An accurate representation of the climatology of the coupled ocean-atmosphere system in global climate models has strong implications for the reliability of projected climate change inferred by these models. Our previous efforts have identified substantial biases of ocean surface wind stress that are fairly common in two generations of the Coupled Model Intercomparison Project (CMIP) models, relative to QuikSCAT climatology. One of the potential causes of the CMIP model biases is the missing representation of large frozen precipitating hydrometeors (i.e., cloud snow) in all CMIP3 and most CMIP5 models, which has not been investigated previously. We examine the impacts of cloud snow on the radiation and atmospheric circulation, air-sea fluxes, and explore the implications to common biases in CMIP models using the National Center for Atmospheric Research coupled Community Earth System Model (CESM) to perform sensitivity experiments with and without cloud snow radiative effects. This study focuses on the impacts of cloud snow in CESM on ocean surface wind stress and air-sea heat fluxes, as well as their relationship with sea surface temperature (SST) and subsurface ocean temperatures in the Pacific sector. It is found that inclusion of the cloud snow parameterization in CESM reduces the surface wind stress and upper ocean temperature (including SST) biases in the tropical and midlatitude Pacific. The differences in the upper ocean temperature with and without the cloud snow parameterization are consistent with the effect of different strength of vertical mixing due to ocean surface wind stress differences but cannot be explained by the differences in net air-sea heat fluxes.

The potential links between ice water path (IWP), radiation, circulation, sea surface temperature (SST), and precipitation over the Pacific and Atlantic Oceans resulting from the falling ice radiative effects (FIREs) are examined from Coupled Model Intercomparison Project phase 5 (CMIP5) and phase 6 (CMIP6) historical simulations. The latter is divided into two subsets with (SON6) and without FIREs (NOS6) in CMIP6. Improvement in nonfalling cloud ice (~20 g m−2) is noticeable over convective regions in CMIP6 relative to CMIP5. The inclusion of FIREs in SON6 subset may contribute to reduce biases of overestimated outgoing longwave radiation and downward surface shortwave and underestimated reflected shortwave at the top of the atmosphere (TOA) by magnitudes of ~8 W m−2 over convective regions against CERES, compared to NOS6 subset. The reduced biases in radiative fluxes in convective regions stabilize the atmosphere and lead to circulation, SST, cloud, and precipitation changes over the trade wind regions, as seen from improved radiative fluxes (~15 W m−2), surface wind stress biases, SST (~0.8 K), and precipitation (1 mm day−1) biases. The significant improvement from NOS6 to SON6 leads to improved multimodel means for CMIP6 relative to CMIP5 for radiation fields over the trade wind regions but the degradation over convective zones is attributed to NOS6 subset. The results suggest that other sources of uncertainty and deficiencies in climate models may play significant roles for reducing discrepancies although FIREs, via radiation‐circulation coupling, may be one of the factors that help to reduce regional biases.

An accurate representation of the land surface temperature (LST) climatology of the coupled land‐atmosphere system has strong implications for the reliability of projected land surface processes and their variability inferred by the global climate models (GCMs) contributed to the Intergovernmental Panel on Climate Change CMIP5. We have identified a substantial underestimation of the total ice water path and biases of surface radiation budget commonly seen in the CMIP models which are highly correlated to the biases of LST over land. One of the potential causes of the CMIP model biases is the missing representation of large frozen precipitating hydrometeors and their radiative effects (i.e., snow) in all CMIP3 and most CMIP5 models. We examine the impacts of snow on the radiation, all‐sky and clear‐sky LST, and air‐land heat fluxes to explore the implications to the common biases in CMIP models by performing sensitivity experiments with and without snow radiation effects using the National Center for Atmospheric Research Community Earth System Model version 1. It is found that an exclusion of the snow radiative effects the CESM1 generates the LST biases (up to 2–3 K) in the midlatitude and high latitude, in particular, in December, January, and February (DJF). All‐sky and clear‐sky LST in model simulations are found to be too cold and are mainly due to underestimated downward surface (longwave) LW radiation in DJF, which is consistent with those in CMIP models. The correlation between the changes of the LST and downward surface LW radiation is very high both in summer and winter seasons.

Most of the global climate models (GCMs) in the Coupled Model Intercomparison Project, phase 5 do not include precipitating ice (aka falling snow) in their radiation calculations. We examine the importance of the radiative effects of precipitating ice on simulated surface wind stress and sea surface temperatures (SSTs) in terms of seasonal variation and in the evolution of central Pacific El Niño (CP‐El Niño) events. Using controlled simulations with the CESM1 model, we show that the exclusion of precipitating ice radiative effects generates a persistent excessive upper‐level radiative cooling and an increasingly unstable atmosphere over convective regions such as the western Pacific and tropical convergence zones. The invigorated convection leads to persistent anomalous low‐level outflows which weaken the easterly trade winds, reducing upper‐ocean mixing and leading to a positive SST bias in the model mean state. In CP‐El Niño events, this means that outflow from the modeled convection in the central Pacific reduces winds to the east, allowing unrealistic eastward propagation of warm SST anomalies following the peak in CP‐El Niño activity. Including the radiative effects of precipitating ice reduces these model biases and improves the simulated life cycle of the CP‐El Niño. Improved simulations of present‐day tropical seasonal variations and CP‐El Niño events would increase the confidence in simulating their future behavior.

Southern Ocean sea-ice cover exerts critical control on local albedo and Antarctic precipitation, but simulated Antarctic sea-ice concentration commonly disagrees with observations. Here we show that the radiative effects of precipitating ice (falling snow) contribute substantially to this discrepancy. Many models exclude these radiative effects, so they underestimate both shortwave albedo and downward longwave radiation. Using two simulations with the climate model CESM1, we show that including falling-snow radiative effects improves the simulations relative to cloud properties from CloudSat-CALIPSO, radiation from CERES-EBAF and sea-ice concentration from passive microwave sensors. From 50-70°S, the simulated sea-ice-area bias is reduced by 2.12 Â 10 6 km 2 (55%) in winter and by 1.17 Â 10 6 km 2 (39%) in summer, mainly because increased wintertime longwave heating restricts sea-ice growth and so reduces summer albedo. Improved Antarctic sea-ice simulations will increase confidence in projected Antarctic sea level contributions and changes in global warming driven by long-term changes in Southern Ocean feedbacks.