Coulometric total carbon dioxide analysis for marine studies: assessment of the quality of total inorganic carbon measurements made during the US Indian Ocean CO sub(2) Survey 1994-1996

• Published in: Marine chemistry Vol. 63; no. 1-2; pp. 21 - 37
• Main Authors: Johnson, K M, Dickson, A G, Eischeid, G, Goyet, C, Guenther, P, Key, R M, Millero, F J, Purkerson, D, Sabine, CL, Schottle, R G, Wallace, DWR, Wilke, R J, Winn, C D
• Format: Journal Article
• Language: English
• Published: 01-12-1998
• Subjects: Marine
• ISSN: 0304-4203

Two single-operator multiparameter metabolic analyzers (SOMMA)-coulometry systems (I and II) for total carbon dioxide (TCO sub(2)) were placed on board the R/V Knorr for the US component of the Indian Ocean CO sub(2) Survey in conjunction with the World Ocean Circulation Experiment-WOCE Hydrographic Program (WHP). The systems were used by six different measurement groups on 10 WHP Cruises beginning in December 1994 and ending in January 1996. A total of 18,828 individual samples were analyzed for TCO sub(2) during the survey. This paper assesses the analytical quality of these data and the effect of several key factors on instrument performance. Data quality is assessed from the accuracy and precision of certified reference material (CRM) analyses from three different CRM batches. The precision of the method was 1.2 mu mol /kg. The mean and standard deviation of the differences between the known TCO sub(2) for the CRM (certified value) and the CRM TCO sub(2) determined by SOMMA-coulometry were -0.91 plus or minus 0.58 (n=470) and -1.01 plus or minus 0.44 (n=513) mu mol/kg for systems I and II, respectively, representing an accuracy of 0.05% for both systems. Measurements of TCO sub(2) made on 12 crossover stations during the survey agreed to within 3 mu mol/kg with an overall mean and standard deviation of the differences of -0.78 plus or minus 1.74 mu mol/kg (n=600). The crossover results are therefore consistent with the precision of the CRM analyses. After 14 months of nearly continuous use, the accurate and the virtually identical performance statistics for the two systems indicate that the cooperative survey effort was extraordinarily successful and will yield a high quality data set capable of fulfilling the objectives of the survey.