Impacts of Model Horizontal Resolution on Mean Sea Surface Temperature Biases in the Community Earth System Model
Gaopeng Xu,Ping Chang,Sanjiv Ramachandran,Gokhan Danabasoglu,Stephen Yeager,Justin Small,Qiuying Zhang,Zhao Jing,Lixin Wu
DOI: https://doi.org/10.1029/2022jc019065
2022-01-01
Journal of Geophysical Research Oceans
Abstract:Impacts of model horizontal resolution on sea surface temperature (SST) biases are studied using high-resolution (HR) and low-resolution (LR) simulations with the Community Earth System Model (CESM) where the nominal resolutions are 0.1 degrees for ocean and sea-ice and 0.25 degrees for atmosphere and land in HR, and 1 degrees for all component models in LR, respectively. Results show that, except within eastern boundary upwelling systems, SST is warmer in HR than LR. Globally averaged surface ocean heat budget analysis indicates that 1 degrees C warmer global-mean SST in HR is mainly attributable to stronger nonlocal vertical mixing and shortwave heat flux, with the former prevailing over the latter in eddy-active regions. In the tropics, nonlocal vertical mixing is slightly more important than shortwave heat flux for the warmer SST in HR. Further analysis shows that the stronger nonlocal mixing in HR can be attributed to differences in both the surface heat flux and shape function strength used in the parameterization. In addition, the shape function shows a nonlinear relationship with surface heat flux in HR and LR, modulated by the eddy-induced vertical heat transport. The stronger shortwave heat flux in HR, on the other hand, is mainly caused by fewer clouds in the tropics. Finally, investigation of ocean advection reveals that the improved western boundary currents in HR also contribute to the reduction of SST biases in eddy-active regions. Plain Language Summary Sea surface temperature (SST) is a key climate variable, through which the atmosphere and ocean are coupled. However, current generation climate models that have nominal horizontal resolutions of similar to 1 degrees generally produce colder-than-observation SST over much of the tropics and midlatitudes. Increasing model resolution to 0.25 degrees or finer can help reduce this cold bias, but the underlying physics is not well understood. By analyzing high-resolution (HR) and low-resolution (LR) Community Earth System Model (CESM) simulations, we find that vertical turbulent processes, particularly the convective component, show dominant contributions to the improved SST bias in the tropics and midlatitudes. The difference in the convective flux between HR and LR CESM is partially attributable to the difference in the surface heat flux, and the other important contributing factor is the warmer subsurface temperature in HR CESM driven by the vertical eddy heat transport. Overall, this study highlights the importance of improving representation of ocean eddy and turbulent processes in climate models.