Tuchen et al. 
looking and a downward-looking 300-kHz ADCP to the CTD 
‚osette frame. The processing of raw data followed the inverse 
solution approach developed by Visbeck (2002). 
Combining Current 
Velocity Measurements 
The final data product is defined on a grid with 12-hour temporal 
and 5-dbar vertical resolution. In a first step, ADCP 
measurements prescribe the time dimension and are 
ınterpolated onto the 5-dbar vertical grid. In a second step, 
pairs of MMP profiles are placed at consecutive time steps of the 
12-hour temporal grid since the time difference between the start 
of up- and down-profiles within one profile pair is 6 hours. 
Due to several overlaps of velocity measurements, it is 
important to state the hierarchy of data chosen at one grid 
point, if more than one instrument provided data at the same 
depth and time: ADCP data overwrite current meter data and 
MMP data. MMP data and current meter data do not overlap as 
ihe MMP is bounded by physical stoppers on the mooring cable 
and current meters are located outside of this range. LADCP data 
are used when no other velocity data are available. 
In a third and final step, linear interpolation in time is applied 
at each depth of the combined data set to fill temporal gaps of up 
to 6 days. However, this method is mainly effective at the MMP 
depth range and used to fill the temporal gaps in between pairs of 
MMP profiles. 
The resulting zonal velocity time series as well as mean 
vertical profiles of zonal and meridional velocity are presented 
ın Figure 1. 
UNCERTAINTY ASSESSMENT 
For the error of the ADCP measurements, it is assumed that the 
observed variability of current velocities on seasonal to interannual 
üme scales exceeds the instrument accuracy. The uncertainty that 
arises due to the compass calibration of the individual moored 
ADCPs is assumed to be unsystematic (Brandt et al., 2021) and is 
iherefore negligible when addressing research questions on climate 
‚elevant time scales covering several mooring periods. 
During a total of seven deployments, the ADCP data coverage 
extended deep enough to overlap with data from the shallowest 
installed current meter at depths between 750 to 850 m. The 
overlap is used to assess the uncertainty by comparing these two 
independent measurements. For five deployments, the cross 
correlation (r) between the zonal velocity component from 
ADCP measurements and from current meter data is above 
c=0.92 demonstrating the coherence between different 
instruments. The correlation for the meridional velocity 
component is generally lower, but still exceeds r=0.71. During 
wo deployments (KPO-1125 and KPO-1210), current meter 
data were partially out of the ADCP range. Although high 
correlations between the horizontal velocity components are 
observed for those periods (r>0.9 and r>0.7 for zonal and 
meridional velocity, respectively), the root mean squared 
(RMS) differences were significantly higher than during 
Zrontiers in Marine Science | www frontiersin oru 
Moored Velocity Observations at 0°N, 23°W 
deployments with full overlap. RMS differences during the 
deployments with full overlap are in the range of 0.02 to 
0.04 m s°* for both velocity components. A comparison 
between overlapping ADCP and MMP data during KPO-1210 
shows good agreement as well (not shown). 
As another assessment of uncertainty, the relative compass 
error between ADCP and current meter observations was 
derived by rotating the current velocity components of one 
instrument in such a way that the highest correlation between 
both current velocity time series was found. Those optimal 
rotation angles lie within 43° (corresponding to an uncertainty 
of about 0.05 m s°* for a velocity of 1 m s”') for the five 
deployments with full overlap. 
During the processing of MMP velocities, no compass bias 
correction was applied. However, in the special case for the 
equator, it is assumed that the highest standard deviation of a 
vertical profile occurs in the zonal velocity component due to the 
high baroclinic structure. Both velocity components are rotated 
towards the angle of maximum standard deviation in the zonal 
component. The applied rotation angles were of the order of 3-5° 
and within the range of typical error estimates of compass angles. 
CONCLUDING REMARKS 
With the presented current velocity data product, we aim to 
provide an important and accessible reference data set against 
which models and reanalyses output could be validated. The time 
series will be helpful for studies focusing on long-term climate 
variability to search for connections with changes in the 
equatorial circulation over the last 20 years. Earlier versions of 
rhis data product have already been used in a variety of studies 
and provided a significant contribution to an overall improved 
understanding of equatorial ocean dynamics. The moored 
observatory at 0°N/23°W is an ongoing example of a successful 
multilateral collaboration extending over more than two decades. 
The authors commit to regularly update the data set after future 
mooring recoveries of which the next one is scheduled for 2023. 
DATA AVAILABILITY STATEMENT 
The combined current velocity data set (v1.0) generated for this 
study (including zonal (U) and meridional (V) current velocity, 
datenum time (DTIME), julian time (JTIME), pressure (P) and 
depth (Z), as well as instrument type (IT)) is provided as a 
NetCDF file. Together with a list of the individual data sets as 
described in Table 1 it is stored in the PANGAEA data repository 
under: https://doi.pangaea.de/10.1594/PANGAEA.941042. 
Future updates of this data set will also be published through 
the PANGAEA repository and linked to the initial version (v1.0) 
of the data set presented in this data report with successive 
version numbering after each mooring recovery. The MATLAB 
code and detailed instructions how to load, merge, combine and 
interpolate the individual data sets are provided via zenodo: 
https://doi.org/10.5281/zenodo.6000560. PIRATA-FR LADCP 
une 2022 | Valume 9 | Article 91097