Total CO2 Measurements
As on previous cruises, TCO2 was determined using automated dynamic headspace sample processors (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 may be found in earlier cruise reports (see Johnson et al. 1995,1996), and further details concerning the coulometric titration can be found in Huffman (1977) and Johnson, King, and Sieburth (1985). The methods used for discrete TCO2 on WOCE sections have been extensively dealt with in previous reports (Johnson et al. 1998a) and need only be briefly summarized.
The AR24 section required modification of the usual sampling procedures. As noted in Section 3.1.2, 4-L sampling bottles were employed on the rosette, limiting the amount of sample available for the carbonate system analysts to one 500-mL bottle. Hence, the TCO2 coulometric titration analysis had to be completed before the partially empty 500-mL bottle was passed to the TALK group for the potentiometric alkalinity titration. There was enough sample to complete both measurements, but not enough time or sample for TCO2 replicate analyses from the same 500-mL sample bottle. The 4-L sampling bottles also made it impossible to draw duplicate samples from the same sampling bottle. Without duplicate samples from the hydrographic stations, standard measures of sample precision (DOE 1994; Johnson et al. 1998b) could not be completed on the AR24 section. Samples were poisoned with 100 L of a 50% solution of HgCl2 and analyzed for TCO2 within 24 hours of collection (DOE 1994).
For sections A24, A20, A22, single or duplicate samples were collected in 300-mL biological oxygen demand (BOD) bottles, poisoned with 100 L of a 50% solution of HgCl2, and analyzed for TCO2 within 24 hours of collection, according to standard operating procedures (DOE 1994). The samples were stored in a dark refrigerator at 4-6°C until approximately 1-2 hours before analysis, when they were removed and placed in a temperature bath at 18-20°C and thermally equilibrated. The SOMMA sample pipette and sample bath were also kept at approximately 20°C. Duplicate samples were usually collected on each cast at the surface and from the bottom waters. For some casts, three sets of duplicates were taken. The duplicates were analyzed within the run of cast samples from which they originated so that the time elapsed between duplicate analyses was 3-12 hours. As per standard operating procedure (DOE 1994), CRM was routinely analyzed according to DOE (1994) guidelines. The CRM was supplied by Dr. Andrew Dickson of the SIO, and for the North Atlanticcruises, batches 33, 36, and 37 were used. The certified values for these batches were TCO2 = 2009.85 µmol/kg @ salinity = 33.781 for batch 33; TCO2 = 2050.21 µmol/kg @ salinity = 35.368 for batch 36; and TCO2 = 2044.15 µmol/kg @ salinity = 34.983 for batch 37. The CRM TCO2 concentration was determined by vacuum-extraction/manometry in the laboratory of C. D. Keeling at SIO.
An accurately known volume of seawater was injected from an automated to-deliver (TD) pipette into a stripping chamber. Following acidification, the resultant CO2 from continuous gas extraction 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 VFC during the titration was displayed). In the coulometer cell, the acid (hydroxyethylcarbamic acid) formed from the reaction of CO2 and ethanolamine is 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) is related by Faradays 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 is 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 is measured in minutes from birth (chronological age) and in mgC titrated since birth (carbon age).
Each system was controlled with an IBM-compatible PC equipped with two RS232 serial ports (coulometer and barometer), a 24-line digital input/output card (solid state relays and valves), and an analog-to-digital card (temperature, conductivity, and pressure sensors). Real Time Devices (located in State College, PA 16803) manufactured the cards. 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 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 coulometers operated in the counts mode, conversions and calculations were made using the SOMMA software rather than the programs and the constants hardwired into the coulometer circuitry.
The SOMMA-coulometry systems were calibrated with pure CO2 (calibration gas) using hardware consisting of an 8-port gas sampling valve (GSV) with two sample loops of known volume [determined gravimetrically by the method of Wilke, Wallace, and Johnson (1993)] connected to the calibration gas through an isolation valve; the vent side of the GSV was 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 prepared titration cell [normally, one cell per day and three sequential calibrations per cell at a carbon age of 39 mgC (mean age @ 6 mgC), with the result of the third calibration taken as the CALFAC if it was consistent with the second (i.e., agreement to ± 0.1% or better)]. Daily gas calibrations were made on both systems throughout the cruises.
The "to-deliver" volume (Vcal) of the sample pipettes was determined (calibrated) gravimetrically prior to the cruise to ± 0.02% or better in October of 1996. The calibration was checked periodically during all cruises by collecting aliquots of deionized water dispensed from the pipette into pre-weighed 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 balance (Sartorius, Gttingen, Germany) as soon as possible. The apparent weight (g) of water collected (Wair) was corrected to the mass in vacuum (Mvac) with the to-deliver volume being Mvac divided by the density of the calibration fluid at the calibration temperature. After the AR24 section in 1996, the system pipettes were dismounted and replaced with chemically cleaned pipettes in March, 1997. For the 1997 sections, the calibration volumes (Vcal) at the calibration temperature (tcal) of the sample pipettes were redetermined to ± 0.01% from a set of calibration samples taken on July 3, 1997, on board the Knorr at the completion of section A24 and were weighed on September 17. The TCO2 pipette volumes for the four North Atlantic sections are summarized in Table 2.
| Section | System S/N | Vcal (mL) | tcal (°C) |
|---|---|---|---|
| AR24 (1996) | 004 | 21.8927 | 19.91 |
| A24/A20/A22 (1997) | 004 | 21.2630 | 19.19 |
| AR24 (1996) | 030 | 21.3733 | 20.91 |
| A24/A20/A22 (1997) | 030 | 25.8544 | 19.52 |
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 × 10-5/°C), and t is the temperature of the pipette at the time of a measurement. The mean pipette temperature on the AR24 section in 1996 was 20.32 ± 0.51°C (n = 948), and on the 1997 North Atlantic Sections it was 19.55 0.52°C (n = 4666).The factory-calibrated coulometers were electronically calibrated independently in the laboratory before the cruise as described in Johnson et al. (1993, 1996) and DOE (1994), and the terms INTec and SLOPEec were obtained and entered into the software for each system. The micromoles of carbon titrated (M), whether extracted from water samples or the gas loops, was
M = [Counts / 4824.45 - (Blank × Tt ) - (INTec × Ti)] / SLOPEec
where 4824.45 (counts/µmol) is a scaling factor obtained from the factory calibration; Tt is the length of the titration in minutes; Blank is the system blank in mol/min; INTec is the intercept from electronic calibration in mol/min; Ti is the time in minutes during the titration where current flow was continuous; and SLOPEec 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 be calculated only from calibrated levels of current flowing continuously.
Unfortunately, the coulometer system 030, which was electronically calibrated prior to the AR24 cruise and again in March 1997, had to be replaced at the start of section A24 in May 1997. However, the replacement coulometer (S/N CBE-9010-V) was calibrated at the factory on March 20, 1997. Hence we assumed that the replacement coulometer was properly calibrated, and we entered the default calibration coefficients into the software (SLOPEec = 1.0 and INTec = 0.0). The system 004 was also recalibrated in March 1997 following the AR24 cruise with nearly identical results to those obtained in October 1996, and it was not recalibrated during the 1997 WOCE sections. The electronic calibration coefficients, along with the mean gas calibration factors determined for the North Atlantic section discrete TCO2 coulometers, are given in Table 3.
Table 3 illustrates an advantage of the independent laboratory electronic calibration procedure. The mean CALFAC for systems 004 and 030 using the laboratory-determined electronic calibration coefficients was approximately 1.0036 (or 99.64% recovery of the theoretical mass of CO2 calibration gas measured coulometrically) vs 1.0053 (99.47% recovery) for the factory-calibrated coulometer. Hence, a small percentage (0.17%) of the less than 100% recovery for known masses of CO2 coulometrically titrated can be explained by a factory-calibration procedure that is apparently slightly less accurate than the laboratory calibration. This difference has been consistent throughout the CO2 survey.
| Section | System S/N | SLOPEec | INTec mol/min | CALFAC(n) | St. dev. | Rel. st. dev. (%) |
|---|---|---|---|---|---|---|
| AR24 | 004 | 0.999372 | 0.002528 | 1.003892(9) | 0.000650 | 0.06 |
| A20/A22/A24 | 004 | 0.998905 | 0.001466 | 1.003361(63) | 0.000740 | 0.07 |
| AR24 | 030 | 0.999306 | 0.003550 | 1.003780(26) | 0.000497 | 0.05 |
| A20/A22/A24 | 030a | 1.000000 | 0.000000 | 1.005344(59) | 0.001369 | 0.13 |
aFactory-calibrated coulometer installed at the beginning of the A24 section in May 1997.
For water samples, the discrete TCO2 concentration in µmol/kg was calculated from
TCO2 = M × CALFAC × [1 / (Vt × p)] × dHg
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 dHg is the correction for sample dilution with bichloride solution (for the AR24 section in 1996 dHg = 1.0002 and for the 1997 sections dHg = 1.000333 ).
One of the SOMMA-Coulometry Systems (S/N 004) was equipped with a conductance cell (Model SBE-4, Sea-Bird Electronics, Inc., Bellevue, WA) for the determination of salinity measurement as described by Johnson et al. (1993). Whenever possible SOMMA and CTD salinity were compared to identify mistrips or other anomalies, but the bottle salinity (furnished by the chief scientist) was used to calculate TCO2.
>Quality control-quality assurance (QC-QA) was assessed from the results of the 275 CRM analyses made using systems 004 and 030 during the four North Atlantic sections. These data are summarized in Table 4, and the temporal distribution of the differences is plotted in Fig. 2 for section AR24 (1996) and in Fig. 3 for sections A24, A20, and A24 (1997).
| Section | System S/N | Δ TCO2 (µmol/kg) | St. dev. (µmol/kg) | n |
|---|---|---|---|---|
| AR24 | 004 | 1.42 | 2.10 | 16 |
| AR24 | 030 | 1.54 | 1.88 | 49 |
| Mean/total | 1.51 | 1.92 | 65 | |
| A24 | 004 | 0.04 | 1.10 | 49 |
| A20 | 004 | 0.23 | 1.20 | 42 |
| A22 | 004 | 0.06 | 0.69 | 17 |
| Mean/total | 0.10 | 1.08 | 108 | |
| A24 | 030 | 0.79 | 1.00 | 48 |
| A20 | 030 | 0.44 | 1.43 | 35 |
| A22 | 030 | 0.26 | 1.22 | 19 |
| Mean/total | 0.57 | 1.21 | 102 | Overall mean/total | 0.61 | 1.47 | 275 |
The overall accuracy of the CRM analyses was better than 1 µmol/kg on both systems for the four North Atlantic sections, with a combined overall mean difference of + 0.61 µmol/kg (n = 275). However, Table 4 shows that on the AR24 section (1996), the mean difference and the standard deviation of the differences were noticeably larger for both systems compared with the 1997 sections (A24/A20/A22). This may be due in part to mechanical problems experienced by the AR24 measurement group, operator procedures, and possibly the relatively short time available to service and re-calibrate the systems prior to the AR24 section. The latter was brought about by the fact that system 004 had been used in the Indian Ocean from 19941996 and was only returned to BNL for service, repair, and re-calibration in the fall of 1996. System 030, which was a newly built system returned to the laboratory after a test cruise in the North Atlantic, also was not returned until the summer of 1996. For the 1997 sections, both systems were available in the laboratory for servicing from January through May of 1997. Indeed, the 1997 WOCE sections represented the only opportunity during the CO2 survey for the BNL measurement group to thoroughly service and test the systems, reagents, and analytical gases in the laboratory with real samples and CRM prior to shipment. As a result, the accuracy and precision of the CRM analyses made in 1997 (see Table 4) probably represent the highest quality possible for these systems under field conditions.
All CRM analyses made on the discrete systems (004 and 030) during the 1997 sections are reported in Table 4. However, for section AR24, two CRM analyses were classified as outliers and dropped from the data set. These were CRM No. 206 run on system 030 on November 23 (difference = +10.17 µmol/kg) at a cell carbon age of 39.5 mgC, and CRM No. 600 on system 030 on November 28 (difference = +7.99 µmol/kg) at a carbon age of 35.7 mgC. One CRM analysis (CRM No. 352) run on system 004 on December 1 is not included in the data set because the titration did not attain an endpoint.
The second phase of the QC-QA procedure was an assessment of precision. As described in the text, duplicate samples could not be taken during the AR24 section in 1996. Hence the only estimate of AR24 sample precision was the standard deviation of the differences between the measured and certified TCO2 on both systems (see Table 4). Because differences from both systems have been combined, the CRM measurements are analogous to the sample duplicates analyzed on each system and should reflect both random and systemic error (bias). The decrease in precision for the CRM analyzed on the AR24 section in 1996 (±1.92 µmol/kg) compared with the CRM analyzed in 1997 (±1.20 µmol/kg) was consistent with the problems described for the 1996 leg. The good agreement in TCO2 between systems in 1996 (see Table 4) suggests that analyzing duplicate seawater samples on each system, as was done in 1997, might have yielded a higher precision than the precision of the CRM differences. Nevertheless, without sample duplicates, the AR24 sample precision must be based on the CRM analyses. Hence the precision of the TCO2 determination for the AR24 section in 1996 was ±1.92 µmol/kg (n = 65). Because procedures and performance varied from 1996 to 1997, separate estimates of sample precision were required for each year; the data for 1997 are given in Table 5.
By 1997 the deployment of two independent SOMMA systems side-by-side was routine, and the conventions employed for the estimation of precision in the earlier WOCE data reports are retained in Table 5. For sections A24, A20, and A22 in 1997, the single-system precision was determined from samples with duplicates analyzed on the same system (either 004 or 030). The sample precision was calculated using duplicates that were analyzed on both systems (004 and 030).
| Section | Mean absolute difference | Pooled standard deviation | |||||
|---|---|---|---|---|---|---|---|
| abs (µmol/kg) | St. dev. | K | Sp2 (µmol/kg) | K | n | d.f | |
| Single-system precision | |||||||
| A24 | 1.08 | 1.01 | 175 | 1.04 | 175 | 350 | 175 |
| A20 | 0.95 | 1.14 | 84 | 1.04 | 84 | 168 | 84 |
| A22 | 0.99 | 0.93 | 71 | 0.96 | 71 | 142 | 71 |
| Sample precision | |||||||
| All | 1.76 | 1.41 | 56 | 1.59 | 61 | 122 | 61 |
Single-system and sample precision have been separately assessed in Table 5 as "between-sample" precision (abs), which is the mean absolute difference between duplicates (n = 2) drawn from the same Niskin bottle; and the pooled standard deviation (Sp2) calculated according to Youden (1951), where K was the number of samples with duplicates analyzed, n was the total number of replicates analyzed from K samples, and n - K was the degrees of freedom (d.f.).
Single-system precision provided a measure of drift in system
As on other sections in the Atlantic Ocean(e.g., A8 and A10) where SOMMA-coulometer systems have been run in parallel, the sample precision was slightly less than the single-system precision. This indicated that changes in system response during the coulometer cell lifetime in 1997 were clearly within the precision of the method (±1.59 µmol/kg), while the slight but consistent decrease in sample precision compared with single-system precision was probably due at least in part to a small bias between the 004 and 030 systems. Although the precision was equivalent for both systems, system 030 gave on average slightly higher results than system 004. For example, the mean ΔTCO2for system 004 CRM was +0.10 µmol/kg, but it was +0.57 µmol/kg for system 030 CRM (see Table 4); while the mean of the seawater samples (n = 56, see Table 5) analyzed on 030 was +1.17 µmol/kg higher than the mean for the same samples analyzed on system 004. Hence the uniformly excellent single-system precision for 1997 cannot be used for sample precision, and analyzing duplicate replicates on each system remains the definitive measure of the overall precision of the 1997 data set and the TCO2 calibration procedures. The two discrete systems should give the same result for the same sample, and the extent to which they differ is a measure of the overall precision of the data set obtained with two independent systems. For TCO2 on the 1997 North Atlantic WOCE sections, the precision of the TCO2 determination was ± 1.59 µmol/kg (K = 56).
The North Atlantic sample precision for all four sections in 1996 and 1997 (±1.92 and ±1.59 µmol/kg, respectively) is in good agreement with the published and unpublished sample precision for other WOCE sections where systems were run in parallel: AE1, 1991 (±1.65 µmol/kg); P6, 1992 (±1.65 µmol/kg); A10, 1993 (±1.92 µmol/kg); A8, 1994 (±1.17 µmol/kg); Indian Ocean, 1995 (±1.20 µmol/kg). During the 1997 North Atlantic sections, a limited number of duplicate samples (K = 6) were analyzed from two different Niskin bottles closed at the same depth, and the mean absolute difference and standard deviation was 0.77 ± 0.50 µmol/kg, which was consistent with earlier findings (e.g., Johnson et al. 1998a; Johnson et al. 2001) that there were likely no significant analytical effects due to gas exchange with the overlying headspace of the Niskin bottles during sampling.
Tables 4 and 5 show an internally consistent data set of high quality with excellent accuracy (< or = 2.0 µmol/kg), high single-system precision (< or = 1.0 µmol/kg), and a slightly higher imprecision for the sample precisions (1.59 - 1.92 µmol/kg). Based on these data, the TCO2 data clearly meet survey criteria for accuracy (< or = 4.0 µmol/kg) and precision, and as with previous data submissions, no correction for instrumental bias or CRM analytical differences has been applied to the TCO2 data.
