4.2 Narrative of TAlk Adjustments

Alkalinity data showed greater differences than DIC data showed between cruises. Crossovers where deviations were greater than 6 µmol/kg were checked in greater detail. Although adjustments of less then 6 µmol/kg are discussed in the following narrative, we chose not to apply these adjustments, given the uncertainty of the overall data set and the reliability of the assessments at that level of accuracy. That is, for highly precise data sets it is possible to recommend adjustments of less than 6 µmol/kg, but the uncertainty for imprecise data is such that adjustment cannot be recommended with confidence.

A01E. There was only one crossover available for evaluation of the A01E alkalinity data. The crossover showed values that were 59 µmol/kg greater than A16N with a standard deviation of 17. Such a large offset along with large deviation make the data highly questionable for our purposes. Therefore, it is recommended that the TAlk for cruise A01E not be used.

A01W. For A01W, there were three crossovers for TAlk. Cruise A01W showed values that were systematically low by 4 to 16 µmol/kg. The regional analysis showed an offset of 14 µmol/kg in the same direction, albeit with a large standard deviation. Therefore, we recommend an upward correction of 14 µmol/kg for the A01W TAlk values. The TAlk data for this cruise are very noisy, with apparent random scatter of ±20 µmol/kg.

A01WR. A01WR had one crossover for TAlk and a significant section of overlap with A01W. The crossover showed a difference of 26 ± 4 µmol/kg with A01W. The crossover occurs in a water depth of 300 m, however. The comparison of overlapping cruise tracks between A01W and A01WR using the MLR technique shows a difference of 12 µmol/kg. This difference is in accord with the proposed adjustment of A01W of 14 µmol/kg and suggests that A01WR TAlk data do not need to be adjusted. The regional MLR showed an offset of -5.6 µmol/kg . Inspection of the data showed significant scatter with depth and changes from station to station of 10-20 µmol/kg. No adjustment is recommended for A01WR, but it must be recognized that the TAlk data for both A01W and A01WR are of dubious quality.

A02. A02 had four crossovers for TAlk. A02 showed a positive anomaly of 16 µmol/kg with A01W, which was reduced to 2 µmol/kg after the proposed adjustment to A01W. A positive anomaly of 11 µmol/kg was also observed with the A20 crossover. However, two A24 crossovers showed an offset of 3 to 4 µmol/kg, and the MLR indicated good agreement. Given the inconsistent results, no adjustment is recommended.

A06 and A07. Based on nine crossovers, the offsets and scatter are such that we do not recommend including the data from A06 or A07 in the synthesis product.

A08. The two crossovers suggested that the values for A08 were 4 to 12 µmol/kg low. The regional MLR was 3 mol/kg low, but we do not feel that there was sufficient evidence to suggest an adjustment.

A09. A single crossover with A17 suggested that the values are 7 µmol/kg high. The regional regression implied a 10-µmol/kg bias in the same direction. A downward correction of 7 µmol/kg is proposed for A09 TAlk.

A10. There were three crossovers with TAlk for A10. A10 showed a large negative bias with respect to A17, but crossovers with A16S and A15 did not indicate a significant offset. The MLR did not suggest an offset either, so no adjustment is recommended.

A14. There were three crossovers of A14 with other cruises where TAlk was measured. It showed a 12-µmol/kg high offset with A08 but was only 3 µmol/kg high relative to A09. Both the tropical and southern MLR analyses suggested that values were high by about 5 to 6 µmol/kg. The internal consistency also suggested that the TAlk could be high by 5 µmol/kg. An adjustment of 5 µmol/kg would bring values in better agreement with all analyses, but because this adjustment was less than the 6-µmol/kg criterion, no adjustment was proposed for this synthesis.

A15. There were four crossovers for A15. A15 showed both 8-µmol/kg high values (with respect to A16S) and 15-µmol/kg low crossover values (with respect to A17), but both had large standard deviations. The regional MLR showed that the TAlk on the cruises was low by about 3 to 4 µmol/kg. Given the inconsistency of the results, no adjustment is recommended.

A16S. A16S had three crossovers. It showed an 8-µmol/kg low bias with A15 and a low bias of similar magnitude with the regional MLR in both the tropical and southern regions (albeit with only 35 data points). The internal consistency showed a low bias in TAlk as well, assuming that DIC and fCO2 are accurate. A 4-µmol/kg increase would put the values in better agreement with the other cruises, but because this is less than the minimum adjustment criteria, no adjustment is recommended.

A17. There were six crossovers for A17. A17 showed both significantly high deviations (25 µmol/kg with respect to A10 and 15 µmol/kg compared to A15) and low deviations (-7 µmol/kg with respect to A09) in the crossover analyses. The regional MLR suggested that the TAlk data are 5 to 6 µmol/kg high compared with data from other cruises in the tropical and southern regions. The internal consistency suggested that the TAlk was high by 8 µmol/kg. A decrease in TAlk of 6 µmol/kg for A17 TAlk would bring the values in better agreement. Because a decrease was suggested for A09 as well, the bias between these cruises remains of the same magnitude. However, this offset of 6 µmol/kg is right at the recommended cutoff, and no adjustment is recommended.

A20. There were three crossovers of A20 with other cruises. The alkalinities for this cruise were measured with very high precision but showed systematic offsets in every procedure, except for the assessment of internal consistency conducted by fCO2 and DIC. The TAlk values were low by 3 to 11 µmol/kg. The 11-µmol/kg offset was observed with A02, whose TAlk was consistently higher than those of other cruises. A correction of 4 µmol/kg would bring values in line, but no adjustments are proposed.

A24. A24 had six crossovers. A24 showed an offset with A01W but showed good agreement in the MLR analysis. If the suggested correction were applied to A01W, the crossover analysis would fall in line with A24. The assessment of internal consistency conducted by using TAlk, DIC, and pH showed an offset with pH; however, the pH data on this cruise were not of highest quality because they were measured by electrode. No adjustment to A24 TAlk data is proposed.

Last Modified: 2017-08-18 12:57 UTC