Assessment of pH dependent errors in spectrophotometric pH measurements of seawater

A recent analysis of full water column hydrographic data revealed a pH-dependent discrepancy between spectrophotometrically measured pH using purified meta-cresol purple and pH calculated from dissolved inorganic carbon (DIC) and total alkalinity (TA). The discrepancy (pH(spec) - pH(TA,DIC)) is approximately - 0.018 and 0.014 at pH 7.4 and 8.2, respectively. This discrepancy has a wide range of implications for marine inorganic carbon measurements, such as establishing robust calibration protocols for pH sensors operating on profiling floats. Here, we conducted a series of lab based experiments to assess the magnitude of pH-dependent errors for spectrophotometric pH measurements in seawater by directly comparing its performance to pH measured by an Ion Sensitive Field Effect Transistor (ISFET) pH sensor known to have Nernstian behavior. Natural seawater was titrated with high CO2 seawater while simultaneously measuring pH using spectrophotometry and an ISFET sensor over a large range in pH (7-8.5) and temperature (5-30 degrees C). The two pH measurements were consistent to better than +/- 0.003 (range) at all temperatures except at 5 and 10 degrees C and very low and high pH, where discrepancies were as large as +/- 0.005. These results demonstrate that pH-dependent errors in spectrophotometric pH measurements can be rejected as the cause of the pH-dependent discrepancy between pH(spec) and PHTA,DIC. The cause of this discrepancy is thus likely due to our incomplete understanding of the marine inorganic carbon model that could include errors in thermodynamic constants, concentrations of major ions in seawater, systematic biases in measurements of TA or DIC, or contributions of organic compounds that are not accounted for in the definition of total alkalinity. This should be a research priority for the inorganic carbon community.