Hydrographic and chemical measurements were made at a total of 67 stations (Table 2) using a Seabird 9/11 Plus system (the "WOCE system" equipped with a SBE 23Y Yellow Springs Instruments (YSI) dissolved oxygen sensor. The data were logged on a 33 Mhz 486 PC and post-processed between stations using SEABIRD's software. Once processed, the data were backed up to a CD-ROM.
Water column sampling was accomplished with a SeaBird Carousel 12-bottle rosette. Duplicate nutrient and single oxygen isotope samples were drawn at roughly standard depths on alternate stations only. Duplicate dissolved oxygen samples were drawn at 50m. In addition, two calibration bottles were tripped at the bottom of each cast for roughly replicate nutrients, salinities and dissolved oxygens. Due to limited numbers of sample bottles, nutrient and oxygen isotope sampling were curtailed after stations 43(save 66) and 23, respectively (see Table 2).
The SeaBird system worked well throughout the cruise. However, sampling was a problem in that there was often not enough water in the 1.7L bottles to fill all of the required sample bottles. Part of this problem was due to the use of 300 ml bottles for the O2 samples. The accuracy of the salinity samples seemed to suffer most, and the suggestion was made to draw the salinity sample first because of its importance in sorting out bottle misfires and leaking.
5a. Processing
The processing of CTD data was initiated by a single command on the PC at the end of the station. This command, called PROCESS, starts a batch job that sequentially passes the data through a number of programs. Most were from SEABIRD's SEASOFT package. A few were custom written at BIO. The following is a summary of the processing procedure [modifications required for the manual bottle trip procedure are indicated in square brackets]:
(0): [Run MARKSCAN to create .BSR (bottle scan records) file from .MRK file, created by hitting (cntrl-f5) after a bottle is tripped.]
(1): Convert raw frequency data to binary pressure, temperature and conductivity using SEABIRD's DATCNV program. [.ROS file is based on either bottle flags within the data stream (normal), or on information from the .BSR file (manual procedure)]
(2): Split the file into the up and down traces using SEABIRD's SPLIT program.
(3): Check downcast for and mark any 'wild' data points with SEABIRD's WILDEDIT program.
(4): Filter downcast conductivity and temperature using SEABIRD's FILTER program. This is a low pass filter and we used a time constant of 0.045 seconds for conductivity and 0.15 seconds for temperature.
(5): Mark downcast scans where the CTD is moving less than the minimum velocity of 0.10 m/s using SEABIRD's LOOPEDIT program.
(6): Align downcast pressure, temperature and conductivity using SEABIRD's ALIGNCTD program by advancing the conductivity signal by 0.01 sec.
(7): Apply the thermal mass correction for the conductivity cell using SEABIRD's CELLTM program.
(8): Compute dissolved oxygen in ml/l using SEABIRD's DERIVE program.
(9): Create WOCE 2-dbar dataset using OSD program PRO-WOCE.
(10): Bin average downcast data to 1.0-dbar intervals using SEABIRD's BINAVG program.
(11): Compute downcast salinity, potential density(sq), potential temperature(q), and depth using SEABIRD's DERIVE program.
(12): Convert the down cast from binary to ASCII using SEABIRD's TRANS program.
(13): Convert downcast to ODF format using PCS program SEAODF.
(14): Create IGOSS message using PCS program ODF_IGOS.
(15): Prepare batch and command files to transfer the data to the VAX and create the input for SEABIRD's ROSSUM program using our customized MAKEFILE program.
(16): Check for bottles, then use ROSSUM to create the rosette summary file.
(17): Convert the resulting .BTL file to a format suitable for ingestion into Quattro PRO (.QAT file) using our customized QPROBTL program.
(18): Create the calibration file of merged up- and down-cast data using OSD program CALIB.
(19): Copy Quattro, downcast, and ODF files to appropriate directories and clean up.
Plots and status info displayed by the SEASAVE program during the acquisition are discarded when the program terminates. The post-processing plotting was not included in the batch job because SEABIRD's SEAPLOT program requires interactive operator attention. Plots produced after each station include T, S, O2, and sq vs. pressure and T vs. S. Attempts to produce section plots with Igor Yakashev's contour package, modified to accept .ODF files, were frustrated by improper header information in the ODF files. "Station" and "Cast" numbers were confused, causing problems to propagate into the processed files. The only foolproof solution appears to be reprocessing of the entire data set, which is presently underway.
5b. Calibration
At the base of each CTD cast two rosette bottles were tripped and a single salinity sample was drawn from each to be analyzed onboard with an Guildline AutoSal salinometer. Assuming that the temperature offset for the WOCE system is negligible, the comparison of the salinity standards against the SeaBird CTD (Table 2a below) shows that the mean offset is small and not significantly different from zero. However, the standard deviation about this mean is unacceptably large (order of magnitude higher), even after accounting for outliers based primarily on replicate sampling. Some of this variability may be due to real time dependence in the properties at the base of each cast, but part may be due to sampling techniques which need to be re-examined.
The performance of the YSI O2 sensor was similar to that on previous cruises. The surface values on the downtrace appeared to be fully equilibrated, thanks to a 3 min. waiting period (as suggested by the manufacturer), but there was usually a large hysteresis between the down- and uptraces in the vicinity of the pycnocline, and there were occasional spikes from the electronics. The hysteresis probably results from a mismatch of the temperature and O2 sensors in the probe and might be improved by applying filters with appropriate phase lags. The dissolved oxygen samples collected from 300-ml calibration bottles were analyzed on board with an Orion 853 oxygen probe, borrowed from Marine Chemistry. The mean offset of the YSI over the entire cruise was significantly different from zero (Table 2a) and the standard deviation about the mean was substantial (0.3 ml/l). A linear regression (Table 2b) of the Orion standard on the YSI values from the CTD provides a calibration for the YSI with a standard error of ±0.3 ml/l (Figure 2a). To assess the quality of the Orion standard, "replicate" O2 samples were obtained from the 50-m rosette bottle, and "near-replicate" samples were drawn from the two bottles at the base of the cast. Overall, these "replicates" agreed to within a mean difference of 0.02 ml/l, but with a standard deviation of 0.11 ml/l (or better, discounting outliers; Table 2a). As a further check on the Orion sensor, standard Winkler analysis was performed on 8 samples drawn from a late CTD cast (CTD65) and "pickled" before their return to BIO 12 hours later. Although the Winkler replicates showed substantial variation (Table 2a), their average was quite consistent with the Orion results (Table 2b: Figure 2b)
Problems/Recommendations:
(1)Efforts should be made to remove the hysteresis between the up and down traces from the YSI sensor by application of filters with various lags.
(2)Salinity samples should be drawn first from small (1.7L) rossette bottles.
(3)Smaller O2 sample bottles should be used
(4)The new CTD computers require WINZIP and OFFICE 97 software.