Sea surface height (SSH) anomalies from the altimeter have been shown to be good proxies for upper ocean heat content, where salinity effects are small. An upper ocean heat budget in the Gulf Stream region (Dong and Kelly, 2004) showed that interannual variations are dominated by contributions from advection and air-sea heat fluxes. Here, an estimate of advection is obtained by removing the contribution of known forcing terms and using a Kalman filter with an "unknown control" term on the residual SSH, similar to the procedure described in Kelly (2004).

The modeled effects on SSH are the static response to air-sea fluxes, the (flat-bottomed) Sverdrup response, and SSH effects of wind-forced Rossby waves in the thermocline. The estimates here are from simplified versions of the method used by Vivier et al. (1999).

A large contribution to errors in inferring advection comes from errors in the net surface heat flux. We used the net surface heat flux from the ECMWF ERA-40 Reanalysis (combined with a scaled version of the NCEP-2 Reanalysis fields after mid-2002). A lack of correlation between the SSH (Figure 1a) and the steric response (Figure 1b) suggests that the interannual variations in the ECMWF product are not realistic. Correlations with SSH were used to reduce the amplitude of the interannual variations in air-sea fluxes and therefore in the steric response (Figure 1c).

Westward propagating SSH anomalies dominate the SSH in the low and midlatitudes. A simple wind-forced Rossby wave model (with phase speed estimated empirically from the SSH) produces a reasonable estimate of the propagating SSH anomalies (Figure 2b). However, errors in the wind stress curl field are proportionately larger than those in the wind stress itself. To correct for these errors, SSH is assimilated into the simple model to produce an adjusted version of the propagating field (Figure 2c).

At latitudes of about 40-44N, the barotropic Sverdrup response becomes important (more than 5% of the nonsteric SSH variance) and it is removed.

Finally, advection and its effects on SSH is estimated from the residual SSH (after removing the steric, Rossby wave, and Sverdrup effects) using a Kalman filter. This analysis of advected SSH shows large-scale variations throughout the North Atlantic (Figure 3); these variations may be the result of wind-forced changes in ocean circulation.

Dong, S. and K.A. Kelly, 2004. Heat budget in the Gulf Stream region: the importance of heat advection and storage, J. Phys. Oceanogr., 34, 1214-1231. [Download manuscript]

Kelly, K. A., 2004. The relationship between oceanic heat transport and surface fluxes in the western North Pacific: 1970-2000, J. Climate, 17, 573-588. [Download manuscript]

Vivier, F., K. A. Kelly, and L. Thompson, 1999. Contributions of waves, wind forcing, and surface heating to sea surface height observations of the Pacific Ocean, JGR, 104,20,767--20,788.