Crossover Analysis for pCO2Measurements Made During WOCE Section P6
Because the pCO2 method was still under development and because of the instrument difficulties experienced during the WOCE Section P6, additional QC-QA assessment was required. Final results of the P6 pCO2 analyses were checked by comparing the deep water results with those obtained on other WOCE cruises that intersected the P6 line. With this test it was assumed that deep and bottom water results have not changed at a given location over the relatively short time interval of a few years separating the different cruises.
The stations selected for each crossover are those which are close to the crossover point and on which carbon measurements were made. The number of stations selected was somewhat subjective but was such that sufficient measurements were present for the analysis without getting too far away from the crossover location. In all cases the stations were within approximately 1° of latitude and longitude of the crossover point. Table 12 lists the stations used for each crossover.
Once the stations were chosen the results were plotted against potential density referenced to 3000 dbar (s3). Only data from pressures greater than 2500 dbar were included in order to minimize the influence of possible lateral gradients. A smooth curve was fitted to the combined station data from each leg so long as there were seven or more data points that could be used for the fit. The fitting curve chosen was a "robust loess" function designed to minimize the influence of outliers (Feely et al. 1999). In cases having fewer than 7 points, linear segments were used to "connect the dots." Only data which had been marked with a quality control flag of 2 (good) or 6 (replicate) were included in the analyses. Reported pCO2 results were converted to fCO2 using the Weiss function (Weiss 1974) and the measured temperature prior to the comparison. fCO2 values are not reported in this NDP.
In order to quantitatively estimate the mean difference between legs, each of the two fitted curves was evaluated at 50 evenly spaced intervals covering the range of space common to the selected stations from both legs. The 50 differences were then averaged. Table 12 summarizes the mean differences and standard deviations for each crossover, and indicates the differences in terms of the cruise leg designations.
Samples from one of the intersecting cruises, P18, were analyzed by Rik Wanninkhof of the Atlantic Oceanographic and Meteorological Laboratory while the other three were analyzed by scientists from Taro Takahashi's group of LDEO. Previous crossover tests have indicated that the P18 fCO2 results may be somewhat low as a result of minor sample loss, and the checks preformed at CDIAC tend to support that finding. Differences between the P6 results and those measured by the LDEO group are well within the precision (Table 12) of the technique.
Based on these checks, the prototype pCO2 method appears to have performed adequately, and no additional corrections to the pCO2 data have been made. Nevertheless, the P6 Section work showed that the pCO2 method could be improved. Before the next deployment in 1994, the pCO2 method incorporated automated and highly accurate measurements of P, Peq, and analytical temperatures throughout, rigorous control of the phase volume ratio and the headspace gas composition prior to equilibration, and new software (Neill et al. 1997). Figure 7 summarizes the analytical results as a contour section plot of the pCO2 data from the WOCE Section P6.
