This documentation discusses the procedures and methods used to measure total carbon dioxide (TCO₂), total alkalinity (TALK), and pH at hydrographic stations during the R/V Maurice Ewing cruise in the South Atlantic Ocean on the A17 WOCE section. Conducted as part of the World Ocean Circulation Experiment (WOCE), this cruise was also a part of the French WOCE program consisting of three expeditions (CITHER 1, 2, and 3) focused on the South Atlantic Ocean. The A17 section was occupied during the CITHER 2 expedition, which began in Montevideo, Uruguay, on January 4, 1994 and finished in Cayenne, French Guyana, on March 21, 1994. During this period the ship stopped in Salvador de Bahia and Recife, Brazil, to take on supplies and exchange personnel. Upon completion of the cruise the ship transited to Fort de France, Martinique. Instructions for accessing the data are provided.

TCO₂ was measured using a single-operator multiparameter metabolic analyzer (SOMMA) coupled to a coulometer for extracting and detecting CO₂ from seawater samples. The overall precision and accuracy of the TCO₂ analyses was ±1.6 µmol/kg. A second carbon system variable, TALK, was determined by potentiometric titration with an overall precision of ±1.7 µmol/kg. During the A17 cruise the carbon system was overdetermined because a third carbonate system variable, pH, was also measured potentiometrically with an overall precision of ±0.003. The underway partial pressure of CO₂ (pCO₂) in surface waters was also continuously measured along the cruise track.

A comparison of A17 TALK with recent data in the South Atlantic Ocean confirms that A17 TALK data need a downward correction of 8 µmol/kg that was integrated in the CDIAC database. The internal consistency study carried out among the four carbon system variables led us to adjust the pH measurements by stations in order to eliminate the difference between TCO₂ measured and TCO₂ calculated from pH and TALK.

The R/V Maurice Ewing A17 data set is available free of charge as a numeric data package (NDP) from the Carbon Dioxide Information Analysis Center. The NDP consists of three oceanographic data files, one FORTRAN 77 data retrieval routine file, and this printed documentation, which describes the contents and format of all files as well as the procedures and methods used to obtain the data.

Keywords: carbon dioxide, TCO₂, total alkalinity, pH, partial pressure of CO₂, carbon cycle, coulometry, potentiometry, hydrographic measurements, World Ocean Circulation Experiment, meridional section, South Atlantic Ocean.

The World Ocean Circulation ExperimentWorld Hydrographic Program (WOCE-WHP) was a major component of the World Climate Research Program. The primary goal of WOCE was to understand the general circulation of the global ocean well enough to be able to model its present state and predict its evolution in relation to long-term changes in the atmosphere. The need for carbon system measurements arose from the serious concern over the rising atmospheric concentrations of carbon dioxide (CO₂). Increasing atmospheric CO₂ may intensify the earths natural greenhouse effect and alter the global climate.

Although CO₂-related measurements - specifically, total CO₂ (TCO₂), total alkalinity (TALK), partial pressure of CO₂ (pCO₂), and pH - were not official WOCE measurements, a coordinated effort was supported as a core component of the Joint Global Ocean Flux Study (JGOFS). This effort received support in the United States from the U.S. Department of Energy (DOE), the National Oceanic and Atmospheric Administration (NOAA), and the National Science Foundation (NSF), and in Spain from the Comisin Interministerial de Ciencia y Tecnologia (CICYT), for WOCE cruises through 1998 to measure the global spatial and temporal distributions of CO₂ and related parameters. Goals were to estimate the meridional transport of inorganic carbon in a manner analogous to oceanic heat transport (Bryden and Hall 1980; Roemmich and Wunsch 1985; Brewer et al. 1989; Holfort et al. 1998; Alvarez et al. 2003; Rosn et al. 2003), and to build a database suitable for carbon-cycle modeling and the estimation of anthropogenic CO₂ increase in the oceans. To obtain a reliable database, Wanninkhof et al. (2003) made a comparison of inorganic carbon system parameters measured in the Atlantic Ocean from 1990 to 1998, recommending small adjustments for consistency among other cruises in the zone. The CO₂ survey took advantage of the sampling opportunities provided by the WOCE cruises during this period, and the final data set covered on the order of 23,000 stations. Wallace (2002) reviewed the goals, conduct, and initial findings of the global CO₂ survey, and recently Sabine et al. (2004) estimated a global oceanic anthropogenic CO₂ sink between 1800 and 1994.

This report discusses results of the research vessel (R/V) Maurice Ewing expedition along the WOCE Section A17, from 4 January to 21 March, 1994 (Fig. 1.1). The cruise, designated as CITHER2_12, was a part of the French WOCE program consisting of three expeditions focusing on the South Atlantic Ocean: CITHER 1 (1993), 2 (1994), and 3 (1995). TCO₂ analysis personnel and support for this expedition were from Brookhaven National Laboratory (BNL), Lamont-Doherty Earth Observation (LDEO), and Battelle Pacific Northwest National Laboratory (PNNL). Analyses of TALK, pH, and nutrients were performed by Spanish scientists from the Consejo Superior de Investigaciones Cientificas (CSIC), Instituto de Investigaciones Marinas of Vigo. The hydrographic work was carried out by French scientists under the direction of Laurent Memery Laboratoire d'Oceanographie Dynamique et de Climatologie (LODYC), University of Pierre et Marie Curie, Paris, France.

The A17 section work will yield a map of the large-scale three-dimensional distribution of temperature, salinity, and chemical constituents, including the carbon system variables. This map will be combined with the results of the remaining French WOCE South Atlantic sections (A6, A7, A13, and A14) and the other South Atlantic WOCE sections measured by CO₂ survey participants (A8, A9, A10, and A11) to provide an extensive reference data set. Knowledge of the measured variables and their initial conditions allow determination of heat and water transports as well as carbon transport and elucidate regional sources and sinks of carbon and fossil fuel carbon. Studies estimating the carbon transport and establishing the anthropogenic CO₂ sources and sinks based on these data have already appeared in the literature (Holfort et al. 1998; Rios et al. 2003). An understanding of anthropogenic CO₂ uptake and transports contributes to the understanding of processes relevant to climate change. The South Atlantic A17 section was especially relevant to CO₂ transport because it focused on the western boundary sections and currents, and provided a description of the water masses and their meridional evolution between 50°S and 10°N (Memery et al. 2000).

Fig. 1.1. Cuise track during the R/V Maurice Ewing Atlantic Ocean survey expedition along WOCE section A17.

The work aboard the R/V Maurice Ewing was supported by the Institut Francais de Recherche pour L'Exploitation de la Mer (IFREMER; Grant 210161), the Institut National des Sciences de l'Univers (INSU), and the Centre de la Recherche Scientifique (CNRS), in the framework of the Programme National d'Etude de la Dynamique du Climat (PNEDC) and its WOCE/France subprogram. The carbon dioxide and nutrients work was supported by DOE (DE-ACO2-76CH00016) and CICYT (Grant ANT93-1156-E). We would like to thank the master, officers, and crew of R/V Maurice Ewing and all the participants on the cruise CITHER-2. Special thanks go to M. Arhan, coordinator of the WOCE-France program CITHER, and L. Memery, chief scientist of cruise CITHER 2. The authors are also especially grateful to the Sonderforschungsbereich 460 (SFB) at the University of Kiel (Dr. F. Schott, Leader), funded by the Deutsche Forschungsgemeinschaft, for their support and assistance in completing the written documentation.

Ship Name	Maurice Ewing
EXPOCODE	3230CITHER2_1-2
WOCE section	A17
Ports of call	Montevideo, Uruguay; Salvador de Bahia and Recife, Brazil; Cayenne, French Guyana
Dates	January 4March 21, 1994
Funding support	CITHER cruise: INSU, CNRS, PNEDC, France TCO2: DOE Alkalinity, pH, and nutrients: CICYT, Spain
Chief scientist	Dr. Laurent Memery, LODYC, Paris, France

Parameters measured, institution, and responsible investigators

Parameter	Institution	Responsible Personnel
CTD, salinity, XBT	LODYC	L. Memery, M. Arhan, H. Mercier
Nutrients	IIM.CSIC	X. Alvarez-Salgado, C. G. Castro
Oxygen	LPO	H. Mercier
CFCs	LODYC	L. Memery
Tritium, He, 14C	LMCE	P. Jean Baptiste
TCO2	BNL/PNNL	L. Bingler, L. Arlen
Total alkalinity, pH	IIM.CSIC	A. F. Rios, G. Roson
Underway pCO2	LDEO	D. Chipman
Brazilian observer	RB	J. A. Fontainha

Participating institutions

BNL	Brookhaven National Laboratory
IIM.CSIC	Instituto de Investigaciones Marinas, CSIC, Vigo, Spain
LDEO	Lamont-Doherty Earth Observatory
LMCE	Laboratoire de Modelisation du Climate et de l'Environnement
LODYC	Laboratoire d'Oceanographie Dynamique et de Climatologie
LPO	Laboratoire de Physique des Oceans
PNNL	Pacific Northwest National Laboratory
RB	Rio de Janeiro, Brazil

Although the cruise took place aboard a U.S. ship, the R/V Maurice Ewing, the A17 section was a part of a three-year-long French hydrographic expedition (CITHER) in the South Atlantic. In order to make the TCO₂ measurements on A17, four U.S. institutions had to combine forces to complete the work. The single-operator multi-parameter metabolic analyzer (SOMMA) and coulometer analytical system came from the Lamont-Doherty Earth Observatory (LDEO) at Columbia University; the TCO₂ group leader L. Bingler was from Battelles Pacific Northwest Laboratory (PNNL); the assistant TCO₂ analyst L. Arlen came from the NOAAs National Marine Fisheries Service (NMFS) Laboratory in Sandy Hook, New Jersey. The training and financial support of the analysts was carried out at and provided by BNL. In addition, PNNL paid for a barometer, which was installed in the LDEO system, and supported the production of a revised instrument manual for the SOMMA-coulometer systems. The TALK was measured by A. F. Rios and G. Rosn of CSIC, Instituto de Investigaciones Marinas of Vigo, Spain. The latter group also measured pH, so that as a result of the A17 cooperative scientific effort, the carbonate system was overdetermined. Underway pCO₂ was also measured by David Chipman, who installed the pCO₂ equipment in Montevideo, just before the start of the expedition

The Maurice Ewing departed Montevideo on January 4, 1994, and headed generally south in direction of the Falkland Islands, where measurements for the main section were to begin just to the north of the islands. On the way measurements for two test stations were taken. Measurements for the main south-north section were started on January 10 beginning on the Falklands Plateau, with station intervals normally of 30 nm decreasing to 9 nm depending upon the topography. The station work was interrupted by a storm for a day during the period of January 14-15. On January 21 the ship turned northwest toward Brazil and then east to sample the Porto Alegre western boundary section, work which was completed on January 26. A storm and problems with the CTD wire and rosette interrupted the work for two days, but by January 31 the ship reached the Vema channel between the Argentine and Brazil basins, thereafter moving northward to continue the A17 section until February 10, when work at station 115 was completed. The ship then transited to Salvador de Bahia, Brazil, making two additional test stations before arriving in port on February 13. In Salvador de Bahia, French CTD and hydrographic personnel were exchanged. The CO₂ and nutrients measurement groups remained on board, however.

The Maurice Ewing departed Salvador de Bahia on February 17, 1994, and commenced sampling the Salvador western boundary section. However, difficulties were again experienced with the CTD and rosette, so that the completion of measurements for the latter section was delayed by two days. Due to these problems a decision was made to transit to Recife to pick up the expert Jean Pierre Gouillou, arriving from France, who was tasked with repairing the CTD and performing the software modification. The personnel transfer was completed during the period February 24-26, and the ship continued with the station samplings. With the removal of the first 500 m of the CTD wire, the CTD and rosette were restored to function, and after February 28 no additional problems were noted. By March 14 the main north-south section had been completed (station no. 210), and by March 15 the last section sampling between the Mid-Atlantic Ridge and Cayenne had begun, with sampling for the last station of the cruise (no. 235) taken just out from Cayenne on March 20. Some of the scientific personnel disembarked in Cayenne on March 21, 1994, whereupon the ship left immediately for Fort de France, Martinique, where the remainder of the scientific and crew disembarked.

A SOMMA (S/N 007) with CO₂ detected by coulometry was used to determine TCO₂ on the A17 section. TALK was determined by potentiometric titration using an automatic potentiometric titrator, Titrino Metrohm, with separate glass and reference electrodes. The pH was determined potentiometrically using a Metrohm Model 654 pH meter, a combination glass electrode, and National Bureau of Standards (NBS) buffers for standardization. The CO₂ samples from more than 50% of the 235 CTD stations occupied during the A17 cruise were always drawn in conjunction with tracer samples (CFCs, tritium, etc.) and the standard WOCE variables (salinity, oxygen, temperature, and nutrients). As on previous cruises, not all stations could be sampled for TCO₂ and TALK because of the time required for analysis of the samples (see Table 3.1 for inorganic carbon sample distribution). However, pH and the WOCE standard variables were measured on all samples.

Water samples were collected using a 32-position rosette with 8-L Niskin bottles developed at the Laboratoire de Physique des Oceans, IFREMER, Brest, France. The rosette was equipped with a Neil-Brown Mark-III CTD-O₂ (see Brown and Morrison 1978). In order to check the pressure measurements and temperature of the CTD on board, inverse thermometers and pressuremeters, type SIS, were mounted in the Niskin bottles to be fired at the bottom. The signal of the CTD was transmitted to the hydrographic data acquisition system of the LPO. This new system, created around a UNIX work station, allowed the user to see in real time the vertical profiles of the variables measured and calculated in order to check the quality of the signal transmitted by the CTD. The set of data transmitted by the CTD with a cadence of 32 cycles per second was recorded on a diskette. After each station, the data profiles were plotted vs pressure following the procedure of Billant (1985).

At the end of each cast, a full suite of water samples were drawn in the following order: CFCs, helium, oxygen, TCO₂, TALK, pH, nutrients, tritium, and salinity.

During the cruise 6778 samples were analyzed for salinity within 12 days of collection using a Guildline PORTASAL salinometer that was calibrated with standard seawater (Batch P123, K15 = 0.99994) produced at Wormley and dated June 10, 1993. The temperature of the thermostat was fixed at 21°C until station 134, and at 22°C from stations 135 to 235. The precision of the salinity determination was 0.002 from 181 pairs of samples taken from two rosette sampling bottles closed at the same depth. The accuracy of the bottle salinity data was 0.001.

Dissolved oxygen was determined by Winkler titration after the technique of Culberson and Huang (1987). The operational conditions and the analytical method, including the calculation of oxygen concentrations, followed the standard WOCE procedure and recommendations given by Culberson et al. (1991) in the WOCE Manual of Operations and Methods. Appropriate corrections for sample density, blanks, and volumetric expansion have been included. The precision of the analyses was 0.78 µmol/kg from 196 pairs of samples taken from two rosette sampling bottles closed at the same depth. In total, 6756 oxygen analyses were completed during the A17 section.

The nutrients nitrate, nitrite, phosphate, and silicate were determined on every bottle closed on the A17 section by segmented flow analysis with a Technicon II Autoanalyzer. The combined nitrate and nitrite were determined after reduction of nitrate to nitrite in a Cd-Cu column according to the procedure of Mourio and Fraga (1985). The method was calibrated by diluting concentrated primary standards of dried salts (KNO₃, KH₂PO₄, and SiF₆Na₂) dissolved in Milli-Q water with aged, filtered, and low-nutrient seawater and analyzing these substandards for each run of samples analyzed on the autoanalyzer. Phosphate was determined according to the procedure of Hansen and Grasshoff (1983) as modified by Alvarez-Salgado et al. (1992). The accuracy of the method was 0.01 µmol/kg. Silicate was determined according to Hansen and Grasshoff (1983) using ascorbic acid as the reducing agent. The accuracy of the method was 0.25 µmol/kg. Quality control and consistency of nutrient measurements can be seen in Groupe CITHER-2 (1996).

Table 3.1 summarizes the carbonate system variables measured on WOCE Section A17.

As on previous cruises, TCO₂ was determined using an automated SOMMA dynamic headspace sample processor (S/N 007) 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. Further details concerning the coulometric titration can be found in Huffman (1977) and Johnson et al. (1985). The methods used for discrete TCO₂ on WOCE sections have been extensively dealt with in previous reports (Johnson et al. 1998a) and only need to be briefly summarized.

Seawater samples were collected in 300-mL ground-glass stoppered bottles and poisoned with 100 L of a 100% saturated solution of HgCl₂ to prevent biological alterations. Prior to analyses, the samples were stored in the dark and thermally equilibrated to within 23°C of the thermostatted SOMMA system (sample pipette and sample bath), which was kept at a constant temperature of approximately 20°C. The analysis of the TCO₂ samples was usually completed within 14 h of collection (see DOE 1994). Duplicate samples were usually collected on each cast at the surface and from the bottom waters and analyzed within the run of cast samples from which they originated. Following standard procedure, certified reference material (CRM) was routinely analyzed during the sample analyses (approximately one CRM for every 30 samples) according to DOE (1994). The CRMs were supplied by Dr. Andrew Dickson of the Scripps Institution of Oceanography, and the A17 cruise analysts were supplied with batch 18. The certified values for batch 18 were S = 35.298 and TCO₂ = 2115.15 µmol/kg. The CRM TCO₂ concentration was determined by vacuum-extraction and manometry in the laboratory of C. D. Keeling at Scripps Institution of Oceanography (SIO).

The SOMMA injected an accurately known volume of seawater from an automated to-deliver (TD) pipette into a stripping chamber. Following acidification of the seawater and continuous gas extraction, the resultant CO₂ was dried and coulometrically titrated on a model 5011 UIC coulometer with a maximum titration current of 50 mA in the counts mode [the number of pulses or counts generated by the coulometers voltage-to-frequency converter (VFC) during the titration was displayed]. 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 (charge in Coulombs) was related by Faradays constant to the number of moles of OH^- generated and thus to the moles of CO₂, which reacted with ethanolamine to form the acid. The age of each titration cell was logged from its birth (time that electrical current is applied to the cell) until its death (time when the current is turned off). The age was measured in minutes from birth (chronological age) and in mgC titrated since birth (carbon age).

The system was controlled with an IBM-compatible PC equipped with two RS232 serial ports (coulometer and barometer), a 24-line digital input/output (I/O) card (solid state relays and valves), and an analog-to-digital (A/D) card (temperature, conductivity, and pressure sensors). The cards were manufactured by Real Time Devices (State College, PA 16803). The SOMMA temperature sensors (model LM34CH, National Semiconductor, Santa Clara, CA), with a voltage output of 10 mV/°F, were calibrated against thermistors certified to 0.02°F prior to the cruise using a certified mercury thermometer. These sensors monitored the temperature of SOMMA components, including the pipette, gas sample loops, and the coulometer cell. The SOMMA software was written in GWBASIC Version 3.20 (Microsoft Corp., Redmond, WA), and the instruments were driven from an options menu appearing on the PC monitor. With the coulometer operated in the counts mode, conversions and calculations were made using the SOMMA software rather than having the programs and the constants hardwired into the coulometer circuitry.

The SOMMA-coulometry systems were calibrated with pure CO₂. The calibration hardware consisted of an eight-port gas sampling valve (GSV) with two sample loops of known volume (determined gravimetrically by the method of Wilke et al. 1993) connected to the calibration gas through an isolation valve with the vent side of the GSV plumbed to a barometer. When a gas loop was filled with CO₂ at known temperature and pressure, the mass (moles) of CO₂ contained therein was calculated, and the ratio of the calculated mass to that determined coulometrically was the calibration factor (CALFAC). The CALFAC was used to correct the subsequent sample titrations for small departures from 100% recoveries (DOE 1994). The standard operating procedure was to make gas calibrations daily for each newly born titration cell (normally, one cell per day). Normally, two or three sequential gas calibrations were run per cell between the carbon ages of 39 mgC with the last CALFAC used for calculation of TCO₂ if it was consistent with the preceding CALFAC (i.e., agreement to ±0.1% or better). The mean CALFAC and the standard deviation of the mean are shown in Table 3.2. The CALFAC for system 007 remained very stable throughout the A17 section (the change in TCO₂ concentration due to change in CALFAC was 0.05% or 1.0 µmol/kg) over the period November 1993 through March 1994. The mean carbon age for the mean CALFAC shown in Table 3.2 was 8.9 ± 5.1 mgC titrated (N = 73).

The to-deliver volume (V_cal) of the sample pipettes was determined (calibrated) gravimetrically in November 1993 prior to the cruise. The calibration was checked periodically (for A17, once weekly) by collecting aliquots of deionized water dispensed from the pipette into preweighed serum bottles. The serum bottles were crimp-sealed and weighed immediately during the on-shore laboratory calibrations, or returned to shore where they were reweighed on a model R300S (Sartorius, Gttingen, Germany) balance as soon as possible. The apparent weight (g) of water collected (W_air) was corrected to the mass in vacuo (M_vac) and the calibrated TD pipette volume (V_cal) was calculated by dividing M_vac by the density of the calibration fluid at the calibration temperature (t_cal). For A17, V_cal was 28.9315 ± 0.0033 mL at a t_cal of 19.81°C (N = 47). The sample volume (V_t) at the pipette temperature was calculated for all A17 samples from the expression

where a_v is the coefficient of volumetric expansion for Pyrex-type glass (1 X 10⁵/°C), and t is the temperature of the pipette at the time of a measurement. The mean pipette temperature or analytical temperature (t) on the A17 section was 19.70 ± 0.29°C.

The factory-calibrated coulometer was electronically calibrated independently in the laboratory in November 1993, prior to the cruise as described in Johnson et al. (1993, 1996) and DOE (1994); and the terms INT_ec and SLOPE_ec were obtained and entered into the software for system 007. The micromoles of carbon titrated (M), whether extracted from water samples or the gas loops, was

where 4824.45 (counts/mol) is a scaling factor obtained from the factory calibration, T_t is the length of the titration in minutes, Blank is the system blank in µmol/min, INT_ec is the intercept from electronic calibration in µmol/min, T_i is the time in minutes during the titration where current flow was continuous, and SLOPE_ec is the slope from electronic calibration. Note that the slope obtained from the electronic calibration procedure applied for the entire length of the titration, but the intercept correction applied only for the period of continuous current flow (usually 34 min) because the intercept can only be calculated from calibrated levels of current flowing continuously. The coulometer electronic calibration should not change over the duration of the cruiseshown for earlier cruises although not without some exceptions (Johnson et al. 1998b)and system 007 was not electronically recalibrated during the A17 section. The electronic and gas calibration coefficients for system 007 are summarized in Table 3.2.

For water samples, the discrete TCO₂ concentration in µmol/kg was calculated from

where p is the density of sea water in g/mL at the measurement temperature and sample salinity calculated from the equation of state given by Millero and Poisson (1981), and d_Hg is the correction for sample dilution with bichloride solution (for A17 d_Hg = 1.000333).

Quality control and quality assurance (QC-QA) were assessed from the results of the 163 CRM analyses made during the A17 section. The mean and standard deviation of the differences between the measured and the certified TCO₂ (measured certified) are given in Table 3.3, and the temporal distribution of the differences is plotted in Fig. 3.1

The overall accuracy of the CRM analyses was better than 1 µmol/kg on system 007 for the A17 section. The precision of the CRM determination is the standard deviation of the differences between the measured and certified CRM TCO₂ (±1.64 µmol/kg, N = 163). The outlier results are summarized in Table 3.4. Because six of the CRMs analyzed on A17 were considered to be outliers -- meaning that the analytical difference (∆TCO₂) between the measured and certified TCO₂ exceeded ±5.0 µmol/kg (measured - certified) -- these data are not included in Table 3.4.

Throughout the WOCE work, care was taken to titrate a limited number of samples in each coulometer cell to avoid excessive cell carbon ages and coulometer-cell-solution exhaustion or failure. In actual practice, this has meant that, on average, no cell was used to titrate more than a single 36-bottle station (a cell age of 35 mgC titrated), and experience has confirmed this practice (Johnson et al. 1998b).

Fig. 3.1. Temporal distribution of differences between measured and certified TCO₂ for analyzed on SOMMA-coulometry system 007 during WOCE section A17.
The differences were calculated by subtracting the certified TCO₂ from the measured TCO₂.

This convention was not followed on the A17 section because, at this point in the program, experimental evidence was needed concerning the actual lifetime of the cells. Hence, the A17 cells were run so that their carbon ages (mgC titrated) routinely exceeded the 35 mgC limit by factors of 1.5 - 2.5. Based on thousands of CRM analyses made during the CO₂ survey and an overall precision of 1.6 µmol/kg for the coulometric determination of TCO₂, an empirical definition of "cell failure" was proposed. Failure was defined as two successive CRM analyses with a difference >5 µmol/kg on a cell whose carbon age exceeded 35 mgC. The 5 µmol/kg limit was chosen because it was equivalent to three standard deviations in precision. These "failures" have been designated as outliers (see Table 3.4). Table 3.4 indicates that two of the A17 cells (on 2.9 and 3.17) exhibited outliers, but that the second CRM analysis at a later carbon age with these cells was accurate. Hence, the sample data obtained with them were not flagged. For failed cells (2.24, 3.19, and 3.22), a quality flag of 3the WHPO questionable measurement flagwas assigned to those samples analyzed between the carbon age at the time of the last accurate CRM analysis and the carbon age at failure or cell death. However, based on WHPO criteria, the flagged measurements could be correct but may be open to interpretation; we have no direct evidence that they are not correct. The data shown in Table 3.4 also suggest that the original decision to set a conservative limit on cell lifetimes of 35 mgC was sound because failures or outliers become more frequent after 35 mgC.

The second phase of the QC-QA procedure was an assessment of precision, which is presented in Table 3.5. The single-system precision was determined from samples with duplicates analyzed on system 007.

Single-system precision has been assessed in Table 3.5 as "between-sample" precision (σ_bs), which is the mean absolute difference between duplicates (n = 2) drawn from the same Niskin bottle, where K is the number of samples with duplicates analyzed.

Although the single-system sample precision (±0.73 µmol/kg) was excellent, it cannot be taken as the precision of the TCO₂ determination for the A17 section for two reasons unique to this cruise:

1.       During section A17, the replicate samples were always analyzed one right after the other. On other WOCE sections, replicate analyses were spaced such that the interval between replicates was >3 but <12 h. This was done to provide a measure of drift (change in system response) during a sequence of sample analyses on the assumption that drift would be reflected in the single-system precision by an increase in the imprecision of the duplicate analyses. Running the duplicates in sequence eliminated the possibility of detecting drift, and sample precision consequently was probably overestimated.

2.       An evaluation of the samples for which duplicates were taken indicated that 10 duplicate pairs exhibited very poor precision (absolute difference between replicates from 7 to 280 µmol/kg). These samples were flagged when the data set was submitted, and they are not included in the precision given in Table 3.5. Further study indicated that 9 of the 10 pairs originated from the surface rosette sample bottle (stations 8, 13, 25, 51, 61, 155, 177, 188, and 230) from 0 to 5 m, and that only one pair originated from a deep bottle (5334 m at station 157). If the flagged results were used to calculate precision, then s_bs was 3.34 ± 18.70 µmol/kg (K = 115) for the rosette surface bottle duplicates and 0.82 ± 1.41 µmol/kg (K = 121) for the nonsurface bottle duplicates. These data suggested that the observed imprecision did not lay with the TCO₂ measurement system. The cause was probably due to an occasional but undetected mechanical problem with the rosette, especially when the rosette bottles were closed at or near the surface. Alternatively, the on-deck sampling procedure at the rosette could have caused the degassing of CO₂ into the Niskin bottle headspace during the time it took to draw the duplicate samples.

For the above reasons, the precision of the TCO₂ determination on the A17 section was taken to be the standard deviation of the CRM differences (measured − certified) or ±1.64 µmol/kg (Table 3.3) instead of the single-system precision of ±1.02 µmol/kg given in Table 3.5.

The final step in the QC-QA procedure was the ship-to-shore comparison. Here, sample duplicates (commonly called the Keeling samples) were analyzed in real time at sea by continuous gas extraction/coulometry and later, after storage, on shore by vacuum extraction/manometry. The Keeling samples were collected in specially provided, threaded 500-mL glass bottles with 4 mL of headspace volume, poisoned with 100 µL of a saturated HgCl₂ solution, and then sealed airtight with a greased ground-glass stopper that was secured to the bottle with a threaded plastic screw cap. The cap was bored out to fit over the top of the stopper and mated to the bottle threads. The airtight seal was made by gently tightening the cap until a secure seal between the stopper and bottle was attained. Overtightening caused the bottles to break immediately or during transit so that considerable care and practice were required to prepare a sample that would survive the journey back to SIO. The manometric analyses for 21 samples collected from 14 stations during section A17 were completed by December 1994 in the SIO laboratory of C. D. Keeling. The results of the comparison are given in Table 3.6. The mean ship-to-shore analytical difference (ship − shore) and the standard deviation of the differences was −0.09 ± 1.50 µmol/kg (N = 21). This was the best agreement between the ship and shore duplicate sample analyses made by any measurement group with or without BNL-supported equipment during the entire CO₂ survey. Prior to and subsequent to the A17 section, the ship-to-shore comparisons had and have consistently yielded slightly lower TCO₂ values (~2 µmol/kg) for samples analyzed in real time aboard ship compared to the reference analyses made at SIO (Wallace 2002).

Table 3.6 is particularly useful in view of the problems with surface bottle precision, which suggested the possibility of mechanical or chemical problems during rosette sampling during the A17 section. Inspection of the data in the table indicates excellent agreement between surface sample duplicates analyzed on ship and on shore and indicates that the incidence of poor precision, for whatever reason, probably did not compromise the accuracy of the A17 TCO₂ data. Indeed, Tables 3.3, 3.5, and 3.6 show that the TCO₂ data set for the A17 section was internally consistent and highly accurate and precise with respect to the both the CRM, the seawater duplicate samples, and the ship-to-shore comparison seawater samples. Hence, no correction for CRM differences has been applied to the data, and the TCO₂ data clearly met survey criterion for accuracy (4 µmol/kg) and precision. The reader is also referred to a recent assessment of TCO₂ data quality in the Atlantic Ocean resulting from comparisons of TCO₂ analyzes from crossover points sampled by different cruises between 1990 and 1998 (Wanninkhof et al. 2003).

TALK was determined with a Titrino Metrohm automatic potentiometric titrator using separate glass working and reference electrodes. Potentiometric titrations were carried out in a covered but not completely closed (headspace present) titration flask to a final pH of 4.4 as described by Perez and Fraga (1987a). The electrodes were standardized using an NBS buffer of pH 7.413, checked using an NBS buffer of 4.008, and acclimated in a seawater solution buffered to a pH of 4.4. To determine the systematic errors produced by variations of the electrode residual liquid-junction potential, titration curves were performed each week in CO₂-free seawater acidified to pH 4.0 with hydrochloric acid as described by Culberson (1981). The titration curves were linearized, and the inverse slope was taken to represent the apparent hydrogen ion activity coefficient. The decimal logarithm difference (ranging from 0.01 to 0.06) between the apparent activity coefficients of the electrode and those given by Mehrbach et al. (1973) at the same salinity and temperature with their electrode was the pH difference added to the final pH of the sample alkalinity titration to make our results equivalent with theirs using the constants of Mehrbach et al. (1973).