Experiments with a pair of MRI coupled models,@in which the same atmospheric model is coupled to a global ocean model (AOGCM) and to a thermodynamical 50m slab ocean model (SGCM), suggests a close relationship between natural variability and CO2-induced warming. They show that a remarkable similarity is found in spatial patterns of the Pacific SST between the first empirical orthogonal function (EOF1) for natural variability and warming anomaly due to increased CO2, the polarity of the warming anomaly patterns referring to EOF1 is opposite between the AOGCM runs and the SGCM runs, and the selection of the polarity of the warming anomaly patterns is independent of initial conditions.
The present study discusses the evolution of the warming anomaly patterns and the selection mechanism of the polarity in the AOGCM experiment, where ocean dynamics is fully included. Here the AOGCM was integrated for 150 years with constant (345ppm) and gradually increasing (at 1% per year) CO2, and for 20 years with doubled CO2 starting from five different initial conditions derived from the 150-year control run (Monte Carlo method).
The EOF1 for the Pacific SST variability in the control run has an ENSO-like spatial pattern and temporal variation, capturing main features of the observed variability. The spatial pattern has a wedge-shaped anomaly in the equatorial Pacific and anomaly of opposite polarity in middle latitudes. Three-dimensional analyses of atmosphere and ocean in the Pacific area reveal that the similarity in the natural variability and CO2-induced warming anomaly is three-dimensionally coherent.
In the CO2 doubling ensemble experiment, a few years after the CO2 doubling, a stationary negative SST anomaly appears near the equator/date line and on the south-western side of the wedge-shaped pattern. However, the response on the north-eastern side is transitional and less stable during the 20 years. The response in the eastern Pacific is mainly caused by the solar radiation/stratus clouds feedback and wind stress/evaporation feedback, which are responsible for the response in the SGCM experiment. On the other hand, enhanced cumulus convection around the maritime continents strengthens the easterly wind stress in the western Pacific, which induce advection and upwelling of cold sea water and results in a La Nina-like response in the western Pacific. The latter effect dominates the former in the present AOGCM experiment, and hence the La Nina-like responses prevail over the whole Pacific region. Since this effect is absent in the SGCM, the present AOGCM experiment suggests that it is the most responsible for the difference in CO2-induced responses between the AOGCM and the SGCM experiments.