Total Carbon Dioxide Measurements
TCO2 was determined on 18,963 samples using two automated single-operator multiparameter metabolic analyzers (SOMMA) with coulometric detection of the CO2 extracted from acidified samples. A description of the SOMMA-coulometry system and its calibration can be found in Johnson et al. 1987; Johnson and Wallace 1992; and Johnson et al. 1993. A schematic diagram of the SOMMA analytical sequence and a complete description of the sampling and analytical methods used for discrete TCO2 on the Indian Ocean WOCE sections appear in Appendix B (Johnson et al. 1998). Further details concerning the coulometric titration can be found in Huffman (1977) and Johnson, King, and Sieburth (1985). The measurements for the Indian Ocean Survey were made on two systems provided by BNL (S/Ns 004 and 006) and a backup by WHOI (S/N 023).
TCO2 samples were collected from approximately every other station [~ 60 nm intervals, 50% of the stations (Figure 2.)] in 300-mL glass biological oxygen demand (BOD) bottles. They were immediately poisoned with 200 L of a 50% saturated solution of HgCl2, thermally equilibrated at 20C for at least 1 h, and analyzed within 24 h of collection (DOE Handbook of Methods 1994). Certified reference material (CRM) samples were routinely analyzed, usually at the beginning and end of the coulometer cell lifetime, according to DOE (1994). As an additional check of internal consistency, duplicate samples were usually collected on each cast at the surface and from the bottom waters. These duplicates were analyzed on the same system within the run of cast samples from which they originated, but the analyses were separated in time usually by ~3 h. Periodically, replicate samples were also drawn for shipboard analysis at sea using coulometry and for later analysis on shore at SIO by manometry. The latter samples, typically designated as the "Keeling samples," consisted of two 500-mL replicate samples collected at two depths (four samples total per station). These were analyzed only if both replicates survived the storage and the return journey to SIO.
Seawater introduced from an automated "to-deliver" (TD) pipette into a stripping chamber was acidified, and the resultant CO2 from continuous gas extraction was dried and coulometrically titrated on a model 5011 UIC coulometer. The coulometer was adjusted to give a maximum titration current of 50 mA, and it was run in the counts mode [the number of pulses or counts generated by the coulometer's voltage-to-frequency converter (VFC)] during the time the titration was displayed and acquired by the computer. In the coulometer cell, the acid (hydroxyethylcarbamic acid) formed from the reaction of CO2 and ethanolamine was titrated coulometrically (electrolytic generation of OH-) with photometric endpoint detection. The product of the time and the current passed through the cell during the titration was related by Faraday's constant to the number of moles of OH- generated and thus to the moles of CO2 that reacted with ethanolamine to form the acid. The age of each titration cell was logged from its birth (time that electrical current was applied to the cell) until its death (time when the current was turned off). The age was measured from birth (chronological age) and in mass of carbon (mgC) titrated since birth (carbon age). The systems were controlled with PCs equipped with RS232 serial ports for the coulometer and the barometer, a 24-line digital input/output (I/O) card for the solid state relays and valves, and an analog-to-digital (A/D) card for the temperature, conductivity, and pressure sensors. These sensors monitored the temperature of the sample pipette, gas sample loops, and, in some cases, the coulometer cell. The controlling software was written in GWBASIC Version 3.20 (Microsoft Corp., Redmond, Wash.), and the instruments were driven from an options menu appearing on the PC monitor.
The TD volume (Vcal) of the sample pipettes was determined gravimetrically prior to the cruise and periodically during the cruise by collecting aliquots of deionized water dispensed from the pipette into pre-weighed serum bottles which were sealed and re-weighed on shore. The apparent weight of water collected (Wair), corrected to the mass in vacuo (Mvac), was divided by the density of the calibration fluid at the calibration temperature to give Vcal. The sample volume (Vt) at the pipette temperature was calculated from the expression
Vt = Vcal [1 + av (t - tcal)] ,
where av is the coefficient of volumetric expansion for Pyrex-type glass (1 X 10-5/C), and t is the temperature of the pipette at the time of a measurement. Vcal for the Indian Ocean CO2 survey cruises and a chronology of the pipette volume determinations appear in Appendix B.
The coulometers were electronically calibrated at BNL prior to the cruises and recalibrated periodically during the cruises (Sections I8SI9S and I5WI4) to check the factory calibration as described in Johnson et al. (1993) and DOE (1994). The results for the electronic intercepts (Intec) and slopes (Slopeec) are given in Appendix B. For all titrations, the micromoles of carbon titrated (M) was
M = [Counts / 4824.45 - (Blank x Tt ) - (Intec x Ti)] / Slopeec ,
where 4824.45 (counts/mol) was the scaling factor obtained from the factory calibration, Tt was the length of the titration in minutes, Blank is the system blank in mol/min, and Ti the time of continuous current flow in minutes.
The SOMMA-coulometry systems were calibrated daily with pure CO2 (calibration gas) by titrating the mass of CO2 contained in two stainless steel gas sample loops of known volume and by analyzing CRM samples supplied by Dr. Andrew Dickson of the SIO. The ratio of the calculated (known) mass of CO2 contained in the gas sample loops to the mass determined coulometrically was the CALFAC (~1.004). A complete history of the calibration results appears in Appendix B. For water and CRM samples, TCO2 concentration in mol/kg was
TCO2 = M x CALFAC x [1 / (Vt x p)] x dHg ,
where p is the density of seawater in g/mL at the analytical t and S calculated from the equation of state given by Millero and Poisson (1981), and dHg is the correction for sample dilution with bichloride solution (for the cruises dHg = 1.000666).
System 006 was equipped with a conductance cell (Model SBE-4, Sea-Bird Electronics, Bellevue, Wash.) for the determination of salinity as described by Johnson et al. (1993). Whenever possible, SOMMA and CTD salinities were compared to identify mis-trips or other anomalies, but the bottle salinities (furnished by the chief scientist) have been used to calculate p throughout.
Three CRM batches were used for the Indian Ocean Survey. The certified TCO2 concentrations were determined by vacuum-extraction/manometry in the laboratory of C. D. Keeling at SIO and are given in Table 8.
Optimal cell and platinum electrode configurations, according to criteria given in Appendix B, were selected on the first section (I8S) and were used on all subsequent cruises.
The quality control-quality assurance (QC-QA) of the coulometric TCO2 determinations was assessed from analyses of 983 CRM samples during the nine Indian Ocean CO2 survey cruises. For both coulometric titration systems (004 and 006) the average TCO2 (measurement minus CRM value) for the whole survey was 0.86 mol/kg and the standard deviation was 1.21 mol/kg. A cruise-by-cruise breakdown of the accuracy and precision of the CRM analyses is given in Appendix B.
The small mean difference between the analyzed and certified TCO2 and the very high precision (1.21 mol/kg) of the differences indicates that the two systems gave very accurate and virtually identical results over the entire survey (see also Fig. 6 in Appendix B).
The second phase of the QC-QA procedure was an assessment of sample precision, which is presented in Table 9. The sample precision was determined from duplicate samples analyzed on each system during sections I8SI9S at the beginning of the survey and I4I5W about half way through the survey. The pooled standard deviation (Sp2), shown in Table 9, is the square root of the pooled variance according to Youden (1951) where K is the number of samples with one replicate analyzed on each system, n is the total number of replicates analyzed from K samples, and n - K is the degree of freedom (d.f.) for the calculation. Precision was calculated this way because TCO2 was analyzed on two different systems, and an estimate of sample precision independent of the analytical system was required. Hence Sp2 is the most conservative estimate of precision and includes all sources of random and systematic error (bias). Bias between systems would increase the imprecision of the measurements, but the excellent agreement between the Sp2 values for natural seawater samples (Table 9) and the high precision of the CRM differences confirms the virtually uniform response, accuracy, and high precision of both systems during the survey. This finding confirms that the precision of the TCO2 analyses during the Indian Ocean CO2 survey was 1.20 mol/kg.
The next phase of the QC-QA procedure was the comparison of replicate samples analyzed at sea and in the shore-based laboratory. Samples from every cruise were analyzed at sea by continuous gas extraction/coulometry, and later, after storage, duplicate samples were analyzed on shore by vacuum extraction/manometry. The results of the analyses are summarized in Table 10.
In general, the reproducibility and the uniformity of the data as a whole, and specifically, the high precision of the manometric analyses shown in Table 10, indicate that the collection and return of the "Keeling samples" was successfully performed by each of the measurement groups. Poor sampling or storage techniques would probably have been manifested in a much higher imprecision for the on-shore replicate analyses and in the differences between the at-sea and on-shore analyses. However, the negative mean difference (-4.16 1.21, n = 9) for the Indian Ocean sections was greater than the mean difference for WOCE sections in other oceans (-1.36 1.37 mol/kg, n = 22). The accuracy of the CRM analyses, the tendency for the coulometric analyses to give slightly lower results, and the reproducibility of the at-sea and on-shore differences are similar everywhere, but the magnitude of the Indian Ocean difference is clearly the largest observed to date. Even if the consistent and slightly negative difference for the CRM is taken into account (-0.86 mol/kg), the at-sea coulometric measurements are approximately 2 mol/kg lower than the manometric method. A suite of samples from the 1997 North Atlantic sections remains to be analyzed. Until these analyses are completed and a thorough statistical evaluation of the entire CO2 survey data set is made, the explanation of the at-sea and on-shore differences, including those found for the Indian Ocean, is not possible.
An additional step in the QA-QC was also undertaken. Inspection of Figure 1 shows points where the cruise tracks cross or nearly cross. The agreement between TCO2 measurements made at these crossover locations ( 100 km) on different cruises was examined by assuming that the temporal and spatial variations in deep-ocean TCO2 are small relative to the measurement accuracy and precision. Hence, deep ocean waters should have the same TCO2 at different times in the absence of internal vertical motion, and because deep ocean motion probably occurs along constant density surfaces (isopycnals), the comparisons of TCO2 measurements were made with reference to density and not depth. Appendix B and Appendix D (Johnson at al. 1998 and Sabine et al. 1999) give a complete description of the statistical procedures used to make the crossover comparisons. Briefly, crossover points were selected for comparison of water samples collected below 2500 m. A smooth curve was fit through the TCO2 data as a function of the density anomaly referenced to 3000 dbar (sigma3) using Cleveland's LOESS smoother (Cleveland and Devlin 1988). A separate fit was performed for the data collected at each of the two intersecting crossover points, but the same tension parameter was used for all of the crossover points so that the smoothing function was consistently applied to all crossover locations. The difference between the two smoothed curves was evaluated at 50 evenly spaced points covering the density range where the two data sets overlapped. A mean and standard deviation for the 50 comparisons was calculated for each crossover point. For TCO2, differences never exceeded 3 mol/kg, and the overall mean and standard deviation of the differences was -0.78 1.74 mol/kg. The latter differences were consistent with the overall precision of the CRM analyses ( 1.2 mol/kg).
Tables 8, 9, and 10 show an internally consistent TCO2 data set for the Indian Ocean with excellent accuracy with respect to the CRM certified values, consistently good precision, no analytical bias between the coulometric titration systems, and crossover agreement to within the precision of the method. However, the agreement between the at-sea and on-shore analyses is not as good as for earlier WOCE sections from other oceans (i.e., the Pacific and the South Atlantic). Based on the accuracy of the CRM analyses and the high precision of the sample analyses, the TCO2 data were not corrected in any way and were deemed to meet survey criteria for accuracy and precision.
