_e Menn et al. 
Verification of the Calibration of HRSST 
and SST Sensors of Two Buoys 
Once calibrated, the two MoSens sensors have been integrated 
in buoys and these buoys have been placed in the calibration 
bath. A platinum 100 © thermometer has been fixed on one of 
them and protected from the air temperature variations with a 
piece of foam (see Figure 6), in order to measure the external 
temperature of the buoy and to try to detect its influence on 
the HRSST and SST measurements. In the calibration mode, 
the buoys acquire data not every second but every 5s after 
having taken off the magnet. Even if the bath temperature is 
very stable, this reduced sampling rate increases slightly the 
measurement uncertainty. 
Two verification series have been performed on the two buoys. 
[he first series was from 1 to 34°C, the buoys being in contact 
with the air in the laboratory. For the second series, from 34 to 
1°C, the buoys were covered with a survival blanket. The goal of 
this second series was to measure the effect, on HRSST and SST 
FIGURE 6 | Buoys in the calibration bath, close to the reference thermometer. 
A pt100 & thermometer is fixed on one of them and protected with a piece of 
"oam 
SVP-BRST Fiducial Reference Network 
analog measurements, of buoy temperatures closer to the water 
temperature. The blanket has been laid to shield the buoy from 
radiation within the room and thus to partially insulate the buoy 
from the room temperature, to enclose the radiations of the bath 
and to limit the air exchanges. 
The results of the first series show that, for the two buoys, 
the amplitude of the deviations is the same as the amplitudes 
measured during the verification of MoSens sensors alone (see 
Figure 7). It means that the integration of MoSens in the buoy 
does not add systematic errors to the HRSST measurements. 
Furthermore, this implies that MoSens sensors can be calibrated 
alone, before integration in the buoys, which is an essential point 
to develop a fiducial reference network. 
The results of the second series are given in Figure7 and 
in Table 3, for the buoy n° Y17-07. The table shows that in 
spite of buoy temperatures different between the two series 
(ambient vs. covered) by as much as 3.2°C, the deviations are 
similar in amplitude to the first series (0.4 mK at 34°C). It 
shows also that these deviations are more dependent on the 
cooling or the warming of the water than of the air temperature, 
because the maximal deviation is obtained at 16°C and at this 
temperature, the difference in external buoy temperatures is only 
0.7°C. Figure 7 shows that: 
At 35°C the points are superimposed because it is the last point 
of the first series and the first point of the second series. 
From 27 to 12°C the deviations show the buoy temperatures 
are higher than the reference temperatures, probably because 
of the thermal inertia of the ensemble MoSens-Buoy, as the 
temperatures of the bath is gradually reduced. 
At 1 and 6°C, the deviation is inversed because the 
temperature has been generated in increasing order. 
The two measurement series realized on the two buoys can 
be used to assess in details the reproducibility of temperature 
measurements. By using the deviation (amb.—cover.) (see 
Table 3 for n° Y17-07), the relation (14) gives another estimation 
of the expanded measurement uncertainty of two buoys. Table 4 
shows the results. The main source of uncertainty comes from 
the reproducibility of measurements impacted by the thermal 
Q.01C 
1.005 
*  Ty1707 - Tref 
El  Ty1824 - Tref 
*  Ty1707 - Tref (covered) 
» Tv1824 - Tref (covered) 
0,000 { 
2.005 
0010 — 
Tref (°C 
FIGURE 7 | Deviations obtained during the verification of HRSST sensors of two buoys during the two series, with the expanded uncertainty of the verificatior 
trontiers in Marine Science | www frontiersin.orr 
Qantembhear 2019 I Valııme AI Article A7£