Nong et al. 
modified for use with Iridium communication (Wong and Riser, 
2011). Presently the ice avoidance algorithm is a feature in the 
float software of several float types. As of December 2019, more 
than 18,000 Argo CTD profiles have been collected from under 
winter sea ice around the Antarctic continent. 
In the Arctic Ocean, the French-Canadian Green Edge Project 
has successfully deployed PROVOR/ARVOR floats with the 
ice avoidance algorithm in Baffın Bay (Smith et al., 2019). 
The PROVOR/ARVOR floats are able to overcome the strong 
pycnocline in the Arctic Ocean because of their large oil reserve. 
For the Arctic Ocean, the parameters of the algorithm were set 
to Zı = 30m, Z2 = 10m, and T,ef = —0.5°C initially, with 
Tref subsequently changed to —1.1°C or —1.3°C, based on sea 
conditions. Other Euro-Argo projects, such as the Monitoring 
the Oceans and Climate Change with Argo (MOCCA) project, 
have also deployed floats in the Arctic Ocean by using the ice 
avoidance algorithm with parameters tuned to local conditions. 
An examination of a map of Argo’s geographical sampling 
density indicates that there is a weak bias toward sampling near 
coasts with major population centers (e.g., the western North 
Pacific, the western North Atlantic, and near Australia), likely 
due to the ease of deploying in these regions. This bias does not 
appear to be severe or likely to affect global statistics derived from 
the data. With the increase in deployments in the Southern Ocean 
in recent years, especially resulting from the Southern Ocean 
Carbon and Climate Observations and Modeling (SOCCOM) 
program (Riser et al., 2018), and the reduction in float divergence 
at low latitudes resulting from the use of Iridium communication, 
Argo is improving its geographical coverage in regions that are 
historically sparse in observations due to difficult logistics. 
Temperature: Accuracy and Issues 
Manufacturer Static Calibration 
Temperature sensors in SBE CTDs are calibrated with respect to 
the International Temperature Scale of 1990 (ITS-90) in stable, 
computer-controlled calibration baths. The basis of temperature 
calibration in the Sea-Bird Scientific metrology lab are two 
NIST-certified primary standards: the Jarrett triple-point of 
water cell (0°C) and the Isotech gallium melt cell (29.76°C). 
These physical standards provide temperature measurements 
with precision to 5 x 107° °C and accuracy to 0.0005°C. 
These standards are then transferred via a standardized, traceable 
procedure to the calibration baths, yielding static accuracy of 
0.002°C for the SBE-41/41CP CTDs. 
Long-Term Sensor Stability 
At Sea-Bird Scientific, long term stability for temperature sensors 
in the SBE-41/41CP is determined from repeat multi-year 
laboratory calibrations of a reference set of sensors, which yield 
a typical stability of 0.0002°C yr7!. Long-term sensor stability 
in the field is more difficult to assess than in the laboratory, 
as there are very few opportunities to retrieve floats from the 
ocean for post-deployment calibrations. Oka (2005) performed 
one such study. They investigated the long-term stability of the 
temperature sensors on the SBE-41 using 3 recovered floats. 
The floats were deployed by JAMSTEC and were in operation 
in the North Pacific Ocean for 2-2.5 years. They calculated 
rontiers in Marine Science | www.frontiersin.or 
Argo Data 1999-2019 
differences from pre- and post-deployment sensor calibration 
by using an SBE-3 standard temperature sensor and an SBE-41 
calibration bath system in JAMSTEC. Their results showed 
positive temperature changes of 1.36 (40.62), 1.58 (40.88), 
and 1.00 (40.93) x 107% °C, respectively. Hence, although 
temperature sensor drifts were detected, the amounts of drift 
were < 0.002°C over several years. 
In another study, Janzen et al. (2008) assessed temperature 
sensor stability in the SBE-41 based on experiments in 
the laboratory and on recovered floats. They conducted 
repeat calibrations on two SBE-41 CTDs over 5 years and 
post-calibrations on 6 recovered floats that had been in 
operation for 2-6 years. They reported that from the repeat 
calibrations on the two SBE-41 CTDs, the standard deviation of 
temperature measurements was 0.001°C, and from the pre- and 
post-calibrations on the 6 recovered floats, negative sensor drifts 
of no > —0.002°C. 
Currently the Argo delayed-mode QC procedure for 
temperature relies on visual inspection of float temperature 
profiles against nearby data to detect errors. After delayed-mode 
inspection, float temperature data are given the manufacturer 
quoted accuracy of 0.002°C. 
Pressure: Accuracy and Issues 
Manufacturer Static Calibration 
All strain gauge pressure sensors used on SBE CTDs for Argo 
floats are calibrated at Sea-Bird Scientific. Calibrations spanning 
both temperature and pressure ranges are necessary, as strain 
gauge pressure sensors have a nominally linear response to 
pressure and a secondary, non-linear response to temperature. 
The pressure-span calibration is performed by using automated 
dead-weight testers. The pressure sensors measure absolute 
pressure, which is converted to gauge pressure by subtracting 
mean atmospheric pressure (equivalent to 14.7 pounds per 
square inch absolute). 
Laboratory pre-deployment testing data from Argo teams 
indicate that the Druck pressure sensor displays a negative bias 
at cold temperatures that is a function of pressure. Therefore, in 
order to satisfy the accuracy requirements of the Argo Program, 
an additional temperature span calibration is performed at Sea- 
Bird Scientific. This extended calibration range improves the 
span correction at high pressures and low temperatures from + 4 
to + 2 dbar for the 2,000-dbar sensors. Repeat calibrations of 10 
sensors returned to Sea-Bird Scientific after more than a year after 
their initial calibration showed shelf drift of + 0.30 dbar per year. 
Long-Term Sensor Stability 
The long-term stability of the pressure sensors can be evaluated 
by checking the time series of sea surface pressure (SP) values that 
are used in delayed-mode pressure adjustments. Floats normally 
collect at least one SP measurement at the end of each cycle while 
transmitting data at the sea surface. These SP readings are gauge 
pressures at sea level and are mostly within 1 dbar of zero if 
the pressure transducer is stable. Therefore, any pressure sensor 
drift will be seen in the SP readings and can be eliminated by 
subtracting SP from the measured pressures (Barker et al., 2011). 
This pressure adjustment is done onboard automatically for some 
Qanteambear 2020 | Valııme 7 | Article 701