Determination of surface water pCO2 requires measurements for a number of parameters in addition to the concentration of CO2 in the carrier gas equilibrated with seawater. With the broad range of data sources and continued evolution of the measurement systems, it has not been possible to ensure that the observations contain all the necessary data elements with desired precisions. In some cases, pCO2 or fCO2 was reported without VCO2, sometimes with the pressure at the time of equilibration, sometimes not. Some files contained only the sea surface temperature and the pCO2 at that temperature, but not the temperature and pCO2 at equilibration. Ideally, the incoming file should contain the dried concentration of CO2 in a parcel of air in equilibrium with seawater (VCO2), the temperature at the time of equilibration, and the equilibration pressure. When the pressure was not reported, we used the climatological value in the nearest box from the NCEP/NCAR Reanalysis II Project file for the month of the observation. We accepted data points that contained pCO2 and sea surface temperature with or without the pressure.
Field data obtained in earlier years do not necessarily have all the required supporting measurements. For example, the temperatures of equilibration and water samples were recorded by hand at intervals of a few to several hours, and hence these values are often interpolated linearly with time between measurements. The salinity was measured only at hydrographic stations, and these values were interpolated between stations. When salinity was not measured during some expeditions, climatological values were assigned. Some data were unrecoverable and lost as a result of changes in computer systems (hardware and software). For the past decade, computer controlled systems for underway surface water pCO2 measurements have been deployed widely, and more complete high quality data sets have been obtained. In constructing this data file, we exercised our personal judgment for accepting and rejecting data, especially those collected in earlier years. For more recent data sets, water flow rates through the equilibrator and the temperatures of equilibrator water are recorded for each pCO2 measurement. They often served as important criteria for identifying satisfactory operation of the equilibrator. Measurements made at reduced or stopped water flow conditions were rejected, as were those made at unusually rapid changes in the temperature of the water in an equilibrator. CO2 gas analyzers are commonly calibrated using three or more gas mixtures of different CO2 concentrations. Those measurements made using only a single calibration gas mixture are subject to unspecified uncertainties, and hence are rejected.
All data points have been examined individually as they were integrated into this database. While not perfect, we have tried to edit out obvious problems, erring on the side of leaving in questionable ones. Considering differences in equilibrator designs, calibration methods and some interpolated parameters, we estimate that the uncertainty of the pCO2 data presented in this data file is about ±2.5 µatm on the average.
Seawater pCO2 values that were computed using the alkalinity, total CO2 concentration, and pH data are not included in this data file because of their large uncertainties and potential biases resulting from different dissociation constants of carbonic, boric and other acids used for the computations.
During many cruises, atmospheric CO2 concentrations were measured concurrently with surface water pCO2. However, we suspect that many of these measurements were contaminated by local sources (such as ship's exhausts). Since atmospheric CO2 concentrations vary from one air mass to another, especially in the northern hemisphere, distinguishing local contamination from natural variability is difficult. Hence, we decided to omit the atmospheric CO2 data from this file. We recommend that the atmospheric CO2 concentration data listed in the GLOBALVIEW-CO2 for computing the sea-air pCO2 difference.