Description of Variables and Methods
The data file met153.dat (File 5) (see description of data files in Part 2) in this numeric data package (NDP) contains the following variables: station numbers; cast numbers; sample numbers; bottle numbers; CTD pressures, temperatures, salinities, and oxygen; potential temperatures; bottle salinities; concentrations of dissolved oxygen, silicate, nitrate, nitrite, and phosphate; TCO2 and TALK concentrations; partial pressure of CO2 (pCO2) measured at 20oC; CFC-113; CClCO4; CFC-12; CFC-11; and quality flags. The station inventory file m153sta.inv (File 4) (Part 2) contains expocodes, section numbers, station numbers, cast numbers, sampling dates (i.e., day, month, year), sampling times, latitude, longitude, and bottom depth for each station.
Water samples were collected by a General Oceanics rosette equipped with 24 10-L bottles mounted on a Neil Brown Mark III CTD instrument (IFMK numbers NB3 and NB2) equipped with an O2 sensor and bottom alarm. Using IFMK software by L. Bellach, pressure, temperature, conductivity, oxygen, and sensor temperature data were recorded on a PC at the full sampling rate of 32 Hz in binary form and at a reduced sampling rate of 3 Hz on a Micro Vax computer. Reversing thermometers were included on bottles 1 and 23 (deepest and mixed-layer sample); each rack consisted of three thermometers. The CTD pressure, temperature, oxygen, and conductivity data were processed and corrected according to laboratory calibrations in 1990 and 1991, and in situ measurements, according to procedures written by Ruhsam (1994) and Siedler and Zenk (1992). Pressure values are expected to be accurate to ± 3 dbar; temperature to ± 0.002oC. Salinity for selected Niskin bottles (about one in every three) was also determined on a Guildline Autosal model 8400A, that was standardized weekly with International Association for the Physical Sciences of the Ocean (IAPSO) water (batch P112). These data were also used to process the CTD data, and the final salinity data are expected to be accurate to ± 0.002 on the Practical Salinity Scale (PSS). Oxygen was determined on each Niskin bottle by the Winkler method as modified by Grasshoff et al. (1983). Duplicates were taken periodically to estimate the accuracy and precision of the entire sampling procedure, which was determined to be ± 1 µ mol/kg. The concentrations of nitrate, nitrite, phosphate, and silicate dissolved in seawater were determined on a continuous flow analyzer: the Alpkem Corporation RFA 300, which was used in conjunction with a data acquisition system supplied by Oregon State University. The analyses were completed within 24 h after sampling.
The TCO2 was determined using an automated coulometric system (SOMMA) (Johnson et al. 1985; Johnson et al. 1987; Johnson and Wallace 1992). Some 753 individual samples, along with 145 duplicates from 29 stations (Fig. 2), were collected in 300-mL precombusted (450oC for 24 h) bottles and immediately poisoned with HgCl2 according to the DOE Handbook of Methods (DOE 1994). Before analysis, samples were kept in darkness until thermally equilibrated to the pipette temperature. CRM supplied by Andrew G. Dickson, of SIO (DOE 1994), were also analyzed. CRMs are filtered sterile salt solutions or seawater spiked with Na2CO3, analyzed for TCO2 concentration by vacuum-extraction/manometry in the laboratory of Charles D. Keeling at SIO.
For analysis, seawater introduced from an automated "To Deliver" pipette into a stripping chamber was acidified, and the resultant CO2, after drying, was coulometrically titrated on a model 5011 UIC coulometer. In the coulometer cell the 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 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. For system calibration, a gas calibration procedure using pure CO2 was built into the SOMMA. The hardware, located upstream of the stripper, consisted of an eight-port Gas Sampling Valve (GSV) with two sample loops connected to a source of pure CO2 through an isolation valve; the vent side of the GSV was plumbed to a barometer. When a gas loop was filled with CO2, the mass (moles) of CO2 contained therein was calculated by dividing the loop volume (V) by the molar volume of CO2 at the ambient T and P. The molar volume of CO2 [V(CO2)] was calculated iteratively from the expression:
V(CO2) = RT / P[1+ B(T) / V(CO2)] ,
were P is the instantaneous barometric pressure, T is the loop temperature, and B(T) is the first virial coefficient for pure CO2. The ratio of the calculated mass to that determined coulometrically was the gas calibration factor (CALFAC) used to correct the subsequent titrations for small departures from 100% theoretical response. The volume of the loops was determined gravimetrically with deionized water by the method of Wilke et al. (1993).
The "to deliver" volume (TDV) of the SOMMA sample pipette was determined gravimetrically with milli-Q deionized water degassed with helium. The thermostatted sample pipette was filled with water at the same temperature and then discharged into preweighed 50-mL serum bottles that were reweighed on a model R300S (Sartorius, G ttingen, Germany) balance. The apparent weight (g) of water collected (Wair) was corrected to the mass in vacuo (Mvac) from
Mvac = Wair + Wair (0.0012/d - 0.0012/8.0) ,
where 0.0012 is the sea level density of air at 1 atm, d is the density of the calibration fluid at the pipette temperature and sample salinity, and 8.0 is the density of the stainless steel weights. The TDV was
TDV = Mvac /d .
The precruise calibrated TDV of the pipette was 28.7113 ± 0.003 mL (n = 8) at 20oC. During the cruise, 52 preweighed serum bottles were filled from the pipette. They were sealed and returned to the laboratory for reweighing. The mean volume from these bottles was 28.7172 ± 0.0096 at 20oC. The mean difference between the precruise and postcruise results was 0.0059 mL which is less than the standard deviation of the 52 postcruise weighings; accordingly, TCO2 was calculated using the precruise volume of 28.7113 mL.
An IBM compatible personal computer with two RS232 serial, a 24-line digital input/output, and analog-to-digital ports was used to control the coulometer, barometer, solid state control relays, and temperature sensors, respectively. The temperature sensors (model LM34CH, National Semiconductor, Santa Clara, California) with a voltage output of 10 mV/oF were calibrated against thermistors certified to 0.01oC (PN CSP60BT103M, Thermometrics, Edison, New Jersey) with a certified mercury thermometer as a secondary standard. These sensors monitored the pipette, gas sample loop, and the coulometer cell temperatures. The barometer, model 216B-101 Digiquartz Transducer (Paroscientific, Inc., Redmond, Washington) was factory-calibrated for pressures between 11.5 and 16.0 psia. The SOMMA software was written in GWBASIC Version 3.20 (Microsoft Corp., Redmond, Washington), and the instrument was driven from an options menu appearing on the personal computer monitor.
Titrations were done with the coulometer in the counts mode: the total charge passed during a titration was displayed as the total number of counts accumulated by the coulometer's voltage- to-frequency converter (VFC). From the factory calibration of the VFC [frequency = 105 pulses (counts) generated per second at 200 mA] and the value of a Faraday (96489 coulomb/mol), a scaling factor of 4.82445 x 103 counts per micromole was derived, and the micromoles (M) titrated were
M = counts/4824.45 - (blank x TT) ,
where TT was the length of the titration in minutes and blank was the system blank in micromole per minute. The total carbon dioxide concentration in µ mol/kg was calculated as follows:
TCO2 = [M(CALFAC)(1000/TDVC x p)] x 1.00017 ,
where CALFAC is the gas calibration factor, TDVC is the "to deliver" volume of the pipette in milliliters corrected for the thermal expansion of glass, p is the density of sea water in kilograms per liter from the equation of state (Millero and Poisson 1981), and 1.00017 corrects for the dilution of the sample by addition of 100 µ L of HgCl2 solution to the 300-mL sample bottle. Precision for a set of analyzed samples was expressed as the square root of the pooled variance (Sp2):
The precision for the A9 samples was estimated from 90 replicates collected from 18 deep-water samples analyzed throughout the cruise. The Sp2 was ± 0.83 µ mol/kg (Johnson et al. 1993). Analytical accuracy was verified by the analysis of 11 CRMs (batch 2) during the cruise; certified TCO2 was 1978.78 ± 0.93 (n = 9) µ mol/kg. The mean and standard deviation for the CRM analyzed at sea on the SOMMA was 1978.1 ± 0.82 µ mol/kg. Table 1 lists the CRM data. Note the excellent agreement between Sp2 obtained from sample duplicates ( ± 0.83 µ mol/kg) and the CRM precision (± 0.82 µ mol/kg) and the close agreement between the CRM results for February 13 and those for March 20 some 6 weeks later. Unfortunately, only 11 CRM from batch 2 were available for analysis. Because the batch 3 CRMs supplied for the cruise were found to be unstable and uncertifiable, data from this batch cannot be used to evaluate the performance of the TCO2 measurement system. Figure 3 summarizes the analytical results as a countour section plot of the TCO2 data from the A9 transect along 19oS.
Table 1. Results of the certified reference material (batch 2) shipboard analyses during
R/V Meteor Cruise 15/3 February - March 1991.
The CRM had a certified TCO2 of 1978.8 µ mol/kg and salinity of 33.361 PSS.
| CRM bottle no. | Date analyzed | System blank µ gC/min | CALFAC | Total CO2 µ mol/kg | 206 | 13.02.91 | 0.050 | 1.004596 | 1977.1 | 213 | 15.02.91 | 0.058 | 1.004005 | 1977.8 | 297 | 15.02.91 | 0.058 | 1.004005 | 1976.9 | 4 | 16.02.91 | 0.034 | 1.009080 | 1979.6 | 308 | 17.02.91 | 0.018 | 1.003203 | 1977.8 | 181 | 19.02.91 | 0.034 | 1.003736 | 1977.6 | 155 | 23.02.91 | 0.036 | 1.002476 | 1977.6 | 259 | 27.02.91 | 0.038 | 1.004088 | 1979.2 | 292 | 06.03.91 | 0.031 | 1.004465 | 1978.3 | 237 | 12.03.91 | 0.030 | 1.003876 | 1978.3 | 156 | 20.03.91 | 0.039 | 1.003915 | 1978.4 | Mean standard deviation | 1978.1 ± 0.82 |
|---|
Replicate samples from ten Niskin bottles at four stations were also collected for later shore- based reference analyses of TCO2 by vacuum extraction and manometry by Charles D. Keeling, SIO. The results (Table 2), extracted from Guenther et al. (1994), were obtained very early in the program for comparing shipboard analyses by coulometry with shore-based analyses of duplicate samples, and many mistakes and false starts were noted. For example, experimentation with the most suitable sample bottle for this purpose was not yet completed, nor storage precautions or expedited shipping procedures been worked out. Therefore, some of the differences in the seven completed comparisons listed in Table 2 probably resulted from unfamiliarity with the new RODAVISS glass bottle stoppers: some breakage was caused by overtightening; some loss of CO2, by undertightening of the stoppers. Nor were the storage conditions optimal for the data quality of the surviving samples: they were kept in a non-air-conditioned cargo hold for the remainder of R/V Meteor Cruise 15/3 and then transported to Brazil before they could be shipped to SIO. Temperature sensors were not included in the shipping crates, as is now standard operating procedure; however, temperatures likely exceeded an unacceptable 30 C in either the cargo hold during the R/V Meteor's return voyage to Brazil or in Brazil prior to shipment to SIO.
Table 2. Comparison of shipboard analyses of total carbon dioxide by coulometry (BNL)
during R/V Meteor Cruise 15/3 with the shore-based reference analyses
by manometry on
duplicate samples by C. D. Keeling,
Scripps Institution of Oceanography (SIO).
| Station no. | Sample date | Niskin no. | Depth (m) | TCO2 BNL µ mol/kg | TCO2 SIO µ mol/kg | Diff. BNL-SIO µ mol/kg | Sal. Diff. BNL-SIO PSS | 143 | 16.02.91 | 318 | 1593 | 2174.58 | 2171.58 | +2.65 | +0.014 | 143 | 16.02.91 | 312 | 2995 | 2180.38 | 2183.44 | -3.06 | +0.020 | 154 | 21.02.91 | 224 | 8 | 2084.12 | 2095.58 | -11.46 | +0.045 | 187 | 03.03.91 | 224 | 7 | 2073.05 | 2075.58 | -2.53 | +0.013 | 187 | 03.03.91 | 318 | 1194 | 2212.87 | 2217.87 | -4.91 | +0.023 | 187 | 03.03.91 | 312 | 2492 | 2188.71 | 2183.74 | +4.97 | +0.034 | 199 | 07.03.91 | 202 | 898 | 2220.60 | 2228.38 | -7.78 | +0.007 | Mean | -3.16 | +0.022 |
|---|
TALK samples were collected and poisoned with 50 µL of saturated solution of HgCl2 in 250-mL, standard borosilicate glass, screw-cap bottles. They were stored at room temperature and returned to Woods Hole for TALK analysis. TALK was determined by potentiometric titration; a method derived from one first described by Dyrssen (1965) and later modified by Bradshaw et al. (1981) was used. The automated titration was performed in a closed cell maintained at constant temperature (25 ± 0.1oC); to be similar to seawater, the ionic strength of the hydrochloric acid solution (0.1 N) was adjusted with NaCl. The ratio of the acid normality over the cell volume was calibrated before and after the sample analysis. The calibration consisted of preparing solutions of known TALK concentration and measuring them as described by Brewer et al. (1986). The precision of the measurements was estimated to be better than 0.1%. The samples were likely exposed to relatively high temperatures during shipment from Brazil to Woods Hole. Some sample bottles were also broken during shipment. Overall precision of the data set was therefore slightly degraded from that expected from measurement error alone. The pooled standard deviation for ten replicate samples was ~0.12%.
pCO2 was measured using an experimental analytical system employing a batch-equilibration, static headspace analysis technique, with detection by gas chromatography and a flame ionization detector. Subsequently, this technique has been greatly improved and a full description is presented by Neill et al. (1995). During the R/V Meteor Cruise 15/3, the developing technique gave less than optimum data quality for this parameter because not all sources of error were known. The data are presented primarily for completeness, and caution must be exercised in their quantitative interpretation.
Briefly, the technique used was based on the static headspace methane method of Johnson et al. (1990). Samples were collected in 60-mL serum bottles rinsed and filled to overflowing at the Niskin bottle. These samples were transported to a box that was purged with a flow of 350 ppm CO2 in argon. A headspace of ~ 5-mL was introduced using a disposable pipette-tip attached to a special tool (Johnson et al. 1990). The headspace was purged briefly with the argon-CO2 mixture, and a septum was placed over the serum bottle neck and crimped tightly with an aluminum cap. Headspace overpressure from crimping was relieved by piercing the septum with a needle for 3 to 4 seconds. The samples were equilibrated for 4 to 6 hours in the dark in a shaking water bath at 20oC. The experiments performed at sea indicated that there was no significant difference in the measured pCO2 of replicate samples equilibrated for periods from 2 to 9 hours. Following equilibration, the septum was pierced with two needles. The longer needle was inserted to the bottom of the serum bottle to dispense a brine solution, while the shorter one penetrated just below the septum into the headspace. Approximately 4 mL of brine solution was injected into the bottle through the longer needle displacing the headspace through the shorter one to purge and fill a small (400 µL) gas sample loop attached to the gas chromatograph. After filling, the loop was allowed to come to atmospheric pressure, temperature and pressure were recorded, and the contents were injected onto a 6 ft x 1/8 in. stainless steel chromatographic column packed with Porapak N. A methanizer column containing a nickel catalyst (Varian Inc.) mounted in the injector block of the gas chromatograph at 325oC on the terminal end of the column was used to quantitatively convert CO2 to CH4 for detection by flame ionization. Carrier gas flow rate was 30 mL/min of ultra high purity nitrogen; the methanizer was supplied with hydrogen from a hydrogen generator at 30 mL/min. The flame ionization detector was supplied with compressed air and hydrogen at 300 and 30 mL/min, respectively.
The variety of septa used were found to leak during equilibration, after they had been pierced with needles. To calculate the partial pressure of CO2 after equilibration, it is usually necessary to measure or calculate the pressure of equilibration because of the phase redistribution of gases dissolved in seawater. Subsequent testing with a wide variety of septa and improved technique showed that the pressure of equilibration can be calculated or measured accurately when the septa do not leak. However, septa used during R/V Meteor Cruise 15/3 consistently leaked, so that the pressure of equilibration was the same as the ambient atmospheric pressure.
To calculate the pCO2 of the equilibrated samples, the area of the CO2 peak was converted to a mole fraction of CO2 within the headspace from temperature and pressure measurements, and a calibration curve was obtained from injections of gas-phase CO2 standards at nominal levels of 250, 350, 750, and 1500 ppmv. Subsequently, these standards were intercalibrated against primary standards maintained at the Lamont-Doherty Earth Observatory. The mole fraction of CO2 at the measured atmospheric pressure was converted to the partial pressure of CO2 after equilibration. From the measured (unequilibrated) sample TCO2 (SOMMA), the original CO2 content of the introduced headspace, and the CO2 content after equilibration, the mass of CO2 transferred from the liquid to the gas phase, or vice versa, was calculated and used to calculate the sample TCO2 after equilibration. From TCO2 and the measured pCO2, the TALK was calculated. TALK was assumed to be conservative, and from the TALK - TCO2 pair the pCO2 of the water sample at 20oC prior to equilibration was calculated by using the thermodynamic constants of Roy et al. (1993), Weiss (1974), and published procedures (DOE 1994). In the data files pCO2 is reported at a standard temperature of 20oC. Actual equilibration temperatures generally ranged from 19.9oC to 20.2oC.
Precision varied throughout the cruise, depending primarily on the status of the catalyst, that had to be reconditioned periodically. Precision on multiple (>3) replicates varied from 0.4 to ~6% and averaged 2%. Accuracy was judged in three ways. First, throughout the cruise the mole fraction of CO2 in air was measured; air was collected in syringes at the bow of the ship during steaming between stations. The mean pCO2 of the air, expressed as a dry air mole fraction, was 353.6 (± 9.74); this compares well with contemporary measurements of 353.5 and 352.9 made by the National Oceanic and Atmospheric Administration Climate Monitoring and Diagnostics Laboratory air sampling network in February and March 1991 on air samples collected at Ascension Island (7o 55'S and 14o 25'W) (Conway et al. 1994). This agreement suggested that although instrument imprecision was high, the overall accuracy of the measurements was consistent with a completely independent set of measurements. Second, assessment of accuracy arose from samples overdetermined for the carbonate system. In this case, the TALK by potentiometric titration for samples from two stations was compared with the TALK calculated from the measured pCO2-TCO2 pair. TALK calculated from the pCO2-TCO2 pair was 4 µ mol/kg (± 12) lower than the measured TALK for these stations. Once again, this suggested relatively good accuracy but poor precision. Third, overall assessment of accuracy was made by comparing the discrete-pCO2 measurements with measurements made by Lamont-Doherty Earth Observatory (LDEO). Station 199 (19oS, 2oW) of Meteor Cruise 15/3 was compared with Station 144 (20oS, 1oW) of the SAVE expedition, at which a discrete-pCO2 profile had been collected during February 1988 by D. Chipman and T. Takahashi of LDEO. The respective profiles show very good agreement within the Angola Basin Deep Waters (>2000 m) (Fig. 4). For this depth range, the R/V Meteor 15/3 mean value of pCO2 at 20oC was 801 (± 10) compared with a SAVE value of 786 (± 11). In the upper waters, systematic differences were noted between the two profiles; however, these can be seen in the TCO2 data as well.
Exploratory measurements of anthropogenic halocarbon compounds [CCl4, CCl2FCClF2 (CFC-113), CCl3F (CFC-11), and CCl2F2 (CFC-12)] were made using a new analytical technique on R/V Meteor Cruise 15/3 (Wallace et al. 1994). The new method was jointly developed by BNL, Bedford Institute of Oceanography (Canada), and Chalmers University of Technology (Sweden). Briefly, it employs a purge-and-trap extraction technique similar to that used in previous CFC analysis systems (Gammon et al. 1982; Wallace and Moore 1985; Bullister and Weiss 1988). The most significant differences from the earlier technique were:
1.The chromatographic column used was a wide-bore DB-624 glass capillary (J&W; 70 m X 0.53 mm OD; 3-µ m film) which gives baseline resolution between the tracer compounds of interest and a variety of natural and anthropogenic halocarbons. 2.The purged volatile compounds were trapped on a short Porapak N column kept at ambient temperature (~20oC). This eliminated the need for taking cryogenic systems to sea. Because CFC-11 and CFC-12 were measured separately by a group from the university of Bremen, the BNL system was optimized for measuring low levels of CCl4, CH3CCl3, and CFC-113 by using a 20-mL water sample and increasing the purge-gas flow to 5 min at 60 mL/min. Extraction efficiency was > 99% for all compounds except CH3CCl3 (~85%). These conditions caused CFC-12 to approach breakthrough on the trap, decreasing precision and accuracy for this compound. Hence, all CFC-12 (and most CFC-11) data in this NDP are based on measurements obtained by using the separate packed column system, which employed low-temperature trapping.
Two unexpected problems were encountered: a partial chromatographic interference for CFC-113 due to extremely high levels of CH3I in tropical near-surface waters and a second, more serious problem, arising from a buildup of water on the column, which caused large negative peaks and an interfering baseline shift in the vicinity of the CFC-113 peak. Both of these problems have subsequently been corrected; however, they greatly reduced the number of samples that could be obtained.
akozyr 9/05/96
