Chapter 3. CTD and Related Measurements

1.0 Scope and field of application
This chapter describes an appropriate method for a SeaBird CTD. The CTD with additional sensors is used to measure continuous profiles of temperature, salinity, dissolved oxygen, downwelling irradiance, beam attenuation and in vivo fluorescence. Other CTD systems are available, the details of which will not be discussed here.Individual research groups have developed a wide variety of methods of handling CTD data, some of which differ significantly from the method presented here. The BATS (Bermuda Atlantic Time-series Study) methods are presented as one example that gives good results in most conditions. As presented, they are specific to the SeaBird CTD and software. Most of the post-cruise processing can easily be modified to the data collected by other CTD systems.

JGOFS also recognizes certain protocols and standards adopted by the World Ocean Circulation Experiment (WOCE). In regard to CTD measurements of other hydrographic properties, we note the availability of the WOCE Operations Manual, particularly Volume 3, The Observational Programme; Section 3.1, WOCE Hydrographic Programme; Part 3.1.3, WHP Operations and Methods. This manual contains the reports and recommendations of a group of experts on calibration and standards, water sampling, CTD methods, etc. This report was published by the WOCE WHP Office in Woods Hole as WOCE WHP Office Report WHPO 91-1 (WOCE Report 68/91, July 1991). Copies are available on request from the SCOR Office at the Department of Earth and Planetary Sciences, The Johns Hopkins University, Baltimore, MD, 21201, USA (OMNET: E.GROSS.SCOR, fax +1-410-516-7933), or directly from the WHP Office, WHOI, Woods Hole, MA 02543 USA.

2.0 Apparatus
The SeaBird CTD instrument package is mounted on a 12 or 24 position General Oceanics Model 1015 rosette that is typically equipped with 12 l Niskin bottles. The package can be deployed on a single conductor hydrowire.

2.1 The Seabird CTD system consists of an SBE 9 underwater CTD unit and an SBE 11 deck unit. There are four principal components: A pressure sensor, a temperature sensor, a flow-through conductivity sensor and a pump for the conductivity cell and oxygen electrode. The temperature and conductivity sensors are connected through a standard Seabird “TC-Duct”. The duct ensures that the same parcel of water is sampled by both sensors which improves the accuracy of the computed salinity. The pump used in this system ensures constant sensor responses since it maintains a constant flow through the “TC-Duct”. The pressure sensor is insulated by standard Sea-Bird methods which reduces thermal errors in this signal.

2.1.1 Pressure: SeaBird model 410K-023 digiquartz pressure sensor with 12-bit A/D temperature compensation. Range: 0–7000 dBar. Depth resolution:0.004% full scale. Response time: 0.001 s.

2.1.2 Temperature: SBE 3–02/F. Range: -5 to 35°C. Accuracy ±0.003°C over a 6 month period. Resolution: 0.0003°C. Response time: 0.082 s at a drop rate of 0.5 m/sec.

2.1.3 Conductivity: (flow-through cell): SBE 4-02/0. Range 0-7 Siemens/meter.Accuracy ±0.003 S/m per year. Resolution: 5 x 10 -5 S/m. Response time: 0.084 s at a 0.5 m/s drop rate with the pump.

2.1.4 Pump: SBE 5-02. Typical flow rate for the BBSR system is approx. 15 ml/s.(The pump is used to control the flow through the conductivity cell to match the response time to the temperature sensor. It is also used to pull water
through the dissolved oxygen sensor.)

2.2 Dissolved Oxygen: (Flow-through cell): SBE 13-02 (Beckman polargraphic type) Range: 0-15 ml/l. Resolution: 0.01 ml/l. Response time: 2 seconds.

2.3 Beam Transmission: Sea Tech, 25 cm path-length. Light source wavelength = 670 nm. Depth range 0–5000 m.

2.4 Downwelling Irradiance (PAR): Biospherical QSP-200L, logarithmic output, irradiance profiling sensor. Uses a spherical irradiance receiver (no cosine collector in use). Spectral response — equal quantum response from 400–700 nm  wavelengths. Depth range: 0–1000 m. Used in conjunction with a Biospherical QSP-170 deck-board unit for measuring surface irradiance (PAR).

2.5 Fluorescence: Sea Tech SN/83 (plastic housing). Three sensitivity settings: 0–3 mg/m3 (used in BATS), 0–10 mg/m3 , and 0–30 mg/m3 . Excitation: 425 nm peak, 200 nm FWHM. Emission: 685 nm peak, 30 nm FWHM. The fluorescence unit is rated to 500 m depth and is only used on the shallow casts. Connecting the fluorescence unit requires disconnecting and rearranging some of the other instruments. The oxygen sensor is disconnected. The transmissometer is plugged into the dissolved oxygen sensor socket, and the fluorometer plugged into the transmissometer socket.The temperature transducer and conductivity cell are returned to SeaBird approximately once/twice a year for routine calibration by the NWRCC. The dissolved oxygen sensor is returned to SeaBird every six months for calibration; however, if the performance of the
cell is found to be suspect, it is returned more frequently. The pressure transducer is calibrated less frequently and it is usual that this calibration is performed during complete CTD maintenance checks or upgrades at SeaBird.

3.0 Data Collection
The CTD package is operated as per SeaBird's suggested methods. The data from the package pass through a SeaBird deck unit and a General Oceanics deck unit before being stored on the hard disk of a PC-compatible portable computer. The CTD is powered with a single conducting electro-mechanical cable. This single conductor is unable to maintain power to the CTD during bottle fires. During this time, the CTD is kept at the desired depth for 90-120 seconds, after which time a software bottle marker is created. Following the mark, the bottle is immediately fired, which takes approximately 20 seconds during which time the CTD is depowered. Once power has returned to the CTD, the package is further maintained at depth for 120 seconds. After this period, the CTD sensors are found to be stable which permits the continuation of the upcast. The data acquisition rate is 24 samples per second (Hz). The SeaBird deck unit averages these data to 2 Hz in real time. Averaging in the time-domain helps reduce salinity spiking. The 2 Hz data are subsequently stored on the PC. After each cast, a CTD log sheet is completely filled out (Figure 1). The ship's position is recorded directly from the GPS and Loran system. We use the Loran TD values rather than the Loran unit's calculated position which is not usually current. Relevant information such as weather conditions are added in the notes section. The file naming convention used for BATS CTD data is as follows:
 

After the cast is complete, these four files are immediately backed up onto floppy disks. SeaBird data acquisition and processing software are used during the cruise for preliminary observations of raw data. The programs are: In addition, the matrix manipulation program Matlab (The Math Works, Inc., 21Elliot Street South Natick, MA 01760 USA) is used for post-cruise calibration of data with the discrete samples.

4.0 Data Processing
Data processing can be done on a UNIX workstation or IBM compatible microcomputer using the SeaBird software and Matlab. The raw 2 Hz data are first converted to an ASCII format. At this stage, a pressure filter is applied which effectively eliminates all scans for which the CTD speed through the water column is less than 0.25 ms -1 . Each profile is then plotted and visually examined for bad data and spikes which are removed. The salinity and dissolved oxygen data are then passed through a 7 point median filter to systematically eliminate spikes. The oxygen data are further smoothed by the application of a 17 point running mean. The necessary sensor corrections are then applied to obtain a calibrated 2 Hz data stream (see below). Finally, for data submission and distribution, the data are bin averaged to 2 dbar resolution.

4.1 Temperature Corrections: The SeaBird temperature sensors (SBE 3-O2/F) are found to have characteristic drift rates. The drift is a linear function of time with a depen-dency on temperature. For each cruise the calibration history of the sensor is used to determine an offset and slope value. The corrected temperature measurement is given by:

4.2 Salt Corrections: The salinity calculated from the conductivity sensor is calibrated using the discrete salinity measurements collected from the Niskin bottles on the rosette. The samples from the entire cruise are combined to give an ensemble of 36 samples in the depth range 0-4200 m. The bottle salinity samples from the upcast are mapped to the downcast CTD salinity trace, at the temperature of the Niskin closure. These matched pairs from all associated casts are grouped together and used to determine a specific salinity correction. The deviation between the bottle salinity and CTD values is regressed against pressure, temperature and the uncorrected CTD salinity using a polynomial relationship: where:
dS = model (measured bottle salinity - CTD salinity)
S = calibrated salinity
R0 = offset
P = gauge pressure (dbar)
T = temperature (°C)
Su = uncorrected CTD salinity
A i , B i , C i = regression coefficients
l, m, n = order of the polynomial functions (usually = 3)

The order of each polynomial is modified for each cast to provide the best fit for the lowest order polynomial. The F-test indicates the statistical significance of the model. The r2 value predicts the amount of variance explained by the model. The r2 value and a graphical examination of the model residuals are used to determine the best form of the polynomial expression. The standard deviation of the residuals is typically less than 0.003. The consequent regression relationship is used to modify the CTD salinity values from the downcast profile and the regression relationship is reported with the CTD data.

4.3 Oxygen Corrections: In early cruises, the oxygen sensor was calibrated before each cruise. Saturated water was made by bubbling air from a SCUBA tank through tap water for 5–10 hours. Oxygen free water was made by adding 3% sodium sulfite. The current (mA), temperature and barometric pressure were recorded for both solu-tions and entered into the SeaBird program OXFIT to calculate the calibration factors for the oxygen sensor. Nevertheless, the oxygen sensor gives a very poor fit to the bottle data, probably because of both pressure and temperature hysteresis effects. There are 36 replicate discrete oxygen samples from 0-4200 m. These oxygen samples from the upcast are mapped to the downcast profile at the temperature if the Niskin closure. These matched pairs from all associated casts are grouped together to determine a single equation for the complete depth range. The measured bottle oxygen values are regressed against temperature, pressure, oxygen current, oxygen temperature and oxygen saturation such that the CTD oxygen is directly predicted by the following equation:

where:
        MO = model CTD oxygen
        R0 = linear offset
        P = pressure (dbar)
        T = temperature (°C)
        OC = oxygen sensor current (mA)
        OS (T,p,S) = oxygen saturation value at measured temperature,salinity and pressure (mmolkg)
        Ai , B i , C i , D i = regression coefficients
        l, m, n, o = order of the polynomial functions (l = 3, rest usually = 2)

The order of each polynomial is determined by comparing successive fits until the correlation coefficients stabilize, and the residuals seem randomly distributed. The standard deviation of the residuals is typically less than 1.5 mmol kg -1 .

4.4 Transmissometer Calibration. The transmissometer shows frequent offsets in deep water which indicate variations in its performance. The theoretical clear water minimum beam attenuation coefficient is 0.364 (Bishop, 1986). We assume that the minimum beam ‘C’ value observed at the BATS site in the depth range 3000-4000 m is representative of a clear water minimum. We equate this minimum value with the
theoretical minimum to determine an offset correction. The correction is given by:

where BAC min =minimum beam ‘C’ for 3000 m<depth<4000 m. This offset is applied to the entire profile.

The Sea Tech transmissometers used on these cruises have had a series of problems, some of them associated with component failures on the deeper casts. Other problems are associated with the temperature compensation unit in the transmissometer. These temperature related problems give rise to a variety of suspect behaviors: 1) high surface values (well beyond normal) that correlate with the time of day (highest at noon), 2) exponential decay within and below the mixed layer, 3) linear or expo-nential decays in the permanent thermocline, and 4) high cast to cast variability, even in deep water. The ability to distinguish between genuine patterns and instrument problems can be difficult.

4.5 Fluorometer Calibration. The fluorometer returns a voltage signal that is processed by the SEASOFT software to a chlorophyll concentration. There is a standard instrument offset which is determined from the voltage reading on deck with the light sensor blocked off. There is a “scale factor” which is determined for each chlorophyll range. The BATS fluorometer is scaled to read chlorophyll from 0 - 1.5 mgI -1 .

In addition to the standard offset, there is a post cruise offset that is applied considering the measured chlorophyll concentration in the water column. This “field offset” is determined using the data from 250 m depth:

This offset procedure is applied to all of the CTD casts on that cruise. Further regression analysis of bottle chlorophyll versus fluorometry or HPLC chlorophyll can also be performed.

5.0 References
Bishop, J. (1986). The correction and suspended particulate matter calibration of Sea Tech
transmissometer. Deep-Sea Research 91, 7761–7764.
SeaBird Electronics, Inc. CTD Data Acquisition Software manual.