An assessment of the sea surface temperature influence on surface wind stress in numerical weather prediction and climate models

The ability of six climate models to capture the observed coupling between SST and surface wind stress in the vicinity of strong midlatitude SST fronts is analyzed. The analysis emphasizes air-sea interactions associated with ocean meanders in the eastward extensions of major western boundary current systems such as the Gulf Stream, Kuroshio, and Agulhas Current. Satellite observations of wind stress from the SeaWinds scatterometer on NASA's Quick Scatterometer and SST from the Advanced Microwave Scanning Radiometer clearly indicate the influence of SST on surface wind stress on scales smaller than about 30 longitude x 10 latitude. Spatially high-pass-filtered SST and wind stress variations are linearly related, with higher SST associated with higher wind stress. The influence of SST on wind stress is also clearly identifiable in the ECMWF operational forecast model, having a grid resolution of 0.35 x 0.35 degrees (T511). However, the coupling coefficient between wind stress and SST, as indicated by the slope of the linear least squares fit, is only half as strong as for satellite observations. The ability to simulate realistic air-sea interactions is present to varying degrees in the coupled climate models examined. The Model for Interdisciplinary Research on Climate 3.2 (MIROC3.2) high-resolution version (HIRES) (1.1 degrees x 1.1 degrees, T106) and the NCAR Community Climate System Model 3.0 (1.4 degrees x 1.4 degrees, T85) are the highest-resolution models considered and produce the most realistic air-sea coupling associated with midlatitude current systems. Coupling coefficients between SST and wind stress in MIROC3.2_HIRES and the NCAR model are at least comparable to those in the ECMWF operational model. The spatial scales of midlatitude SST variations and SST-induced wind perturbations in MIROC3.2_HIRES are comparable to those of satellite observations. The spatial scales of SST variability in the NCAR model are larger than those in the ECMWF model and satellite observations, and hence the spatial scales of SST-induced perturbations in the wind fields are larger. It is found that the ability of climate models to simulate air-sea interactions degrades with decreasing grid resolution. SST anomalies in the GFDL Climate Model 2.0 (CM2.0) (2.0 degrees x 2.5 degrees), Met Office Third Hadley Centre Coupled Ocean-Atmosphere General Circulation Model (HadCM3) (2.5 degrees x 3.8 degrees), and MIROC3.2 medium-resolution version (MEDRES) (2.8 degrees x 2.8 degrees, T42) have larger spatial scales and are more geographically confined than in the higher-resolution models. The GISS Model E20/Russell (4.0 degrees x 5.0 degrees) is unable to resolve the midlatitude ocean eddies that produce prominent air-sea interaction. Notably, MIROC3.2_MEDRES exhibits much weaker coupling between wind stress and SST than does the higher vertical and horizontal resolution version of the same model. GFDL CM2.0 and Met Office HadCM3 exhibit a linear relationship between SST and wind stress. However, coupling coefficients for the Met Office model are significantly weaker than in the GFDL and higher-resolution models. In addition to model grid resolution (both vertical and horizontal), deficiencies in the parameterization of boundary layer processes may be responsible for some of these differences in air-sea coupling between models and observations.