Internal Consistency of GLODAP Indian Ocean Database

Adapted From:
Sabine, C.L., R.M. Key, K.M. Johnson, F.J. Millero, A. Poisson, J.L. Sarmiento, D.W.R. Wallace and C.D. Winn. 1999. Anthropogenic CO₂ inventory of the Indian Ocean. Global Biogeochemical Cycles, 13:179-198.

The internal consistency of the Indian Ocean cruises (Table 1) was examined by comparing carbon values in the deep waters (pressure > 2500 dB) at the intersections of different legs following the procedures described in the crossover study.  The mean and standard deviation of the difference in TALK and TCO₂ at the 35 intersections identified in Figure 2 are shown in Figure 3. The long-term stability of the WOCE/JGOFS measurements can be estimated from the first 17 crossover results. The mean of the absolute values for the leg-to-leg differences was less than the estimated accuracy for both TCO₂ (1.8±0.8 µmol/kg) and TALK (2.4±1.6 µmol/kg). Although there is only one reliable crossover point between the WOCE/JGOFS cruises and the CIVA1 (I6S) cruise, the differences for both parameters are within the estimated accuracy of the measurements. Results from the analysis of Certified Reference Materials (CRMs) on the CIVA1 cruise also support the quality of the measurements. Some of the older INDIGO cruises, however, did appear to have offsets relative to the WOCE/JGOFS and CIVA1 data. INDIGO I and II TALK values averaged 6.5 µmol/kg high and 6.8 µmol/kg low, respectively, while the INDIGO III TALK values showed no clear offset. The INDIGO TCO₂ values were all consistently high relative to WOCE/JGOFS and CIVA1, with differences of 10.7, 9.4 and 6.4 µmol/kg, respectively. These offsets are consistent with differences observed between at-sea values and replicate samples run at C.D. Keeling's shore-based TCO₂ facility. Since the INDIGO cruises were run prior to the introduction of CRMs, these offsets were presumed to be calibration differences and each leg was adjusted to bring the values in line with the remaining cruises. The dotted boxes in Figure 3 show the original offsets at the crossovers. The solid boxes show the final offsets for the GLODAP database. The mean of the absolute values for the leg-to-leg differences for all 35 crossover analyses suggest that the final data set is internally consistent to 2.2 and 3.0 µmol/kg for TCO₂ and TALK respectively.

GEOSECS

Since CRMs were not available at the time of GEOSECS, the only way to infer consistency with the WOCE data set is to assume the deep water carbon distributions have not changed since GEOSECS survey. The most reliable way to compare the two data sets is to examine the difference between the predicted TCO₂ and the measured TCO₂ (excess CO₂) in deep waters. The basic assumption with this technique is that the correlation between the different hydrographic parameters in the deep waters does not change with time. Given the long residence time of the deep and bottom waters in the ocean, this should be a reasonable assumption. This technique has the advantage that it implicitly accounts for the possibility of real variability in hydrographic properties between the two expeditions which would not be taken into account by simply comparing carbon profiles.

Examination of the excess CO₂ values in waters that should be free of anthropogenic CO₂ [pressures > 2000 dbar, containing no detectable chlorofluorocarbons (CFCs)] revealed that the GEOSECS values were 22.5 ± 3 µmol/kg higher than the comparable WOCE measurements. This difference is comparable to the correction of -18 ± 7 µmol/kg noted by Weiss et al. (1983) to make the TCO₂ measurements consistent with the TALK and discrete pCO₂ measurements based on the Merbach et al. (1973) dissociation constants. Additional support for an adjustment of the original GEOSECS data comes from C. D. Keeling's shore-based analysis of TCO₂ samples collected on both the GEOSECS and the WOCE/JGOFS expeditions. Weiss et al. (1983) point out that the shore-based analyses were systematically smaller than the at-sea measurements by 16.5 ± 5 µmol/kg during GEOSECS. Comparable comparisons between the WOCE/JGOFS at-sea measurements and Keeling's shore-based analyses indicate that the shore-based samples are approximately 5 µmol/kg higher than the at-sea values. Together, the GEOSECS-Keeling-WOCE/JGOFS combination suggests an offset of 21.5 µmol/kg between GEOSECS and WOCE/JGOFS at-sea measurements. It is also important to note that there is no indication of a depth- or concentration-dependent correction for the GEOSECS data. The WOCE shore-based comparison, based only on samples collected at the surface, is within 1 µmol/kg of the deep comparison described above. Based on these results, a constant correction of the -22.5 µmol/kg should be applied to the original reported GEOSECS TCO₂ values to improve the consistency with theWOCE/JGOFS data sets.

References

• Merbach, C., C.H. Culberson, J.E. Hawley and R.M. Pytkowicz. 1973. Measurements of the apparent dissociation constants of carbonic acid in seawater at atmospheric pressure. Limnology and Oceanography. 18:897-907.
• Weiss, R.F., W.S. Broecker, H. Craig and D. Spencer. 1983. GEOSECS Indian Ocean Expedition Vol. 5, Hydrographic Data 1977-1978, US Government Printing Office, Washington DC, 48 pp.