Pacific Subtropical Cell Response to Reduced Equatorial Dissipation

Abstract

Abstract Equatorial turbulent diffusivities resulting from breaking gravity waves may be more than a factor of 10 less than those in the midlatitudes. A coupled general circulation model with a layered isopycnal coordinate ocean is used to assess Pacific climate sensitivity to a latitudinally varying background diapycnal diffusivity with extremely low values near the equator. The control experiments have a minimum upper-ocean diffusivity of 10−5 m2 s−1 and are initialized from present-day conditions. The average depth of the σθ = 26.4 interface (z26.4) in the Pacific increases by ∼140 m after 500 yr of coupled model integration. This corresponds to a warming trend in the upper ocean. Low equatorial diffusivities reduce the z26.4 bias by ∼30%. Isopycnal surfaces are elevated from the eastern boundary up to midlatitudes by cooling in the upper several hundred meters, partially compensated by freshening. Entrainment of intermediate water masses from below σθ = 26.4 decreases by ∼1.5 Sv (1 Sv ≡ 106 m3 s−1), mainly in the western tropical Pacific. The Pacific heat uptake (30°S–30°N) from the atmosphere reduces by ∼0.1 PW. This is associated with warmer entrainment temperatures in the eastern equatorial Pacific upwelling region. Equatorward heat transport from the Southern Ocean increases by ∼0.07 PW. Reducing the upper-ocean background diffusivity uniformly to 10−6 m2 s−1 cools the upper ocean from the tropics, but warms and freshens from the midlatitudes. Enhanced convergence into the Pacific of water lighter than σθ = 26.4 compensates the reduction in upwelling of intermediate waters in the tropics. Basin-averaged z26.4 bias increases in the low background case. These results demonstrate basin-scale sensitivity to the observed suppression of equatorial background dissipation. This has clear implications for understanding oceanic heat uptake in the Pacific as well as other important aspects of the climate system. Diapycnal diffusivities due to truncation errors and other numerical artifacts in ocean models may need to be less than 10−6 m2 s−1 in order to accurately represent this effect in climate models.