WHP CRUISE SUMMARY INFORMATION WOCE section designation A06 & A07 Expedition designation (EXPOCODE) 35A3CITHER1_1-2 Chief Scientist(s) and their affiliation Alain MORLIERE( ORSTOM), Christian COLIN (ORSTOM) Dates 1993.01.02 - 1993.03.19 Ship L'ATALANTE Ports of call Pointe Noire (Congo) to Natal (Brazil), Cayenne (French Guyana) to Abidjan (Ivory Coast) Number of stations 224 Geographic boundaries of the stations 8°20.42''N 51°19.50''W 10°50.39''E 5°39.43''S Floats and drifters deployed none Moorings deployed or recovered none Contributing Authors (In order of appearance) T. Mueller A. Billant P. Branellec M. Arhan CITHER 1 CRUISE HIGHLIGHTS Expedition Designation: The CITHER 1 cruise is a french contribution to WOCE Programme. This cruise describes WHP Lines A6 and A7 Scientist in charge of the cruise: Claude OUDOT, Institut Français de Recherche Scientifique pour le Developpement en Cooperation (ORSTOM) Chief Scientists: Leg 1: Alain MORLIERE, Institut Français de Recherche Scientifique pour le Developpement en Cooperation (ORSTOM) Leg 2: Christian COLIN, Institut Français de Recherche Scientifique pour le Developpement en Cooperation (ORSTOM) Ship: R/V L'ATALANTE Ports of call: Leg 1: Part 1: Pointe Noire (Congo) to Natal (Brazil): WHP Section A7 Part 2: Natal (Brazil) to Cayenne (French Guyana) Leg 2: Part 1: Cayenne (French Guyana) to Abidjan (Ivory Coast): WHP Section A6 Part 2: Abidjan (Ivory Coast) to Pointe Noire (Congo) Cruise Dates: Leg 1: Part 1: January 2 (Pointe Noire) to January 23 (Natal), 1993 Part 2: January 26 (Natal) to February 10 (Cayenne), 1993 Leg 2: Part 1: February 13 (Cayenne) to March 8 (Abidjan), 1993 Part 2: March 10 (Abidjan) to March 19 (Pointe Noire), 1993 Cruise Summary Cruise Track The cruise track and station locations are shown above. Sampling accomplished Water sampling on the cruise included measurements of salinity both by CTD and water bottle samples, CTD and bottle sample oxygen determinations, CTD temperature, and nutrients (silicate, phosphate, nitrate, nitrite). Tracer analyses were made for CFC-11 and CFC-12 as well as sampling for tritium/helium. Besides water sampling was made for measurements of CO2 system parameters (TCO2, pH, fugacity of CO2), dissolved gases (nitrogen, argon, methane and nitrous oxide). Type and number of stations During the two legs of the cruise a total of 224 CTDO/Rosette stations were occupied using a 32-bottle IFREMER rosette equipped with 8 liters PVC water sampling bottles. The usual spacing of stations was 30 nm, except over the continental slope (4 to 5 nm) and the abyssal plains (40 nm). List of Principal Investigators The parameters with the principal investigators and their affiliation are listed in Table 1. Table 1: List of measured parameters and the Principal Investigators for each. Parameter Sampling Group Principal Investigator CTDO2 / Rosette LPO/IFREMER-Brest M. Arhan / H. Mercier S, O2 LPO/IFREMER-Brest M. Arhan / H. Mercier NO3, NO2, PO4, Si(OH)4 ORSTOM-Brest C. Oudot LOC/UBO-Brest P. Morin CFC-11, CFC-12 ORSTOM/LODYC-Paris C. Andrie Tritium, Helium LMCE-Saclay P. Jean-Baptiste CO2 system ORSTOM-Brest C. Oudot Dissolved gases (N2, Ar)ORSTOM-Brest C. Oudot Trace gases (N2O, CH4) LOC/UBO-Brest M. Guevel ADCP ORSTOM/LODYC-Paris A. Morliere LPO/IFREMER-Brest H. Mercier PEGASUS IFM-Kiel F. Schott ORSTOM-Cayenne C. Colin Preliminary results The R/V L'ATALANTE departed Pointe Noire, Congo for the WHP Section A7 on January 2nd, 1993. The first station near 5°04 N, 10°40 E (bottom depth = 2100 m) was to test one of the two CTD systems and its rosette water sampling equipment. The CTDs are EG&G Neil Brown Mark III equipped with Beckman dissolved oxygen sensor. The first CTD equipment was replaced by the second one at station 83 (January 29, 1993) owing to problems with the conductivity sensor. All the CTD temperature, pressure and conductivity sensors were calibrated at the IFREMER calibration facility both before and after the cruise. The conductivity and oxygen sensors were also calibrated at sea using data from the analyses of the salinity and oxygen samples collected at each station. Water samples were collected from 32 PVC sampler bottles (capacity 8 liters)) mounted on the two-storied IFREMER Rosette sampler. The water sample conductivity measurements and oxygen titrations were made in a constant temperature (20°C) portable laboratory. Additional samples were also collected from each PVC bottle for the shipboard analysis of nutrients (silicate, phosphate, nitrate, nitrite) and chlorofluorocarbons CFC-11 and CFC-12 (every other station until station number 66, every station beyond and until the last station). Helium and tritium samples were also collected at many of the stations (a total of 58): the analysis of these samples will be later carried out in a shore-based laboratory. Other samples were also collected from PVC bottles for the shipboard analysis of dissolved gases (nitrogen - argon - total CO2 - methane and nitrous oxide) and the determination of pH and fugacity of CO2 (in surface water and in atmosphere). The phytoplankton biomass (chlorophyll) was also sampled for shore-based analysis. Underway ADCP and thermosalinograph data were recorded along the track of the ship (10 154 nm). Twelve PEGASUS profilings were done near the western coast in the boundary currents. Problems During the first leg (station number 83) we must have to replace the CTD system: shift and noises of the conductivity sensor. The second CTD system will be used until the end of the cruise without problems. Through the cruise we used successively three Guildline salinometers: one Autosal and two Portasal. The problems were a shift of the calibration between the stations; or drift within a series of measurements. The later acquired Portasal model has given the best results and was used to measure all the salinities during the leg 2. With the analytical measurements of the tracers, the most serious problem was the CFC contaminations from the PVC sampling bottles, mainly due to the grease of the stopcocks. A few special stations (5) were made to test the contaminations, by closing all the bottles at the same depth where the CFC concentrations were the lowest (generally around 2500 m depth). The mean contamination is estimated to about 0.005 ± 0.002 pmol/l for F-12 and to about 0.008 ± 0.002 pmol/l for F-11. List of the cruise participants The list of the cruise participants is given in Table 2. STATION SUMMARY The station positions, time, etc are tabulated in a summary file. This file (CITHER1.SUM) is reported on attached pages (numbered 1 to 12) and on attached floppy disk in MS-DOS format (ASCII characters). The parameter numbers are defined in Table 3. Table 2: Cruise participants PARTICIPANTS ROLE AFFILIATION LEG Chantal Andrie CFCs ORSTOM/LODYC-Paris 1 - 2 Michel Arhan CTDO2 LPO/IFREMER-Brest 1 Sabine Arnault Tritium, Helium ORSTOM/LODYC-Paris 2 François Baurand Nutrients ORSTOM-Brest 1 - 2 Andre Billant S, O2 LPO/IFREMER-Brest 2 Jean-Michel Bore CTDO2 ORSTOM-Cayenne 1 - 2 Bernard Bourles CTDO2 ORSTOM-Cayenne 1 - 2 Pierre Branellec S, O2 LPO/IFREMER-Brest 1 Elisabete Braga Oxygen IOUSP-Sao Paulo 2 Remy Chuchla Oxygen ORSTOM-Cayenne 1 Souleymane Cissoko CTDO2 CRO-Abidjan 2 Christian Colin Chief Scientist,Pegasus ORSTOM-Cayenne 2 Daniel Corre CTDO2 ORSTOM-Brest 2 François Dangu Salinity - CTDO2 ORSTOM-Cayenne 1 - 2 Nathalie Daniault CTDO2 LPO/IFREMER-Brest 1 Andre Dapoigny Tritium, Helium LMCE/CEN-Saclay 1 Alain Dessier CO2, N2, Ar ORSTOM-Brest 2 Jean-Pierre Girardot CTDO2 LPO/IFREMER-Brest 2 Jean-Pierre Gouillou CTDO2 LPO/IFREMER-Brest 1 Yves Gouriou CTDO2 ORSTOM-Brest 1 - 2 Stephanie Gueneley Nutrients ORSTOM-Brest 1 Mickael Guevel Trace gases LOC/UBO-Brest 1 - 2 Catherine Hemon CTDO2 LPO/IFREMER-Brest 2 Philippe Hisard Salinity ORSTOM-Brest 2 Philippe Jean-Baptiste Tritium, Helium LMCE/CEN-Saclay 2 Milton Kampel CTDO2 INPE-Brazil 1 Lamine Keita CTDO2 CERESCOR-Conakry 2 Jean-Jacques Lechauve CTDO2 ORSTOM-Brest 1 Jerome Lecomte CO2, N2, Ar ORSTOM-Cayenne 1 - 2 Nathalie Lefevre CO2 Fugacity LODYC-Paris 2 Jean-François Maguer Nutrients LOC/UBO-Brest 1 Jean-François Makaya CTDO2 ORSTOM-Pte Noire 1 Laurent Memery CFCs LODYC-Paris 1 Herle Mercier CTDO2, ADCP LPO/IFREMER-Brest 2 Marie-Jose Messias CFCs LODYC-Paris 2 Pascal Morin Nutrients LOC/UBO-Brest 2 Alain Morliere Chief scientist, ADCP ORSTOM/LODYC-Paris 1 Claude Oudot CO2, N2, Ar ORSTOM-Brest 1 - 2 Christophe Peignon CO2, N2, Ar ORSTOM-Lome 1 Jean-Paul Rebert Tritium, Helium ORSTOM-Brest 1 Joerg Reppin Pegasus IFM-Kiel 1 Birane Samb CTDO2 CRO-Dakar 2 Jean-François Ternon CFCs ORSTOM-Brest 1 - 2 Mohideen Wafar Nutrients LOC/UBO-Brest 1 Table 3: Parameter numbers in the CITHER1.SUM file Parameter Parameter Unit Number 1 Salinity PSS-78 2 Oxygen µmol/kg 3 Silicate µmol/kg 4 Nitrate µmol/kg 5 Nitrite µmol/kg 6 Phosphate µmol/kg 7 Freon 11 (CFC-11) pmol/kg 8 Freon 12 (CFC-12) pmol/kg 9 Tritium TU 10 Helium nmol/kg 11 Delta Helium-3 % 12 13 14 14 15 16 17 18 19 20 21 22 23 Total Carbon µmol/kg 24 Total Alkalinity µmol/kg 25 Fugacity fCO2 Pa - µatm 26 pH None 33 Nitrogen µmol/kg 15 Argon nmol/kg 33 Nitrous oxide nmol/kg 31 Methane nmol/kg 34 Chlorophyll a µg/l 35 Phaeophytin µg/l Thomas J. Mueller Institut fuer Meereskunde an der Universitaet Kiel Duesternbrooker Weg 20 24105 KIEL, Germany e-mail: tmueller@ifm.uni-kiel.de Woods Hole, MA, U.S.A. 02 Oct, 1996 WOCE Hydrographic Programme Data Quality Expert (DQE) Report on CTD data One Time Sections A6, A7 N/O L'Atalante Jan - Mar 1993 CTD-O2 data Introduction The French written cruise report consists of four volumes: - Volume 1 with general cruise information. Also, procedures of calibration and processing of 'En Route' data, ship borne ADCP, and some PEGASUS stations are described. - Volume 2 (Le Groupe CITHER-1, 1994) with a description of CTD-O2 data calibration, processing and extensive hard copy displays. - Volumes 3 and 4 with geochemical measurements. Also, an English written cruise report is available at the WHPO (6 pages plus - .SUM file). The present DQE report deals with the CTD-O2 data from A6 and A7. It consists of three parts: (A) A brief summary of the French written A6/A7 CTD-O2 data report (Le Groupe CITHER-1, 1994; GC1 henceforth) which describes the procedures of laboratory calibrations, data acquisition and processing, in-situ calibrations and verifications. Along with this summary, I have included (and flagged as such) some comments at the end of sections where appropriate. No figures and tables are available in electronic form from the above report, and therefore reference is made to figures and tables as they appear in the report. (B) A report of evaluation of the A6 and A7 CTD-O2 data as they were available at the WHP-O in September 1996. (C) Recommendations Part A. Campagne CITHER-1 of R/V L'ATALANTE (2 janvier-19 mars 1993). Recueil de donnees, Volume 2: CTD-O2 (English summary by DQE with comments added at ends of sections; sections, figures and tables are numbered as in the French report) I The CITHER-1 Group To obtain WOCE one time zonal sections A6 (along 07N30') and A7 (04S30') is one among other French contributions to WOCE. The cruise in 1993 was divided into two legs. In addition to stations along A6 and A7, two meridional sections were obtained between A6 and A7, along 035°W and 004°W. PI's for CTD-O2/rosette were Michel Arhan (leg 1) and Herle Mercier (leg 2); see Table 1 in the report for other PI's. II Cruise participants with respect to CTD-O2 work see Table 2 III Calibration of CTD-O2 measurements (A. Billant and P. Branellec, LPO) 1. Acqusition of CTD-O2 data A total of 223 stations with two Mark III CTD-O2 systems were obtained along with a 36x8 l bottle rosette PASH 6000 developed by LPO. For locations of stations see Figure 1. Major events (i) Section A7 was interrupted westbound after Stat. 77 before the vessel entered the 200 nm EEZ of Brazil. Prior to continuate A7, L'ATALANTE had to call port of Natal, Brazil, to pick up a Brazilian observer. Five days later, A7 was continued with Stat. 78 as repeat station on the position of Sta. 77. (ii) The first CTD-O2, S/N 2521, was replaced due to problems with the conductivity sensor after Stat. 82 by the second CTD-O2, S/N 2782. (iii) Stations 27, 75, 118 and 190 were taken in between WHP stations, with bottle being closed at special depths for calibration and test purposes. Data acquisition and processing The CTD's data cycles were transferred to the computer at a 32 Hz rate and on- line processed. Processed data then were stored on magnetic tape. Two steps of processing were applied. First, each data value was compared with the one in the preceding cycle. If the absolute difference of a value to the preceding one exceeded a certain amount (see table below), the complete cycle was omitted. The parameters for this comparison were: Pressure 0.5 dbar Temperature 0.032 K for pressure < 1500 dbar 0.005 K for pressure > 1500 dbar Conductivity 0.032 mS/cm for pressure < 1500 dbar 0.005 mS/cm for pressure > 1500 dbar Oxygen curr. 0.010 UA Oxygen temp. 0.3 K Next, cycles were averaged in pressure intervals. The intervals were chosen such that of all data cycles at least 25% were kept as 'good' and contribute to the average. For a lowering speed of 1 m/s, this means that at least 8 cycles contribute to an average over 1 dbar. Only, lowering profiles are considered. DQE's comments on section 1: From the French report, I understand that the original data set is not stored but only the (single value) de-spiked and averaged cycles with no other processing steps being applied before or afterwards. If this is true, I see some principal problems with this procedure. Although such a procedure may not affect very much CTDs that behave well, and although the non-averaged data may not be available any longer (as I understand the report), let me describe some steps necessary in processing open sensor CTD data. (i) the de-spiking method as described above can only recognize single spikes. It also is problematic in that it compares only with preceding values. If two or more spikes occur in turn (which to my experience may happen) these are smeared into the average during the averaging process; they can never be re-identified, and it is hard to detect and remove such 'bad' averages. (ii) Before averaging or low pass filtering, other important processing steps are performed for 'open sensor' CTD's by other institutes like WHOI (see Yang and Millard, 199xx) and IFM Kiel. They are not described for A6 and A7. The steps are: - create (if not already available) a cycle number or time and keep it throughout the processing. - check the (single value) despiked series for further spikes. - apply a low pass filter to the pressure series; this matches the pressure sensor resolution (0.1 dbar) to the lowering speed which at 1 m/s requires a resolution of 0.03 dbar. - monotonize the profile with respect to pressure; conductivity and oxygen sensor respond quite differently under different lowering speeds. Even better would be to first apply a 'minimum lowering speed' criterion to the profile and then monotonizing. - match the time constants of the (combined) temperature signal and the conductivity sensor. This can be done either 'by eye' looking at salinity spikes in sharp gradient regions, or more objectively by looking at the coherence and phase spectra. - apply a low pass filter to 0.5 dbar response and average on 0.5 dbar intervals. - apply the (static) calibrations for pressure, temperature and conductivity. - apply a low pass filter to 2 dbar response. - apply the correction for the dynamic response of the pressure sensor to temperature changes - average on 2 dbar intervals - calculate follow up quantities (salinty, pot. temperatur, pot. density) - apply the calibration of the oxygen sensor. 2. Sampling Sampling was done with a 36 x 8 l bottle rosette PASH 6000 developed by LPO. Bottles were closed on the way up (see Fig. 2, 3). A total of 6269 samples for salinity and 6460 samples of analysis of dissolved oxygen were taken. 12 bottles carried reversing temperature and pressure sensors made by SIS. Samples from bottles were drawn according to the instructions in the WOCE operation manual. DQE's comment on section 2 ok 3. Sample analysis for salinity and dissolved oxygen 3.1 Salinity Samples for salinity were drawn to 125 ml flasks, stored in a constant temperature (20°C ± 1 K) laboratory and analyzed within 20 h to 30 h. Standard seawater, batch P120 (K15=0.99985) from Wormley by 06 April 1992, was used to standardize the salinometers. Standardizations were performed before analysis started each day. After 36 bottles, standardization was verified and the result noted in a log. Each sample was rinsed three times before measuring and read three times. Due to stability problems of order 0.003 psu within a series of 36 bottles, salinometers were changed: Stat ID Stability 36 samples 001 to 010 PORTASAL A 0.001 psu 011 to 018 AUTOSAL 8400 B <0.003 psu 019 to 119 PORTASAL A 0.001 psu 120 to 223 PORTASAL B <0.001 psu Whenever unstable conditions were observed, standard seawater was used and salinity linearly corrected for drift. At four (non-WHP) stations, bottles were closed at same depths to get multiple samples for comparison. The maximum deviations from the means were less 0.003 psu. From the following statistics it follows that the precision is better 0.002 psu. Test stations: Salinity Stat depth Bottles close Stand. dev 27 2000 32 0.0009 75 4400 26 0.0018 118 2500 27 0.0011 190 1000 24 0.0016 Figures 4 and 5 show the results from 275 double samples from pairs of bottles taken throughout the cruise from the whole water column. Of these, 51% differ by less than 0.001 psu, and 85% by less than 0.003 psu. This result is not significantly improved when only samples from deeper than 980 dbar are considered. DQE's comment on section 3.1 All salinity measurements were done and reported thoroughly. As the comparisons of oxygen measurements (see 3.2 below) from the same test stations with significantly improved results from deeper levels show, the relative high value in salinity precision seems not to be due to mistakes in sampling but to the trouble with drifts in all 3 salinometers, rather. Nevertheless, from the high number of samples one may expect a good calibration the CTD's salinities. 3.2 Dissolved Oxygen Samples for oxygen were drawn after those for CFC's and helium into flasks of 120 ml. Temperature of the sample was measured before rinsing the flask three times. Samples were measured along the guidelines of the WOCE Operations Manual in constant temperature (20°C ± 1 K) laboratory. The method included to automatically detect the inflection. Multiple samples from same depths at three test stations show that a precision of 0.01 ml/l is expected. Test stations: Oxygen Stat depth Bottles close Stand. dev 27 2000 32 0.003 75 4400 26 0.007 190 1000 24 0.009 In figures 6 and 7 the results from 297 double samples from pair of bottles throughout the cruise and the water column are displayed. Of all double samples, 39% agree to within 0.005 ml/l, and 70% to within 0.015 ml/l. This result is much improved if one restricts to the 213 samples from depths larger than 980 m: then, even 45% agree to within 0.005 ml/l. For depths larger 2480 m, the standard deviation is 0.013 ml/l. DQE's comment on section 3.2 As the multiple and the double samples show, oxygen measurements meet the requirements of the WHP. 4. CTD pressure sensor calibration Both CTD's carried a Paine strain gauge sensor. These sensors routinely are calibrated at IFREMER's calibration center which is certified by the 'Bureau National de Metrologie' (BNM). A dead weight tester made by 'Desgranges et Huot' with an accuracy of ± 0.75 dbar at 6000 dbar is used. 4.1 Calibration under laboratory conditions (20°C) Pre- and post cruise calibrations were made for both CTD's with repeated loading (upper panels in fig. 8, 9) and unloading (lower panels) cycles. Third order polynomials have residuals less 2 dbar. 4.2 Static temperature effects Pressure sensor temperature was measured during the profiles. Laboratory calibrations at 7 different temperatures that cover the range are available. The effect is less 5 dbar. The additional corrections are necessary after having applied the 20°C basic calibration less than 3 dbar. The inner sensor temperature is modeled for a typical decent and hatched in figure 10. 4.3 Dynamic effects of temperature changes The dynamic responses to about 20 K temperature shocks were measured in the laboratory for both CTD's (fig. 11). The corrections applied for CTD profiles assume a single shock of this order within the thermocline, a lowering speed of 1m/s, 13 minutes at maximum pressure before the up-profile starts, and a 1 minute stop to close a bottle. 4.4 Corrections of pressure measurements Taking the 20 C basic 3rd order regressions at the 400 dbar interval calibration points, the corrections for the effects of both, static and dynamic temperature corrections are added. For the combined effects, a 5th order polynomial regression is applied to all pressure measurements (fig. 12, 13: loading mode in upper panels, unloading mode in lower panels). 4.5 Verifications after corrections For both CTDs, the differences at the surface before and after the profile corresponded well to the overall laboratory calibrations displayed in figures 12 and 13. Reversing electronic pressure sensors of SIS were used on the up profile. Pre- and post cruise calibrations were performed at 2.5°C at 7 points between 0 dbar and 6000 dbar. The corrected values of CTD and SIS sensors compare well within 2 dbar which may be assumed to be the overall accuracy of pressure measurements for WHP cruises A6 and A7. DQE's comment on section 4 Both sensors show a major change in their response characteristics at pressures larger than 4500 dbar in the post cruise calibration (fig. 8, 9) which appears strange to me. While the pre cruise calibration has the 3rd order polynomial response as it is typical for the Paine sensor, the post cruise calibrations for both sensors are more or less parabolic. The effect results in an order 3.5 dbar change for CTD2521 at 5400 dbar, which is the maximum pressure during the cruise; the effect is less for CTD2782. I wonder if such a change in the response characteristics found in other sensor calibrations from this period of time in which case they might indicate a shift in reference rather than CTD sensors. Hysteresis may depend on the maximum pressure to which the sensor was exposed before unloading, with maximum hysteresis being expected at the high end of the range at 6000 dbar. During these calibrations, the maximum pressure was kept to 6000 dbar. This excludes check of hysteresis effects at lower maximum pressures. However, since hysteresis was less than about 1.5 dbar at all pressures this will have a minor effect on the final calibration. The corrections for static temperature responses could better have been applied directly by linear interpolation since the inner temperature was measured, as I understand. However, the effect will be small, anyway. The same holds for the dynamic response. All corrections are modeled empirically into one 5th order polynomial for each, loading and unloading mode. As the comparison of corrected CTD pressures with corrected SIS pressures shows this method was able to meet the WHP requirements for CTD pressure measurements. 5. CTD temperature sensor calibration The measurements of a high precision Rosemount and that of a fast response NTC resistance are combined to standard MKIIIB temperature output at a resolution of 5 mK. 5.1 Operational mode CTD temperature sensors are routinely calibrated at IFREMER before and after a cruise. During calibration, the CTD is completely immersed into the temperature stabilized calibration bath. Temperature readings are compared to a reference Rosemount sensor which ITS90 calibration is traced back on a regular basis to the BNM. Both CTDs were in use since 1982 with changes in calibration not exceeding 10 mK. While CTD2521 stayed stable during the cruise (fig. 16a), CTD2782 showed a clear offset of 2 mK at 0°C and 8 mK at 25°C (fig. 16b). The uncertainty of CTD2782 is 2 mK up to 5°C, and 4 mK for larger temperatures. 5.2 Verification after correction Seven reversing electronic thermometers made by SIS and calibrated, both before and after the cruise, were used throughout the cruise. After the change of CTDs between stations 82 and 83, a 'jump' in the difference to all SIS sensors is observed (15 mK ± 1 mK) that corresponds well to the difference in the CTD laboratory calibration at 2°C (16 mK; see fig. 17 for temperature range 2.5 to 5°C and fig 18 for the 1°C to 2.5°C range). Final offsets between SIS and CTD are probably due to a pressure effect on the SIS sensors. For stations 1 to 82, accuracy as derived from figures 17 and 18 is of order 1 mK, over the whole cruise 2 mK. DQE's comment on section 5 From the calibration curve of CTD2521, its uncertainty seems to be of the order of 1 mK. As for CTD2782, it might be interesting to search for similiar 'jumps' in earlier calibrations. Accuracy of CTD temperatures as estimated from pre- and post cruise calibrations, and from comparisons with the seven SIS thermometers seems better than 2 mK, thus meeting WHP requirements. 6. CTD conductivity sensor in-situ calibration 6.1 Operational mode The conductivity sensor output is averaged while bottles are closed. This average is subject to the cell's pressure and temperature correction. The result is compared to in-situ conductivity values as derived from bottle salinities. A first order linear polynomial regression is calculated for stations or groups of stations: COR=C0 + C1*COS Outliers are removed until all differences are within 2.8*STDEV, STDEV being the standard deviation. 6.2 Station grouping CTD2782 stayed rather stable for large groups of stations. CTD2521, however, needed a station by station calibration from station 57 on until its exchange after station 82. Since the linear coefficient C1 did not change when calculated for stations 1 to 56 or station 1 to 77, the change in calibration was totally due to the offset C0. Thus, taking C1 as fixed, C0 was adjusted for station 57 to 77. For stations 78 (after the call of port) to 82, both coefficients were calculated station by station. See table III-1 for a complete listing of coefficients. 6.3 Overview profile calibration With the 5580 samples (89%) used for the calibration (see fig. 19, 20 for conductivity; fig. 21 for salinity), the overall standard deviation of the residuals is 0.0023 mS/cm. Only station group 204 to 219 is slightly worse (0.0029 mS/cm). Overall the cells' in-situ calibrations are close to WHP standards. 6.4 Verification Stations 31 and 119 were repeated with a different CTD at stations 223 and 156, respectively. Also, positions of stations 211 and 145 are close to SAVE station 45 and TTO station 63, respectively. All 4 theta-S diagrams coincide well in the deep sea with salinity deviations of just 0.001 psu. DQE's comment on section 6 The method applied to determine the calibration coefficients is well established. Comparison in theta-S space of two 'cross stations' of this cruise and two 'cross stations' with stations from SAVE and TTO establish an accuracy in salinity close to 0.001 psu meeting WHP standards. 7. CTD dissolved oxygen sensor in-situ calibration 7.1 Operational modes The calibration of the oxygen sensor followed the method described first by Millard (1982, see GC1 for the complete reference). The formula models the effects of temperature, inner and outer temperature difference and pressure, and salinity through the saturation formula by Krause (1984, see CG1 for the complete reference) on the electrical current (OC) that is measured in the cell. Compared are averages of OC over a 15 dbar interval from those depths of the lowering profile where sample oxygen were measured. The calibration coefficients are determined for groups of stations. 7.2 Units of dissolved oxygen The calibration is performed and reported in units of ml/l. All units are converted then to Umol/Kg keeping those values in ml/l. 7.3 Station grouping Three sensors were used: Stat. CTD Oxygen sensor 001-069 2521 A 070-082 2521 B 083-223 2782 C Sensor A, in addition to Millard's regression needed a 5th order polynomial regression in pressure. Sensor B needed a calibration by stations. Only sensor C was stable over large parts. See Tables III-2 and III-3 for coefficients and details. 7.4 Overview of profile calibration The results are presented in figures 24 and 25. A total of 6052 samples (93.7%) were used in the calibration procedure. Of these 42.4% have residuals less 0.025 ml/l, and 83.9% less 0.075 ml/l with a standard deviation of 0.066 ml/l. Disregarding samples from depths less 980 dbar, this result improves to 49.8% and 92.2%, respectively and a standard deviation of 0.041. The subset of stations 70 to 223 has an overall (all depths) standard deviation of 0.046 ml/l. 7.5 Verification One station pair (Stat. 119, 156) from this cruise with different sensors, and two SAVE stations can be compared (fig. 26, 27). The obvious differences between stations 119 and 156 also show up in other chemical parameters, and thus probably are due to a change in deep water masses at that position during the cruise. Stations 218 and 130 compare well with SAVE station 158 and TTO station 25. DQE's comment on section 7 The formula used to model the oxygen sensor response did not account for the sensor's speed through the water as requested in a later version in the WHP Operations and Methods Handbook. Nevertheless, the standard deviations reported for the residuals of the sensor calibration meet well the WHP requirements. Part B. CTD data evaluation 8. Basics A6 and A7 data available at the WHP-O were: -.SUM file -.WCT CTD data -.HY2 bottle data and additional two meridional sections linking A6 and A7. CTD data were on 1 dbar intervals. WHP requirements are 2 dbar intervals; the higher vertical resolution has led to problems with computer (PC) storage and computing time using the programs kindly provided by R. Millard, WHOI. CTDTMP and CTDSAL in the CTD files are reported with 4 decimal places, however with tailoring zeros. This is not WHP standard. Also, the quality byte for oxygen was set to zero throughout the CTD-files. Although the overall quality of the data set is expected to meet WHP standards, the remarks above and the quick evaluation below will show that some revision of the data needs to be made. I therefore restrict to the (more problematic section A7 plus some meridional stations (Stat. 1 - 99); nevertheless, all recommendations made below also hold for A6. The set of DQE programs allows to compare the CTD files with the CTD values in the bottle file. Only data flagged as 'good' were used. The following checks including some blow-up figures (not always shown) were made: - theta-CTDSAL, overall in the east and in the west - theta-CTDOXY, overall in the east and in the west - deviations CTDSAL and SALNTY on pressure levels by station - deviations CTDOXY(downcast) and OXYGEN on pressure levels by stations - same by pressure in station groups (waterfall plots) - noise level in the deep ocean - static stability in profiles 9. Theta-CTDSAL, Theta-SALNTY These plots are grouped for Stat. 1-50, and 41 -91. For stations 1 - 50 in the eastern basin, the overall plot (*Fig. 28a) shows extremely low salinities at the surface as a result of the Congo River plume. At least two non-flagged CTDSAL outliers from the upcast at the high end are detectable (and marked in *fig. 28a). Others are identified at lower temperatures (*Fig.28b). In the deep ocean (*Fig. 28c), some SALNTY values are aside the bunch. An example (*Fig 28c) shows that large deviations between samples and the CTD are observed at Stat. 9. This station needs to be compared directly with neighboring stations for the salinity calibration. A more careful check will later identify other stations with calibration offsets. In *Fig. 29a to 29.c the same is repeated for Stat. 41 to 91. Again, some few outliers of SALNTY are identified in the deep ocean. Overall, flags need to be checked. 10. Theta-Oxygen Station groups 1 to 51 (*Fig. 30a-c) and 41 to 91 (*Fig. 31 a-c), both show some extreme non-flagged spikes (Stat. 7, Stat. 38) in CTDOXY and some bad non- flagged values in the samples. Also, some CTDOXY profiles look rather noisy. Overall, flags need to be set/checked. 11. Residuals in calibration 11.1 CTDSAL In *figure 32a these differences are plotted as single dots by STNNBR for all depths (upper panel), for depths larger 1000 dbar (middle) and by pressure (lower panel). Also included are the mean differences for each station (bold line). *Fig. 32b gives a blow-up of the upper and lower panels of *Fig. 32a. Some non-flagged outliers are marked. The marked minima and the maxima of the bold line in the *Fig. 32a (middle panel) identify those stations, where the differences between CTDSAL and SALNTY need a check of the CTDSAL calibration by comparing neighbouring deep CTD stations: This is recommended for the following stations: 009, 010, 023, 033, 035, 048, 076, 077, 078. A more severe problem is obvious from *Fig. 32b: It shows a bias in CTDSAL calibration at pressures higher than 4000 dbar. Perhaps, the pressure compensation that has been applied is not sufficient. To my experience, these sensors may need additional corrections to the linear one applied to the compensated raw data. While *Fig. 32 allows one to identify stations with suspicious overall calibration, the waterfall plots in *Fig. 33a to 33i give insight to the residuals' distribution over single profiles. Although the resolution is sparse, some stations can be identified to have a systematic bias against the samples on that station. This holds for almost all stations which have samples from depths larger than 4000 dbar (as seen already in *Fig. 32). In *Fig. 33a and 33b, the subset shallow stations may have calibration problems: stations 005, 006, 010 and 097. 11.2 CTDOXY In *Figure 34, the residuals between the CTD downcast and the sample oxygen are shown. Some non-flagged outliers are marked (*Fig. 34a). With better resolution, *Fig. 34b (middle) shows the station mean residuals well within ± 5 Umol/Kg for pressures > 1000 dbar. Problems may occur at the beginning (Sta.6), and only a few other stations. I recommend comparison of neighboring stations in the deep ocean: 57, 58, 88, 95 and maybe 86. Station 6 is shallow and may checked against station 008. In the waterfall plots of *Fig. 35 those stations are marked that over wider parts of a profile show a bias in the residuals. At these stations, the CTDOXY should be compared to neighboring stations to verify the calibration. 12. Noise level in CTD profiles Since the data are provided on a 1-dbar interval rather than on 2-dbar intervals, the noise level maybe expected higher than usual for 2-dbar WOCE data. The method calculates means and rms over 2 - 12 dbar high pass filtered data. For the deep ocean (*Fig. 36a), the rms of CTDSAL is well below 0.001 psu (upper panel), that of CTDOXY generally below 0.5 Umol/Kg (middle panel). The mean rms for salinity is 0.0004 psu is slightly higher than for other WOCE cruises with low values in the deep eastern basin (stations 10 to 50) and high values between station 55 and 86 reflecting more variability in the deep western basin. The station averaged rms for oxygen (0.24 Umol/Kg is twice as high as the so far best WOCE cruises show probably reflecting the fact that the sensor's speed through the water column was not taken into account during the calibration. Some stations (around 20, 43, 51, and 75) peak in scatter and may be re- examined. Part C. Recommendations Resubmit the data set subject to: ** check for the calibration procedure of CTDSAL for high pressures ** incorporate the oxygen sensor's speed through the water column into the calibration to improve the noise level. ** deliver downcasts at: 2dbar intervals 4digit places for CTDTMP, CTDSAL, SALNTY (no zeros tailoring) ** set flags for CTDOXY ** carefully check all flags for SALNTY, CTDSAL, CTDOXY; setting flags may make use of the known standard deviations for the calibration. I'm prepared to inspect the complete data set when resubmitted. Acknowledgements The WHP-O at WHOI again has been a friendly and effective host. Software used for part B of this evaluation, was kindly made available by Bob Millard; special thanks to him for his helpful guidance. This work was supported by the Bundesminister fuer Bildung und Wissenschaft, Bonn, Germany, under grant WOCE IV. References Le Groupe CITHER-1: Campagne CITHER-1 N/O L'ATALANTE (2 janvier-19 mars 1993). Recueil de donnees, Vol 2: CTD-O2. Rap. Interne LPO 94-04, Laboratoire de Physique des Oceans, IFREMER, Brest, France, 1994. Millard, R.R. and K.E. Yang. CTD calibration and processing methods used at WHOI. WHOI Techn. Rep. 93-44, 1993 For further references see Le Groupe CITHER-1 (1994), there especially Billant (1985) for CTD calibration methods as applied at IFREMER; Billant (1990) for SIS pressure meter characteristics; Millard (1982) for the calibration of the oxygen sensor. FIGURES Fig. 1: Position géographique des 223 stations de la campagne CITHER 1 Les principaux 'évènements' intervenus en cours de campagne sont répertoriés. Fig. 2: Coupes synoptiques indiquant le niveau des prélèvements à chaque station sur les radiales 4°30S et 35°W. Fig. 3: Coupes synoptiques indiquant le niveau des prélèvements à chaque station sur les radiales 7°30S et 4°W. Fig. 4: Ecarts de salinité entre deux bouteilles fermées au même niveau: a) en fonction du numéro de station à laquelle a été réalisé le doublet, b) en fonction de la pression à laquelle a été réalisé le doublet. Fig. 5: Histogramme des écarts de salinité: a) pour les 275 doublets de la campagne, b) pour les 209 doublets réalisés à pression supérieure à 980 dbars. Fig. 6: Ecarts en oxygène entre deux bouteilles fermées au même niveau: a) en fonction du numéro de station à laquelle a été réalisé le doublet, b) en fonction de la pression à laquelle a été réalisé le doublet. Fig. 7: Histogramme des écarts en oxygène: a) pour les 275 doublets de la campagne, b) pour les 209 doublets réalisés à pression supérieure à 980 dbars. Fig. 8: Répartition des écarts, tous les 400 dbars, entre la pression de référence et la pression indiquée par le capteur Neil-Brown (sonde 2521) lors de l'étalonnage pré- et post- campagne à la température de 20°C: a) cycles montée en pression (profil descente), b) cycles descente en pression (profil montée). Le courbe de degré 3 qui réduit ces écarts est représentée. Fig. 9: Répartition des écarts, tous les 400 dbars, entre la pression de référence et la pression indiquée par le capteur Neil-Brown (sonde 2782) lors de l'étalonnage pré- et post- campagne à la température de 20°C: a) cycles montée en pression (profil descente), b) cycles descente en pression (profil montée). Le courbe de degré 3 qui réduit ces écarts est représentée. Fig. 10: Ecarts, tous les 1000 dbars, entre la pression référence et la pression indiquée par le capteur Neil-Brown à différentes températures expérimentales. Les limites de la surface pointillée sont, d'une part, la courbe obtenue à la température à 20°C et, d'autre part, celle d'une température à la température équivalente interne du capteur Neil-Brown mesurée sur les profils "bathysonde": cette surface correspond à la correction de température statique. Fig. 11: Etude de l'effet dynamique de température sur les capteurs de pression Neil-Brown (2521 et 2782) en laboratoire. Après immersion de la sonde dans un bain plus froid, les paramètres pression, température et température interne du capteur de pression sont représentés en fonction du temps. Le choc thermique provoque un décalage de l'indication de pression qui atteint environ 5 dbars après 30 minutes. Fig. 12: Ecarts, tous les 400 dbars, entre la pression de référence et la pression indiquée par le capteur Neil-Brown (sonde 2521) après correction de la linéarité du capteur à 20° (figure 8), de l'influence de température statique (figure 10) et de l'effet dynamique de température (figure 11). a) montée en pression (profil descente), b) descente en pression (profil montée). La courbe de degré 5 qui corrige la pression sur les profils est représentée. Fig. 13: Ecarts, tous les 400 dbars, entre la pression de référence et la pression indiquée par le capteur Neil-Brown (sonde 2782) après correction de la linéarité du capteur à 20° (figure 9), de l'influence de température statique (figure 10) et de l'effet dynamique de température (figure 11). a) montée en pression (profil descente), b) descente en pression (profil montée). La courbe de degré 5 qui corrige la pression sur les profils est représentée. Fig. 14: Ecarts obtenus, à chaque station, entre la lecture de 3 pressiomètres SIS et la pression indiquée par le capteur Neil-Brown en fonction de la pression d'observation. Les écarts, concernant les deux sondes utilisées pendant la campagne sont différenciés. Les courbes (en trait plein pour la sonde 2521 et en pointillé pour la sonde 2782) représentent la correction d'étalonnage à apporter à la lecture des deux instruments comparés (SIS et Neil-Brown). Lorsque les étalonnages pré- et post- campagne du pressiomètre sont différents, deux courbes sont présentées. Les points comparés à ces courbes montrent que, après correction, la pression SIS est égale à la pression CTD à 2 dbars près (le pressiomètre 6199 est devenu défectueux en cours de campagne). Fig. 15: même légende que la figure 14 pour une autre série de 3 pressiomètres. - A noter le mauvais fonctionnement intermittent du pressiomètre 6196. - Dans le cas du pressiomètre 6137, les écarts observés après correction sont de l'ordre de 4 dbars : cette différence est attribuée à un étalonnage incorrect du pressiomètre. Fig. 16: Ecarts entre la température de référence et la température indiquée par le capteur Neil-Brown lors de l'étalonnage pré- et post- campagne: a) sonde 2521, b) sonde 2782. La courbe de degré 2 qui corrige la température sur les profils est représentée. Fig. 17: Ecarts obtenus, à chaque station, entre la lecture de 3 thermomètres SIS, et la température indiquée par le sonde Neil-Brown: la température expérimentale est comprise entre 2.5 et 5.0°C. Les segments de droites représentent la correction d'étalonnage à apporter à l'indication du capteur Neil-Brown additionnée à celle du thermomètre SIS. La dérive des thermomètres a été compensée à raison de 0.001° entre les stations 1 et 82 et de 0.002°C entre les stations 83 et 223. Le décalage des points par rapport à ces segments de droites est attribué à un effet de pression sur le thermomètre SIS. Fig. 18: même légende que figure 17 pour 4 autres thermomètres. (entre 1°C et 2.5°C) Fig. 19: Ecarts entre la conductivité des 5580 échantillons validés et la conductivité 'bathysonde', au niveau du prélèvement, après recalage: a) en fonction du numéro de la station concernée, b) en fonction de la pression au niveau du prélèvement. Fig. 20: Histogramme des écarts entre la conductivité des échantillons et la conductivité 'bathysonde', au niveau du prélèvement, après recalage: a) pour la totalité des 5580 échantillons validés sur la campagne, b) pour les 3852 échantillons validés et prélevés à pression supérieure à 980 dbars. Fig. 21: même légende que figure 20 pour les écarts en salinité. Fig. 22: Comparaison de diagrammes theta-S tracés d'après les données de la campagne CITHER 1. Dans les deux cas, les stations ont été réalisées à la même position géographique avec unesonde différente. Fig. 23: Comparaison de diagrammes theta-S de la campagne CITHER 1 avec les données d'autres campagnes obtenues à une position géographique proche: a) station 211 de CITHER 1 et station 45 de SAVE (leg 2) (données 'bathysonde'), b) station 145 de CITHER 1 et station 63 de TTO-TAS (données 'rosette'). Fig. 24: Ecarts entre la valeur d'oxygène mesurée sur les 6052 échantillons validés et celle du profil descente 'bathysonde' à la pression du prélèvement, après recalage: a) en fonction du numéro de la station concernée, b) en fonction de la pression au niveau du prélèvement. Fig. 25: Histogramme des écarts en oxygène entre la valeur mesurée sur les échantillons validés et celle du profil descente 'bathysonde' à la pression du prélèvement, après recalage: a) pour la totalité des 6052 échantillons validés sur la campagne, b) pour les 4387 échantillons validés et prélevés à pression supérieure à 980 dbars. Fig. 26: Profils d'oxygène dissous obtenus à la campagne CITHER 1. Les stations 119 et 156 ont été réalisé à la même position géographique avec deux sondes différentes. L'oxygène mesuré sur les prélèvements de chaque station est reporté sur les profils avec un signe distinctif. Fig. 27: Profils d'oxygène dissous obtenus aux stations 218 et 130 de CITHER 1. Les valeurs d'oxygène mesurées sur les prélèvements de ces 2 stations sont indiquées. Pour comparaison, les mesures d'oxygène extraites de stations, réalisées à une position géographique proche, au cours d'autres campagnes sont portées sur ces figures. a) les valeurs de la station SAVE 158 (leg 3) sont les données 'bathysonde', b) les valeurs de la station TTO-TAS 25 sont les données 'rosette'. *Figures 28 - 32 are excluded in this report. TABLE III-1 Bilan de la calibration des profils de conductivité de la campagne CITHER 1 Sonde Station Nombre Nombre Déviation Coefficients utilisée ou d'échantillons d'échantillons Standard groupe considérés retenus par (0-6000) C1 C0 le calcu ------------------------------------------------------------------------------ 2521 1=>56 1367 1187 0.0024 0.999357 0.0320 57 32 28 " 0.0290 58 32 29 " 0.0227 59 32 28 " 0.0233 60 32 27 " 0.0239 61 32 29 " 0.0245 62 32 31 " 0.0252 63 32 28 " 0.0258 64 32 29 " 0.0264 65 32 27 " 0.0270 66 32 29 " 0.0277 67 32 30 " 0.0283 68 32 30 " 0.0271 69 32 26 " 0.0277 70 32 28 " 0.0284 71 32 31 " 0.0290 72 32 28 " 0.0296 73 32 30 " 0.0302 74 32 30 " 0.0309 2782 75 32 32 0.0017 0.999022 0.0423 2521 76 32 30 0.999357 0.0303 77 32 27 " 0.0310 78 32 19 0.999492 0.0415 79 32 29 0.0024 0.999716 0.0258 80 32 30 0.0021 0.999520 0.0332 81 32 29 0.0021 0.999382 0.0323 82 32 29 0.0019 0.999096 0.0405 2782 83=>91 237 211 0.0022 0.999781 0.0063 92=>118 769 686 0.0021 0.999695 0.0072 2521 119 32 29 0.0013 0.999862 0.0379 2782 120=>203 2425 2164 0.0021 0.999589 0.0112 204=>219 479 442 0.0029 0.999545 0.0106 220=>223 128 117 0.0020 0.999687 0.0106 TABLE III-2 Bilan de la calibration des profils d'oxygène dissous de la campagne CITHER 1 Capteur Station Nombre Nombre Déviation Standard Coefficients utilisé ou d'échantillons d'échantillons 0-6000 0-1000 1000- SOC OXPC OXTC OXC2 groupe considérés retenus par 6000 le calcul ------------------------------------------------------------------------------------------------------------------- Capteur 1=>11 189 184 0.179 0.244 0.053 0.0356 0.000193 -0.0169 3.552 Correction A 12 30 30 0.103 0.165 0.044 0.0394 0.000165 -0.0334 0.769 supplé- 13 30 30 0.093 0.185 0.045 0.0404 0.000153 -0.0246 1.511 mentaire 14 31 31 0.095 0.156 0.038 0.0398 0.000163 -0.0227 1.497 par 15 31 31 0.098 0.209 0.035 0.0409 0.000149 -0.0252 1677 polynome 16 32 32 0.131 0.240 0.029 0.0403 0.000155 -0.0260 1.553 de degré 17=>21 160 143 0.061 0.120 0.045 0.0468 0.000122 -0.0348 2.678 5 22 32 32 0.076 0.135 0.027 0.0408 0.000154 -0.0239 1.791 23 29 29 0.059 0.103 0.032 0.0425 0.000147 -0.0261 1.311 24 32 31 0.054 0.088 0.023 0.0404 0.000162 -0.0237 1.420 25 31 31 0.037 0.070 0.025 0.0444 0.000141 -0.0285 1.341 26et27 35 34 0.063 0.111 0.030 0.0422 0.000155 -0.0256 1.347 28=>67 1251 1216 0.082 0.127 0.046 0.0430 0.000149 -0.0267 1.402 68* 32 27 0.102 0.148 0.118 0.0430 0.000128 -0.0270 1.210 69* 32 31 0.049 0.064 0.046 0.0440 0.000137 -0.0272 0.972 Capteur 70 32 32 0.062 0.111 0.043 0.0658 0.000138 -0.0348 0.596 B 71 32 31 0.051 0.087 0.041 0.0712 0.000131 -0.0335 0.933 72 32 30 0.065 0.113 0.053 0.0750 0.000125 -0.0349 1.111 73 32 31 0.047 0.047 0.047 0.0732 0.000131 -0.0343 0.820 74 32 32 0.036 0.042 0.035 0.0712 0.000139 -0.0328 1.102 75** 32 29 * Les profils 68 et 69 sont partiellement inexploitables. ** Le profil 75 est totalement inexploitable TABLE III - 3 Bilan de la calibration des profils d'oxyène dissous de la campagne CITHER I Capteur Station Nombre Nombre Déviation Standard Coefficients utilisé ou d'échantillons d'échantillons 0-6000 0-1000 1000- SOC OXPC OXTC OXC2 groupe considérés retenus par 6000 le calcul --------------------------------------------------------------------------------------------------------- Capteur 76 32 31 0.037 0.043 0.037 0.0683 0.000143 -0.0326 0.799 B 77 32 31 0.031 0.033 0.031 0.0698 0.000142 -0.0341 0.657 78 32 32 0.066 0.077 0.063 0.0698 0.000140 -0.0331 0.938 79 32 32 0.044 0.080 0.029 0.0689 0.000144 -0.0328 0.687 80 32 29 0.016 0.008 0.018 0.0694 0.000144 -0.0331 0.521 81 32 29 0.048 0.043 0.050 0.0703 0.000142 -0.0333 0.817 82 32 32 0.052 0.082 0.042 0.0707 0.000143 -0.0345 0.753 Capteur 83=>91 236 221 0.059 0.086 0.043 0.0566 0.000148 -0.0307 0.563 C 92=>118 769 717 0.054 0.074 0.044 0.0559 0.000149 -0.0295 0.658 Capteur 119 32 31 0.046 0.083 0.027 0.0679 0.000157 -0.0316 0.698 B Capteur 120=>203 2423 2213 0.045 0.058 0.040 0.0562 0.000147 -0.0304 0.609 C 204=>223 607 557 0.037 0.048 0.033 0.0551 0.000149 -0.0294 0.642 ------------------------------------------------------------------------------- Comments on the DQE recommendations for the CTD-O2 data of WHP lines A6 and A7 (M. Arhan, A. Billant) The DQE considered the data as meeting the WHP standard, yet made several recommendations (Part C of the report). ** Check for the calibration procedure of CTDSAL for high pressures. We have checked the calibration procedure: It is the one recommended in the WHP operations manual, and described in the Unesco Technical Paper in Marine Science nb 54 (1988). When using this procedure, some depth-dependency of the residuals at high pressures (> 5000 dbar) cannot be avoided (as an example, see figure 3.8 of the UNESCO report) at least in certain oceanic area. ** Oxygen sensor speed: No accurate measurement of the time was available on that cruise, for which the in situ reference parameter was pressure. We usually remove the heave effect from the oxygen profiles by a ~10dbar running mean. ** Four digit places for CTDTMP, CTDSAL, SALNTY: As said in the cover letter, we can create new exchange files at this format if you judge it necessary. ** Set flags for CTDOXY: These are oxygen values from the down-profiles, averaged over a 15 dbar pressure range centered at the pressures of bottle triggering. These values are compared with the water sample data and, in case of a discrepancy exceeding 2.8 standard deviation, we choose to flag the bottle value, not the CTD one. This is a matter of convention, and the DQE is right in pointing out that, in some cases, the high difference is caused by inaccurate CTD values. As these CTDOXY values are only used for the calibration, we did not judge it necessary to examine the problematic cases to decide which parameter should be flagged. Had we done it, the choice could only have been subjective in most cases. ** Carefully check all flags for SALNTY, CTDSAL, CTDOXY. (See the set of figures with the problematic points marked). In several property-property plots (e.g. 28b, c, d), some points are found slightly aside of the main << cloud of points >>, although the difference << CTD minus water sample >> was less than 2.8 standard deviations, and the values were therefore not flagged. Again, this is a matter of convention. In several other plots (e.g. 30a, 31a, 32a, 33, 34a), differences CTD-water sample were reported, although the water sample data were flagged to either 5 or 3. This leads to apparent problems (only apparent, because the data were flagged). For instance, the value -9 was set when there was no data, with a flag of 5 (absence of data) in the WS files. Taking into account the value -9 leads to several differences at ~44 (~35 -(-9)) in figure 32a, or ~209 (= 200 -(-9)) on figure 34a. The same cause leads to horizontal lines on the <>, and to points aside of the <
> in the property-property plots. In particular, although CTDOXY was not measured at station 75 (all flags at 5) and was only partially present at stations 68, 69 (sensor problems), erroneous points for these stations are reported on figure 34a.