Data quality of the TRITON CT sensors

The drift of CT (conductivity and temperature) sensors for mooring type (SBE37IM) was investigated by the pre-calibration and post-calibration data measured before the deployment in 1999 and after the recovery in 2000 of TRITON buoy.   The drift of temperature sensors was quite small within 3 mK from post-calibration data except for 2 sensors (Fig. 1).   The drift of conductivity sensors is shown in Figure 2 and Table 1, and could be classified by installed depth into two regimes of shallow depth range: above 200 meters and below 250 meters.   The drift of conductivity shows large positive drift in the shallower layer.  

After 1-year-mooring, the drift of conductivity in the shallower layer (above 100 meters) was 0.010 S/m (0.065psu) at 6S/m in average, and that in the thermocline layer (125-200 meters) was 0.0053 S/m (0.034psu) at 6S/m in average. Conversely, in the deeper layer (below 250 meters), the drift of conductivity was quite small (less than 0.0001 S/m with standard deviation of 0.001 S/m), and the conductivity sensors were quite stable.  

The time evolution of the drift of conductivity except for moored in the deeper layer seems to be exponentially increasing after 4-month- (120days) mooring in the surface (0-100 meters) and subsurface (125-250 meters) layers.   In the surface layer, above all, the drift of conductivity seems to increase rapidly after 8 months (240 days).   The drift in group of sensors near 90 days was calculated by three-month moored sensors in 1998.   A pinhole in temperature sensors was found after the recovery and the serial number were all young (less than 250 days), so that we guess that the negative drift around 90 days may not reflect the real drift during mooring but rather due to somewhat a hardware-oriented electric drift.   The results from in-situ comparison of TRITON salinity data with ship-board CTD system (SBE9/11 plus) also supported the positive conductivity drift with time.

Table 1  The drift of conductivity sensors by pre and post calibration data. The data are based on the sensors deployed in 1999 and recovered in 2000.

Average period of moorings Surface layer Subsurface layer Deep layer
(Days) (0-100 m) (125-200 m) (250-750 m)
Ave. S.D. N.D. Ave. S.D. N.D. Ave. S.D. N.D.
13 2.50 5.73 5 -2.75 2.54 3 0.62 0.29 4
95 -0.92 3.36 17 -1.05 2.47 11 -4.75 5.50 15
121 2.06 1.11 18 1.45 1.85 12 0.16 1.14 15
253 4.41 1.44 9 2.75 2.63 6 -0.01 0.65 8
373 9.99 5.42 20 5.27 4.64 13 0.02 1.00 15

Unit in this table is mS/m, and 1mS/m corresponds to 0.0065 psu at 6 S/m and at 30 degree-C. “Ave.”, “S.D.” and “N.D.” mean average, standard deviation and number of data respectively. Shadowed values are calculated with few numbers (less than 6) of data.

Figure 1  The drift of temperature sensors by pre and post calibration data based on deployed in 1999 and recovered in 2000.   The horizontal axis shows the days of moored. The negative days means the temperature residuals of pre calibration by JAMSTEC from the manufacturer's calibration.

Figure 2   The drifts of conductivity sensors classified by the depths of installed.   The data used to plot are the same as in Table 1.   Horizontal axis is the days of moored.   It is apparent that the drift in the surface layer occurred to positive direction, while in the deeper layer of 250-750 db, the drift was not recognized.