Discrete pCO₂ Measurements

The discrete pCO₂ data were measured by Dr. Rik Wanninkhof's group from NOAA/AOML only on the P16N_2006 cruises. No discrete pCO₂ measurements were made on P16S_2005. The pCO₂ system is patterned after the instrument described in Chipman et al. (1993) and is discussed in detail in Wanninkhof and Thoning (1993) and Chen et al. (1995). The major difference between the two systems is that the Wanninkhof instrument uses a LI-COR (model 6262) non-dispersive infrared analyzer, while the Chipman instrument uses a gas chromatograph (GC) with a flame ionization detector.

Samples were drawn from Niskin bottles into 500-mL volumetric flasks using Tygon tubing with a silicone adapter that fit over the petcock to avoid contamination of DOM samples. Bottles were rinsed while inverted and filled from the bottom, overflowing half a volume taking care not to entrain any bubbles. About 5 mL of water was withdrawn to allow for expansion of the water as it warms and to provide space for the stopper, tubing, and frit of the analytical system. Saturated mercuric chloride solution (HgCl₂)(0.2 mL) was added as a preservative. The sample bottles were sealed with a screw cap containing a polyethylene liner. The samples were stored in coolers at room temperature usually for no more than 5 h. Generally, when samples were taken from the Niskin bottles, flasks were drawn on all the Niskins including four duplicates. Two of the duplicates were analyzed at different analysis temperatures. The duplicates run at different temperatures were normalized to 20°C and compared. Normalization was performed using the constants and procedures as outlined in Peng et al. 1987 as incorporated in the GW BASIC data reduction program. Three types of duplicates were taken. The average difference:

Standard deviation and number for the three types are listed below:

The omitted values were due to a problem in analysis in one of the duplicates.

Using the constants as refit by Dickson and Millero and the salinity dependence of borate as proposed by Dickson gave an average difference of 1 %, that is, these constants yielded worse agreement in temperature normalization than using the constants listed in Peng et al. (1987).

Once the samples reached the analyses temperature, a 50-mL headspace was created by displacing the water using a compressed standard gas with a CO₂ mixing ratio close to the anticipated pCO₂ of the water. The headspace is circulated in a closed loop through the infrared analyzer that measures CO₂ and water vapor levels in the sample cell. The samples are equilibrated until the running mean of 20 consecutive 1-second readings from the analyzer differs by less than 0.1 ppm (parts per million by volume). This equilibration takes about 10 min. An expandable volume in the circulation loop near the flask consisting of a small, deflated balloon keeps the headspace of the flask at room pressure.

In order to maintain analytical accuracy, a set of 6 gas standards is run through the analyzer before and after every 10 seawater samples. The standards were obtained from Scott-Marin and referenced against primary standards purchased from C.D. Keeling in 1991, which are on the World Meteorological Organization (WMO)-78 scale. The cylinder serial numbers and mole fractions of CO₂ with balance artificial air are:

The calculation of pCO₂ in water from the headspace measurement involves several steps. The CO₂ concentrations in the headspace are determined via a second-degree polynomial fit using the nearest three standard concentrations. Corrections for the water vapor concentration, the barometric pressure, and the changes induced in the carbonate equilibrium by the headspace-water mass transfer are made. The corrected results are reported at the analytical temperature and at a reference temperature of 20°C.

No instrumental problems occurred during the cruise. The relatively time-consuming analyses and the presence of only one analyst limited the spatial coverage. Sampling and analyses focused on precision and accuracy rather than high throughput.