Abstract
The authors present a study for determining the seasonal net surface heat flux over the tropical Pacific Ocean using an adjoint technique. A simple tropical ocean model with thermodynamics is chosen and the seasonal sea surface temperature (SST) observations are assimilated. A least-squares fitting of the model state to data is used. The cost function has a misfit term that measures the difference between the modeled and observed SST and two additional terms that penalize the departure of the estimated parameters from their prior information.
The adjoint method ensures that the flux pattern obtained is consistent with the model's dynamics and thermodynamics and is also in agreement with observations. Comparisons with heat flux atlases of Oberhuber and Fu et al. show that our adjoint calculations have successfully captured the main seasonal signals of the surface heat flux distribution over the tropical Pacific Ocean although, not surprisingly, some differences exist. The differences are examined from both thermodynamic and air-sea interaction viewpoints. Two experiments are conducted to study the effects of the prior information on the optimal solution.
