€ Menn et al. 
TABLE 6 | Results of the comparison made at sea, between buoys transmitted 
jalues. CTD and SBE 35 measurements 
Value Sst Ttrans- Ttrans- SSTcor- 
transmitted corrected Tetd Tsbe35 Tsbe35 
538758002 16.35 16.382 —0.048 —0.047 —0.014 
SST 58019 16.35 16.389 —0.048 —0.047 —0.008 
RSST 58002 16.391 —0.007 —0.006 
ARSST 58019 16.398 0.000 0.001 
"he first (last) two lines show the results for the SST analog (HRSST sensors, respectively). 
calculated in the metrology laboratory, they are within the 
calculated expanded uncertainties of buoys SST sensors, and also 
within the estimated standard deviations from the percentiles 
transmitted by the buoys. 
This comparison at sea shows firstly that the temperature 
values transmitted by the buoys are not erroneous. 
Secondly, regarding the HRSST sensors, the deviations 
obtained in comparison to two independent instruments, 
are small and probably representative of the dispersion 
of the medium temperatures. They are inferior to the 
deviations obtained with the SST analog sensors, even 
after linear correction of SST transmitted values, probably 
because of their better response time, resolution and 
calibration uncertainties. 
After this comparison, buoys were released in an eddy feature. 
[he details of this deployment are given by Poli et al. (2018). After 
initial deployment on 26th April 2018, the buoys were initially 
separated by <1km and they remained within 10km of each 
other until 23rd May. After that, they quickly diverged until the 
first one ran ashore. At the time of initial writing of this paper, 
the second buoy was still drifting with its drogue, five and a half 
month after its deployment. 
According to Poli et al. (2018), this comparison showed that 
‘once freely drifting, the buoys observe that the SST spread 
within 5min is usually smaller than 0.1 K, especially when the 
sea-state is well-mixed and the buoys are within an eddy core. 
Che availability of percentiles from the 5-min distribution of SST 
sampled at 1 Hz (by a sensor with a fast response time) should 
help users improve their data processing chain to move toward 
an ensemble approach.” The results of this other paper suggest 
also that “it is important to consider the sea-state mixing and the 
ocean surface circulation to understand the representativeness of 
the in-situ SST data, as they both affect observed SST variations 
(within the day and within 5min). Consequently, they may 
both be worth taking into account in the process of satellite 
SST cal/val.” 
CONCLUSION 
The goal of this study relates to the conception and the 
metrological characterization of new surface drifting buoys, 
design to comply with the requirements of SST satellites 
measurements validation and to link through comparison these 
measurements to the SI. This linkage can be achieved by the 
rontiers in Marine Science | www.frontiersin.or 
SVP-BRST Fiducial Reference Network 
calibration of each buoy and the assessment of the instrumental 
measurement uncertainty, taking into account all the elements of 
the temperature calibration chain. 
Calibrating individually 100 drifting buoys in a calibration 
bath is time-consuming and unrealistic. This study shows that 
it is possible to calibrate the sensors and their conditioning 
electronic circuits beforehand, without adding significant 
errors or uncertainties to in situ measurements even once 
the sensors have been integrated in buoys, and to keep the 
instrumental uncertainty under the tolerance of 0.01°C. This 
was possible through the design of the MoSens modules 
which include high resolution temperature sensors and 
hydrostatic pressure sensors. The concept of high resolution 
includes the possibility to make temperature measurements 
with a repeatability close to a milli-degree, a fast thermal 
response time measured in laboratory and a fast sample rate 
(1 Hz). 
The measurements made on the two buoys have also enabled 
the improvement of the calibration of the SST analog sensors. 
If, initially, their measurement errors are already included 
in the +0.1°C tolerance, it is possible, by using slope and 
offset correction coefficients, to obtain instrumental expanded 
measurements uncertainties inferior to 15 mK. With these 
corrections, in situ comparisons have shown that it is possible 
to reduce the deviation of 0.047 to —0.014°C for one sensor 
and —0.008°C for the other. However, this correction procedure 
requires each buoy to be placed in the calibration bath. This 
is not feasible for an industrial process to ensure repeated 
accuracy. Also, one must bear in mind that the large size of 
the SST analog sensors makes them much slower to respond to 
seawater variations than smaller HRSST sensors, as shown in 
this paper. 
The temperature-dependence of the MoSens pressure sensor 
has also been studied. It can lead to errors of -+0.15 
dbar in the temperature range 0-35°C. These errors can 
be compensated with average slope and offset coeflicients to 
improve the determination of HRSST measurements depth 
during calm sea conditions. In rough sea conditions, this 
sensor provides an indication of the sea-state, which is 
essential to understand the deviations between satellites and 
buoys temperature measurements. The relationship between 
information contained in the high-frequency data and the sea 
state should be explored in future work. 
The specifications of two prototypes measured 
in laboratory, have been confirmed during the 
initial deployment at sea by a comparison to a 
reference thermometer SBE 35 and a CTD profiler, 
demonstrating also the good transmission of data and 
the very good trueness of HRSST measurements in a 
homogeneous medium. 
Future prospects include deploying at least 100 SVP-BRST 
units, with the aim of closing the metrological loop with buoys 
that would be recovered from sea, to be verified in a calibration 
bath. An experiment will also be carried out with a SVP-BRST 
buoy kept over a long duration at a fixed position at sea, next to 
a reference moored buoy. This will allow to determine whether 
Zantembear 2019 I Valıme A 1 Article S7£