The Coupled Ocean—Atmosphere Modeling Problem in the Tropical Pacific and Asian Monsoon Regions

Abstract

Two ocean formulations, one a simple, 50-m slab ocean and another a coarse-resolution global ocean general circulation model (GCM), are coupled to a global atmospheric GCM. To determine what part of the simulation errors is introduced by the atmospheric model and what part arises from limitations of the ocean formulation, results are compared to observations as well as to integrations involving the ocean GCM run with observed atmospheric forcing and the atmospheric GCM run with specified observed sea surface temperatures (SSTs). The tropical Indian and Pacific regions are studied because of the associations involving the dynamically coupled ocean-atmosphere system in those regions related to large-scale tropics and global interannual variability. Analysis of the net surface heat flux leads to the conclusion that limitations in the ocean formulations contribute more to errors in the coupled climate simulations than inherent deficiencies in the atmospheric model. In spite of the limitations of the ocean formulations in simulating SST, the atmospheric model simulates most major features associated with the low-level wind fields in the tropical Indian and Pacific regions with differences consistent with the SSTs supplied by the ocean models. The implication is that increased quality of the ocean simulation will result in substantial improvements of the coupled climate model simulations even without upgrades to the atmospheric model. The point is raised that a coupled model with surface fluxes computed interactively produces a more internally consistent climate simulation than could be expected from the same atmospheric model forced with observed SSTs, even if the SSTs computed by the coupled model do not exactly match the observed values. Surface fluxes, then, are not ?absolute? values and must be interpreted as compensatory products of the limitations (and strengths) of the respective media in simulating SST patterns. Thus, even the present generation of imperfect coupled models can provide insight into some of the processes, mechanisms, and sensitivities of the coupled climate system that no other research tool can.