3.5 Internal Consistency Checks
For cruises in which more than two inorganic carbon system parameters were measured, it is possible to assess the agreement of the inorganic carbon data by an internal consistency check. With knowledge of the apparent carbonate dissociation constants and two carbon system parameters, the other two can be calculated. Although there remains controversy about the accuracy of the carbonate-dissociation constants, there is increasing evidence that those of Merhbach et al. (1973) as refit by Dickson and Millero (1987) yield consistent results (Lee et al. 2000, Millero et al. submitted). Internal consistency checks were performed for cruises where three or four parameters were measured. The results are presented in Table 10.
Of the cruises listed in Table 10, A06 and A07 are not recommended for use in the synthesis because of inconsistent data and because of the absence of appropriate metadata. A17 and A24 show significant differences in measured and calculated TAlk and DIC. For A24 this appears to be an issue with the pH measurements because the calculation with fCO2 shows reasonable agreement. For A17, R쮳 and Perez (1999) show better agreement, having used the original Mehrbach (1973) constants. Because the crossover differences in DIC and TAlk for A17 with other cruises are smaller than the internal consistency offset, we attribute this to issues with the applied internal consistency method. When the constants used in our analysis are applied, the difference is consistent over the entire DIC range (Fig. 6) but with a trend with latitude. The pH data for A20, A22, and A24 were measured by potentiometry and, according the responsible investigator (Millero), do not have the accuracy for a robust crossover analysis.