same map. The western boundary current systems
were averaged on the same scales as mid-ocean
circulation, but with full awareness that this 2°6°
average, as originally specified in the WOCE plan,
cannot capture the true spatial scale or the velocity
amplitude of the western boundary circulation features. Each boundary current requires a binning
strategy that is unique and is not presented in this
overview for every boundary current system. An
example of the effects of reducing the averaging
scale in the western North Pacific is discussed
below. Also, bin size affects the computation of the
scale-dependent dynamical quantities such as divergence (Poulain, 1993) and the relative vorticity. A
more sophisticated binning routine would have to
be used if stable gradients of the mean velocity are
desired, an objective not in the WOCE plan.
Finally, the drifters followed the two-dimensional circulation at 15 m depth and this velocity
field could be strongly convergent or divergent due
to the sinking or rising of water vertically. Drifters
were usually released in large-scale divergent areas
along merchant shipping lines in the tropics or
regions near coasts from where they travelled to
the mid-gyres that were typically convergent. The
resulting sampling of the velocity field thus could
occur in a biased fashion if the tendency to remain
in a convergent region was greater than the windproduced slip to transit the region. Regions of
strong convergence could, thus, be sampled more
often than regions of divergence, and if the velocity
in either region were stronger, then a bias would
result from the ensemble mean. Within the
Kuroshio, south of Kyushu, this biased sampling
led to the result that the ensemble mean drifter current was about 30 cm s
91 larger than the average
current derived from a combination of current
meters, hydrography and satellite altimeter sea-level
observations (Uchida et al., 1998). Apparently,
drifters converged towards the velocity axis of the
Kuroshio. This is another reason why the detailed
structures of each of the strong western boundary
current regions were not discussed; each boundary
current would require a different velocity data set
to assess the potential biases of drifter sampling.
Satellite altimeter data offer a technique for testing the sampling characteristics of the variance of
the velocity also in the open ocean, although they
cannot provide a test for the mean. This test was
carried out in the North Pacific in a comparison
of the variance of geostrophic currents normal to
the TOPEX/POSEIDON altimeter track. The
geostrophic component of the drifter velocity was
computed by subtracting the Ekman component
according to the formulae specified in Section
4.1.4 (Ralph and Niiler, 1999). The geostrophic
velocity normal to the satellite track from drifter
data was binned on a nominal 2°3° spatial scale.
Then bins were chosen with more than 33, 5-day
average observations and where more than five
drifters produced the data. If data from only one
or two drifters existed in a bin, the data could not
be representative of average conditions, as these
drifters could have resided in very special features
such as a single cold or warm eddy. This sampling
is equivalent to 164, 2-day average observations
on a 2°6° resolution, as displayed on Fig. 4.1.4.
Using the above criteria, the ensemble average
variance from drifters was within the standard
error of the track normal geostrophic velocity
variance computed from the TOPEX/POSEIDON
satellite altimeters from the period 1992–98 (Barth
et al., 2001). This comparison was important for
testing the scheme of averaging both drifter and
altimeter data and for the future considerations of
merging of two global circulation data sets in such
a way that biases between the two data sets would
be minimized.
4.1.3 The global mean velocity and
velocity variance
The principal objective of WOCE was to observe
the world ocean circulation. The 15-m depth ensemble average drifter velocity of the world shows
the major surface current systems (Fig. 4.1.5a, see
Plate 4.1.5, p. 300). Apparent on Fig. 4.1.5a are
the strong poleward-flowing western boundary
currents in each ocean and the strong zonal
currents in the tropics that flow in the direction of
the winds. In the Norwegian Sea, there is a strong
poleward current along the eastern boundary
(Poulain et al., 1996). The North Pacific Equatorial Countercurrent, flowing counter to the wind, is
evident in the 5–10°N latitude band, spanning the
entire North Pacific basin. Mariners could approximately locate most of these currents a century
ago from the ship-drift charts (Bowditch, 1966;
Richardson, 1989). The observations used in Fig.
4.1.5a were made with calibrated instruments,
providing a nearly 10-fold improvement in accuracy of individual observations over those of ship
SECTION 4 THE GLOBAL FLOW FIELD
198
were averaged on the same scales as mid-ocean
circulation, but with full awareness that this 2°6°
average, as originally specified in the WOCE plan,
cannot capture the true spatial scale or the velocity
amplitude of the western boundary circulation features. Each boundary current requires a binning
strategy that is unique and is not presented in this
overview for every boundary current system. An
example of the effects of reducing the averaging
scale in the western North Pacific is discussed
below. Also, bin size affects the computation of the
scale-dependent dynamical quantities such as divergence (Poulain, 1993) and the relative vorticity. A
more sophisticated binning routine would have to
be used if stable gradients of the mean velocity are
desired, an objective not in the WOCE plan.
Finally, the drifters followed the two-dimensional circulation at 15 m depth and this velocity
field could be strongly convergent or divergent due
to the sinking or rising of water vertically. Drifters
were usually released in large-scale divergent areas
along merchant shipping lines in the tropics or
regions near coasts from where they travelled to
the mid-gyres that were typically convergent. The
resulting sampling of the velocity field thus could
occur in a biased fashion if the tendency to remain
in a convergent region was greater than the windproduced slip to transit the region. Regions of
strong convergence could, thus, be sampled more
often than regions of divergence, and if the velocity
in either region were stronger, then a bias would
result from the ensemble mean. Within the
Kuroshio, south of Kyushu, this biased sampling
led to the result that the ensemble mean drifter current was about 30 cm s
91 larger than the average
current derived from a combination of current
meters, hydrography and satellite altimeter sea-level
observations (Uchida et al., 1998). Apparently,
drifters converged towards the velocity axis of the
Kuroshio. This is another reason why the detailed
structures of each of the strong western boundary
current regions were not discussed; each boundary
current would require a different velocity data set
to assess the potential biases of drifter sampling.
Satellite altimeter data offer a technique for testing the sampling characteristics of the variance of
the velocity also in the open ocean, although they
cannot provide a test for the mean. This test was
carried out in the North Pacific in a comparison
of the variance of geostrophic currents normal to
the TOPEX/POSEIDON altimeter track. The
geostrophic component of the drifter velocity was
computed by subtracting the Ekman component
according to the formulae specified in Section
4.1.4 (Ralph and Niiler, 1999). The geostrophic
velocity normal to the satellite track from drifter
data was binned on a nominal 2°3° spatial scale.
Then bins were chosen with more than 33, 5-day
average observations and where more than five
drifters produced the data. If data from only one
or two drifters existed in a bin, the data could not
be representative of average conditions, as these
drifters could have resided in very special features
such as a single cold or warm eddy. This sampling
is equivalent to 164, 2-day average observations
on a 2°6° resolution, as displayed on Fig. 4.1.4.
Using the above criteria, the ensemble average
variance from drifters was within the standard
error of the track normal geostrophic velocity
variance computed from the TOPEX/POSEIDON
satellite altimeters from the period 1992–98 (Barth
et al., 2001). This comparison was important for
testing the scheme of averaging both drifter and
altimeter data and for the future considerations of
merging of two global circulation data sets in such
a way that biases between the two data sets would
be minimized.
4.1.3 The global mean velocity and
velocity variance
The principal objective of WOCE was to observe
the world ocean circulation. The 15-m depth ensemble average drifter velocity of the world shows
the major surface current systems (Fig. 4.1.5a, see
Plate 4.1.5, p. 300). Apparent on Fig. 4.1.5a are
the strong poleward-flowing western boundary
currents in each ocean and the strong zonal
currents in the tropics that flow in the direction of
the winds. In the Norwegian Sea, there is a strong
poleward current along the eastern boundary
(Poulain et al., 1996). The North Pacific Equatorial Countercurrent, flowing counter to the wind, is
evident in the 5–10°N latitude band, spanning the
entire North Pacific basin. Mariners could approximately locate most of these currents a century
ago from the ship-drift charts (Bowditch, 1966;
Richardson, 1989). The observations used in Fig.
4.1.5a were made with calibrated instruments,
providing a nearly 10-fold improvement in accuracy of individual observations over those of ship
SECTION 4 THE GLOBAL FLOW FIELD
198