Hawaii Ocean Time-series (HOT)
in the School of Ocean and Earth Science and Technology at the University of Hawai'i at Manoa

Sensor Correction & Calibration

SBE 3plus

Sea-Bird Electronics

FTP View Data
To assist in the interpretation of the data, it can be displayed using the Hawaii Ocean Time-series Data Organization & Graphical System (HOT-DOGS©).

Pressure

The pressure calibration strategy employed a high-quality quartz pressure transducer as a transfer standard. Periodic recalibrations of this lab standard were performed with a primary pressure standard. The transfer standard was used to check the CTD pressure transducers. The corrections applied to the CTD pressures included a constant offset determined at the time that the CTD first enters the water on each cast, and a pressure-dependent offset, obtained from semi-annual bench tests between the CTD sensor and the transfer standard.

The transfer standard is a Paroscientific Model 760 pressure gauge equipped with a 10,000-PSI transducer. This instrument was purchased in March 1988, and was originally calibrated against a primary standard. Subsequent recalibrations have been performed every 2.5 years on average either at the Northwest Regional Calibration Center, at the Scripps Institute of Oceanography or at Fluke Electronics (DH Instruments Division). The latest calibrations were conducted at the Scripps Institute of Oceanography in April 1999, May 2001, May 2003, and July 2005; and at Fluke in July 2009 and November 2012. The standard stopped working soon after the January 2014 bench tests, and was replaced by a new Digiquartz portable standard Paroscientific SN 136923 in May 2016. There was no other standard available to conduct bench tests during the February 2014 to April 2016 period.

CTD pressure transducer bench tests were done using an Ametek T-100 pump and a manifold to apply pressure simultaniously to the CTD pressure transducer and to the transfer standard. All these tests had points at six pressure levels between 0 and 4500 dbar, increasing and decreasing pressures. Pressure tests were not conducted between February 2014 and April 2016 because the transfer standard malfunctioned and could not be repaired.

Pressure transducer #101430 was used during cruises HOT-299 through -305. Pressure transducer #75434 was used from HOT-306 to -308. A correction of 0.16 dbar and -0.861 dbar were applied to the pressure offset at 0 dbar during data collection for casts conducted with sensor #101430, and #75434, respectively. However, a more accurate offset was later determined for the time that the CTD first enters the water on each cast.

For sensor #101430, the mean before-cast pressure by the end of September 2018 was -0.6 dbar, nearly 0.10 dbar higher than the mean of 0 dbar offset from the July 2018 calibration (-0.68 dbar). For sensor #75434, the mean before-cast pressure during HOT-306 through -308 was 1.3 dbar, nearly 0.3 dbar higher than the mean of 0 dbar offset from the July 2018 calibration (1.06 dbar). These differences are because before the pressure tests, the CTD is powered on 24 hours for full stabilization. The on-deck pressures are recorded only about 10 min after the CTD is powered on. Pressure stabilization tests conducted in our lab have shown that our CTD pressure sensors change by up to 0.8 dbar during the first 10 minutes after applying power to the CTD, and the pressure continues to change a few tenths of a decibar until reaching full stabilization a few hours later.

The maximum difference between before-cast and after-cast on-deck pressure for sensor #101430 for all deep casts during 2018 cruises that used this sensor was about 0.10 dbar, slightly higher the mean hysteresis measured in the lab in July 2018 and January 2019 (0.04 and 0.05 dbar, respectively). The hysteresis measured during the July 2018 and January 2019 bench test for sensor #75434 was 0.09 dbar, which is smaller than the maximum difference between before-cast and after-cast on-deck pressure, about 0.35 dbar.

The 0-4500 dbar pressure offset and hysteresis from the bench tests have been within expected values and increased only slightly for both sensors relative to the February 2018 test. A linear pressure-dependent offset was applied during data collection for sensor #101430, and #75434 to correct for the 0-4500 dbar span offset of about 0.49, and 0.14 dbar, respectively, from the February 2018 bench test.

Temperature

Five Sea-Bird SBE-3-Plus temperature transducers #1416, #2454, #2907, #4448, #5519 were used during the 2018 HOT and WHOTS cruises. The history of the sensors, as well as the procedures followed to obtain the sensor drift from the Sea-Bird calibrations are well-documented in previous HOT Data Reports (Fujieki et al., 2019, 2018, 2017, 2016, 2015, 2014, 2013, 2012, 2011, 2010, 2008, 2007, 2006, 2005, 2004, 2002, Santiago-Mandujano et al., 2000, Tupas et al., 1993, 1994, 1995, 1997, 1998, 1999, Karl et al. 1996). Calibration coefficients obtained at Sea-Bird for these sensors after 2017 and used in the drift estimates were used in the following formula that gives the temperature (in Deg C) as a function of the frequency signal (f):

temperature = 1/{a + b[ln(fo/f)] + c[ln2(fo/f)] + d[ln3(fo/f)]} - 273.15

For each sensor, the final calibration consists of two parts: first, a single "baseline" calibration is chosen from among the ensemble of calibrations during the year; second, for each cruise a temperature-independent offset is applied to remove the temporal trend due to sensor drift. The offset, a linear function of time, is calculated by least squares fit to the 0-30 Deg C average of each calibration during the year. The maximum drift correction in 2018 was less than 1.1 x 10-4 Deg C for the data collected with these sensors. The baseline calibration is selected as the one for which the trend-corrected average from 0-5 Deg C is nearest to the ensemble mean of these averages.

A small residual pressure effect on the temperature sensors documented in Tupas et al. (1997) has been removed from measurments obtained with our sensors. Another correction to our temperature measurements was for the viscous heating of the sensor tip due to the water flow. This correction is thoroughly documented in Tupas et al. (1997).

Dual sensors were used during each of the 2018 cruises. The temperature differences between sensor pairs were calculated for each cast to evaluate the quality of the data, and to identify possible problems with the sensors. Means and standard deviations of the differences in 2-dbar bins were calculated from the ensemble of all casts at Station ALOHA for each cruise. Both sensors performed correctly during the 2018 cruises, showing temperature differences within expected values. The mean temperature difference as a function of pressure was typically less than 1 x 10-3 Deg C, with a standard deviation of less than 0.5 x 10-3 Deg C below 500 dbar. The largest variability was observed in the thermocline, with standard deviation values up to 5 x 10-3 Deg C.

Conductivity

Five conductivity sensors were used during the 2018 cruises, including #2218, #2959, #3162, #3984, and #4687. Sensor #2218 was sent to Sea-Bird for evaluation after showing offsets against its sensor pair during the deep casts of HOT-306, but nothing wrong was found with the sensor. The history of our sensors is well documented in previous HOT Data Reports (Fujieki et al., 2019, 2018, 2017, 2016, 2015, 2014, 2013, 2012, 2011, 2010, 2008, 2007, 2006, 2005, 2004, 2002, Santiago-Mandujano et al., 2000, Tupas et al., 1993, 1994, 1995, 1997, 1998, 1999, Karl et al. 1996). Dual sensors were used during each of the 2018 cruises.

For each sensor, the nominal calibrations were used for data acquisition, and a final calibration was determined empirically from salinities of discrete water samples acquired during each cast. Before empirical calibration, conductivity was corrected for thermal inertia of the glass conductivity cell as described in Chiswell et. al. (1990).

Procedures for preliminary screening of bottle samples and empirical calibration of the conductivity cell are described in Tupas et al. (1993, 1994a). For cruises HOT-299 through -308, the standard deviation cutoff values for screening of bottle salinity samples were: 0.0034 (0-150 dbar), 0.0050 (151-500 dbar), 0.0019 (501- 1050 dbar), and 0.0009 (1051-5000 dbar).

A least squares fit (ΔC = b0 + b1C + b2C2) to the CTD-bottle conductivity differences was used. None of the cruises required a quadratic calibration. The calibrations were best below 500 dbar because the weaker vertical salinity gradients at depth lead to less error when the bottle and CTD pressures are slightly mismatched.

The final step of conductivity calibration was a cast-dependent bias correction as described in Tupas et. al. (1993) to allow for drift during each cruise or for sudden offsets due to fouling. Note that a change of 1 x 10-4 Siemens m-1 in conductivity is approximately equivalent to 0.001 in salinity.

Conductivity differences between sensor pairs were calculated the same as for the temperature sensors. The range of variability as a function of pressure was about ± 1 x 10-4 Siemens m-1, with a standard deviation of less than 0.5 x 10-4 Siemens m-1 below 500 dbar, from the ensemble of all the cruise casts. The largest variability was in the halocline, with standard deviations reaching up to 5 x 10-4 Siemens m-1 between 50 and 300 dbar.

Oxygen

During the 2018 cruises, our four Sea-Bird SBE-43 oxygen sensors were used: #43262, #43982, #43918 and #431601. The history of our sensors is documented in previous HOT Data Reports (Fujieki et al., 2019, 2018, 2017, 2016, 2015, 2014, 2013, 2012, 2011, 2010, 2008, 2007, 2006, 2005). Sensor #43982 displayed bad data during the HOT-300 cruise, and it was sent to Sea-Bird for evaluation, where its membrane was found punctured/ballooned. The sensor’s lid and membrane assembly were replaced. Sensor #431601 failed during the HOT-306 cruise, and it was sent to Sea-Bird for evaluation, where its membrane was found to be damaged. The sensor’s lid and membrane assembly were replaced.

Water bottle oxygen data were screened and the oxygen sensors were empirically calibrated following procedures described previously (Winn et. al. 1991; Tupas et. al., 1993). The calibration procedure follows Owens and Millard (1985), and consists of fitting a non-linear equation to the CTD oxygen current and oxygen temperature. The bottle values of dissolved oxygen and the downcast CTD observations at the potential density of each bottle trip were grouped together for each cruise to find the best set of parameters with a non-linear least squares algorithm. Two sets of parameters were usually obtained per HOT cruise, corresponding to the casts at Station 1 and 2 (calibration coefficients from cast 2 are also used to calibrate the cast at station 6, 50 and 52). The calibration procedure for the Sea-Bird SBE-43 sensors is documented in Santiago-Mandujano et. al. (2001). No oxygen samples were collected during cruise WHOTS-15 which used the same sensors used during HOT-305, therefore, coefficients from this cruise were used to calibrate WHOTS- 15 oxygen data.

Dual sensors were used during cruises, but only the sensor whose data were deemed more reliable is reported.