Pacific Warm Pool Temperature Regulation during TOGA COARE: Upper Ocean Feedback*
Additional Document Info
The Hasselmann feedback model was applied to hindcast western Pacific warm pool sea surface temperatures (SST) with heat flux observations obtained near 2 degrees S, 156 degrees E from October 1992 to February 1993 during the Tropical Ocean Global Atmosphere Coupled Ocean-Atmosphere Response Experiment (TOGA COARE). The model versus observed SST correlations were greater than 0.85. Two important feedback (or damping) timescales emerged, with e-folding times of lambda(-1) = 0.2 days and 8 days, fitting to the diurnal and subdiurnal variations, respectively. Distinct mixed layer depth scales were also found for the respective timescales. A time-varying depth parameter with a median of similar to 5 m was derived for the shorter timescale and varied with the observed daily minimum mixed layer depth. A constant similar to 16 m was optimal for the longer timescale, which is similar to the time-averaged observed mixed layer depth of 14.8 In and the Monin-Obukhov scale of similar to 17 m. This bears on the choice of mixed layer parameters for climate model simulations of warm pool conditions observed in TOGA COARE, The low-frequency time- and depth-scale parameters give a negative feedback of about 95 W m(-2)degrees C-1, which is significantly greater than previous studies have indicated, This restoring influence was treated separately from fluxes across the air-sea interface such as latent, radiative, and sensible heat loss or cloud shading, and is thus attributed to oceanic mixed layer processes. The frequency band where the damping or feedback becomes important is defined by omega approximate to lambda, which is found to coincide with the diurnal cycle and the similar to 50-day Madden-Julian oscillations for the respective lambda(-1) timescales. This indicates a possible dynamic connection between the surface heat forcing and mixed layer dissipation timescales, which the authors suggest might be accounted for if the dissipation is parameterized as being proportional to the amplitude of SST variations.