DART SURFACE DRIFTER DATABASE
Scientific
document
The surface drifter
data contained in this database come from various buoy
designs, including the modified CODE, CODE/Tz, SVP/OCM drifters. Most
systems (e.g., the
modified CODE
type) measure the surface currents within the first top meter of water
column. Others were drogued at 15 m depth. All
drifters measure Sea Surface Temperature
(SST). Some drifters measure temperature at various levels in the water
column with a thermistor chain (CODE/Tz). Others
measure water optical properties (SVP/OCM).
The drifters were released in the Middle Adriatic between
March 2006 and August 2006
during oceanographic surveys conducted onboard NRV Alliance (DART06a and DART06b
sea trials).
All drifters were tracked by, and transmitted data to, the Argos
Data Collection and Location System (DCLS) onboard the NOAA polar-orbiting satellites.
Transmissions were programmed every 90 seconds. In
addition, some drifters were fitted with GPS receivers to obtain more accurate position at 1 h intervals.
2.
Data Reduction and editing
2.1 Drifter Data
reduction
The data for each drifter were read, reduced and written into individual
files. The sensor data were processed and reduced in the following way. The
sensor data (i.e., time, voltage, temperature and GPS position) records
telemetered during a single satellite overpass were decompressed, that is,
each record was repeated by a number of times equal to a given compression
index and the repeated records were shifted back in time by successive 90 s
increments. For the drifters without GPS receiver, the data were then sorted
in increasing sequential order and the median values were estimated. These
median statistics were assigned to the drifter Argos location and were written
into the output raw file. For the passes with good sensor data but for which
no drifter location was provided by Service Argos, the output raw latitude and
longitude were assigned with the NaN default value. For the drifters equipped
with a GPS receiver, the position and sensor times are identical and the
sampling interval is constant (1 h). No median operator was applied to
the data.
During the reading and reduction process, the times were converted into MATLAB
time (serial days where 1 corresponds to 01-Jan-0000). The deployment
coordinates (time, latitude, longitude and bucket SST) were added to the
drifter time series as the initial record. A location class 4 was assigned to
this record.
2.2
Determination
of time of last good fix and type of death
The type of
dysfunction
or the circumstances of the termination receipt of good quality
oceanographic
data have been carefully investigated by examining the suspect records
in the context of their proximity to the coast line, the values and the
probability that they were picked up by seafarers. Thus, the time for
the
last good fix was determined and the type of death was classified into
one of five categories: Recovered (intentionally), Grounded; Picked-up;
Battery
failure; Unknown. For drifter equipped with GPS receivers, the
start and end times of the Argos and GPS position time series were also
identified.
2.3 Data
editing
The position and
temperature data
were edited through automatic and manual procedures. The
criteria utilized in the automatic procedure for the position data are
based on
a maximum distance of 1000 m, a maximum speed of 50 cm/s and a maximum
angle of
45 degrees, between successive points. This means that the longitude
and
latitude of a point are flagged with NaN if (i) the distances
with
the previous
and successive points are greater than the limit; (ii) the
previous or the
successive velocities are greater than the limit and (iii) the angles
formed
with the previous and successive points are both within 180+/-45
degrees
. This procedure
is iterated twice. Thereafter, a manual procedure is performed to
eliminate the remaining residual outliers visualized on the drifter
tracks.
The temperature
data
were edited using the following conditions: the temperatures are
flagged
with NaN if (i) the temperature gradient between successive points
exceeds
1/8 degrees/hour
and (ii) the temperature difference
with respect to a blackman
running mean exceeds 5, 4, 3, 2 and 1 degree (in iterative successive
runs). The
editing based on the maximum
gradient is then repeated. Finally the manual procedure is performed
to remove remaining outliers appearing in time series plots.
3. Data
interpolation and filtering
The despiked data
were
interpolated onto regular 2-hour intervals using an optimum analysis
technique
known as Kriging.
The
Kriging used here employed an analytic function fit to the empirical
structure
function
computed from the entire despiked data set. Both the interpolated
value and an estimate of its accuracy were
computed.
The interpolated
positions and temperature were then low-pass filtered with a Hamming
filter
with cut-off period at 36 hours
in
order to remove high frequency current components, especially the
strong
tidal and inertial currents. The low-pass time series were finally
subsampled
every 6 hours and the velocity was computed by finite centered
differencing
the 6-hourly interpolated/filtered position data. The processed data
files
contain 6-hourly values of position, velocity and
temperature.
The velocity for the first and last records of each drifter, the
temperature
after failure of the SST sensor, and all the variables during temporary
grounding, were assigned the NaN default value. If the time
difference
between the interpolated point and the closest edited observation is
larger
than 3 days, the corresponding velocity was assigned NaN in order
to
avoid meaningless interpolated velocity estimates in large data gaps.