At the end of each rosette deployment water samples were drawn from the bottles in the following order: Oxygen; Salinity. Bottle trip depths are summarized as a function of station in Figure 7. The correspondence between individual sample containers and the rosette bottle from which the sample was drawn was recorded on the sample log for the cast. This log also included any comments or anomalous conditions noted about the rosette and bottles.
Normal sampling practice included opening the drain valve before opening the air vent on the bottle, indicating an air leak if water escaped. This observation together with other diagnostic comments (e.g., "lanyard caught in lid", "valve left open") that might later prove useful in determining sample integrity were routinely noted on the sample log.
Once individual samples had been drawn and properly prepared, they were distributed to their laboratory for analysis. Oxygen and salinity analyses were performed on computer-assisted (PC) analytical equipment and then transferred to the data processing PC for centralized data analysis. The analyst for a specific property was responsible for delivering their results to the CTD data processor for inclusion in the cruise database.
The first stage of bottle data processing consisted of verifying and validating individual samples, and checking the sample log (the sample inventory) for consistency. At this stage, bottle tripping problems were usually resolved, sometimes resulting in changes to the pressure, temperature and other CTD properties associated with the bottle. Note that the rosette bottle number was the primary identification for all samples taken from the bottle, as well as for the CTD data associated with the bottle. As all CTD trips were retained (whether confirmed or not), resolving bottle tripping problems simply consisted of assigning the right rosette bottle number to the right CTD rosette trip. Diagnostic comments from the sample log were then translated into preliminary WOCE quality codes, together with appropriate comments. Each code indicating a potential problem would be investigated.
The second stage of processing would begin once all the samples for a cast had been accounted for. All samples for bottles suspected of leaking were checked to see if the property was consistent with the profile for the cast, with adjacent stations and where applicable, with the CTD data. All comments from the analysts were examined and turned into appropriate water sample codes. Oxygen flask numbers were verified, as each flask is individually calibrated and significantly affects the calculated O2 concentration.
The third stage of processing would continue throughout the cruise (and indeed until the data set is considered "final"). Various property-property plots and vertical sections were examined for both consistency within a cast and consistency with adjacent stations. In conjunction with this process the analysts would review (and sometimes revise) their data as additional calibration or diagnostic results became available. Assignment of a WHP water sample code to an anomalous sample value was typically achieved through consensus.
WHP water bottle quality flags were assigned with the following additional interpretations: 3: An air leak large enough to produce an observable effect on a sample is identified by a code of 3 on the bottle and a code of 4 on the oxygen. (Small air leaks may have no observable effect, or may only affect gas samples.)
WHP water sample quality flags were assigned using the following criteria: 1: The sample for this measurement was drawn from a bottle, but the results of the analysis were not (yet) received. 2: Acceptable measurement. 3: Questionable measurement. The data did not fit the station profile or adjacent station comparisons (or possibly CTD data comparisons). No notes from the analyst indicated a problem. The data could be correct, but are open to interpretation. 4: Bad measurement. Does not fit the station profile, adjacent stations or CTD data. There were analytical notes indicating a problem, but data values were reported. Sampling and analytical errors were also coded as 4. 5: Not reported. There should always be a reason associated with a code of 5, usually that the sample was lost, contaminated or rendered unusable. 9: The sample for this measurement was not drawn.
Salinity samples were drawn into 200 ml Oceanic Scientific bottles after 3 rinses, and were sealed with custom-made plastic insert thimbles and screw caps. This assembly produces very low container dissolution and sample evaporation. As loose inserts were found, they were replaced to ensure a continued airtight seal. Salinity was determined after a box of samples had equilibrated to laboratory temperature, usually within 8-12 hours of collection.
Two Guildline Autosal Model 8400A salinometers (Numbers 1 and 5) located in a temperature-controlled laboratory were used to measure salinities. The salinometers contained interfaces for computer-aided measurement. A computer (PC) prompted the analyst for control functions (changing sample, flushing) while it made continuous measurements and logged results. The salinometer cell was flushed four times then until successive readings met software criteria for consistency, then 3 successive measurements were made and averaged for a final result.
The salinometer was standardized for each cast with IAPSO Standard Seawater (SSW) Batch #127 for stations 60-86 (K=0.99990, S=34.996), Batch #129 for stations 87-138 (K=0.99996, S=34.998), Batch #131 for stations 139-146 (K=0.99986, S=34.999) once every 48 samples or 48 hours, whichever is less. The program ASAL13 requires an additional standard of 10-low Ocean Scientific water (batch 30L7, K=0.87332, S=30.071). The estimated accuracy of bottle salinities run at sea is usually better than 0.002 psu relative to the particular Standard Seawater batch used. PSS-78 salinity [UNES81] was then calculated for each sample from the measured conductivity ratios, and the results merged with the cruise database. Salinometer 5 was used on stations for the entire cruise. Temperature control of the salinity area was good.
Samples were collected for dissolved oxygen analyses soon after the rosette sampler was brought on board. Nominal 125 ml volume-calibrated iodine flasks were rinsed once with minimal agitation, then a 10 second inverted rinse with laminar flow from the drawing tube along the sides of the flask, then filled via a drawing tube, and allowed to overflow for at least 3 flask volumes. Reagents were added to fix the oxygen before stoppering. The flasks were shaken twice; immediately after drawing, and then again after 20 minutes, to assure thorough dispersion of the MnO(OH)2 precipitate. The samples were analyzed within 4-36 hours of collection.
Dissolved oxygen analyses were performed with an automated oxygen titrator using photometric end-point detection based on the absorption of 365 nm wavelength ultra-violet light. Thiosulfate was dispensed by a Dosimat 665 buret driver fitted with a 1.0 ml buret. A whole-bottle modified-Winkler titration was used following the technique of Carpenter [Carp65] with modifications by Culberson et. al. [Culb91], but with higher concentrations of potassium iodate standard (approximately 0.012N) and thiosulfate solution (50 gm/l) [Brew74]. Standard solutions prepared from pre-weighed potassium iodate crystals were run at the beginning of each session of analyses, which typically included from 1 to 3 stations. Several standards were made up during the cruise and compared to assure that the results were reproducible, and to preclude the possibility of a weighing error. Reagent/distilled water blanks were determined to account for oxidizing or reducing materials in the reagents. The auto-titrator generally performed very well. The samples were titrated and the data logged by the PC control software. The data were then used to update the cruise database.
Blanks, and thiosulfate normalities corrected to 20 C, calculated from each standardization, were plotted versus time, and were reviewed for possible problems. New thiosulfate normalities were recalculated after the blanks had been smoothed. These normalities were then smoothed, and the oxygen data was recalculated.
Oxygen flasks were calibrated gravimetrically with degassed deionized water (DIW) to determine flask volumes. This is done at AOML before using flasks. During ACCE2, two flasks were determined to have suspicious volumes (numbers 32 and 61) and were removed from the sampling cases unitl recalibration at AOML could be performed.
All volumetric glassware used in preparing standards is calibrated as well as the 10ml Dosimat buret used to dispense standard Iodate solution. Iodate standards are pre-weighed in AOML's chemistry division to a nominal weight of 0.44 grams and exact normality calculated at sea. Potassium Iodate (KIO 3) is obtained from Johnson Matthey Chemical Co. and is reported by the suppliers to be greater than 99.4 percent pure. All other reagents are "reagent grade" and are tested for levels of oxidizing and reducing impurities prior to use.