The role of pH measurements in modern oceanic CO sub(2)-system characterizations: Precision and thermodynamic consistency

In May 1992, surface seawater samples were collected along an equatorial transit (130 to 100 degree W) and analyzes for total hydrogen ion concentration (expressed as spectrophotometric pH sub(T)) total dissolved inorganic carbon (coulometric C sub(T)), and total alkilinity (potentiometric A sub(T) and spectrophotometric A sub(T)). This data set, which presents a striking view of the chemical signature of tropical instability waves, is also unusual in that its "overdetermination" of the CO sub(2)-system in seawater includes both potentiometric A sub(T) data and the first spectrophotometric A sub(T) data collected at sea using a double-wavelength, one-step acid addition method. Our data set indicates that spectrophotometric A sub(T) measurements (A sub(Tspec)) are both precise ( plus or minus 1.8 mu mol/kg) and accurate; the mean observed difference between A sub(Tspec) and A sub(T) obtained potentiometrically in this study (A sub(Tpot)) is 1 mu mol/kg. As only two of the three measured parameters are required to characterize the CO sub(2)-system in seawater, this analytical redundancy of our analyses (C sub(T), A sub(T), pH) afforded an opportunity to evaluate, in terms of thermodynamic consistency, a recent calibration of m-cresol purple, a pH indicator. Using various carbonic acid dissociation constants, measured parameters were combined in pairs (C sub(T)-A sub(T), pH-C sub(T), and pH-A sub(T)) to calculate a third parameter for comparison with the shipboard measurements. Depending on the selected set of carbonic acid dissociation constants, the average offset between directly measured and predicted values of A sub(T) and C sub(T) was as small as plus or minus 1 mu mol/kg. The results of this study indicate that the present 25 degree C mCP calibration, paired with the 25 degree C combined dissociation constants of Hansson and Mehrbach, produces accurate predictions of A sub(T) and C sub(T). Extensive replication of the shipboard spectrophotometric measurements made it possible to examine directly the sensitivity of derived parameters to variations or errors in input A sub(T) and pH. In accord with earlier theoretical treatments of this question, the results presented demonstrate that pH is imprecisely predicted from A sub(T) and C sub(T), whereas imprecision in measured pH-at the level typical of spectrophotometric measurements ( plus or minus 0.0004)-contributes negligibly ( plus or minus 0.3 mu mol/kg) to imprecision in derived A sub(T) and C sub(T).