| description abstract | The present generation of global coupled ocean?atmosphere GCMs contains considerable systematic errors both in terms of net surface heat flux and simulated SSTs. Here, a global coupled GCM is used to illustrate how systematic errors in the separate coupled model components (atmosphere and ocean) contribute to the simulations of net surface heat flux and SST when the components are coupled together. Features of the coupled model simulation are a combination of errors in the component models and errors introduced due to the dynamic interaction, both local and nonlocal, between atmosphere and ocean. Various regions and latitudinal zones are examined to determine the processes that produce the net surface heat fluxes and SSTs in the coupled simulation. In the coupled model, a good simulation of net surface heat flux does not always produce a correspondingly accurate simulation of SST. Alterations of surface winds and/or ocean currents can introduce SST errors and consequent compensating surface fluxes that have apparent agreement with observed estimates (e.g., near 60°N in the North Atlantic). Additionally, an SST error that occurs due to a combination of surface flux errors from atmosphere and ocean components in the coupled simulation, as well as an alteration of the ocean surface currents, can produce a better agreement of the net surface fluxes in the coupled model with observed estimates in spite of the large SST errors (e.g., near 50°N in the Atlantic and Pacific). Conversely, a good simulation of SST in the coupled model can be associated with surface heat flux errors also due to dynamic adjustments in the atmosphere and ocean in the coupled simulation (e.g., near 20°N and 20°S). A high-quality coupled model simulation does not necessarily require a precise reproduction of observed net surface heat fluxes, even though accurate observed surface fluxes are necessary to calibrate model parameterizations in the components and to provide an index of model performance. Rather, improved coupled model simulations must rely on improvement of the entire thermodynamic and dynamic simulations (and verification of state variables) in the components separately and when coupled. | |
