3.4 Other Inorganic Carbon System Parameters
The primary uses for pH and fCO2 measurements in this work are to determine the internal consistency of the inorganic carbon measurements when three or four carbon parameters were measured and to calculate TAlk or DIC on the cruises when these state variables were not measured. In these cases, DIC and pH (or fCO2) are used to determine TAlk, and TAlk and pH (or fCO2) are used to calculate DIC. The fCO2 and pH data on cruises were compared by using the same crossover analyses as DIC and TAlk, but no recommendations are made regarding adjustments because the paucity of fCO2 and pH data makes it difficult to sufficiently evaluate these parameters.
Electrodes or spectrophotometry were used to measure pH. The spectrophotometric pH measurements have excellent precision, of 0.001 or better (McElligott et al. 1998). Because pH is very sensitive to changes in TAlk and DIC, it can be used to discern minute changes in them. For example, for North Atlantic deep water with DIC = 2192 µmol/kg, TAlk = 2349 µmol/kg, and pHsw (25°C) = 7.7417, a 1-µmol/kg change in DIC will result in a 0.0023 change in pH. However, the accuracy of the measurements is not well quantified and is currently under debate (DelValls and Dickson 1998, Millero et al. submitted). CRMs were used to normalize pH values on cruises A13, A14, and A17 by calculating pH from the certified DIC and TAlk values and from the dissociation constants recommended by Mehrbach et al. (1973). The measured pH values were then normalized to the mean calculated CRM value for each cruise (Ríos and Perez 1999)
Eleven of the 24 cruises had pH measurements. Different analysis techniques were used, and the pH values were recorded on different reference scales. Metadata provided with the pH measurements often were not sufficient to fully comprehend methods and corrections. A summary of the metadata is provided in Table 9. Much of the pH data were reported at 25°C on the seawater scale [pHsw (25°C)]. The data from A16N, A17C, and A14 that were not reported at 25°C and on the SW scale were corrected to pHsw (25°C) in order to perform the crossover analyses. The conversion was done by using the program of Lewis, Wallace, and Allison (1998) and by applying the carbonate dissociation constants of Mehrbach et al. (1973) as refitted by Dickson and Millero (1987). Because the temperature dependence of pH is not well known, these adjustments (and therefore the crossover comparison) should be viewed with caution.
A16N data were provided on the seawater scale at 20°C. The correction to 25°C was performed by using the measured DIC and pHsw (20°C) and the Lewis, Wallace, and Allison (1998) program. The A14 and A17 data were provided on the National Bureau of Standards (NBS) scale at 15°C. Correction to 25°C on the seawater scale was done by first calculating fCO2 from DIC and pHnbs (15°C), and then by using the DIC and calculated fCO2 to determine pHsw (25°C). Both the original reported pH data and the converted data are presented in the working synthesis data files.
3.4.2 Fugacity of CO2
The fugacity of CO2 was measured throughout the water column on seven cruises. Like pH, the measurements were performed at a fixed temperature. Agreement between cruises was good (see Table 4), considering that there were no liquid reference materials and that gas standards used by different groups were not intercalibrated. The precision of the measurements depends on the methodology, but in general it was better than 2 µatm for surface waters (<500 µatm) and within 1% of the measured value for deep water. The fCO2 (20°C) is very sensitive to changes in TAlk and DIC, particularly in deep water. For example, for North Atlantic deep water with DIC = 2192 µmol/kg, TAlk = 2349 µmol/kg, and fCO2 (20C) = 757 µatm, a 1 µmol/kg change in DIC will result in a 5-µatm change in fCO2 (20C). Thus a commonly quoted DIC precision of 2 µmol/kg corresponds to a 10 µatm uncertainty in fCO2 (20°C) for deep water. The only crossover that exhibited significant differences was that of A16S and A8 (see Table 5). Based on the internal consistency analyses (see Sect. 3.5), the difference in measured TAlk and calculated TAlk [from DIC and fCO2 (20C)] for A16S of -3.6 µmol/kg suggests that the A16S deep-water fCO2 (20°C) measurements are low by 10 to 15 µatm.