DQE of WOCE P6E Hydrographic Data (A. Mantyla)

27 February 1994

The three legs of WOCE line P6 (E, C, and W) at about 32S in the Pacific, will be a very valuable data set to help fill in the most barren strip of Marsden Squares in the Pacific. The P6 data is generally quite good, but there are a couple of areas that could stand improvement (see below). The PI’s have done a thorough job in sorting out rosette-trip problems, but there may be a few more that might be resolved upon re-examination, tabulated below. The data originators probably went overboard in flagging slightly anomalous data points as "bad" data, while the "questionable" data code might have been more appropriate. I’ve softened or changed some of those flags, but for the most part, the originators codes have been left as is.

There were a number of instances where "leaky" bottle were identified to have bad water samples by the salinity check sample, but only the salt and nutrients were flagged, while the oxygen was accepted as ok because it fortuitously appeared to fit in the profile. In the case of verified leakers, all water samples should be flagged doubtful. The oxygen precision in general, while better than many historical data sets was not up to WOCE expectations. The poor precision was seen in the duplicate trip data, mixed layer data, and numerous "bumps" in the deep profiles that were not supported by the CTD O2 data. Although some of the errors were likely due to sampling errors, the rest must be due to the specific analytical techniques used on the cruise. As an example of the excellent precision achieved elsewhere, the CalCOFI cruises, using the Carpenter whole bottle titration method, routinely report mixed layer oxygens that differ by no more than 0.01 ml/l. P6 mixed layer oxygens commonly had a spread of several times that value. I would urge that the analysts seriously reconsider their methodology and give the Carpenter system a try.

Salinity results also were not as sharp as WOCE expects, but in this case, I believe the source of the problem lies in the sample drawing stage, rather than in the analyses. According to the cruise report, salinity bottles were only rinsed twice, while nutrients, a part per hundred number, were given 3 rinses. Two salinity rinses is ok if the historical salinity precision of .003 to .005 is the goal, but 3 rinses are a must to reach the WOCE goal of .001. Comparisons with nearby JUNO stations show that the desired WOCE targets are possible, the deep O2 and salt profiles are distinctly smoother (like the CTD profiles on both cruises) on JUNO. That doesn’t mean that the P6 data is bad, just that it is possible to do a little better. My impression was that the bumpiest profiles were from stations that started within a few hours of midnight, GMT.

The nutrient data looked quiet good, in spite of noticeable offsets from crossing WOCE lines, particularly in the silicate data. I think the cruise to cruise nutrient differences are an indication that the current nutrient methodology is not yet up to

WOCE expectations, but both cruise results are as good as any that I’ve seen.

The nutrient data appear to have been evaluated independently from the other data. Some nutrient bumps were flagged doubtful, but had supporting consistent oxygen inflections that were in turn verified by the CTD O2 probe, confirming that the nutrient anomalies were probably real. A few flags have been changed on that basis, but for the most part, I’ve gone along with the original flags.

The ".sum" files are incomplete and should be finished. This format is not very popular to outside groups; coming from the source at WHOI, one would expect to see a model example of how it should be done! Only the cast start time positions are tabulated. The WOCE guidelines expect a position for the cast down time (which is when the water sampling begins) and a cast end time position. A bottom depth and depth above the bottom should also be tabulated A few obvious goofs are tabulated below but the table needs to be fleshed out and proofed carefully.

The following are some specific problems that should be looked at:

Station 8, btl. 2: Data listed at 2038db unlikely, but would probably be ok at assumed unlisted trip depth between 2038 and 1588db, perhaps near 1888db. Bottles 3 and 4 both listed at 1588db, suspect rosette hang-up started with bottle 2. Recommend data originators check to see if the data listed at a shallower depth would be ok and then change data flags.

Station 9: Numerous trip errors, not all resolved. O2’s not listed below 869db;

they should be with appropriate quality codes. The data from

bottles 2 and 9 are listed at odd pressure intervals, data would look ok if listed at pressures midway between bracketing bottles. Data from bottles 20 and 21 are essentially the same, suggesting both tripped at the same depth, perhaps near 815db according to the salinity gradient. If trip depth can’t be resolved, recommend flag all water samples from bottle 20 uncertain also.

Station 10, btl. 17: Salt, silica and O2 suggests data would be ok if listed at pressure about half way down to next reported pressure. If done, change quality flag to ok.

Station 11, btl. 1: Increase in trench bottom O2 probably is real, see SCORPIO stas. 83 and 90. The trench is the main conduit to the bottom

Eastern Pacific basins.

Station 13: Suspect sample drawing errors: salts from bottles 6 and 7 flagged poor, would be ok in reverse order. No salt listed from deepest bottle. O2 from 20 was probably drawn from bottle 21. O2’s from 6 and 7 clearly would fit better deeper in the water column. On

second thought, salt, O2, and silica from bottle 7 would be fine one depth deeper; if that were done, the quality flags could be changed to ok.

Station 16, btls. 8 and 9: Both salt and O2 indicate they tripped one depth deeper; nutrients ok at either depth. If moved accept data as ok. O2 from bottle 3 probably drawn from bottle 2, must "u" it at present depth.

Station 17, btl. 8: Salt, O2 and nuts indicate bottle probably tripped at unlisted depth about half way down to next depth. Data ok if moved; if left as is, flag nutrients doubtful also.

Station 19, btls. 5 and 6: Salt, O2, silica would fit better with adjacent station and CTD O2 if water data moved up one depth. If done, change flags to ok.

Station 31, btl. 6: Nutrients originally flagged bad, probably because they look a little high. However, O2 inflection agrees with nutrient profile, and the bottle O2 is confirmed ok by the CTD O2, so I’ve flagged all of the data as ok.

Station 32 salts: All appear high compared to adjacent stations THETA - S curves and the CTD, suspect faulty salinometer run. Recommend all be flagged questionable (not done yet).

Station 41, btls. 3 and 4: Comparison of salinities with the CTD suggests samples were reversed in sampling. However, O2 was also bad on bottle 4, so only that depth was flagged, but bottle 3 salt could be also.

Station 55, btl. 32: Is the reported oxygen of 14.120 ml/l a typo or the actual measurement? Seems unlikely to titrate a factor of 3 times too high.

Station 64, btl. 21 O2: Clearly too high, but would agree with CTD O2 if a transposition error had occurred: 5.93 ml/l recorded instead of 5.39 ml/l. Should check with the original sources.

.sum file, sta. 14: Probably on the 7th, rather than 6th.

.sum file, sta 17: BO 8th, not 7th.

.sum.file, stas. 21 and 22: Must have been on the 9th.

.sum file, sta. 32: bottom date wrong, must be 13.

.sum file, sta. 69: Longitude seems unlikely, 9 deg. off?

G.3. DQE of WOCE P6W Hydrographic Data