Underway Data Processing Methods

1.       Navigate to Underway Data folder and join into a single file, do so using Advanced File Joiner or similar:  http://download.cnet.com/Advanced-File-Joiner/3000-2094_4-169639.html

a.     CalCOFI collects its own data while aboard SIO ships (New Horizon).  These are saved in the Events folder of the appropriate cruise.  A single file name is similar to: 0901NHMETEV_06012009.csv.  Although each filename is labeled by date, date data my not correlate.  This is of no particular consequence, just be aware of it.

b.     When a cruise is aboard a NOAA ship, it is normal to receive a CD-ROM at the cruise with a variety of SCS data.  These data formats are highly variable by ship and by year.  It will usually take a combination of files to find all that you will require (i.e., one file type for TSG, one for fluorescence, one for lat long).  Just merge each file type, then cut and paste columns into the correct order.  Make all SCS data available on the web.

c.     The resultant  file becomes the RAW UNDERWAY DATA for that cruise.  This file contains all unedited data from the Shipboard Computer System (SCS).  This file may include navigational, meteorological, oceanographic, and other mechanical data.  In your local working directory  (e.g., ../UnderwayData/2009/0911NH) save raw data in tab delimited ASCII format and name as YYMMSS_UnderwayRaw.txt  (e.g., 0901NH_RawUnderway.txt)

d.      *Links to SCS information

                               i.    http://oceanexplorer.noaa.gov/technology/tools/scs/scs.html

                               ii.    http://www.unols.org/meetings/2006/200610inm/SessionIIIa/SessionIIIa_Shields.pdf

                               iii.    http://sdsioa.ucsd.edu/

2.       Open merged file and crop data columns to include the following, in order and with the following column headers:

a.     DATE_UTC, TIME_UTC, LA, LO, ST, SA, FL, FI  (Date(UTC), Time(UTC), Lat, Lon, Temp, Salinity, Fluor, Pump)

                                          i.    *Note – There is the possibility that power gets toggled on/off to underway sensors during the cruise.  This results in changes in the total number of data columns (i.e., the first days may have 100 columns of data and the next few only 98).  Attention  is needed to make sure that when cropping data that the correct data is sub sampled.

                                         ii.     **Note – Different ships have different underway column headers/variable designations/configurations.  Consult Underway Data Formats (link) for more information.

3.        Delete rows with:

a.     pump speed not within than manufacturer’s specifications

                                          i.     Doubling the flow rate from what we recommend will have at least one main problem. This flow rate will probably be too high for the conductivity cell and overtime you can see a degradation of the conductivity readings due to this flow rate. What can happen is that the high flow rate may slowly remove the platinum on the electrode within the conductivity cell, especially in areas with a lots of debris and particulate in the water. This will change the area of the conductivity cell which can cause the readings to be high of correct. I am unsure how long this process would take but it is something we have seen before. I really suggest using the recommended flow rate for this instrument. ~SBE Technical Support

b.     no spatial coordinates

c.     erroneous values

                                          i.    *Note – There is the possibility that one or more columns may be missing some or all of the data.   There are several workarounds depending on the missing variable. Consult WTF Moments (link) for more information.

                                         ii.    **Note – With upwards of 100,000 rows of data it is inevitable that most bad values will go unnoticed at this point.  Subsequent data processing steps (i.e., plotting) will make it easier to find errors and to remove them.  Just try to find the large and obvious discrepancies. 

4.    Save cropped file as 0901NH_UnderwayCrop.txt  in same folder as the raw file.

5.    It is necessary to compare and calculate correction regressions based on the comparison of CTD to Underway Data.  Navigate to the final CTD Downcast .csv file folder and join all csvs into a single file (as in Step 1).  These csv files are the ones with IEH bottle data added to them.  In your local working directory  (e.g., ../UnderwayData/2009/0901NH) save raw data in tab delimited ASCII format and name as YYMMSS_CTDd.txt (e.g., 0901NH_CTDd.txt).  If the YYMM_CTDd.csv file is available open in Excel and crop columns accordingly (#7).  

6.    Open file with MS Excel and select the first column ‘A’ and Text-to-Column the Data with comma delimiter.

a.   *Note – DO NOT save this file as YYMMSS_CTDd.txt while in Excel.  A host of things can go wrong.  Data will be Cut and Pasted into a new file shortly.  If you must you can save it as some tempfile.xlsx.

7.        Crop data columns to include the following, in order and with the following column headers:

a.      Date_Time_UTC, Lat_Dec, Lon_Dec, Temp0, Temp1, BTL_Depth, BTL_Salt, BTL_Chl

                                          i.    *Note – If Underway Data was only available with Date and Time in PST select Date_Time_PST instead.

8.    Sort Ascending (Smallest to Largest) based on BTL_Depth and Delete all depths greater than 2m, all rows with no BTL_Depth values, and all Header rows except for one which needs to be cut and pasted back at the top after we are finished sorting.

9.    Sort Ascending based on Date_Time.

10.  Delete any rows that are missing any information.

11.  Delete any duplicate stations.  You can tell if two subsequent rows are from the same station by the time, delete  the deeper.

12.  Make sure Header row is at the top.

13.  After the Temp 1 column, insert 2 additional blank columns.

a.   Average the two Temp columns into the first blank column, then copy and paste ‘values only’ into the second.  Delete the Temp0, Temp1, and first column.  Header the remaining column as TempAvg.

14.      Add three extra headers/columns  at the end of the existing

a.      …, ST, SA, FL, FI (…, T, S, Fluro, Pump)

15.  Open a previously completed copy of YYMMSS_CTDvsUnderway.xlsx.  Select, Copy, and Paste all rows and columns from the sorted and corrected …CTDd.txt file into the recently opened xlsx file.  Save and RENAME the xlsx file accordingly (i.e., NEW YYMMSS ).  Make sure no old data lingers in the spreadsheet as this would cause errors in the fit of the regression.

a.   *Note – We reuse the xlsx file because it already has multiple graphs set up.  You will of course need to resize the data sets for each graph as necessary, and rename the graph titles.  But if you followed the above steps the columns will be in the right places and everything should fall into place.

16.  Open the …UnderwayCrop.txt in MS Excel, Tab delimited.  Arrange both files in Tile View.

17.  Using this new …CTDvsUnderway.xlsx file’s Date_Time as a guide, scroll down the …UnderwayCrop.txt and find a corresponding date and time.  There will be more than one row of data for the time you are trying to match.  Try not to bias the selection of the Underway data and just pick the first one you see. Select the four columns, Temp, Salinity, Fluor, Pump, from UnderwayCrop and Copy them, and Paste them to the xlsx file in the last four columns ST, SA, etc.  Do this for all the rows in the CTDvsUnderway file.

a.     Because the number of rows will change from cruise to cruise, the graph data source will need to be adjusted accordingly.  This is important as the linear regression equations shown on the graphs are used to correct the Underway Data in subsequent steps.  Save file.

18.  For export to MATLAB it is necessary to Cut the xlsx data (numbers no graph info) and paste it into a new UltraEdit .txt file. Name it accordingly (YYMMSS_CTDvsUnderway.txt).  This is done to remove any .xlsx formatting that may cause MATLAB to choke on import.

19.  Start Matlab and load a previously completed .m file as a template (e.g., 0901NH.m).  The .m file will have incorrect variable names and notes pertaining to that specific cruise.  Change and update accordingly.  Also make sure you are working in your current underway folder.

a.     Import UnderwayCrop.txt and follow along the notes in the script for error checking and saving data as a .mat file.

b.      Import …CTDvsUnderway.txt.  There will be an error with this as the Date_Time will not ingest properly.  Ignore it, the date is not important for plotting  unless GPS has failed and you need the data and time to correlate with the Event Log GPS to get some coordinates to make these maps.

c.      An important step to be aware of is toward the end of the script it asks you to use dlmwrite function to export the final underway data product.

20.     The template m-file will guide you through the entire process.  When you get to the regression corrections take the values from the xlsx file graphs that you generated earlier.  When you make the Matlab plots they should print into your working folder.

21.     Open a previous yymmplot.ai file in Adobe Illustrator and the first plot from Matlab (i.e., t0901NHpath.eps).  Save the .ai file with the current cruise YYMM.

a.     In dealing with the .eps file from Matlab you will find that all we really need is the color path and the gradient bar frame and numbers.  Ungroup and delete all you don’t need (coastline, grid, tick marks, gradient color, etc.)

b.     Since the path is thousands of small circles, select them all, group them, then copy them to the clipboard. Delete the preexisting path in the .ai file and make sure you paste the new one in the correct .ai layer and resize to fit the station layout / cruise path.  This isn’t an exacting process, just make it fit best you can.  Make sure to lock/unlock layers as you go to prevent the pasting / editing in the wrong place.

c.     Go back to the eps and copy the gradient bar box, tickmarks, and numbers.  Since each cruise’s graph has a different range of values these gradient bars change.  The colors stay the same (if you followed the Matlab m-file correctly) but we need to make the new numbers fit the template.   Delete the old box and numbers and paste in the new and resize accordingly.  Resizing will skew the font so once everything is in the right place check the font setting and set to 18pt Myriad Pro with 100% font stretching in both directions.  You will also have to change the decimals and numbers a bit as Matlab tend to separate the numbers after the decimal in eps format.  Just delete the end and retype the number using the left half.

d.     Unlock and Edit  the Title layer to the correct cruise designation and dates and title.

e.     In Illustrator to save as pdf “save a copy”, to save as png Use Export (white background, medium resolution, anti aliased) .  Open each png in Photoshop and shrink to 75 pixels keep proportionality and rename with a         ” _t” at the end for the web thumb.

31.   Upload all data and images to calcofi.com FTP Underway_Data directory.

32.   Edit calcofi.com Joomla site with new underway data accordingly.

 ISUS NITRATE HISTORY

The original methods description was written Feb 2010 by J. Wilkinson.  Changes to the method or instrument are listed below.

Changes

ISUS Nitrate Sensor

SUMMARY: Since November 2004, a Satlantic ISUS nitrate sensor has been integrated with a Seabird 911+ CTD-Rosette system deployed on CalCOFI cruises. Cruises typically occupy 75 stations, collecting approximately 1400 discrete seawater samples throughout the water column. The discrete seawater samples are analyzed at-sea for nitrate, nitrite, silicate, phosphate and ammonia within 24 hours of collection. The ISUS voltage data are processed along with other sensor data using Seabird’s SBE Data Processing Suite. Processed CTD-ISUS data are merged with bottle data. The ISUS voltages are plotted versus corresponding nitrate data, generating a voltage-to-nitrate regression. These regression coefficients are applied to all ISUS voltages, converting voltages to estimated nitrate.

1. Principle

The Satlantic ISUS (In Situ Ultraviolet Spectrophotometer) is a real-time, chemical-free ultraviolet spectrophotometer detecting absorption characteristics of inorganic compounds in the UV light spectrum. The ISUS uses the UV (200-400 nm) absorption characteristics of nitrate and bromide to provide in situ measurements of their concentrations in solution. The sensor has four key components: a stable UV light source, a UV spectrophotometer, a bifurcated fibre optic sampling probe, and a processing microcomputer housed in a pressure case rated to 1000 meters. The ISUS measures the in situ absorption spectrum and then uses the calibrated coefficients and a least-squares curve fitting routine to calculate an absorption spectrum matching the measured spectrum. It then calculates the concentrations of nitrate and bromide required to generate the matching spectrum. This response is exported to the Seabird CTD as voltage logged with other sensor data at 24Hz.

2. CTD Integration

3. Data Processing

4. Calculations

4.1 Linear regression of ISUS voltage vs discreet nitrate data generates cruise-average correction coefficients.

4.2 BtlVsCTD calculates individual station regressions of ISUS voltage vs discreet nitrate data. This ‘on-the-fly’ linear regression generates station-specific corrections coefficients which are applied to the specific cast.

Both cruise and station-corrected nitrate estimates (and the coefficients) are tabulated in the final csvs.

5. Equipment/Supplies

·         Satlantic ISUS v2 Nitrate Sensor

·         Three 12v Wet-labs rechargeable battery packs

·         ISUS analog signal to Seabird 9 interface cable

·         ISUS power to battery cable

·         ISUS Rs-232 interface cable to download internal data files

·         Windows laptop with serial interface to program the ISUS and download data.

·         Alcohol & cotton swabs

·         Nutrient collection tubes for seawater samples

·         Seal QuAAtro nutrient analyzer & in-house analyst

6. References

·         Johnson, K.S.; & L.J. Coletti. 2002. In situ ultraviolet spectrophotometry for high resolution and long-term monitoring of nitrate, bromide and bisulfide in the ocean. Deep Sea Research I 49: 1291-1305.

·         Maillet, Gary and Geoff MacIntyre. 2009 Real-Time Monitoring of Nitrate With the Satlantic-ISUS Sensor. Online at: http://www.meds-sdmm.dfo-mpo.gc.ca/isdm-gdsi/azmp-pmza/documents/docs/bulletin_6_10.pdf

·         Satlantic Incorporated. 2005. MBARI-ISUS V2 Operation Manual, Document Number: SAT-DN-272, Revision G.1, August 2006

PRIMARY PRODUCTIVITY CHANGES

A description of the method was written February 2010 by D. Wolgast.  Changes to the method are listed below.

Radioactive Specific Activity Changes

Radioactivity added

Methodological Changes

During a cruise a 24 bottle rosette is used to collect seawater salinity samples of ~225 mL from all bottles closed from pre-designated depths.  A Guildline Instruments Portasal™ Salinometer (8410A) makes the precise conductivity comparisons between the water samples and a reference water standard.  From these comparisons salinities are then calculated and logged using PC based software that averages data that meet replicate criteria. Concurrent with the water sampling, a Sea-Bird Electronics CTD (SBE 911 plus) profiles in situ data.  Data processing software is used to compare the bottle salts to in situ CTD measurements.  This is useful to confirm bottle closure, monitor CTD performance, and to select the best salinity data for future comparison.  

Salinity bottles are collected in ~250 ml borosilicate KIMAX-35 bottles.  These square cross sectioned bottles have screw tops and lids with separate plastic thimbles to prevent leakage and evaporation. Numbered stackable divider boxes hold 24 numbered bottles to match the 24 Niskins on the rosette water sampler. Bottles are filled with previously unused sample water to limit salt crystallization and to pre-leech silicate from the glass.  

Numbered salt bottles are drawn from corresponding Niskin bottles and filled to the shoulder after three ~40 ml rinses. The last fill is done without interruption until overflowing; then ~10 ml is poured out over the thimble and the bottle is stoppered and capped. Samples are left to equilibrate to room temperate for >8 hours prior to measurement. Before being placed in the salinometer samples are: gently inverted 3 times to remove any possible stratification, wiped around the bottom of the cap to draw out as much water as possible that may be trapped under and around the cap, wiped around the thimble and threaded neck once the cap is removed to eliminate any excess water or salt precipitate, and flushed by dumping out ~10 ml into a collection bucket.  

Guildline Portasal Salinometer model 8410A is used to make the precise conductivity comparisons between the water samples and reference water standards.  Two of these instruments are taken on each cruise.  Guildline specifications state an accuracy of  ±0.003 PSU (same set point temperature as standization and within -2°C and +4°C of ambient), and a precision of 0.0003 PSU.  

Bath temperature must be within 2 degrees below and 4 degrees above ambient. To check bath temperature set point press the TSET key. Compare set point to actual by pressing ENTER for actual bath temperature. Set point and actual temperature must be within 0.02°C. Press UP ARROW key to view temperatures of the dual bath thermistors (TH1 and TH2). TH1 and TH2 must agree to within 0.04°C.  

The Portasal must be powered up for >3 hours to ensure bath temperature regulation has begun before Reference values can be calibrated. While the the FUNCTION switch is in STDBY pressing REF key will display alternating Reference readings plus (+), minus (-), and Reference readings. After several cycles, when + and – values are within 1 unit of each other, pressing the COND key will initiate salinometer self-calibration.  

With the FUNCTION switch set to ZERO, pressing the COND key will initiate the ZERO calibration process. When satisfied that the displayed value is stable, press the ZERO key. When the subsequent displayed number is stable (but not necessarily zero) press the COND key. The display should then read 0.00000. Setting the FUNCTION switch back to STDBY will ready the Portasal for Standardization.  

Deep sea water is collected on cruises from bottles tripped at depths greater than 300m in 10 liter polyethtylene jerry jugs. The seawater is then transfered to 50 liter carboys at the lab for filtration. During the transfer 10 ml hypochlorate (bleach) is added to inhibit biological activity.   Concentrated high salinity seawater is prepared by evaporating deep sea water (>300m) by heating the seawater at about 90°C in an oven overnight. A volume of 1800 ml will evaporate down to about 1000 ml in 24 hours, final volume can be adjusted with DI water, this will give a salinity of around 60psu. Filter the concentrate through GF/F filter before using it for adjusting the sub-standard. Filter the deep seawater through GF/F to 46 liter mark. Add 10 ml laundry bleach to inhibit biological activity. Measure the salinity of the filtered water. Adjustment of the substandard salinity is usually done in two steps. For the first adjustment, add concentrate to the filtered sea water to adjust to the desired salinity using the relationship:

(vol. sw)(sal. sw) + (vol. conc.)(sal. conc.) = (vol. sw + vol. conc.)( final sal. sub)

The salinity of the concentrate will be too high to measure on the salinometer. If you have prepared the concentrate as above you can do your initial adjustment using an estimated salinity of 60 for the concentrate. The final salinity of the sub should be close to your standard value ± 0.015 PSU. For the initial adjustment it works well to add about 90% of the concentrate volume calculated above. Be sure you have thoroughly mixed the solution by stirring gently. A motorized paddle wheel is used to thoroughly stir the mixture. Measure the salinity of the sub after the first adjustment. Using the equation above you can now calculate the apparent salinity of the concentrate, by using the salinity of the concentrate as an unknown, along with the known values for the initial salinity of the seawater, and the adjusted value of the substandard as just measured. Applying the calculated salinity value for the concentrate, use the equation again to calculate a new volume of concentrate to be added to get to the final desired salinity of the substandard. If your initial concentrate addition was to large you may have to dilute with DI water using zero for salinity concentrate in the equation. When you have arrived at the desired salinity for the sub-standard add a layer dimethylpolysiloxane to prevent evaporation resulting in salinity changes. 
 

Cut and Paste from WHP Operations and Methods July 1991 

Standard Seawater prepared by Ocean Scientic International Ltd. (OSI) is the recognized standard for the calibration of instruments measuring conductivity (salinity). This water is natural surface water which has been collected in the North Atlantic and carefully ltered and diluted with distilled water to yield seawater with a conductivity ratio near unity and a salinity near 35. The seawater is sealed in glass vials and labeled with the date, batch number, K15 value and chlorinity. Mantyla (1987) has compared batches of this seawater (P-29 to P-102) and found inter-batch dierences as great as about 0.003 in some of the older batches. Since batch P- 93 however, inter-batch dierences have not exceeded about 0.001.

It is recommended that a single batch of SSW be used during each cruise and that it be identied in the cruise report. It is the responsibility of the salinity analyst and chief scientist to ensure that the quality of the salinity observations are the highest attainable. It follows, therefore, that the inter-batch differences described by Mantyla (1987) should be used to correct the nal salinities before they are reported. It has also been noted that the conductivity of some Standard Seawater changes with time. Vials more than 4-5 years old should be compared with fresher Standards to determine possible changes in conductivity due to aging.

In addition to using IAPSO SSW, a batch of aforementioned substandard water has been bottled and sealed in glass vials.  This is done to decrease dependence on IAPSO standard while maintaining acceptable standardization accuracies.

Before standardization can begin the conductivity cell must be flushed rigorously as it had been filled with detergent or 50% ethanol between uses. To flush, attach the peristaltic pump to the Portasal sampling tube. Turn on both the Portasal and peristaltic pumps, and open the valve on the on the substandard feed line. Flush repeatedly (~10 flushes) until the cell is free of contamination. It is convenient to do this during Portasal configuration. Begin the salt program on the salt computer, and complete requisite data fields.

With the conductivity cell full of substandard and a stable value being displayed turn the FUNCTION switch to READ. When displayed value is stable press STD key. When the Portasal display reads STD STANDARDIZE press ENTER key. The display will show the conductivity of the standard being used. Press ENTER key to proceed (if values of substandard need to be adjusted use ARROW keys.) The BATCH # will be displayed and should match what is being used. If values need to be changed modify with the ARROW keys, pressing the ENTER key will ready Portasal for standardization. The display will read ENTER WHEN READY, press ENTER key. The Portasal will begin measurement and display a substandard value. When stable press the COND key. This will terminate the standardization and the displayed conductivity ratio should match that of your sub-standard within ± 0.00001. Pressing the ENTER key on the salt computer will save it to file.

At the beginning of each sample run the Portasal is standardized with substandard.  However, every other day IAPSO and Wolga standards are also sampled.  All three standards are run before and after the sample run in order to ascertain instrument drift.  

The SIO-CalCOFI-authored conductivity recording software, PSal.exe, is a Windows-based data acquisition program that records conductivity values from discreet salinity samples. Averaging five or more conductivity readings from the Guildline Portasal, the operator saves a stable reading. The flow cell is flushed and refilled and additional readings are performed. A pair of conductivity values which agree within 0.0010 standard deviation are recorded and salinity calculated. The software compares this calculated bottle measurement to the matching CTD salinity when available. This comparison is a good indicator of sensor performance and bottle sampling accuracy.
The data are saved as salt run files which may contain more than one station. Keeping the salt runs combined is necessary to calculate the drift-over-time calculated from any change in the end standard or substandard reading. The data are also save in a single, combined, database-friendly csv for database processing (in development).

A SBE 9ll+ CTD is equipped with dual conductivity and temperature sensors.  These are routinely calibrated (~6 months), but because of superior stability the pressure sensor in calibrated much less frequently (~24 months).

The SBE 11plus Deck Unit applies a real time alignment correction to conductivity during data collection, and the accompanying software generates a marker file for each profile containing CTD data at the instant of each Niskin closure.
Seasoft-processed CTD salinities are processed and imported into station csvs for comparison to bottle salinities.

CalCOFI-authored Windows program will process the bottle sample conductivity data files generated by the SalReCap.exe data acquisition program. Using beginning and end standards, a linear drift is applied to each conductivity pair before salinity (PSU) is calculated. DECODR combines all stations selected into a single text file for data-quality assessment. When CTD salinities for matching depths are available, DECODR compares the bottle & CTD salinities, calculating & reporting the differences. DECODR, as an option, will save the bottle salinity values into each station’s csv. DECODR can generate data products using bottle or CTD data such as IEH or data reports.

DISSOLVED OXYGEN

SUMMARY: The amount of dissolved oxygen in seawater is measured using the Carpenter modification of the Winkler method. Carpenters modification (1965) was designed to increase the accuracy of the original method devised by Winkler in 1889. Using Carpenters modification, the significant loss of iodine is reduced and air oxidation of iodide is minimized. Rather than using the visible color of the iodine-starch complex as an indicator of the titration end-point, we use an automated titrator that measures the absorption of ultraviolet light by the tri-iodide ion, which is centered at a wavelength of 350 nm.

1. Principle

Manganous chloride solution is added to a known quantity of seawater and is immediately followed by the addition of sodium hydroxide iodide solution.  Manganous hydroxide is oxidized by the dissolved oxygen in the seawater sample and precipitates forming hydrated tetravalent oxides of manganese.

Mn⁺² + 2OH^– ———————————————-> Mn(OH)₂ (solid)
2Mn(OH)₂ + O₂ ——————————————–> 2MnO(OH)₂ (solid)

Upon acidification of the sample, the manganese hydroxides dissolve and the tetravalent manganese in MnO(OH)₂ acts as an oxidizing agent, setting free iodine from the iodide ions.

Mn(OH)₂ + 2H⁺ ——————————————-> Mn⁺² + 2H₂O
MnO(OH)₂ +4H⁺ +2I ————————————> Mn⁺² +I₂ +3H₂O

The liberated iodine, equivalent to the dissolved oxygen present in the sample, is then titrated with a standardized sodium thiosulfate solution and the dissolved oxygen present in the sample is calculated.  The reaction is as follows:

I₂ + 2S₂O₃ ————————————————–> 2I + S₄O₆

2. Reagent Preparation

2.1.  The manganous chloride solution (3M) is prepared by dissolving 600g of reagent grade manganous chloride tetrahydrate, MnCl₂•4H20, in Milli-Q water to a final volume of 1 liter.  This solution is then filtered using 47mm glass fiber filters (Whatman GF/F).

2.2  The sodium hydroxide (8M)-sodium iodide (4M) solution is prepared by first dissolving 600g sodium iodide (NaI) in approximately 600ml Milli-Q water.    After the NaI is dissolved, 320g of NaOH is added slowly (caution-the solution will get hot) and the volume is adjusted to 1 liter with Milli-Q.  The solution is then filtered through a GF/F.

2.3.  Sulfuric acid solution (10N) is prepared by slowly adding 280ml of reagent grade concentrated sulfuric acid, H₂SO₄, to 770ml of Milli-Q water.   This should be prepared with caution as is gets very hot.

2.4.     The sodium thiosulfate solution (0.2N) is prepared by dissolving 50g sodium thiosulfate pentahydrate (Na₂S₂O₃.5H₂0) and 0.1g anhydrous sodium carbonate (Na₂CO₃) in Milli-Q water to a final volume of 1 liter.  This solution is prepared approximately 2 weeks before use and stored in an amber glass bottles.

2.5.  The potassium iodate standard (0.0100N) is prepared by first drying potassium iodate(KIO₃) in a drying oven for approximately one hour.  Once the KIO₃ is dried, carefully measure out 0.3567g, using a 5-place balance, and dissolve in Milli-Q water to a final volume of 1 liter.

3. Sample Drawing

3.1.     Oxygen samples are always drawn first from the Niskin bottles and should be drawn as soon as possible to avoid contamination from atmospheric oxygen.  Approximately 6 inches of Tygon tubing connected to a temperature probe via a y-connector is slipped onto the discharge valve of the Niskin.

3.2.  A calibrated volumetric flask is rinsed three times, then with the seawater still flowing, the end of the tygon tubing is placed into the flask nearing the bottom.  The sample is then overflowed with twice the sample volume while making sure that there are no bubbles in the tubing during the overflow process.  The tubing is then carefully removed from the sample flask to prevent the influx of bubbles.

3.3. Immediately after drawing the sample, 1 ml of manganous chloride solution is added into the flask.  This is followed by the addition of 1ml of sodium hydroxide-sodium iodide solution to the sample.  Both dispensers should be purged to remove air bubbles prior to the addition of these reagents.

3.4.  The stopper is then carefully placed in the bottle to avoid the trapping of air and the temperature of the seawater at the time of sample draw is recorded.

3.5.  After all samples are drawn, they are shaken vigorously to disperse the precipitate uniformly through the flask.  This process is repeated again after the precipitate has settled to the bottom of the flask, or after at least ten minutes.
 

4. Standardization of thiosulfate

4.1.  Proper care in the setup of the auto-titrator is required before running blanks, standards and samples.  The UV lamp is turned on at least 30 minutes prior to the run and should have a stable voltage of 2.4-2.5 volts for the run.  The dosimat tubing is carefully purged so that the lines are completely free of air bubbles, and the water bath inside the auto-titrator is clean and filled with Milli-Q water prior to analysis.  Make sure, prior to your first run, any thiosulfide solution is rinsed off the tip of the line with Milli-Q after purging.

4.2.  Using a Metrohm 655 Dosimat, dispense 10 ml of the standard potassium iodate solution into a clean oxygen flask.  Add a stir bar and rinse down the sides with a small amount of Milli-Q water.  Add 1 ml of the 10N sulfuric acid solution and swirl to ensure the solution is well mixed before adding the pickling reagents.

4.3.  Add 1 ml sodium hydroxide-sodium iodide solution to the acidified flask and swirl gently.  Then add 1ml of Manganous chloride solution, swirl gently, and fill the solution to the neck of the flask with Milli-Q water. 

4.4.  The UV detector on the auto-titrator measures the transmission of ultra-violet light through the standard (as well as seawater sample and blank) as a Metrohm 665 Dosimat dispenses thiosulfate at increasingly slower rates.  The endpoint is reached when no further change in absorption is detected by the detector.  At this point all of the iodine has been consumed.
 

5. Blank determination

5.1.  Using the Metrohm 655 Dosimat, dispense 1 ml of the standard Potassium Iodate solution into a clean oxygen flask.  Add a stir bar and rinse down the sides with a small amount of Milli-Q water.  Add 1 ml of the 10N sulfuric acid solution and swirl to ensure the solution is well mixed before adding the pickling reagents.

5.2.  Add 1 ml sodium hydroxide-sodium iodide solution to the acicified flask and swirl gently.  Then add 1ml of Manganous chloride solution, swirl gently, and fill the solution to the neck of the flask with Milli-Q water.  The solution is then titrated to the end-point as described in section 4.3 above.

5.3.  A second 1 ml aliquot is added to the same solution which is then titrated to a second end-point.  The difference between the first and second titration is used as the reagent blank.
 

6. Sample analysis

6.1.  Samples are analyzed after all of the precipitate settles to the bottom of the flask, after the second shake.  The top of the flask is wiped with a kimwipe to remove moisture containing excess reagent around the stopper and then the stopper is carefully removed.

6.2.  1ml of 10N sulfuric acid is added to the sample and a stir bar is placed inside the flask.  The flask is then secured inside the clean water bath.  The tip of the thiosulfate dispenser is placed inside the sample flask and the automated titration can begin with the use of an auto-titrator program.

7. Certified Standard Comparison

7.1 Presented here are the results of a comparison of several certified standards with a CalCOFI prepared standard (weighed and diluted up to set normality in 1 liter volumetric)standard.   Three certified iodate solutions, (Table 1) were tested using the same 0.2N thiosulphate solution.   Iodate concentrations are back calculated from defined thiosulphate concentrations for comparison purposes.   It is noteworthy that the Acculute and Fisher solutions had to be diluted before use.  The same volumetric was used for all dilutions, the same Dosimat and piston was used for all titrations, extensive rinsing between uses.  Titration n=9 as dictated by the maximum number of samples that could be acquired from the 100ml portion for the OSIL standard.  CalCOFI values are a result of cruise and shore based titrations to demonstrate an integrated real use sampling.  Differences between all standards represent less than   +/-  0.5 percent of the signal. 

 7.2 These results are consistent with previously published methods comparison for precision of oxygen measurements¹ (WOCE Report 73/91) and previous replicate analysis to verify precision with auto-titration methods².  Although the Ocean Data Facility (ODF) performed the 1991 comparisons for Scripps and was different than CalCOFI, the Winkler techniques used then and now are the same.  Presently, CalCOFI methods employ a UV end point auto-titrator made by ODF that produces a precision of 0.005-.01 ml/L.  The difference between high and low samples in replicate analysis equals ~0.5% with a standard deviation typically under 0.010 ml/L.  See references:

1. oceaninformatics.ucsd.edu/calcofi/docs/CCConf06Wolgast.ppt
2. odf.ucsd.edu/index.php?id=56

8. Calculation and Expression of Results

8.1.  The auto-titrator uses a UV detector that detects changes in voltage as thiosulfate is added to the sample.  The volume of thiosulfate added is recorded at an endpoint once there is no change in voltage.  The end point is determined by a least squares fit using a group of data points just prior to the end point, where the slope of the titration curve is steep, and a group of data points just after the endpoint, where the slope of the curve is close to zero.  The intersection of the two lines is taken as the endpoint.

8.2.  The calculation of dissolved oxygen follows the same principle oulined by Carpenter (1965).  Our results are expressed in mL/L.
 

O₂(ml/L)=      —————————  –  ———

Where:
R= Sample titration (mL)
Rblk = Blank value (mL)
VIO₃ = Volume of KIO₃ standard (mL)
NIO₃ – Normality of KIO₃ standard
E = 5,598 mL O2/equivalent
Rstd = Volume used to titrate standard
Vb – Volume of sample bottle
Vreg = Volume of reagents
DOreg = Oxygen added in reagents
 

9. Equipment/Supplies

• Volumetrically calibrated 100ml Glass Erlenmeyer flasks with paired ground glass stoppers
• 3 – 1ml Brinkman reagent dispensers
• Tygon tubing
• Fisher Scientific UV Longwave Pencil Lamp, 365 nm and power supply, 115 VAC
• 2 Metrohm Dosimat 665 Automatic Burets
• 10ml Metrohm Dosimat Exchange unit
• 1ml Metrohm Dosimat Exchange unit
• Metrohm Dosimat Keypad
• Spare 1ml and 10ml dispensor pistons for the Metrohm Dosimat Exchange units
• Scripps/STS Auto-titrator Unit and Software
• PC Computer
• Waterproof sampling thermometer
• Concentrated Sulfuric Acid, H₂SO₄, ACS Grade
• Manganous Chloride Tetrahydrate, MnCl₂•4H₂0, ACS Grade
• Sodium Hydroxide, NaOH, ACS Grade
• Sodium Thiosulfate, Na₂S2O₃•5H₂0, ACS Grade
• Potassium Iodate, Dry high purity KIO₃, Alfa Aesar
• Granular Sodium Iodide, EMD Chemicals via VWR
• Magnetic Stir Bars

10. References

• Anderson, G. C., compiler, 1971. “Oxygen Analysis,” Marine Technician’s Handbook, SIO Ref. No. 71-8, Sea Grant Pub. No. 9.
• Carpenter, J. H., 1965. The Chesapeake Bay Institute technique for the Winkler dissolved oxygen method.  Limnol. Oceanogr., 10: 141-143.
• Culberson, C. H. 1991. Dissolved oxygen. WHP Operations and Methods — July 1991.
• Parsons, T. R., Y. Maita, C. M. Lalli, 1984. A Manual of Chemical and Biological Methods for Seawater Analysis. Pergamon Press Ltd., 3-28.

• oceaninformatics.ucsd.edu/calcofi/docs/CCConf06Wolgast.ppt