Shore-based Reference Analyses
The replicate samples from 100 Niskin bottles at 18 stations were collected for shore-based reference analyses at the laboratory of C. D. Keeling of SIO. The TCO2 measurements were produced by vacuum extraction/manometric analysis and the TALK values by potentiometric titration. Both measurements were performed under controlled laboratory conditions using standards. The replicate sample standard deviation (s) for this large data set of 71 unflagged pairs is 1.0 µmol/kg after omitting the three replicate pairs with deltas greater then 3s (a replicate sample standard deviation calculated from the set of analyses on duplicate samples) (Guenther et al. 1994b).
Substantial reduction in the calculated s of the ship-minus-shore comparison is made by omitting 24 comparisons of singlet replicate samples, plus 5 more that are greater than 3s for either the replicate pairs or the comparison difference. For the 66 remaining comparisons, the average ship-minus-shore difference is 2.12.4 µmol/kg (Guenther et al. 1994b). Figure 4 shows the ship- minus-shore differences for all available surface and deep data from the TUNES-3 Expedition. The plotted data indicate a bias for surface data relative to deep data; surface data show better agreement between shipboard and shore-based data than do deep data.
Table 1 and Table 2, reprinted from Guenther et al. 1994b, summarize the replicate shore-based measurements of TCO2 and TALK and their comparisons with shipboard measurements.
