Poster: Circulation of the North Atlantic Ocean During the 1990's as Determine by Lagrangian Drifters

1. We have assembled a new North Atlantic climatology of near-surface velocity and eddy kinetic energy (EKE) based exclusively on 15-m
drogued surface drifter trajectories measured between 1990 and 2000. Our objective is to define the state of the North Atlantic surface
circulation during the 1990’s and provide a reference point for both synoptic circulation studies (e.g. theWOCE Atlantic Circulation and
Climate Experiment) and studies of circulation variability associated with inter-decadal climate variability. In this poster we present
decadal-mean circulation statistics computed on a one-degree grid and compare these results with satellite altimetry measurements
and with previous drifter-based studies in the North Atlantic.
The satellite-tracked surface drifters
used in this study are similar in construc-
tion to the WOCE/Tropical Ocean-Global
Atmosphere (TOGA) Lagrangian drifter
described by Sybrandy and Niiler (1991).
All drifters were tracked using ARGOS and
were fitted with a submerged flexible
drogue which hung suspended at a central
depth of 15 m beneath a buoyant surface
float [Figure 1]. Approximately 1500
individual drifter trajectories totaling
nearly 300,000 drifter-days of position and
velocity information were acquired from
the archives of the Global Drifter Data
Assembly Center at NOAA/AOML in
Miami, Florida, U.S.A. Initial processing of
the data at AOML, including quality
control and temporal interpolation, is
described in detail in Hansen and Herman
(1989) and Hansen and Poulain (1996).
Drifter trajectories were truncated when
drogues detached as indicated by an
onboard submergence sensor or strain
gauge. The final data product consists of 6-
hourly interpolated position, velocity, and
surface temperature measurements.
The drifter trajectories used in this
analysis are shown in Figure 2. The
population of drifters in the North Atlantic
increased during the 1990’s and does not
exhibit any particular seasonal bias
[Figure 3]. The regional distribution of drifter trajectories varies signifi-
cantly throughout the decade [Figure 4]. Note the shift in observational
emphasis from the Gulf Stream region (1990-1992) to the eastern subtropi-
cal gyre (1992-1995) to the subpolar gyre (1995-1998) and finally to the
western tropical Atlantic and Caribbean Sea (1997-1999).
A summary of the fastest- and slowest-moving drifters is shown in Figure
5. The fastest drifter motions were found near the equator, along the
western boundary, in the eastward Gulf Stream extension, on the eastern
and western coasts of Greenland, and in a narrow eastward band corre-
sponding to the Azores Current. The greatest number of slow-moving
drifters were found in the eastern subtropical Atlantic.
1000˚W 900˚W 800˚W 700˚W 600˚W 500˚W 400˚W 300˚W 200˚W 100˚W 100˚E
10˚S
10˚N
20˚N
30˚N
40˚N
50˚N
60˚N
70˚N
All Trajectories 1990-2000
0˚
0˚
Fabric Drogue
Surface Float
Subsurface Float
PhotographcourtesyofMarkSwenson,NOAA/AOML
FIGURE 1: AWOCE/TOGA
Lagrangian drifter on the surface
shortly after being deployed. The
fabric drogue is weighted and will
quickly sink to a vertical position
beneath the surface float.
FIGURE 2: A
composite dia-
gram of all
drifter trajecto-
ries used in the
present analy-
sis. All trajec-
tory segments
shown were
measured be-
tween January
1990 and June
1999. Note the
very poorly
sampled region
in the south-
eastern sub-
tropical gyre.
1990 1995 2000
Year
0
50
100
150
200
NumberofObservations(1000's)
J F M A M J J A S O N D
Month
a b
Speed > 40 cm/s
80W 60W 40W 20W 0
Speed < 10 cm/s
80W 60W 40W 20W 0
0
20N
40N
60N
a b
Overview
FIGURE 3: Temporal distribu-
tion of drifter data presented
in histogram form. The verti-
cal axis indicates the number
of 6-hourly measurements of
position and velocity. (a) Dis-
tribution by year from 1990
onwards (through June 1999). (b) Distribution by month.
FIGURE 5: A summary of the fastest and slowest drifters in the present database. Only
trajectory segments meeting the specified speed criterion for a contiguous 36-hour period
are plotted. (a) Fast drifters, with speeds exceeding 40 cm/s. The fastest drifters were found
near the equator, along the western boundary, in the eastward Gulf Stream extension, on
the eastern and western coasts of Greenland, and in a narrow eastward band correspond-
ing to the Azores Current. (b) Slow drifters, with speeds less than 10 cm/s. The greatest
number of slow-moving drifters were found in the eastern subtropical Atlantic.
Using historical hydrographic observations Curry and McCartney (2000)
describe a relationship between baroclinic ocean transport in the
subtropical gyre and the phase of the atmospheric North Atlantic Oscilla-
tion (NAO). We now consider whether the present drifter data, in combina-
tion with previous drifter observations, can be used to directly measure
interdecadal changes in surface circulation strength or character.
Richardson (1983) synthesized North Atlantic surface drifter measure-
ments during the period 1977-1980. This time interval coincides with a
period during which the NAO index (Hurrell, 1995) was relatively low
[Figure 10]. In contrast, the first half of the 1990’s were characterized by
particularly large values of the NAO index. Large values of the NAO index
correspond to stronger westerly winds and tend to result in a more
northerly Gulf Stream position (Taylor and Stephens, 1998).We hypoth-
esize that variations in the baroclinic transport of the subtropical gyre/Gulf
Stream system associated with these NAO index extrema should result in
observable changes in the near surface circulation.
1998
Gulf Stream
1
Newfoundland
Basin
3
Gulf Stream
Extension
2
North Atlantic
Current
4
70W 50W 30W
40N
50N
0
500
1000
1500
2000
2500
Gulf Stream Gulf Stream
Extension
Newfoundland
Basin
North Atlantic
Current
1990-2000 (This study)
1977-1980 (Richardson, 1983)
EKE(cm2/s2)
1 32 4
Acknowledgements
Eddy kinetic energy derived fromTOPEX and ERS altimeters was provided by Nicolas Ducet of
CLS Space Oceanography Division, France. Processing of the raw drifter data was performed at the
Global Drifter Data Assembly Center at NOAA/AOML under the direction of Mark Swenson and
Mayra Pazos.This work was supported by the National Oceanic and Atmospheric Administration as
part of a cooperative project with Dr. Robert Cheney of the NOAA Laboratory for Satellite Altimetry.
References
Hansen, D.V. and A. Herman, Temporal sampling requirements surface drifter buoys in the tropical Pacific, J. Atmos.
Ocean. Tech., 6, 599-607, 1989.
Hansen, D.V. and P.-M. Poulain, Quality control and interpolations of WOCE/TOGA drifter data, J. Atmos. Ocean.
Tech., 13, 900-909, 1996.
Hurrell, J. W., Decadal trends in the North Atlantic Oscillation: Regional temperatures and precipitation. Science,
269, 676-679, 1995.
Curry, R.G. and M.S. McCartney, Ocean Gyre Circulation Changes Associated with the North Atlantic Oscillation,
submitted to J. Phys. Oceanogr., 2000.
Richardson, P. L., Eddy kinetic energy in the North Atlantic from surface drifters, J. Geophys. Res., 88, 4355-4367, 1983.
Sybrandy, A. L. and P. P. Niiler, WOCE/TOGA Lagrangian drifter construction manual, SIO Ref. 91/6, WOCE Rep. 63,
58 pp., Scripps Institution of Oceanography, La Jolla, Calif., 1991.
Taylor, A. H., and J. A. Stephens,The North Atlantic Oscillation and the latitude of the Gulf Stream,Tellus, 50A, 134-142, 1998.
As additional drifter data from the 1990’s become available (or are made
known to us by others) we will continue to update and improve this
climatology. We are presently extending our study of the drifter-derived
quasi-Eulerian velocity and EKE fields to include joint analysis of contem-
porary hydrographic data, TOPEX and ERS altimetry, and surface wind
products. Drifter-derived surface velocity and EKE data on one- and two-
degree grids are available for general use by the oceanographic commu-
nity. Please contact the lead author for details (dfratantoni@whoi.edu).
FIGURE 4: Temporal and spatial distribution of drifter
trajectories for each year from 1990-1999. Note the shift
in observational emphasis from the Gulf Stream region
(1990-1992) to the eastern subtropical gyre (1992-1995)
to the subpolar gyre (1995-1998) and finally to the west-
ern tropical Atlantic and Caribbean Sea (1997-1999).
We compared the Richardson (1983) quasi-Eulerian circulation statistics
with those resulting from our 1990-2000 analysis. There are significant
differences in both data volume and spatial distribution between the two
climatologies (the 1990-2000 climatology contains almost an order of
magnitude more data than the 1977-1980 analysis). We tried to minimize
the importance of these differences by concentrating our comparisons in a
region encompassing the Gulf Stream and the North Atlantic Current,
features that are reasonable well sampled in both analyses. In addition, we
focused our comparisons on EKE rather than on the mean velocity field. As
found by Richardson (1983), EKE exceeds the energy of the mean circula-
tion over most of the North Atlantic by a factor of about 10-20. This strong
variability makes it difficult to accurately resolve the mean circulation
without enormous quantities of data. The present 1990-2000 climatology is
sufficiently data-dense to compute meaningful means over large areas —
the 1977-1980 dataset is not.
-10
0
10
20
30
40
50
U(cm/s)
0
500
1000
1500
2000
2500
3000
EKE(cm2/s2)
30354045
Latitude
30354045
Latitude
1990-2000
1977-1980
55W65W
55W65W
1990-2000
1977-1980
1990-2000
1977-1980
Gulf Stream Sections at 65W and 55W
We computed EKE within four spatial subregions [Figure 11] using both
climatologies. In the Gulf Stream, Gulf Stream Extension, and Newfound-
land Basin subregions we find an increase in EKE in the 1990-2000
climatology relative to the 1977-1980 measurements [Figure 12]. However,
within the computed 90% confidence limits there has been no significant
change in the regionally-averaged EKE over the last 20 years. Note that the
Richardson (1983) climatology includes both drogued and undrogued
drifter trajectories. This suggests that the 1977-1980 EKE values could be
overestimates of the actual EKE, particularly in the Gulf Stream region
where large wind stress and strong synoptic atmospheric variability may
account for much of the motion of an undrogued drifter.
We compared sections of zonal velocity and EKE across the Gulf Stream at
55W and 65W [Figure 13] and found little difference in the structure of the
zonal Gulf Stream jet. There is a slight northward shift in the location of the
mean jet and the associated EKE maximum at 55W in the 1990-2000
climatology relative to the 1977-1980 realization.While the sense of this shift is
consistent with our expectations based on the phase of the NAO (e.g.Taylor
and Stephens, 1998), the magnitude of the shift is not statistically significant.
To summarize, we find that Lagrangian observations of surface velocity
and EKE in the vicinity of the Gulf Stream do not show a significant change
in circulation strength or character from the late 1970’s to the 1990’s.
Although the differences we observed in regional eddy variability and in
the structure of the Gulf Stream jet are suggestive, the available data are
insufficient to prove or disprove our initial hypothesis.
-6
-4
-2
0
2
4
6
1960 1970 1980 1990 2000
Year
NAOIndex
North Atlantic Oscillation Index
This study
Richardson (1983)
5-year mean
FIGURE 10: The North At-
lantic Oscillation (NAO)
index for the past 40 years.
The black curve corre-
sponds to a 5-year running
mean. Intervals corre-
sponding to the Richardson
(1983) surface drifter
analysis (1977-1980) and
the present analysis (1990-
2000) are shaded.
FIGURE 13: A comparison of the meridional structure of zonal velocity (upper panels)
and eddy kinetic energy (EKE; lower panels) between the present surface drifter clima-
tology (1990-2000; red) and the previous 1977-1980 (blue) climatology of Richardson
(1983). There is a slight (but statistically insignificant) northward shift of the Gulf
Stream and its associated EKE maximum at 55W.
FIGURE 11: Eddy kinetic energy (EKE)
was computed within four rectangular
subregions corresponding
to areas of relatively dis-
tinct circulation character.
(1) the Gulf Stream; (2)
the eastward
Gulf Stream
Extension; (3)
the Newfound-
land Basin, and (4) the
North Atlantic Current. Tra-
jectories from a single year
of the 1990-2000 drifter cli-
matology are shown.
FIGURE 12: A compari-
son of eddy kinetic en-
ergy (EKE) between the
present surface drifter
climatology (1990-
2000; red) and the pre-
vious 1977-1980 (blue)
climatology of
Richardson (1983). EKE
was computed in four
subregions (see Figure
11). The vertical black
bars denote the 90%
confidence interval.
Drifter velocity was computed using a cubic spline
function at each 6-hourly interpolated position. The
resulting velocities were grouped into spatial and temporal
bins to construct quasi-Eulerian fields of velocity and eddy
kinetic energy (EKE). EKE was defined as one-half the sum
of the zonal and meridional velocity variances within each
grid box. The number of individual velocity observations in
each one-degree square is shown in Figure 6.
The decadal-mean surface velocity field for the North
Atlantic constructed at one-degree resolution for the period
January 1990 – June 1999 is shown in Figure 7a. The
corresponding EKE field is shown in Figure 7b. Note the
enhanced EKE in the vicinity of the Gulf Stream down-
stream of the New England Seamounts (40N), and in the
Labrador Sea near theWest Greenland Current (60N). There
is also a zonal band of elevated EKE near 34N associated
with the Azores Current. More detailed views of the surface
circulation in the subpolar gyre, the Gulf Stream region,
and the Caribbean Sea are shown in Figure 8.
In Figure 9 we compare EKE computed from our surface
drifter climatology with EKE derived from TOPEX and ERS
altimetry during the period 1992-1998. The general spatial
distributions of EKE are similar with highest values
located within the Gulf Stream downstream of the New
England Seamounts. The enhanced EKE in the Labrador
Sea seen in Figure 9a is absent in the altimeter-derived
EKE field. This is probably due to the choice of a constant
(rather than latitude-dependent) horizontal lengthscale in
the altimeter EKE calculations. RMS sea level anomalies
from ERS and TOPEX confirm the region of enhanced
variability in the vicinity of the West Greenland Current as
seen in the drifter-based EKE field.
70W 60W 50W 40W 30W 20W 10W
30N
40N
50N
60N
Surface Drifter Climatology
1990-2000
Eddy Kinetic Energy (cm2/s2)
TOPEX and ERS Altimetry
1992-1997
Variance Axes
1000 cm2/s2
a
Data Source: Saskia Esselborn, Institut fur Meereskunde, Hamburg, Germany
300 400 500 750 1000 1500 2000 2500 3000
70W 60W 50W 40W 30W 20W 10W
30N
40N
50N
60N
b
FIGURE 8: Detailed view of mean velocity field in three regions. (a) The Labrador Sea and Subpolar
Gyre. (b) The Gulf Stream and its eastward extension.(c) The Caribbean Sea.
60W 50W 40W 30W 20W 10W 0 10E
50N
60N
70N
50 cm/s
10 cm/s
80W
30N
40N
50N
70W 60W 50W 40W 30W
50 cm/s
10 cm/s
90W 80W 70W 60W
10N
20N
30N
0 15 30 100
50 cm/s
10 cm/s
FIGURE 6: Number of individual 6-hourly velocity observations per one-degree square.
FIGURE 7: (a) A decadal-mean surface velocity field for the North Atlantic computed by aver-
aging northward and eastward drifter velocities into one-degree square bins over the period
January 1990 – June 1999.Vectors are only shown for bins containing more than 100 individual
observations. Error ellipses correspond to one standard error in the zonal and meridional di-
rections.We assumed a 10-day Lagrangian integral timescale in the error computations.
FIGURE 7: (b) The corresponding field of eddy kinetic energy (EKE). Regions with insufficient
data (less than 100 observations per one-degree square) are shaded gray. Note the enhanced
EKE in the vicinity of the Gulf Stream downstream of the New England Seamounts (40N), and
in the Labrador Sea near the West Greenland Current (60N).
90W 80W 70W 60W 50W 40W 30W 20W 10W 0
0
10N
20N
30N
40N
50N
60N
70N
<100
100-200
200-500
>500
80W 70W 60W 50W 40W 30W 20W 10W 0 10E
20N
30N
40N
50N
60N
70N
0 15 30 100
50 cm/s
10 cm/s
Mean Surface Velocity (cm/s)
80W 70W 60W 50W 40W 30W 20W 10W 0 10E
20N
30N
40N
50N
60N
70N
0 100 200 300 400 500 750 1000 1500 2000 2500 3000
Eddy Kinetic Energy (cm2/s2) FIGURE 9: Eddy kinetic energy (EKE) in the subtropical and
subpolar gyres computed from (a) the present surface drifter
climatology, and (b) a blend of ERS and TOPEX satellite altim-
etry for an overlapping time period. The contour intervals and
shading are identical.