Paul Dodd

Amelie Meyer

Zoe Koenig

Adam Cooper

Lars H. Smedsrud

Morven Muilwijk

Victoire Rerolle

Nikita Kusse-Tiuz

Jonas Miguet

Ingrid Onarheim

Peter Davis

Achim Randelhoff

Shipboard CTD profiles in MatLab format, with complete and corrected 'metadata' (time position etc…). Standard quality control applied and post cruise calibration applied (see below).

Variables:

```

Name Size

C1_ship 4091x51 Conductivity sensor 1

C2_ship 4091x51 Conductivity sensor 2

P_ship 4091x1 Pressure (dbar)

S1_ship 4091x51 Practical salinity sensor 1

S2_ship 4091x51 Practical salinity sensor 2

T1_ship 4091x51 In situ temperature (oC) sensor 1

T2_ship 4091x51 In situ temperature (oC) sensor 2

lat_ship 1x51 CTD station latitude

lon_ship 1x51 CTD station longitude

sound_vel 4091x51 Water sounds velocity (m/s)

station_ship 1x51 CTD station number

time_ship 1x51 CTD station time

version 1x56 Data file version

```

Notes about quality control applied:

First attempted to use information from the pre/post temperature and conductivity  data sheets from SBE following the procedure in their application note 31.  But the results were rather poor when compared to bottle salinity data. There were 2 main problems: 1) A long time had elapsed between the cruise and pre/post calibrations. 2) There was a big jump in both the conductivity sensors between stations 16 and 17 (end of leg 1 and start of leg 2). This jump means that an assumption of even drift in conductivity slope with time is not valid. The conductivity slope corrections estimated using SBE’s application note 31 pre/post method did improve the accuracy of salinity measurements relative to bottle salinities, but had a  negative effect on the precision (as estimated from the difference between the salinities measured by sensor pairs 1 and 2.

We accepted the temperature drift offset corrections (effectively zero) suggested by the pre/post calibration, but rejected the conductivity slope values derived from pre/post calibration and calculated conductivity slope corrections using bottle salinity values instead. We used the alternative procedure described in SBE application note 31 for calibration using on board bottle salinity measurements. A good number of bottle salinity results were available from deep water masses throughout the cruise. We estimated two slope corrections for each sensor. One for stations 1 to 16 during leg1 before the jump (applied in postcruise1-16.con) and one for stations after the jump (applied in postcruise17-51.con). Below are the conductivity slopes relative to bottle data estimated before any conductivity slope correction was applied (remember that pre/post calibration suggested that no temperature correction is required):

Stations 1-16 Primary conductivity: 0.99999052

Stations 1-16 Secondary conductivity: 1.00012038

Stations 17-51 Primary conductivity: 1.00021485

Stations 17-51 Secondary conductivity: 1.00022678

After applying these slope calibrations we get the following residual values:

Stations 1-16 Primary conductivity: 1.00000004

Stations 1-16 Secondary conductivity: 1.00000001

Stations 17-51 Primary conductivity: 0.99999999

Stations 17-51 Secondary conductivity: 1.00000003

Typical differences between the salinity measured by the sensor pairs are now less than 0.001.

If both sensor pairs were calibrated using the pre/post method there would be some merit in averaging the values from both sensor pairs to achieve the best possible absolute accuracy (they could be considered independent measurements). Here, however both sensors have been adjusted to agree with the bottle data, so we do not think there is much to gain by averaging them, rather use one and revert to the other if some spike or other bad data is encountered.

We have done some elementary data clean up - mostly just deleting the tops of profiles where the CTD wasn’t soaked for long enough to get accurate values. Bad data points were replaced with NaNs and not interpolated.

CEOS IDN DIF

9.8.4