outputs to analyze the linkage of the projected land surface temperature with radiative effects of hydrometeors via surface energy budget. We focus on the physical process related to the falling ice (snow) radiative effects (FIREs) using controlled CESM1-CAM5 simulations, with an emphasis on the geographic distribution changes of projected land surface temperature and the associated relationships with surface energy budget components under a global warming scenario. We discuss how inclusion of FIREs, compared to exclusion of FIREs, can substantially change the simulated surface energy budgets and land surface temperature focusing on boreal winter (December, January and February, DJF), which is chosen due to the strongest signal in the seasonal cycle, and how the seasonal cycle of changes are impacted by FIREs for a few selected regions. In section 2, we describe the controlled simulations and analysis methods. Results are presented in section 3. We discuss and conclude major findings in section 4.