The following evaluation of annual sensor performance shows the effects of sensor problems indicated earlier (Sect. 3) on individual cruises. Data from early cruises, which used the Beckman or YSI sensors, had to be flagged bad or suspect due to those problems. A noticeable improvement in the data quality is evident starting in 1996, when dual sensors were being used, and when new procedures were implemented in late 1995 including checking for possible sensor problems, such as membrane and electrolyte conditions before every cruise. The use of SBE 43 sensors starting in 1999 further improved the data quality.
A Beckman oxygen sensor SN 13-0186 was used to measure dissolved oxygen during cruises 1 through 12. Table 4.1 shows the standard deviations of differences between the calibrated CTD and bottle oxygen values.
Table 4.1: Standard Deviations of Calibrated CTD Values - Bottle Values for data between 0 and 4500 dbar.
|
HOT
|
Oxygen
(mol kg-1)
|
| 1 |
9.3 |
| 2 |
14.4 |
| 3 |
6.9 |
| 4 |
0.8 |
| 5 |
3.4 |
| 6 |
3.0 |
| 7 |
5.2 |
| 8 |
2.5 |
| 9 |
1.9 |
| 10 |
3.6 |
| 11 |
2.7 |
| 12 |
3.1 |
Beckman sensor drift problems addressed above (Sect. 2.1) were present during these cruises. The sensor was calibrated assuming that the sensor does not drift, when in fact, in some cruises, the drift was large between casts. HOT-1 appears to be the worst cruise, where the difference between the CTD oxygen trace and the bottle values is as high as 25 µmol kg-1 in places. On other cruises, e.g., HOT-9 and -11, the sensor was apparently much more stable, and could be calibrated satisfactorily.
Figure 4.1. CTD oxygen data quality per cast (vertical bars) and per 2-dbar bin at Station ALOHA during HOT cruises 1 through 12. The tick marks along the x-axis separate different casts. Casts from the same cruise are separated by vertical black lines.
Figure 4.1 shows graphically the CTD oxygen data quality for every 2-dbar bin in each cast at Station ALOHA for cruises 1 through 12. Quality flag colors are green for good data, orange for suspect, red for bad, and white for uncalibrated; gray indicates that no data was taken. As mentioned earlier, only casts with bottle samples are flagged as calibrated. Cruise 1 had many bad data due to sensor drift problems as explained earlier. Cruise 5 did not have oxygen samples deeper than 650 dbar available for calibration, thus all CTD values below this depth were wildly in error because there were no deep bottle data to constrain it. All these data were flagged bad.
Back to Table of Contents
