New technologies developed since the 1980's have contributed to this understanding. BIOS scientists have aided their development by deploying and recovering new technologies in the ocean during their test phases. Satellite launches and advances in computing power have enabled researchers to realistically simulate ocean and atmospheric circulation, and small-scale features such as ocean currents and eddies. One such eddy of warm water had a significant effect on sea level around Bermuda for several weeks in 2002. Seagoing and laboratory instruments for the measurement of dissolved inorganic carbon (DIC), alkalinity, underway seawater pCO₂ and suspended POM analyses exist in the Marine Biogeochemistry Laboratory at BIOS.

Samples for DIC and alkalinity are drawn from Niskin samplers into individually numbered, clean 1 dm3, or 0.5 dm3 size Pyrex glass reagent bottles, using established gas sampling protocols (Dickson and Goyet, 1994). DIC samples are typically taken after dissolved oxygen sampling on routine CTD hydrocasts. Care is taken to minimize turbulence and prevent the retention of air bubbles in the bottles when sampling. A headspace of <1% of the bottle volume is left to allow for water expansion and all samples are poisoned with 100 or 200 µl of saturated HgCl₂ solution to prevent biological alteration. Bottles are closed with ground-glass stoppers and sealed with Apiezon silicon vacuum grease. Rubber bands were placed around the lip of the bottle and stopper to provide positive closure.

The most commonly used methods for determining DIC are based on extracting CO₂ from natural waters. A known volume or weight of seawater sample is acidified with phosphoric acid. This converts all bicarbonate, and carbonate to free CO₂, which is then transferred via a carrier gas to a detector, which is usually a coulometer or gas chromatograph. Three instruments for the measurement of dissolved inorganic carbon (DIC) are available in the lab at BIOS.

A SOMMA-Coulometer system (Figure 1) has been used at BIOS for the last 15 years, providing highly precise (~0.02%; ~0.4 µmoles kg^-1) and accurate DIC measurements (Bates et al., 1996). Since August 1991, we have used a SOMMA (Single-Operator Multi-Metabolic Analyzer) to control the pipetting and extraction of seawater samples. Our system is similar to the one described by Johnson et al. (1993). The SOMMA is interfaced with a personal computer and coupled to a CO₂ coulometric detector (model 5011, supplied by UIC Coulometrics Inc.). This method offers highly reproducible and rapid sample analyses (15 to 20 minutes).

During analysis, a known volume of seawater sample is acidified with 8.8% v/v reagent grade phosphoric acid, and converting all carbonate species to free CO₂. The evolved CO₂ is then extracted from seawater using an inert carrier gas such as ultra high purity N₂ (99.9995%) or Helium (99.999%), and absorbed by a coulometer cell solution, containing ethanolamine, dimethylsulfoxide (DMSO) and thymolphthalein indicator. Reaction between CO₂ and ethanolamine produces hydroxyethylcarbamic acid, causing a colour change in the solution. Hydroxyl ions are electrochemically generated to maintain the absorbing solution at a constant, colorimetrically defined pH. Current generated by this titration is related by the Faraday constant to the moles of CO₂ absorbed by the solution (Johnson et al., 1993). Since the electrical calibration of the coulometer is not perfectly accurate, the SOMMA system was routinely calibrated with known volumes of ultra high purity CO2 gas (99.998%) contained within two stainless steel loops (Johnson et al., 1993; Wilke et al., 1993).

The SOMMA-coulometer instrument has been widely used for ocean carbon cycle studies over the last 15 years. It is transportable, rugged, sea-going, and reliable; calibration and certification procedures were completed, tested, and published in the early 1990's. The current SOMMA system semi-automatically controls the sample handling and extraction of CO₂ from seawater samples. It is coupled to a coulometer (A; Figure 1) to detect the amount of extracted CO₂.

The SOMMA system components include sample dispensing and extraction, gas calibration (H) and barometer, sample temperature regulation (J), electronic interfacing (I), and conductance cell (C). The SOMMA system is calibrated with pure CO2 gas: a known volume of gas contained within a stainless steel loop attached to the gas sampling valve (H) is carried to the coulometer for detection. A SeaBird (SB Electronic, Inc.) conductance cell can be used to determine the salinity of a sample cell (C). The SOMMA system coulometric cell solutions get exhausted after some time, and therefore, preparation of new cell solutions and recalibration of the SOMMA system is necessary (note: this has to be done daily and takes ~1-2 hours including gas calibration). The coulometric method does not readily lend itself to autonomous applications.

A means of assessing the accuracy of DIC data is provided by routine analysis of seawater certified reference materials (CRM standards, supplied by Andrew Dickson, Scripps Institution of Oceanography). CRM's have a certified DIC concentration determined by an extraction/manometric technique at C.D. Keeling’s laboratory, Scripps Institution of Oceanography (now continued by A.G. Dickson). An accuracy of better than 0.025% (~0.45 µmoles kg-1) was maintained for DIC analyses at the BATS time-series site between 1991 and 2005.

A second, combined DIC and alkalinity VINDTA system (model VINDTA 3C; http://www.marianda.com/prod02.htm) also has similar precision and accuracy (Figure 2). Both instruments are based on coulometric detection and have a sample throughput of ~4 samples per hour. The VINDTA 3 C is built by Marianda Co. in Kiel, Germany by Dr. Ludger Mintrop.

The VINDTA 3C system is based on the coulometric procedure outlined above for the SOMMA system. The VINDTA 3C has a precision of ~0.3 µmoles kg-1 (~0.015%). Unlike the SOMMA system, it is calibrated with CRM samples, but a CO₂ gas calibration module can be easily fitted. The VINDTA 3C is controlled by user-friendly Labview software.

A third DIC analyzer is based on using a LiCOR 6262 NDIR analyzer as the detector (Figure 3). This instrument has a precision of ~0.07% (~1.5 µmoles kg-1) and will also be available for the research program. The prototype AMICA builds upon a proof of concept system first built by Gernot Friederich at MBARI in 1995 (designed for high frequency sampling of the Central California coast upwelling system; Walz and Friederich, 1996; Friederich et al., 2002) and copied by Prof. Jon Sharp at the University of Delaware in 2001. The prototype instrument used by Sharp was used for estuarine surveys with a precision of ~2-3 moles kg^-1. A second copy was built at Delaware in 2004 with minor hardware improvements. At BIOS, the AMICA prototype has been able to give replicate precision on the order of <1-1.5 µmoles kg-1 for oceanic and metabolic samples.

The AMICA prototype DIC system can analyze a wide range of natural water samples; from small volume, pore water samples to open ocean seawater samples. The DIC analyzer components include sample dispensing and extraction, and detection of CO₂ using a LiCOR IR analyzer (Figure 3). Water samples (G; refer to Figure 3) are drawn into a Kloehn digital syringe (A) with a 2.5 ml syringe using a 1 ml draw. The syringe and stripping chamber (B) are rinsed three times to waste (H); the two-way pinch valve (B’) is open to the waste (H) and closed to the dryer (C) for this operation. After the rinses are completed, 5% phosphoric acid (F) is added to the empty chamber (B) and the pinch clamp (B’) switches so that the waste (H) line is closed and the gas will pass into the dryer (C). A 1 ml sample is then injected slowly into the stripper chamber and DIC converted to CO₂.

The carrier gas (O2 or N2 can be used) transfers the evolved CO₂ through a magnesium perchlorate dryer (C) to the detector cell of the LiCOR (D). A mass flow controller calibrated for O₂ (or N₂ or other inert gases) (I) maintains a steady flow of carrier gas into the stripper chamber and the LiCOR cell allowing an integration of the CO₂ signal; a faster flow rate is set during the rinse steps. A laptop computer currently controls the sample dispensing and analysis using a Visual Basic program with communication to the digital relay switches (K) via the parallel port. The relay switches (K) toggle the acid pump (E), the pinch clamp (B’), and the signal from the serial port between the mass flow controller (I) and the Kloehn syringe computer (A). A second serial port

The most commonly used methods for determining alkalinity (TA) is by potentiometric titration, similar to methods first described by Dyrssen (1965) and modified by subsequent workers (e.g. Edmond, 1971; Bradshaw et al., 1981; Brewer et al. 1986, Dickson and Goyet, 1994). Although updated, the alkalinity determination is similar to those methods developed decades earlier. In our lab, one automated VINDTA system (model VINDTA 2S; built by L. Mintrop at Marianda Co.) is used for the measurement of alkalinity (Figure 4). The combined VINDTA system (Figure 2) is also used for alkalinity measurements. Manual titration systems are also present in the lab and now used for test purposes.

In principle, the VINDTA titration systems consisted of a water-jacketed, glass titration cell that is closed with a glass top, containing separate glass and calomel reference electrodes, a thermister (readable to 0.1°C), and a capillary tube that supplies acid from a burette. A pH meter (Brinkman model Titrino), readable to 0.1 mV, monitors the titration and a motor driven piston burette (reproducible to + 0.001 cm³), delivers acid to the cell. Both cell and seawater samples were typically maintained at known temperatures using a thermostated circulator.

A known amount of seawater is dispensed into the cell, and titrated with hydrochloric acid past the carbonic acid endpoint. The acid titrant, approximately 0.10-0.12N hydrochloric acid (HCl), is prepared in a solution of alkimetric standard sodium chloride (NaCl) of approximately 0.6 mol dm^-3, its ionic strength adjusted to that of seawater and calibrated against solutions of Na₂CO₃ (Goyet and Hacker, 1992). Titration data past the carbonic acid end point (~4.5 pH) were used to calculate TA, using a non-linear least squares approach. The analytical precision of the measurement was better than 0.05% (0.4 µmoles kg^-1). The accuracy of the measurement is maintained by routine analyses of CRM samples.

We have three underway seawater pCO₂ systems in the lab. One is installed on the M/V Oleander (Figure 5), one installed on the R/V Weatherbird II, and a third older system now used in the lab. Another autonomous surface seawater pCO₂ system will be available for measurements from the underway seawater stream on one of the ships. More detail on this sensor is available here.

The M/V Oleander transits between New Jersey and Bermuda every week. Seawater is pumped from the Oleander through a NOAA thermosalinograph. Automated underway measurements of CO₂ mole fractions in dry air (xCO₂) were made using a Licor NDIR analyzer (model 6261 or 7000) coupled to a equilibrator that allowed gases in the headspace air and seawater to equilibrate. Seawater is continuously pumped into an equilibrator at flow rates of 8 to 12 L min^-1. During analysis, a small amount of air is recirculated between the LiCor cell and the equilibrator. Atmospheric air is pumped from an intake on the bow of the ship. Seawater equilibrated air and atmospheric air passed through mass flow controllers and a reverse flow naphion dryer (PermaPure flushed with pure nitrogen gas) to remove water vapour, before entering the sample cell of the Licor (Figure 5). Three gas standards (compressed CO₂-in-dry-air standards with mole fraction values of 250, 370, and 450 ppm), one reference gas (N₂), three air samples, and multiple headspace samples from the equilibrator were analyzed every three hours. CO₂ in air standards are calibrated with World Meteorological Organization (WMO) standards of CMDL. Typical precision for each measurement cycle of equilibrated headspace air and atmospheric air (during tests; 6 replicates per cycle) were +0.4 µatm and +0.1 µatm (one standard deviation), respectively.