A Discussion on the Errors in the Surface Heat Fluxes Simulated by a Coupled GCM

This paper contrasts the sea surface temperature (SST) and surface heat flux errors in the Tropical Pacific simulated by the University of California, Los Angeles, coupled atmosphere–ocean general circulation model (CGCM) and by its atmospheric component (AGCM) using prescribed SSTs. The usefulness of such a comparison is discussed in view of the sensitivities of the coupled system. Off the equator, the CGCM simulates more realistic surface heat fluxes than the AGCM, except in the eastern Pacific south of the equator where the coupled model produces a spurious intertropical convergence zone. The AGCM errors are dominated by excessive latent heat flux, except in the stratus regions along the coasts of California and Peru where errors are dominated by excessive shortwave flux. The CGCM tends to balance the AGCM errors by either correctly decreasing the evaporation at the expense of cold SST biases or erroneously increasing the evaporation at the expense of warm SST biases. At the equator, errors in simulated SSTs are amplified by the feedbacks of the coupled system. Over the western equatorial Pacific, the CGCM produces a cold SST bias that is a manifestation of a spuriously elongated cold tongue. The AGCM produces realistic values of surface heat flux. Over the cold tongue in the eastern equatorial Pacific, the CGCM simulates realistic annual variations in SST. In the simulation, however, the relationship between variations in SST and surface latent heat flux corresponds to a negative feedback, while in the observation it corresponds to a positive feedback. Such an erroneous feature of the CGCM is linked to deficiencies in the simulation of the cross-equatorial component of the surface wind. The reasons for the success in the simulation of SST in the equatorial cold tongue despite the erroneous surface heat flux are examined.