P. Poli et al.: SVP-BRST: genesis, design, and initial results 
205 
www.ocean-sci.net/15/199/2019/ 
Ocean Sci., 15,199-214, 2019 
6200500 6200504 4100736 6200505 4100800 6200513 
00 h 01 h 02 h 03 h 04 h 05 h 06 h 07 h 08 h 09 h 10 h 11 h 12 h 13 h 14 h 15 h 16 h 17 h 18 h 19 h 20 h 21 h 22 h 23 h 
Mean solar local time (MSLT) 
Figure 3. Mean differences (a) between the two SST sensors, with the number of data records shown in panel (b), for HRSST-2 SVP-BS 
buoys that reported for at least 250 days (WMO identifier indicated in legend), as a function of mean solar local time (horizontal axis). 
la Marine (SHOM) metrology lab, which used seven verifica 
tions points (between 2 and 32 °C, at steps of 5 °C), and lab. 
no. 3 refers to the Scottish Association for Marine Science 
(SAMS) metrology lab, which used three verification points 
(0, 10, and 20 °C). 
To remove the impact of different dates, the last column 
shows the estimated temporal drifts. The drift results vary 
in magnitude between the probes and the laboratories. This 
is probably mainly because of different choices for the ver 
ification temperatures, though other factors may have also 
played a role, such as probe resolution, probe response time, 
and temperature laboratory influence on the measurements 
(with the electronics not immersed in water), among others. 
However, all the results found here suggest negative trends, 
around — O.OlKyear -1 for lab. no. 2 and —0.005 Kyear -1 
for lab. no. 3. 
Note that it cannot be ruled out that the probes, once re 
moved from the buoys, did respond differently than during 
the initial calibration setup. Indeed, the temperature varia 
tions being looked at are very small, and any influence of 
the acquisition electronics may affect the results. The exact 
environment used for housing the electronics during calibra 
tion of the initial probes, as well as during the verifications, 
even if specified in the initial calibration sheets, cannot be 
replicated with certainty. 
Consequently, these results are to be taken with caution, 
and the importance of the calibration apparatus stands out as 
being an important part of the traceability. However, should 
the negative trend (cooling) be confirmed, it would have an 
impact on the exploitation of the SST drifter data for satel 
lite cal/val, as well as corrections that are made to global 
datasets. Recent adjustments have actually recognized buoys 
as being cooler than ships in terms of SST (Huang et al., 
2015), in line with earlier findings (e.g., Emery et al., 2001; 
Rayner et al., 2010), though no difference was made espe 
cially for drifting buoys as a function of their “age”. The 
three recovered buoys achieved lifetimes of (respectively) 
580, 515, and 453 days (see Table 1). These durations are 
close to or above the average drifter lifetime of 450 days 
(Lumpkin et al., 2012). Considering all the estimated tem 
poral drifts shown in Table 3, the temperature biases of these 
drifters (averaged over the mission duration) would range be 
tween —0.002 and —0.010 K. 
In conclusion, given the importance of drifting buoy SST 
in climate studies, the impossibility of putting together firm 
metrology results indicates that a better-documented calibra 
tion protocol is needed for the measurement of SST by these 
platforms, both to ensure initial calibration and calibration 
verification several years afterwards. 
2.4 Evaluation of HRSST-1 and HRSST-2 drifters 
The analysis of O’Carroll et al. (2008) identified the stan 
dard deviation of error of the drifting buoy network to be 
0.23 K. An interpretation of this finding is it is equivalent 
to the standard uncertainty of the error distribution. An al 
ternate approach to the method of O’Carroll et al. (2008) 
is to derive a theoretical uncertainty estimate for the satel 
lite SST (Bulgin et al., 2016), which can then be validated 
using satellite/drifter differences (Lean and Saunders, 2013; 
Bulgin et al., 2016; Neilsen-Englyst et al., 2018). The con-