Testing an ocean carbon model with observed sea surface pCO 2 and dissolved inorganic carbon in the tropical Pacific Ocean

A basin‐scale carbon model for the tropical Pacific has been tested against in situ observations of CO 2 partial pressure (pCO 2 ) and dissolved inorganic carbon. Best agreement between model and observations occurs when gas exchange is enhanced at low wind speeds and when frictional smoothing as a parameterization of mesoscale eddy stirring/mixing is minimal. However, different realizations of the biological pump are not equally sensitive to the friction parameters, and it is not possible to completely isolate the effects of physics and biology. The model ocean shows substantial interannual variability in pCO 2 and CO 2 flux which is strongly correlated with the Multivariate ENSO Index. Interannual variability is similar to other models and suggests a relatively small role for the ocean in interannual variability of atmospheric CO 2 growth. There are significant areas where CO 2 remained supersaturated throughout the 1997–1998 El Niño and there was net outgassing from the “Wyrtki Box” at all times, but the net flux from the full model domain was near zero at the peak of the event. Testing the model against ship‐based observations produces a credible four‐dimensional field of the tropical ocean carbon system. Sampling this field with methods analogous to those used in empirical reconstructions of CO 2 flux suggests that those methods can underestimate the interannual variability by up to a factor of ∼2.5 depending on the grid resolution used. Models and observations are not currently adequate to state with confidence that undersampled mesoscale variability does not affect the variability of the regional aggregate flux.