A Framework to Decompose Wind-Driven Biases in Climate Models Applied to CCSM/CESM in the Eastern Pacific

Abstract

Annual cycle biases in climate models are suspected to be largely wind driven along the equator, with winds first driving SST changes that then influence the overlying atmospheric circulation. This study utilizes an experimental approach to test the hypothesis that seasonally varying climatological wind stress directly contributes to the SST and ITCZ biases in the eastern equatorial Pacific. Results show that removing the wind stress annual cycle from the ocean forcing, without constraining the atmosphere and ocean dynamics or buoyancy coupling in the NCAR CCSM4/CESM1.2.0 models, results in a remarkable reduction in the SST annual cycle and springtime ITCZ biases. Improvements in the SST occur primarily because wind-driven errors in the variability of horizontal temperature advection are damped. The ITCZ problem is closely tied to biases in the wind-driven near-equatorial SST. Additional model experiments and analyses reveal that the contributions from zonal and meridional wind stress to the biases are locally forced within 10°S–10°N and additive, suggesting that the biases are driven by independent processes. The zonal and meridional components drive different aspects of the SST annual cycle bias and contribute to the springtime ITCZ bias in different zonal locations. Both the atmosphere and ocean components of the model, separately, are shown to produce unfavorable ocean surface conditions for the simulation of a realistic springtime ITCZ, deeming this a coupled problem. Results show that wind stress may act as a pathway for process-based errors in climate models to directly drive SST and ITCZ biases.