Analysis of General Circulation Model Sea-Surface Temperature Anomaly Simulations Using a Linear Model. Part II: Eigenanalysis

Abstract

An eigenanalysis of the barotropic vorticity equation linearized about 300 mb climatological flow from a general circulation model (the NCAR CCM) control simulation is described. The goal is to determine if the known behavior of the Community Climate Model (CCM) during equatorial Pacific sea-surface temperature anomaly experiments can be interpreted in terms of the linear modes calculated in the analysis. Given that modes which would be expected to dominate steadily forced flow would either have rapid growth rates or neutral growth rates and long periods, inspection of the spectrum that results from the eigenanalysis suggests no one mode is likely to be markedly more important than all others. However, the structures of some of the leading modes are similar to anomalies produced by sea-surface temperatures in the CCM. This is especially true of the most rapidly growing mode, which is also more sensitive to steady forcing than any other mode. The (undamped) e-folding time of the fastest growing mode, 15 days, is rather lengthy. Using the eigenfunctions of the adjoint system, some of the forced linear solutions, which Part I of this paper showed to resemble CCM solutions, are expanded in the eigenbasis. Those modes with long periods control these solutions but no one mode dominates. The structure of the adjoint eigenfunction which corresponds to the fastest growing eigenfunction has most of its amplitude in southern Asia and the North Indian Ocean. The fastest growing mode, which relies more on meridional than zonal gradients for its energy source, is shown to be rather insensitive to details of the model formulation.