The optimal carbonate dissociation constants for determining surface waterpCO(2) from alkalinity and total inorganic carbon

In many numerical ocean chemistry models, total dissolved inorganic carbon, DIC and total alkalinity, TA are transported between subsurface boxes. The partial pressure pCO(2) is subsequently calculated from TA and DIC in the surface box in order to account for air-sea exchange of carbon dioxide. The conversion is commonly performed by solving the thermodynamic relationship s for equilibria between carbonate, bicarbonate, and aqueous CO2 using carb onate dissociation constants. Four independent determinations of the consta nts have been made for seawater in the past 50 years. These results have be en corrected, refit, and combined by others creating a virtual cottage indu stry of laboratory and field verification, and cross-checks. Here, we show that, based on field observations in three ocean basins, the calculated sur face water pCO(2) from TA and DIC corresponds best with the measured pCO(2) if the constants proposed by Mehrbach et al. [Mehrbach, C., Culberson, C.H ., Hawley, J.E., Pytkowicz, R.M., 1973. Measurement of the apparent dissoci ation constants of carbonic acid in seawater at atmospheric pressure. Limno logy and Oceanography 18, 897-907] as refit by Dickson and Millero [Dickson , A.G., Millero, F.J., 1987. A comparison of the equilibrium constants for the dissociation of carbonic acid in seawater media. Deep-Sea Res. 34, 1733 -1743] are used. This suggests that the K-1/K-2 ratio of Mehrbach et al. fo r a pCO(2) range of 280-450 mu atm is correct but it does not necessarily i mply that the absolute values of K-1 and K-2 are correct or that these cons tants work as well at higher pCO(2). An independent cross-check with pCO(2) measurements made at constant temperature (20 degrees C) is in agreement w ith these conclusions. The differences in calculated pCO(2) can be as great as 30 mu atm depending which constants are used. If the models are forced with a prescribed atmospheric pCO(2) they will end up having surface concen tration of TA that are up to 20 mu eq kg(-1) too high or DIC concentration up to 20 mu mol kg(-1) too low if constants other than those of Mehrbach et al. are used. This makes comparisons between observation and models proble matic. (C) 1999 Published by Elsevier Science B.V.