The tropical eastern Pacific seasonal cycle: Assessment of errors and mechanisms in IPCC AR4 coupled ocean - Atmosphere general circulation models

Warmer SST and more rain in the Northern Hemisphere are observed year-round in the tropical eastern Pacific with southerly wind crossing the equator toward the atmospheric heating. The southerlies are minimal during boreal spring, when two precipitation maxima straddle the equator. Fourteen atmosphere ocean coupled GCMs from the Coupled Model Intercomparison Project (CMIP3) and one coupled regional model are evaluated against observations with simple metrics that diagnose the seasonal cycle and meridional migration of warm SST and rain. Intermodel correlations of the metrics elucidate common coupled physics. These models variously simulate the climatology of SST and ITCZ rain. In 8 out of 15 models the ITCZ alternates symmetrically between the hemispheres with the seasons. This seasonally alternating ITCZ error generates two wind speed maxima per year -one northerly and one southerly -resulting in spurious cooling in March and a cool SST error of the equatorial ocean. Most models have too much rain in the Southern Hemisphere so that SST and rain are too symmetric about the equator in the annual mean. Weak meridional wind on the equator near the South American coast (2 degrees S -2 degrees N, 80 degrees-90 degrees W) explains the warm SST error there. Northeasterly wind jets blow over the Central American isthmus in winter and cool the SST in the eastern Pacific warm pool. In some models the strength of these winds contributes to the early demise of their northern ITCZ relative to observations. The February -April northerly wind bias on the equator is correlated to the antecedent December -February Central American Pacific wind speed at -0.88. The representation of southern-tropical stratus clouds affects the underlying SST through solar radiation, but its effect on the meridional atmospheric circulation is difficult to discern from the multimodel ensemble, indicating that errors other than the simulation of stratus clouds are also important for accurate simulation of the meridional asymmetry. This study identifies several features to be improved in atmospheric and coupled GCMs, including the northeasterly cross-Central American wind in winter and meridional wind on the equator. Improved simulation of the seasonal cycle of meridional wind could alleviate biases in equatorial SST and improve simulation of ENSO and its teleconnections.