Pro&.Oceonog.Vol. 22, pp. 171 - 204,
1989.
Printed in Oteat Britain. All rights n~erved.
0079 - 6611 / 89 $0.00 + .50
© 1989 Pe~mnon Pre~ pic
Large scale circulation of the North Pacific Ocean
DEAN ROm4MICH a n d TRACT McCALLtST~
Marine Life Research Group, Scripps Institution of Oceanography,
La Jolla, California 92093, U.S.A.
(Submitted: January 88
Final Version Accepted: December 88)
A b s t r a c t - A least squares inversionprocedure is used to estimate the large scale circulation and transport
of the subtropical and subpolar Ncxth Pacific Ocean from a modern data set of long hych-ographic transects.
Initially a deep surface of known motion is specified using information derived from abyssal property
distribetious, moored current meter observations, and basin scale topographic constraints. A geostruphic
solution is obtained which conserves mass while deviating as little as possible in a least squares sense fr~n the
initial field. The sensitivity of the solution is tested with regard to changes in the initial field and to the addition
of conservation constraints in layers.
It is found that about 10 Sv of abyssal water flows northward across 24 °N, princilxdly between the dateline
and 160°E, in the deepest part of the Northwest Pacific Basin. The flow turns westward acxoas 1520]3 and then
mostly northward again near the Izu-Ogasawara Ridge and the coast of Japan. It then feeds a strong deep amicyclonic recireulation bcmeath the cyclonic subpolar gyre in the Northwest Pacific Basin. The abyssal waters
near the western boundary region are found to have a strong component of flow that is upward and across
isupycnal surfaces. Here, the abyssal waters complete an impcxtant loop in the giol~l thermohaline circulation,
entering as bottom water from the South Pacific and returning southward in a less dense and shallower layer.
Deep flow into the Northeast Pacific Basin, and circulation within that basin, appear to be weak, maldng it
remote from the main pathway of deep water renewal.
The circulation of the subtropical and subpolar gyres dominates transpcxt in the upp~ layers. The
subtro~cal gyre appears to penetrate to about 1500-2000 m on both sides of the Izu-Ogasawara Ridge, which
blocks deeper flow betwee~ the Philippine Basin and the Northwest Pacific Basin. The Kuroshio is estimated
to carry about 32 Sv northward in the East China Sea. Farther east, as the thermocline slopes upward toward
the eastern bonndary, the eastward flow is evon shallower. In terms of eddy activity, three regimes are observed
at 24°N. Peak-to-trongh eddy fluctuations in geoslrophicaily balanced sea level diminish from about 40 cm in
the west to about 5 a n in the east. Overall, the western boundary of the ocean is about 25 om higher than the
eaatern boundary in the 24°N section.
Patterus of heat aad freshwater flux determined in the Ncxth Pacific are in accord with those from air-sea
heat flux estimates and hydrological data although the magnitudes axe in some cases different. There is large
heat loss in the western ocean amotmting to about 9.6 x 1014W and modest heat gain elsewhere. Heat Iranspcxt
across 24°N is estimated to be 7.5 x 10 I' W. The subpolar ocean has a large excess of precipitation and runoff
over evapccation, about 5.6 x 10s m 3 s"1noah of 35°N, while in the subtropics there is excess evaporation, about
2.7 x 105 m 3 s -1 between 24°N and 35°N.
CONTENTS
1.
2.
Inl~oduction
Data Selection
2.1.24 ° and 47°N
2.2. 35°N
2.3. 137°E
2A. 152°E
2.5. 175°W
2.6. 152° W
172
173
173
173
174
176
176
176
171
172
D. ROEMMICHand T. McCALLISTER
3.
4.
5.
6.
7.
8.
Featuresof the Large Scale Circulation: Specifyingthe Initial VelocityField
ComputationalTeelmiques
Results
5.1. Deep circulation and transport
5.2. Transportand circdafion in upper ocean and thea'moclinelayers
5.3. Fluxes of heat and freshwater
Discussion
Acknowledgments
References
176
181
187
187
194
197
200
203
203
1. INTRODUCTION
During mid- 1985, two transoceanic CTD/hydrographic sections were completed in the North
Pacific by R/V Thomas Thompson. The first was a crossing of the central subtropical gyre from
east to west, principally along latitude 24 ° 14" N. A total of 216 deep casts were made from late
March through early June. Then, on the return voyage in August and September, 115 deep stations
were occupied in the central subpolar gyre, principally along latitude 47°N. These two transects,
when combined with other recent transects along 35 ° N, 137 ° E, 152 ° E, 175 ° W, and 152 ° W,
constitute an extensive modern survey of the subtropical and subpolar gyres.
The present analysis is a synthesis of the modem data set, with the aim of presenting a
quantitative description of the large scale circulation and transport of the North Pacific Ocean.
A description of this type has not previously been attempted. The data set has been collected by
many investigators and has been analyzed in individual regional studies. Here the goal is to
consider the data set in its entirety. Estimation of circulation in the individual regions should be
improved by requiring a consistent circulation for the ocean basin as a whole.
Previous studies of basin-scale geostrophic transport include the analysis of the South Pacific
by ~
(1987) and that of WtrNsoI and GRANT(1982) in the North Atlantic. In each of these, the
data consist of collections of long intersecting hydrographic sections, mostly terminating at land
boundaries. RInD (1987) used the ocean bottom as an initial reference level for geostrophic
calculations. He made barotropic adjustments to the flow in order to balance the total mass field
and, in addition, so that the direction of flow would be consistent with inferences based on
property distributions. There was not sufficient information to specify the barotropic component
everywhere. WtrNscH and GRANT(1982) experimented with reference levels at 2000 decibars and
the ocean bottom, using an inverse method to balance mass in selected layers. They used two
reference surfaces because there was no justification for selecting a particular surface and the two
solutions yielded very different circulation patterns in the deep ocean.
The approach here will combine elements of each of these analyses and will diverge from them
in a couple of significant respects. The inverse method will be used to balance mass, but we will
assert that there is sufficient knowledge of the circulation, a priori, to guess a level of known
motion. The initial guess will be based on basin-scale tracer information, on flow constraints
imposed by ocean ridges, and on a substantial collection of direct measurements of velocity. The
inverse method provides a solution which satisfies a set of mass conservation constraints while
deviating as little as possible, in a least squares sense, from the initial field. In contrast to the study
by REn~ (1987) of the South Pacific, one of the models examined here will include mass
conservation in a set of density layers as well as the total mass, and the estimates of transport will
include the ageostrophic Ekman flux. An additional aspect of this analysis, as detailed in the
Largescalecirculationof the North Pacific
173
section on computational techniques, is a method of interpolation near the intersection of crossing
tracks which insures the internal consistency of the data set at the crossing points.
Certainly the data base is incomplete and is non-uniform in spatial coverage and
resolution. A high degree of accuracy cannot yet be achieved in estimating the large scale
circulation. Rather, we believe that the present analysis yields the best estimate of circulation
consistent with the data in hand, while identifyingareas of particular interest in the North Pacific
for further study and showing where additional measurements will be most effective. With the
prospect of an intensive program of hydrographic exploration and direct measurements of current
in the next decade, it is important to see what can be learned from the present data and what is
required from the new observations in order to resolve outstanding questions.
2. DATASELECTION
A list of the stations used in this analysis is given in Table 1 and station locations are shown
in Fig. 1. We began with the 240N and 470N sections, which were designed for transport studies
such as this one, and then added a number of other appropriate sections in order to resolve smaller
areas of the ocean basin. There are some problems with incorporating these additional sections,
and these should be kept in mind when interpreting the results. In particular, if station tracks do
not extend into shallow water at boundaries, or if stations are not closely spaced enough to resolve
flows near major topographic features, then any transport along these boundaries or features is
poorly estimated. In addition, if stations are far apart then eddy fluxes are not resolved. A brief
description of the data base follows.
2.1. 24° N and 47° N
The station tracks (Fig. 1) extend into shallow water at ocean boundaries and at mid-ocean
ridges. Along the 240 N section, first and last stations were near the 100 m isobath. Station
separations were 10 km or less at the eastern and western boundaries, including the Kuroshio
crossing in the East China Sea, and on each side of the Hawaii Ridge, the Izu-Ogasawara Ridge,
and the Ryukyu Is along 24°N and the Emperor Seamounts along 47°N. Elsewhere, stations were
spaced about 80 km apart over the abyssal plains and about 50 km apart over rougher topography.
All stations extended to the ocean bottom except in the Mariana Trench, using a Neil Brown
Instrument Systems Mark III CTD and a rosette water sampler fitted with 36 ten-liter Niskin
bottles. The 24"N section is described by Roma~ca and MCCALLISr~(1988) and the 470N
section by T,~Lrra,and JoYcE (1988). Potential temperature profiles from 240N and 47°N are
shown in Figs 2a and c. Bathymetryalong 24°Nis from precision depth recorder soundings which
were digitized and logged approximately 6 times per hour. In addition to the CTD/hydrographic
data, profiles of absolute velocity in the upper ocean were obtained by T. JoYcEusing an acoustic
doppler shear profiler plus the ship's navigation data.
2.2. 35°N
The INDOPAC Expedition (SCRIPPSII~m-trnoN OF ~ O 6 1 ~ a ' W t , 1978; KENYON,1983)
was the only deep zonal hydrographic transect of the North Pacific prior to 1985. Data consist
of 98 stations alternating between casts to 1000 m and casts to the ocean bottom. We have used
the water sample data in preference to data from the early model of STD which was used on this
174
D. ROEMMICHand T. McCALLISTER
60"N
40"
2~0o ~
120eE
140 °
160"
180 °
~60°
140('
[20°W
Fie. la. Station locations. Transects are named according to their nominal latitudes or longitudes.
Numbering refers to the 12 enclosed areas in which mass conservation constraints m'e imposed.
60"N
40 °
|2~PE
MIO"
160 °
180 °
160°
140°
120°W
140 °
120°W
Fio. lb. The 3000 misobeth.
60" N
40"
,~1, - ""
zoo
I'~'~'~f ~
120OE
:l
140 °
J
160 °
J. . . . . . . . . . . . .
180 °
...
160 °
D]~. lc. The 5000 m isobath.
Large scale eir~dation of the North Pacific
175
cruise. Because of the large separation between deep casts, about 200 km, possible flows near
the Emperor Seamounts and the ocean boundaries were poorly sampled. The first cast was well
offshore of California in water of depth 2900 m. Likewise, station 98 was well east of the Japanese
coast in water of depth 2300 m and the deepest water sample at that station was from only about
700 m. The Kuroshio follows the Japanese coast to about 350N, where it separates and flows
along a meandering eastward path. We searched NODC files and located a station made by
R / V Ryofu Maru (Table 1) about 80 km west of INDOPAC station 98, preceeding it by only 4 days
in time. This station was included in the analysis. Note that in the potential temperature section
(Fig. 2b), the isotherms slope upward very sharply west of station 98, indicating that most of the
Kuroshio was indeed flowing around the INDOPAC section. We suspect that even with the
supplementary station, the Kuroshio transport across 35 ° N is underestimated.
TABLI~ 1. Hydrographic sections used
Section
Vessel
No. of
Stations
Dates
Refm'onee c~ source
24* N
Thomas Thompson
206
3/85-6/85
ROEMMICH
and McCALLISTI~ (1988)
35* N
Thomas Washington
Thomas Washington
Ryofu Maru
73
25
1
3/76-4/76
7/77
4/76
SIO (1978)
SIO (1978)
NODC er. 1847 no. 1
47* N
Thomas Thompson
113
8/85-9/85
TXLLEYand JoYcE (1988)
137" E
Ryofu Maru
12
7/81-7/81
NODC or. 2545 no. 86-97
152° E
Thomas Washington
Komahashi
Koshu Maru
22
1
1
5/81-5/81
5/33
4/32
Nm.m~ et al. (1985)
NODC or. 3 no. 446
NODC or. 3 no. 124
175" W
Thomas Thompson
Thomas Thompson
23
24
7/82-8/82
10/83o11/83
WARRm~
and OWENS(1985)
JoYcE(1988)
152" W
Thomas Washington
97
5/84-6/84
MARTINet al. (1987)
2.3. 137 ° E
The Japan Meteorological Agency has occupied a section from the Japan coast to New Guinea
twice a year since 1967 (e.g. MAStrZAWAand NAO~SAKA,1975). Unfortunately there are relatively
few deep stations in most of these surveys. We elected to use data from summer 1981 because
the section from Japan to 24"N contained mostly deep stations, including those in the Kuroshio,
and because this cruise coincided with current meter observations reported by FUK~AWA and
~OrO
(1986) during a non-meander phase of the Kuroshio. Potential temperature is shown
in Fig. 2d. Stations are rather far apart, 50 to 110 km, and this may lead to substantial error in the
estimation of shear below 1000 m near the continental slope. Bathymetry in this and the
subsequent sections is from a global bathymetric data set (5' by 5' resolution) released by the
National Geodetic Data Center as ETOP05.
176
D. ROEMMICHand T. McCALLISTER
2.4. 1 5 2 ° E
An array of moored current meters was maintained along 152 ° E between latitudes 29 ° N and
41°N from July 1980 to June 1982 (Nm_~, Sca~,~rz and L ~ , 1985). CTD sections were occupied
on each yearly cruise. We have elected to use the CTD section from the mid-point, i.e. May 1981.
All stations extended to the ocean bottom. Some stations were located along longitude 150 ° E
rather than 152 ° E (Fig. 1). For the present analysis, all stations have been projected onto the
nominal track along 152 ° E. In order to join this section with the 24 ° N and 47 ° N sections, we
included the nearest stations from those two sections and also included two historical stations
between 28 ° N and 24 ° N in order to help bridge this gap (Fig. 1 and Table 1). Potential
temperature is shown in Fig. 2e.
2.5. 1 7 5 ° W
This section was occupied in two parts. The first part, from the Aleutian Islands to 45°N in
August, 1982, accompanied the recovery of an array of five current meter moorings which were
maintained for 14 months. These datawere described by WARRENand OWENS(1985). The second
part, in November, 1983 (JOVCE, 1987), extended from about 43°N to about 28°N and there were
four additional moorings along this line. All stations in both segments extended to the ocean
bottom, For our purposes there are two gaps in the section, the first between the two segments
and the second between 28°N and 24°N (see Fig. 1), where we included a station from the 24°N
section. We were unable to find any deep stations to fill these gaps in the NODC files. Estimates
of geostrophic shear across the gaps in the upper ocean should be useful, but because of the
topography, the deep shear may be poorly estimated. Potential temperature is shown in Fig. 2f,
2.6. 1 5 2 ° W
The MARATHON expedition along 152°W included closely spaced stations from Alaska to
Hawaii (MARTIN,TALLEYand D~SZOEKe, 1987). Alternate stations extended to 1500 m and to the
ocean bottom. Potential temperature from this section is shown in Fig. 2g. Three-year records
of current were obtained from current meter moorings along 152 ° W at 42 ° N and 28 ° N (HtJ and
NmmR, 1987). Moorings at 39°N and 35°N were maintained for two years.
3. FEATURES OF THE LARGE SCALE CIRCULATION: SPECIFYINGTHE INITIALVELOCITYFIELD
Each of the wind-driven gyres of the North Pacific has its signature in the potential
temperature sections (Figs 2a-g). The subtropical gyre is marked by a central thermocline
depression, or bowl, while a doming of the thermocline marks the subpolar gyre. The steep sides
of these features in the west locate the western boundary currents: the poleward flowing Kuroshio
in the subtropical gyre and the equatorward Oyashio in the subpolar gyre. In the 24 ° N section,
Fig. 2a, the strong downward slope of isotherms from the western boundary in the East China Sea
marks the northward flowing Kuroshio over the Okinawa Trough. Across the ocean interior the
isotherms flatten and then slope upward to the east, signalling the broad southward return flow.
The California Current is visible as an accentuation of this upward slope in the easternmost few
hundred kilometers. The isotherm slopes are much weaker or nonexistent below 1000 m. REID
and ARTmm (1975) found that there was evidence of the subtropical gyre in geopotential
anomalies down to about 2500 m, though at this level the gyre was split in two by the Izu-
6000
I~0
141°E
. . . . . . . . .
150 °
I . . . . . . . . .
_....----.~/~.--
......
1713°E
I . . . . . . . . .
--
IOO e
I . . . . . . . . .
r'foow
I . . . . . . . . .
- ......... _ ....
~
r,ro,
I . . . . . . . . .
160 °
leo"
17o"
tar,
- ..........................
I . . . . . . . . .
150 °
I . . . . . . . . .
140 °
t .
130 °
.
.
.
|9.9.ow
.
--"I
-.
-'_
.
.
.
.
6000
.
0m
.
II
~
,
-
-
;
t
~.
160 °
* .........
~--',,i-,
1700E
I .........
180°
170~N
, .........
..........
t .........
"---'--~-1.2,
~-~.~'~2:"
....
i .........
160°
150 °
-. . . . . . . .
~
1,40°
i .........
.---%.
Fio. 2c. Potential teanpea'atme in the 47°N transect.
. . . . . . . . .
1460EI50 °
.
'c. . . . . . . . . . . . . . .
~..~'~ .........
~
i .........
. . . . .
-- . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Fio. 2a. Potential temperature in the 24°N transect.
1.2 . . . . . . . . . . . . . . . . . . . . ...--.-.. . . . . . . . . .
me-
Fio. 2b. Potential tcmpcratm'e in the 35°N transect.
160 °
I . . . . . . . . .
~
......
me"
~
~5
125°W
, .....
;
liX)O
~
.,d
Z
ft
;
178
D. ROEMMICHand T. McCALLISTER
Om~
~
~
Om-
'
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..................
,.. .....................
:
4000
_---
4000 ~ -
.
,ooo_
,ooo_
I
I
33"N
I
I
[
I
I
I
30 °
t
I
25°N
FIo. 2d. Potential temperature in the 137°E Iransect,
..... ,, .
~--
.
.
.
.
I
]
I-J
I
I
I
I
I
I
420N 4 0 °
.
.
.
.
.
.
.
3
,- ......
.
.
.
.
.
.
.
t
]
i
I
I
30 °
.
.
.
.
.
.
.
.
.................................
, .
.
.
.
.
---f-----
.
.
.
.
.
.
........
.
or. .........
n
~
_.
----.-i
_
~
----
6000 ~
_
...
..~
,
I
,
,
i
,
,
,
,
,
,
1
40*
50*N
30*
25°N
Fto. 2£ Potential temperature in the 175°W transect,
o~
:-
.................................................................
2ooo
i i,'ll. . . . . . . . . . . . . . . . . . . . . . . .
:!,Z::i:::
4000
6000
i
L
I
250N
Fig. 2e. Potential temperature in the 152°E transect.
2000
4000
,
;
i i I I I l l l l l l l l l l l [ l l l l l [
56°N
50 °
40*
I l l l l l l l l l l l .
300
220N
Flo, 2g. Potential temperature in the 152°W transect.
Large scale circulationof the North Pacific
179
Ogasawara Ridge.
In the 47°N section across the subpolar gyre, Fig. 2c, the isotherms generally slope upward
from the western boundary to about 1580E and then slope downward farther to the east. In the
eastern half of the ocean, the slopes and the associated geostrophic shear become quite small
below about 2500 m. In the west, the slopes persist to the ocean bottom. The dome and bowl
structure of the gyres is also seen in the meridional sections, particularly the interior sections along
175"W and 152°W, Figs 2 f a n d g.
After leaving the East China Sea, the Kuroshio flows alongshore over the continental slope
south of Japan except during periods when a large meander forms between 136°E and the IzuOgasawara Ridge at about 140°E. The July, 1981 section at 137°E, Fig. 2d, was made during
a non-meander period (FuKASAWAand ~ c r r o ,
1986), with the Kuroshio near shore at about
33°N. In our study region, the Izu-Ogasawara Ridge forms a solid barrier below a depth of about
2000 m, confining the flow to the two sides of the ridge. In the upper ocean, after crossing the
ridge, the Kuroshio separates from the coast and flows eastward, meandering and eddying along
a path close to 35 ° N. The Kuroshio Extension can still be clearly seen in the strong isotherm
slopes along 152°E, Fig. 2e, centered at about 35030 ' N. The current becomes more diffuse farther
to the east, with some southward recirculation, and in the 175 ° W and 152 ° W sections the
remaining eastward shear is spread over a wide latitude band.
In the 24°N section, between about 1000 m and 2500 m, the shear in meridional geostrophic
velocity integrated across the ocean is very small. But below 2500 m the shearis again substantial,
a signal of the meridional flow of abyssal waters linking the deep North Pacific to the global
thermohaline circulation. It is known that the saline water of the abyssal Pacific can be traced
back to its North Atlantic origin (REIDand LYNN, 1971) and that there is no source of deep water
in the North Pacific (e.g. WARREN, 1981). WtrNscn, Ho and GRANT(1983) determined from the
SCORPIO sections along 28 ° S and 43 ° S that the abyssal waters of the South Pacific flow
northward, with balancing southward flow in overlying layers. BRo~cr~, T ~ m
and
T ~ m
(1985) obtained a similar vertical structure to the net meridional flow of deep water
masses on the basis of property distributions. Abyssal property distributions plotted by MAma'tA
and RInD (1983) indicate northward spreading of bottom waters in both the eastern and western
basins of the North Pacific. At 24°N, geostrophic transport relative to the ocean bottom produces
an unrealistic estimate of this meridional recirculation, with large southward transport of abyssal
waters for which there is no northern source. In contrast, a reference level at mid-depth, where
the shear is very weak, produces a plausible pattern of northward flow in the deepest water of both
basins, with southward flow overlying. We have chosen a mid-depth isopycnal, sigrna-2 equals
36.9, as a reference surface. This isopycnal is located at about 2000 m. The choice is arbitrary,
but because the waters at mid-depth are weakly sheared, moderate variations in the level produce
small changes in estimated deep transport.
The northward spreading of bottom water also appears at 35°N in the maps of ~
and
REID (1983). However, because of the small number of deep stations in the 35 ° N section, and
particularly because of the lack of resolution near topographic features and boundaries, the deep
shear is not well sampled by this section. Indeed, the deep shear as measured by these data is very
small. We believe the data are useful for estimating upper ocean transport away from the eastern
and western boundaries, but not for deep transport or boundary transport. At 35°N, we will use
the same reference surface, sigma-2 equals 36.9, as at 24 ° N.
The 470N section is close to the northern boundary of the Pacific and there is likely to be little
net meridional transport at this latitude in any layer. In the east, deep shear is weak and the
absolute meridional velocity is likely to be weak. The abyssal property distributions (MArcrrLA
180
D. ROEMMICHand T. MCCALLISTER
and REID, 1983) indicate northward spreading along the western boundary at this latitude. We will
introduce this tendency into the initial field by imposing a northward velocity of 0.5 cm s "1 at
4000 m from the western boundary to 149 ° E. Weaker tracer signals suggest southward
recirculation in the area of the Emperor seamounts (e.g. in silicate) and weak northward flow in
the interior of the northeastern basin. In the initial field, we will specify a southward flow of
0.2 cm s -1 from 160 ° E to 175 ° E and a northward flow of the same magnitude from 175 ° W to
160°W. These magnitudes are quite small and were chosen to be comparable to the horizontally
smoothed shear between about 2000 and 4000 m. In any event, we will see that the final adjusted
velocities will be significantly larger. The initial field of velocity is set to zero at 4000 m in
longitude bands other than those specified above.
As was mentioned in the discussion of data selection, all of the meridional sections used in
this analysis were accompanied by long-term moored current meter observations covering
segments of the sections. The current meter data will provide a basis for referring the geostrophic
shear calculations in these sections. In the presence of energetic fluctuating currents, some of the
current meter records are not long enough to produce reliable estimates of the mean. Separations
between adjacent moorings may be, in some cases, greater than the horizontal scale of mean
currents, particularly in areas with rough topography or near ocean boundaries. Further, the
hydrographic data set is not a time mean, but is rather a group of snapshots. In spite of these
shortcomings, there is valuable information in the measured currents with respect to referring the
geostrophic calculations. The mean currents often do appear to have scales which exceed the
separation between moorings. In spite of the variability, the mean shear in the current meter
records is often similar to the geostrophic shear from individual hydrographic sections. We will
try to interpret the mooring records in terms of the information contained in the array as a whole,
establishing the magnitude and direction of deep flows on large scales or extracting from the
arrays a surface where a component of the velocity is small.
Perhaps the most unexpected information comes from the deep current meters along 152° E.
NmmR, S ~
and LEE (1985) found westward flow at 4000 m in all six of the current meters
which returned more than 22 months of data between 41°N and 30°N. At 28°N, the zonal flow
was negligible. This finding is reinforced by that of Sci-~rrz (1987), who found strong stable
westward flow at 4000 m in the same latitude range along 165°E. We will approximate the results
along 152 ° E (see Nm.RR, Scrmirrz and Lr~. 1985, Fig. 1) by setting a westward velocity of
3 cm s-1 at 35 ° N, decreasing linearly to zero at 42 ° N and at 28 ° N. Between 28 ° N and 24 ° N,
where there were no moorings, we will use zero at 4000 m. This prescribed velocity at 4000 m
will then be used to refer the subsequent geostrophic calculations.
At 136 ° E, year-long records from three moorings below the Kuroshio in 1981-82 were
analyzed by FUKASAWAand Tm~AMOTO(1986). These moorings have subsequently been mainrained through the present time (Fui~snwn, Tm~aOTO and Tnmn, 1986). At depths just below
2000 m, during the non-meander period coinciding with the summer 1981 hydrographic section
along 137 ° E, mean flow was westward directly below the Kuroshio axis and eastward just
offshore. The mooring separation was small compared to the separation of the 137 ° E hydrographic stations (Fig. 2d), and the main point here is simply that the flow was relatively weak at
2000 m. There was no evidence of strong penetration of the eastward flowing Kuroshio. Rather,
the pattern is one of bottom intensification of westward flow, with a vertical penetration of the
bottom currents of order 2000 m (FuKASAWA,T ~ O T O and TAmA, 1986). We will use, as the
initial estimate, zero velocity on sigma-2 equals 36.9. In the subsequent calculations, lxansport
across the Izu-Ogasawara Ridge will be required to be zero for layers deeper than 2000 m because
the ridge is nearly continuous below that level in the latitude range of our study (Fig. 1b). In terms
Large scale circulation of the North Pacific
181
of net layer transports, the results would be much the same if any deep reference surface were used
in conjunction with these constraints of no deep flow through the ridge. It is only the pattern of
deep flow west of the ridge which is sensitive to this choice.
Two moored arrays along 175 ° W were accompanied by CTD surveys covering different
latitude bands as shown in Fig. 1 and Table 1: The northern array, discussed by W ~
and
~ s
(1985), consisted of five moorings between 51 ° N and 46 ° N maintained for 14 months.
It showed westward flow at 3000 m close to the northern boundary at 51°N with eastward flow
at two mooring sites farther offshore and then weak flow beyond. Referring to Fig. 2f, the
potential temperature profiles indicate that the geostrophic shear at 3000 m is westward with
decreasing depth close to the boundary, with a dome in the isotherms at about 50"30'N; then the
shear is eastward with decreasing depth farther offshore; then it weakens beyond. The mooring
data seem consistent with the geostrophic shear of the CTD surveys and with historical dynamic
topography (REID and Aarrnm, 1975) and suggest strongly that a deep reference surface is more
appropriate in this region than one at mid-depth or shallower. We will use a reference velocity
of 1.5 cms-' westward at 3000 m from the boundary to 510N, 1.5 cms -1 eastward at 500 3 0 ' N
and 49025 ' N, interpolating linearly between these values, and then tapering to zero at 48°N and
south to 42°N. Mooring data in the southern segment show mean eastward flow from 39 ° N to
31 ° N with a maximum of 1.5 cm s-' at 35 ° N. This eastward flow is poorly resolved by the army,
but is consistent with the eastward spreading of abyssal properties from the Northwest Pacific
Basin to the Northeast Pacific Basin via the large gap between the Emperor Seamounts and the
Hawaii Ridge. We use the measured mean currents at 4000 m to specify the initial field of
velocity, interpolating between mooring locations (Table 2). South of 29 ° N, the initial velocity
is set to zero at 4000 m.
Along 1520W, moorings were maintained for 3 years at 28* N and 410N and for 2 years at
35 ° N and 39°N. Following Hu and N m ~ (1987) and Niiler (personal communication), it appears
from viewing the current meter data together with the hydrography that the zonal velocity at 28 *N
is westward below about 140 m and eastward above that level, while to the north this reference
surface slopes downward, to approximately 300 m at 350 N, 500 m at 39°N, and 600 m at 410N.
The sloping reference level is congruent with the notion that the center of the subtropical gyre
tends poleward with increasing depth (e.g. REID and ARrrItra, 1975). I f a deep surface is chosen
as a reference for geostrophic computation, the structure of the large scale shear field is such that
the surface of zero velocity corresponding to the gyre center appears a few hundred meters deeper
than the level indicated by the current meters. Equivalently, the reference level based on the upper
level current meter data will, through the thermal wind balance, produce a net westward flow in
the abyssal waters. In order to reproduce this result in our initial field of velocity, we specify a
westward velocity of 0.2 cm s" from 42 ° N to 24 ° N. South of 28 ° N, the upper level surface of
no zonal motion continues to slope upward and the transition from eastward to westward surface
flow occurs near 240N (Rmn, 1961). North of 41°N, there is no current meter data along the
152°W line. Property distributions (REID and MArrrrLA, 1983) indicate eastward spreading of
abyssal waters along the northern boundary, as is the case farther west. We will specify an
eastward flow of 0.5 cm s-' from the northern boundary to 54 ° N, consistent with the measured
net eastward flow farther to the west.
4. COMPUTATIONAL TECHNIQUES
In preference to traditional station-pair calculation of geostrophic shear, the method of
objective mapping was used to produce evenly spaced grids of the measured quantifies, with the
182
D. ROEMMICHand T. McCALLISTER
shear in geostrophic velocity being derived from gridded values of specific volume anomaly. The
application of objective mapping to hydrographic sections was described by ROF_~iMICrI(1983).
Traditional station-pair analysis yields poor results in regions where there is a large change in
bottom depth between stations. In that case, traditional methods either ignore the water below
the deepest common depth of a pair of stations or extrapolate in order to estimate the vertical shear
below this level. The former procedure produces a biased (low) estimate of transport since it is
equivalent to assuming no flow below the deepest common depth. Vertical extrapolation can at
best take into account only the vertical covariance of the data while ignoring the horizontal
covariance. The objective mapping technique utilizes both the vertical and the horizontal
covariance in a consistent manner. As long as the station spacing is small compared to the scale
of features being mapped, this method will produce significantly better estimates than any vertical
extrapolation or other ad hoc procedure. In most of the modern surveys, such as those at 24 ° N
and 470 N, a design criterion was to minimize the station spacing in order to obtain coherent
sampling. In producing the gridded fields, very low data noise was specified in order to avoid
artificial smoothing of the measurements.
A second advantage of the objective mapping technique is realized in the analysis of nonsimultaneous crossing sections. WtJNSCH and GRANT0982) noted that a transient eddy in one
member of a crossing pair of sections could result in a discrepancy of order 30 Sv* in mass
conservation near the crossing point. This led them to use a subset of the entire North Atlantic
dataset, in order to reduce the apparent inconsistency of the complete dataset, and to accept the
remaining discrepancies as a form of model error. Certainly it is true that individual sections are
an inadequate representation of a smooth mean field. But when there are multiple sections near
a given point, then the estimate of a mean field should be enhanced rather than degraded in the
vicinity of this point, precisely because there is more than one realization. In other words, crossing
sections should constitute an asset rather than an inconsistency if the analysis technique is capable
of exploiting it. This is readily done with objective interpolation. Near the crossing point, stations
from different cruises are incorporated in the estimation process. Any transient feature in either
section will appear in both sections, with its magnitude reduced by averaging. The ideal case
would include many sections passing near a given point in order to estimate the mean field with
known small errors. The problem of combining data from different cruises and years remains a
serious one. The dataset may still contain inconsistencies if, for example, two snapshots of a
western boundary current at different locations sample different states of the current. We have,
however, eliminated a significant form of inconsistency in the treatment of data from crossing
sections.
In each hydrographic transect, the initial field of velocity was specified at some level, as
explained in the previous discussion and in Table 2. Starting from the specified level, the full
geostrophic velocity field was then obtained by integrating the gridded values of geostrophic
shear to the ocean surface and to the bottom. Ekman transport was computed, and was added to
the surface layer transport, using tabulations of wind stress given by HELLE~aANand ROSm~SrF_~
(1983).
In order to consider the transport in different water masses, the ocean was divided into a series
of 5 layers separated by isopycnal surfaces, as defined in Table 3. For each layer, the mass
transport was obtained at each grid point by integration of the velocity from the bottom to the top
of the layer. Surface-to-bottom mass transport and heat and salt transports were similarly
calculated. Note that while we will refer to layers of finite thickness, the integrations involving
velocity, density, salinity, and temperature within the layers retain the vertical structure of the
measurements. Thus, the model is of multiple layers having continuous stratification, not
* lSv = 106m~s".
Large scale circulation of the North Pacific
183
TABLE 2. The initial field of velocity for models la and If. The inilial field is linearly interpolated
between the points where it is specified. N.Bnd., W.Bnd., and E.Bnd. refe~ to the northean, western,
and eastern boundaries of the sections
24°N
velocity zero on sigma-2 equals 36.9
35°N
velocity zero on sigma-2 equals 36.9
W.Bnd.
149°E
150°E
1550E
160°E
0.0
0.0
-0.2
-0.20.0
470N
4000 m
0.5
180 °
1750W
160°W
155°W
E.Bnd.
470N
4000 m
0.0
0.2
0.2
0.0
0.0
240N
1370E
1750E
velocity zero on sigma-2 equals 36.9
152"E
4000 m
175"W 3000 m
175"W 4000 m
152"W 4000 m
420N
350N
280N
0.0
-3.0
0.0
0.0
N.Bnd.
50"30'N
49*30' N
480N
42"N
-1.5
1.5
1.5
0.0
0.0
41*N
39"N
35"N
31*N
29"N
24"N
0.0
-0.5
0.2
1.5
0.3
0.0
N.Bnd.
54"N
53"N
43"N
42"N
24"N
0.5
0.5
0.0
0.0
-0.2
-0.2
TABLE3. Density layers. The set of 5 layers on the left is the set referred to in the transpoR summaries
of Figme 3 and Table 4. The top two of these layers were used for the added mass conservation
constraints of model II. The set of 14 layers on the right is used for the finer transport census of Figure
4. As an aid in the interpretation of transport maps, typical values of potential tempemuae on the layer
interfaces are: 6.5"C on sigma-them = 26.8, 4.0°C on sigma-them = 27.3, 1.5 oC on sigma-2 = 36.96,
and 1.06"C on sigma-4 = 45.885
No.
Top Surface
Bottom Surface
Top Surface
Bottom Surface
ocean surface
sigma-theta = 26.8
ocean surface
sigma-theta =24.3
2
sigma-them = 26.8
sigma-them = 27.3
sigma-them = 24.3
sJtnna-theta = 26.0
3
sigma-them = 27.3
sigma-2 = 36.96
sigma-them = 26.0
sigma-them = 26.8
sigma-them = 27.3
1
4
sigma-2 = 36.96
sigma-4 = 45.885
sigma-theta = 26.8
5
sigma-4 = 45.885
ocean bottom
signm-theta = 27.3
sigma-2 = 36.69
6
sigma-2 = 36.69
sigma-2 = 36.8
7
stgeaa-2 = 36.8
sJtnna-2 = 36.9
8
stgma-2 = 36.9
sigma-2 = 36.96
9
~t, ma-2 = 36.96
s~c,ma--4 = 45.82
10
sigma-4 = 45.82
sigma-4 = 45.85
11
s~tnna-4 = 45.85
sJ£ma-4 = 45.87
12
sigma-4 = 45.87
ssgma-4 = 45.885
13
stgma-4 = 45.885
sjmna-4 = 45.90
14
sjt,ma-4 = 45.90
ocean bottom
184
D. ROBMMICHand T. McCALLISTER
multiple uniform layers.
The data set displayed in Fig. 1 divides the North Pacific into 12 areas, which are numbered
in that figure. We wish to constrain the flow in order to have no net mass transport in each area,
and thus each of the 12 areas provides a single mass conservation constraint. Formally the
constraints are written as in WUNSCH(1978). The transport due to solution velocities is required
to correct the net transport imbalances of the initial field. The initial field includes the sum of
geostrophic and Ekman components. A 13th constraint is provided by the enclosure of the North
Pacific basin, and accordingly, the mass transport across 24 ° N is required to be approximately
zero. Additional constraints result from basin topography. In the segment of the 24°N section
crossing the East China Sea, transport must be small below sigma-them equals 27.3. Although
this section extends to nearly 2000 m in the Okinawa Trough (Fig. 2a), the controlling sill for
outflow near Tokara Strait has a depth of only about 600 m. Farther to the east, the Izu-Ogasawara
Ridge is a nearly continuous barrier below 2000 m from the coast of Japan south to 24 ° N
(Fig. lb). Water in layers 1-3 can traverse the ridge, but deeper layers are blocked. There is no
water denser than layer 4 on the west side of the ridge. Transport in layer 4 across 240N is required
to be small between the ridge and the western boundary. Similarly, transport in layer 4 across
1370 E together with the segment of 24 ° N between 137 ° E and the ridge is also small. With the
addition of topographic constraints, the total number of constraints in this first model is 16.
The solution was obtained by a least squares inversion process as in Wtmscn (1978) and
ROEMMICn (1981), and the reader is referred to the earlier work for details. The solution consists
of a correction velocity to be applied uniformly in depth at each grid point (at each station pair
in the previous work). It is the smallest correction to the initial field, in aleast squares sense, which
satisfies the given constraints. The problem is underdetermined and thus the solution is dependent
on the specification of the initial field of velocity. The intent here was to put as much information
as possible into the initial field. The solution is a small adjustment of this field in order to satisfy
the requirements of mass conservation and of no flow through solid barriers.
As in traditional water mass and geostrophic analysis, the quality of the solution depends on
the information in the initial field. A good initial guess is required for a good solution. There are
several advantages of the inversion procedure over traditional techniques. First, as we will
demonstrate below, one may formally examine the dependence of solutions on initial assumptions and on the nature and number of constraints. Second, the least squares method of solution
is explicit and optimal. Third, the subjective analyst can at best trade o f f a handful of constraints
in seeking consistent solutions. The least squares solutions are not similarly limited, and in basin
scale models with an increasing data base the limitation is a serious one. In the present problem,
we have a certain amount of information on which to base an estimate of the initial field. It is not
sufficient for stable statistical purposes. In order for our solution to be useful and to exploit the
advantages of the inversion formalism, we must set down the assumptions of the initial field and
demonstrate the sensitivity of the solution to these assumptions. We have tried to do the former
in the previous section and the latter in the section which follows. As more data are accumulated,
the sensitivity calculation can be replaced by a statistical error analysis.
We have used the Singular Value Decomposition form of solution (WImscn, 1978), with rank
K equals 13. The rank is the number of eigenvalues retained from decomposition of the matrix
derived from layer thicknesses and may be thought of as the number of linearly independent
constraints in the problem. Choosing the rank 16 solution would satisfy all constraints perfectly,
an unrealistic choice for a problem with noise in both the data and the geostrophic model. The
rank 13 solution has correction velocities which are smooth in space and fairly s m a l l - - typically
of order 1 cm s-1 near the boundaries and much less in the interior. The problem is weighted so
Large scale ci~datlon
TABLE 4. T r a u s p ~ ,
of the North Pacific
185
i n S v e r d r u p s , f r o m t h e i n i t i a l f i e l d ( m o d e l 0), a n d f r o m t h e s o l u t i o n s o f m o d e l s
Ia, Ib, a n d a . T r a u s I X ~ i s g i v e n f o r e a c h s e g m e n t i n a g i v e a area. P o s i t i v e t r a n s p o r t i n d i c a t e s n e t f l o w
i n t o t h e area; n e g a t i v e i n d i c a t e s n e t f l o w o u t o f t h e a r e ~ T h u s , f o r e x a m p l e , i n a r e a 1 t h e s e g m ~ t o f
t h e 2 4 ° N t r a n s e c t b e t w e e n O k i n a w a a n d 1 3 7 ° E h a s 3.3 S v e n t e r i n g t h e a r e a ( n o r t h w a r d t r a n s p o r t ) i n
l a y ~ I in m o d e l Ia
Modd 0
Lay~
Mode,l IA
1
2
3
4
5
Total
1
2
3
4
5
Total
24°N: Okinawa-137°E
24°N: East China Sea
1370E: Japan-24°N
Area 1 Total
1.8
29.2
-33.3
-2.3
0.4
2.9
-9.2
-5.9
-0.4
-0.9
-5.2
-6.4
-1.4
0.0
2.4
1.0
0.0
0.0
0.0
0.0
0.4
31.2
-45.2
-13.6
3.3
28.9
-33.3
-1.0
1.1
2.8
-9.1
-5.2
3.4
-1.0
-5.2
-2.7
5.3
0.0
2.0
7.2
0.0
0.0
0.0
0.0
13.1
30.7
-45.5
-1.7
24*N: Izu Ridge- 152"E
24"N: 137"E-IzuRidge
35°N: Jalmm-152"E
152°E: 35°N-24"N
137°E: Japan-24.N
Area 2 Total
-13.7
9.0
-18.5
20.9
33.3
31.0
-2.3
2.1
-3.1
10.5
9.2
16.3
-1.5
1.7
-1.7
28.2
5.2
32.0
3.1
-4.6
1.1
26.8
-2.4
24.0
3.4
0.0
0.8
11.1
0.0
15.3
-11.1
8.2
-21.4
97.6
45.2
118.5
-19.1
9.3
-21.1
13.0
33.3
15.4
-4.8
2.2
-4.9
6.5
9.1
8.1
-13.9
3.0
-8.7
12.1
5.2
-2.3
-11.3
-1.9
-8.3
3.3
-2.0
-20.0
-0.8
0.0
-3.0
1.6
0.0
-2.3
-49.9
12.6
-45.9
36.5
45.5
-1.2
24°N: 152°B-175°W
35°N: 152"B-175°W
175°W: 35°N-24°N
152°E: 35°N-24°N
Area 3 Total
-8.1
10.8
-7.8
-20.9
-25.9
-0.6
1.8
-2.0
-10.5
-11.3
0.9
1.3
-9.9
-28.2
-35.9
-0.7
-1.6
-11.3
-26.8
-40.4
4.7
-0.1
-0.0
-ILl
-6.5
-3.8
12.2
-31.1
-97.6
-120.2
-4.6
11.9
-5.8
-13.0
-11.4
1.2
2.5
-0.8
-6.5
-3.6
9.4
4.1
-4.8
-12.1
-3.5
9.4
1.8
-3.8
-3.3
4.0
10.0
0.5
0.6
-1.6
9.5
25.4
20.9
-14.7
-36.5
-5.0
24"N: Hawaii Ridge-152oW
24°N: 175*W-HawaiiRidge
35°N: 175"W-152"W
152°W: 35°N-24"N
175°W: 35°N-24°N
Area 4 Total
-8.4
-5.0
3.9
-1.2
7.8
-2.9
-3.2
0.2
1.0
2.7
2.0
2.8
-1.5
-0.1
0.6
6.7
9.9
15.7
2.5
-1.9
-0.3
6.2
11.3
17.9
0.0
-0.3
-0.3
0.7
0.0
0.2
-10.6
-7.0
4.9
15.2
31.1
33.7
-9.2
-5.6
3.8
-2.2
5.8
-7.4
-3.6
-0.1
0.9
2.1
0.8
0.1
-3.6
-1.6
0.4
4.0
4.8
4.0
-0.2
-3.8
-0.7
1.9
3.8
1.0
0.0
-0.4
-0.3
0.3
-0.6
-1.0
-16.5
-11.5
4.0
6.1
14.7
-3.2
240N: 152°W-Calif.
35°N: 1520W-Calif.
152"W: 35oN-24ON
Area 5 Total
-14.9
9.0
1.2
-4.7
0.4
1.0
-2.7
-1.4
1.5
-0.7
-6.7
-5.9
-7.1
-0.1
-6.2
-13.4
0.6
0.0
-0.8
-0.1
-19.5
9.2
-15.2
-25.5
-14.4
9.6
2.2
-2.5
0.8
1.5
-2.1
0.2
3.4
1.5
-4.0
0.9
-5.0
3.1
-1.9
-3.8
0.8
0.0
-0.3
0.5
-14.4
15.7
-6.1
-4.8
47°N: Japan-152°E
35°N: Japan-152°E
152°E: 42"N-35°N
Area 6 Total
3.1
18.5
-34.7
-13.1
3.7
3.1
-7.7
-1.0
-0.2
1.7
7.8
9.3
-6.4
-1.1
26.1
18.7
-0.1
..0.8
10.6
9.7
0.2
21.4
2.1
23.6
1.3
21.1
-35.9
-13.5
0.7
4.9
-9.2
-3.6
-10.5
8.7
2.0
0.2
-20.1
8.3
18.2
6.4
-0.7
3.0
8.2
10.5
-29.2
45.9
-16.7
0.0
47°N: 152*E-175"W
35°N: 152°E-175°W
175°W: 47°N-35°N
152"E: 42°N-35oN
Area 7 Total
2.4
-10.8
-19.9
34.7
6.3
6.4
-1.8
-8.4
7.7
3.9
27.3
-1.3
-5.7
-7.8
12.6
52.2
1.6
-8.3
-26.1
19.4
0.0
0.1
0.0
-10.6
-10.5
88.3
-12.2
-42.2
-2.1
31.7
0.5
-11.9
-18.8
35.9
5.7
1.4
-2.5
-7.2
9.2
0.9
8.0
-4.1
-1.1
-2.0
0.9
23.0
-1.8
-0.9
-18.2
2.1
0.0
-0.5
0.0
-8.2
-8.8
32.9
-20.9
-28.0
16.7
0.8
47°N: 175°W-152"W
35°N: 175°W-152"W
152°W: 47°N-35"N
175°W: 47°N-35"N
Area 8 Total
-3.4
-3.8
-10.3
19.9
2.4
-4.4
-1.0
-3.6
8.4
-0.6
-7.8
-0.6
3.3
5.7
0.6
-7.1
0.3
3.8
8.3
5.1
0.0
0.3
0.0
0.0
0.3
-22.7
-4.9
-6.8
42.2
7.8
-2.9
-3.8
-10.7
-18.8
1.4
-3.4
-0.9
-4.1
7.2
-1.2
-3.8
-0.4
1.6
1.1
-1.5
-2.0
0.7
1.0
0.9
0.6
0.0
0.3
0.0
0.0
0.3
-12.2
-4.0
-12.2
28.0
-0.4
47°N: 152°W-Calif.
35°N: 152°W-Calif.
152°W: 47°N-35°N
Area 9 Total
-2.7
-9.0
10.3
-1.4
-2.8
-1.0
3.6
-0.2
-2.2
0.7
-3.3
-4.9
-0.4
0.1
-3.8
-4.1
0.0
0.0
0.0
0.0
-8.0
-9.2
6.8
-10.5
-1.8
-9.6
10.7
-0.7
-1.2
-1.5
4.1
1.3
2.9
-1.5
-1.6
-0.1
3.2
-3.1
-1.0
-0.9
0.0
0.0
0.0
0.0
3.2
-15.7
12.2
-0.3
47"N: Japan- 175"W
175°W: Alaska-47°N
Area 10 Total
-5.5
0.4
-5.1
-10.0
0.5
-9.6
-27.1
-6.4
-33.6
-45.9
-6.4
-52.3
0.1
-1.0
-1.0
-88.4
-13.0
-101.4
-1.8
1.0
-0.8
-2.2
2.3
0.2
2.4
0.6
3.0
-2.8
2.7
-0.1
0.7
-0.4
0.4
-3.7
6.3
2.6
47°N: 175°W-152°W
152"W: Alaska-47°N
1750W: AI~,,,k--47"N
Area 11 Total
3.4
2.9
-0.4
5.9
4.4
1.4
-0.5
5.3
7.8
-2.7
6.4
11.5
7.1
-3.2
6.4
10.3
0.0
0.0
1.0
1.0
22.7
-1.7
13.0
34.1
2.9
2.8
-1.0
4.7
3.4
1.0
-2.3
2.1
3.8
-3.8
-0.6
-0.5
2.0
-4.3
-2.7
-5.0
0.0
0.0
0.4
0.4
12.2
-4.3
-6.3
1.6
47°N: 1520W-Waah.
152"W: A/aska-47°N
Area 12 Total
2.7
-2.9
-0.2
2.8
-1.4
1.4
2.2
2.7
4.9
0.4
3.2
3.6
0.0
0.0
0.0
8.0
1.7
9.7
1.8
-2.8
-1.0
1.2
-1.0
0.2
-2.9
3.8
0.9
-3.2
4.3
1.1
0.0
0.0
0.0
-3.2
4.3
1.2
186
D. ROEMMICHand T. McCALLISTER
TABLE 4 (contd):
M o d e l IB
Layer
M o d e l I1
1
2
3
4
5
Total
1
2
3
4
5
Total
24"N: Okinawa- 137"E
24°N: East China Sea
137°E: Japan=24°N
Area 1 Total
3.6
29.0
-33.4
-0.8
1.2
2.9
-9.2
-5.1
3.8
-0.9
-5.4
-2.6
5.1
0.0
1.8
6.9
0.0
0.0
0.0
0.0
13,6
31.0
--46.2
- 1.6
2. I
30.6
-32.5
0.3
0,9
5,0
-7.5
-1.6
2.1
2,2
-4.2
0.1
1.1
0.0
-0.4
0.6
0.0
0.0
0.0
0.0
6.1
37.8
-44.5
-0.6
24"N: lzu Ridge-152"E
24"N: 137°E-I~a Ridge
35"N: lapan-152*E
152"E: 35"No24°N
137"E: Japan-24°N
Area 2 Total
-17.2
9.6
-21.4
11.5
33.4
15.9
-3.9
2.3
-5.2
5.6
9.2
8.1
-9.4
3.3
-9.6
9.1
5.4
-1.1
-6.0
-1.9
-9.5
- 1.1
-1.8
-20.3
0.7
0-0
-3.5
-0.9
0.0
-3.8
35.8
13.3
-49.1
24.2
46.2
-1.2
-20.0
6.1
-20.5
13.2
32.4
12.2
-5.4
1.1
-4.7
6.6
7,5
5.1
-15.5
0.6
-9.6
12.7
4"2
-7.6
-7.0
-0.7
-9.6
5.5
0.4
-11.4
1.1
0.0
-3.6
3.0
0.0
0.5
-46.8
8.1
--48.1
41.0
44,5
-1.3
24"N: 152°E-175*W
35"N: 152°E-175°W
175"W: 35*N-24*N
152"E: 35"N-24*N
Area 3 Total
-5.9
12.1
-6.1
11.5
-11.4
0.5
2.6
-1.0
5.6
-3.5
6.1
4.5
-5.6
9.1
-4.0
5.5
2.3
-4.9
-i.1
4.0
7.9
0.6
0.5
-0.9
10.0
14.2
22.1
-17.0
24"2
-5.0
-4.8
17.9
-5.5
-13.2
-5.6
1.0
5.0
-0.6
-6.6
-l.2
8.5
t0,3
-4.0
-12.7
2.0
6.3
0.3
-3.2
-5.5
-2.1
7.1
-2.8
-0.6
-3.0
1.8
18.0
30.8
-12.7
-41.0
-5,0
24"N: Hawaii Ridge- 152"W
24*N: 175*W-Hawaii Ridge
35"N: 175°W-152*W
152"W: 35*N-24*N
175"W: 35*N-24*N
Area 4 Total
-9.4
-5.7
3.9
-2.3
6.1
-7.4
-3.7
-0.2
1.0
2.1
1.0
0.1
-4.1
-2.0
0.6
3.7
5.6
3.9
-0.8
-4.2
-0.2
1.4
4.9
1.2
0.0
-0.4
-0.3
0.3
-0.5
-1.0
-17.9
-12.5
5.0
5.2
17.0
-3.2
-9.4
-5.8
5.9
-2.9
5.5
-7.7
-3.7
-0.2
1.7
1.7
0,6
0.1
-4.2
-2.2
3.4
2.2
4.0
3.2
-0.7
-4.3
5.5
-0.9
3.2
2.8
0.0
-0.4
-0.2
-0.0
-0.6
-1.2
-18.1
-12.9
15.3
0.1
12.7
-2.8
24°N: 152*W-Calif.
35*N: 152*W-Calif.
152"W: 35*N-24*N
Area 5 Total
-t4.6
9.7
2.3
-2.6
0.6
1.6
-2.1
0.2
2.6
1.8
-3.7
0,7
-5.8
3.6
-1.4
-3.6
0.7
0.0
-0.3
0.5
-16.4
16.7
-5.2
-4.8
-13.7
7.9
2.9
-3.0
0.6
0.0
-I ~7
-1.2
7,3
-4.5
-2.2
0.6
1.8
-5.7
0.9
-3.0
1.5
0.0
0.0
1.5
-2.6
-2.3
-0.1
-5.0
47"N: Japan-152*E
35"N: Japan-152°E
152"E: 42°N-35°N
Area 6 Total
2.1
21.4
-37.2
-13.7
2.0
5.1
-10.3
-3.2
-6.2
9.6
-2.4
1.0
-14.4
9.5
12.3
7.4
-0.5
3.5
5.5
8.6
-17.1
49.1
-32.0
-0.0
-0.8
20.5
-35.2
-15.5
-1,3
4.8
-8.4
-5.0
-15.5
9.7
5.0
-0.8
-22.9
9.7
22.3
9.1
-0.9
3.6
9.5
12.2
-41.4
48.1
-6.8
-0.1
47"N: 152"E-175*W
35"N: 152°E-175*W
175"W: 47*N-35*N
152°E: 42*N-35*N
Area 7 Total
0.1
-12.1
-19.0
37.2
6.3
0.4
-2.6
-7.4
10.3
0.7
4.1
-4.5
-2.0
2.4
-0.0
17.0
-2.3
-2.3
-12.3
0.1
0.0
-0.6
0.0
-5.5
-6.2
21.6
-22.1
-30.7
32.0
0.8
0.8
-17.9
-18.0
35.2
0.2
2.5
-5.0
-6~6
8.4
-0.8
11.7
-10.3
i. 1
-5,0
-2,5
30.6
-0.3
2,7
-22.3
10.7
0.0
2.8
0.0
-9.5
-6.7
45.5
-30.8
-20.8
6.8
0.8
47°N: 175*W-152°W
35"N: 175°W-152*W
152"W: 47*N-35*N
175"W: 47*N-35*N
Area 8 Total
-3.0
-3.9
-10.8
-19.0
1.4
-3.5
-1.0
-4.2
7,4
-1.2
~t.2
-0.6
1,3
2.0
-1.5
-2.5
0.2
0.6
2.3
0.6
0.0
0.3
0.0
0.0
0.3
-13.1
-5.0
-13.1
30.7
-0.4
-2.3
-4.9
-10.7
18.0
0.2
-.1.9
-1 7
-3,6
6.6
-0.6
1.5
..3,4
3.3
-1.1
0.3
4.9
-5.5
2.8
-2,7
-0.6
0.0
0,2
0.0
0.0
0.2
2.2
-15.3
-8.2
20.8
-0.5
47"N: 152*W-CAfiff,
35"N: 152"W-Calif,
152"W: 47*N-35*N
Area 9 Total
-1.7
-9.7
10.8
-0.7
-1.2
-1.6
4.7
2.4
3.0
-1.8
-1.3
-0.1
3.3
-3.6
-0.6
-0.9
0.0
0.0
0.0
0.0
3.3
-16.7
13.4
-0.3
-3.3
-7.9
10.7
-0.5
-3.6
0.0
3.6
-0.0
-4,3
4.5
-3,3
•3.1
0.2
5.7
-3.8
3.2
0.0
0.0
0.0
0.0
-10.9
2.3
8.2
-0.4
47"N: Japan-175*W
175"W: Ala~ka-47*N
Area 10 total
-2.2
1.0
-1.2
-2,4
2.4
0.0
2.2
0.8
3,0
-2.6
3.1
0.6
0.5
-0.3
0.2
-4.5
7.0
2.6
-0.0
! .4
1,4
-1.0
2.9
1.8
3,8
0.8
4.5
-7.7
2.2
-5.6
0.9
-0.7
0.2
-4.2
6.5
2.4
47"N: 175"W- 152"W
152"W:/~ I~ilca,.47*N
175*W: Alaska-47*N
Area 11 Total
3.0
2.8
-1.0
4.7
3.5
1.0
-2.4
2.1
4.2
-3.9
-0.8
-0.5
2.5
-4.4
-3.1
-5.0
0.0
0.0
0,3
0.3
13,1
-4.4
-7.0
1.6
2.3
3.0
-1.4
3.8
1.9
2.4
-2.8
1.a
-t .5
2.2
-9.7
.O 1
-4.9
2.4
-2.2
-4.7
0.0
0.0
0.7
0.7
-2.2
10.0
-6.5
1.2
47"N: 152*W-Wash.
152*W: Alaska-47*N
Area 12 Total
1.7
-2.8
-1.0
1.2
- 1.0
0.2
-3.0
3.9
0.9
-3.3
4.4
1.I
0.0
-0.0
0-0
-3.3
4.4
1.2
3.3
- 3.0
0.3
3.6
-2.4
1.2
4.3
-2.2
2.t
-0.2
-2~4
-2.6
0.0
0.0
0.0
10.9
- 10,0
1.0
Large scale circulation of the North Pacific
187
that the solution should not be biased toward large corrections in deep water (Rom~tlc~, 1981).
Residual mass imbalances are small. Table 4 shows initial and final transports from the rank 13
solution in each of the 12 enclosed areas numbered in Fig. 1. This solution will be referred to as
Model Ia.
The sensitivity of the solution to the formulation of the problem will be examined in two ways.
The first question is: can additional information be obtained from layer-by-layer conservation of
mass7 It does not seem reasonable to require conservation of mass in the deep layers of the North
Pacific. In analyzing trans-Atlantic data, ROEMMXCHand WUNSCH(1985) and others imposed mass
conservation in deep layers because it is thought that the flux of North Atlantic Deep Water
through to the South Atlantic and beyond is large compared to the rate at which deep water is lost
by mixing in the subtropical North Atlantic. The North Pacific is quite different since it lacks a
source of deep water and it is an enclosed basin. Any net flow into the basin in a deep layer must
be balanced by flow out in a different layer. Over much of the ocean the deep flow is quite sluggish
and in any given area the amount of water flowing through the area may be comparable to the
amount flowing across isopycnal layer boundaries within the area. Surface and thermocline
layers, on the other hand, have much greater flux through the areas because the size of the gyres
is large compared to the size of the areas shown in Fig.1. In these layers, one might reason that
the flow through an area is much greater than the water mass conversion within the area. Model
II is obtained by requiring mass conservation in the upper two layers in each of the 12 areas of
Fig. 1. The total number of constraints is now 40 and we have chosen the rank 20 solution. Note
that although the number of constraints has increased by 24, the rank has increased by only 7,
indicating that most of the added constraints are redundant. Some additional calculations were
made using up to 14 layers, but the amount of independent information did not increase
substantially, so the smaller number is retained for simplicity. Results of Model II are discussed
in the following section and are shown in Table 4.
The second question is: how sensitive are the solutions to the initial specification of velocity?
The most unusual and interesting feature of the initial field is the large westward flow indicated
by the deep current meter array at 152°E (Nm.m~, S ~
and 1.~, 1985). Spacing between
moorings may be large compared to the scale of the mean flows near the latitude of the Kuroshio
Extension, and the westward mean transport may be substantially overestimated. On the other
hand, the westward flow extends across many moorings and is supported by additional measurements of strong deep westward velocity in the array at 165°E (ScmarIz, 1987). Therefore, for test
purposes we will suppose that the westward sign is correct but that the velocity has been
overestimated by a factor of 2 in Table 2. The initial field of westward velocity along 152 ° E at
4000 m was therefore decreased to 1.5 cm s 1 at 35 ° N, tapering linearly to zero at 42 ° N and
28 ° N. Model I was rerun with this modification and will be designated as Model Ib.
5. RESULTS
5.1. Deep circulation and transport
We will begin by discussing transport in the deepest layers, where the shear is weakest and
the transport is most sensitive to the initial field of velocity. For each of the five density layers,
Fig. 3 displays the integral of transport, from model Ia, along every segment of each transect,
bounded by crossing transects and/or major ridges. Arrows are located arbitrarily in the segments
for display purposes and it should be kept in mind that the transports are accumulated from waters
flowing in both directions. The absence of an arrow in a particular segment indicates that there
188
D. ROEMMICH and T. McCALLISTER
60*N
40*
20*
120*E
1400
160 °
1800
160 °
140"
t20"W
FIG. 3a. Geostrophic transport (Sv.) in layer 1 from model Ia.
r~_
60* N
T_
....
,"
40*
".-~.~
+,.o
.-" ...... *i~ . . . . . . . . . . i ........ ~ : ~ ...... I. . . . . . . . . . . . . .
.<.';'"
~.'~ 7.2
':: ............... *~g ........... ~
~,,.,
2o°
--;" 9.t
q
~ ' . 6.5
*
Izo*E
laO*
4.1
*6:9 ....................... *,.5 .....
- - ~ 0.8
.... i~ 'l. . . . . . . . . . . . . .
i60"
180°
~,.?....,
~
2.i
A? r ./.
160"
~
0.8
,
)aO°
.........
J
'
J
Izo*w
FIG. 3b. Geostrophic transport (Sv.) in layer 2 from model la.
60*N
40*
20 o
I20°E
.
.
.
.
140 °
_.
160 °
m
t80 °
160 °
140 °
FIG. 3c. Geostrophic transport (Sv.) in layer 3 from model la.
120*W
189
Large scale circula~on of the North Pacific
1
SOLUTION Ia
r~'LAYER
'~
"
, ,.
".3
,3_.
•
~ ................. :....
,.:-...... ....... ~;io ....... i. :o
,0.
51/
~ .~cl8.2
,o.~
~ 1 ....
,-,.o
!
•:i ................ ~,~ ........... i ..... ~.~. ........................... ~i .......
. ' ~-." z.o ~.3.3
-~3.8
~2
~-1.9
" .-'"".
."~," _i i.,!.9 • !
~
~:.
.(..
•
....'
,'~.f, r,i;;,
............ i;x,
............ i............ ~-~.; i~ ....... ; ............ i;ori
,
~o.
I20"E
140 =
160"
180 °
160 °
140 °
120°W
FiG. 3d. Geostrophic lransport (Sv.) in layer 4 f r o m model Ia.
60" N
40"
20 o
120°E
.., ............, ..... ~:~+.- ...............,.6:~ ......................~.:7..,, .........y,.
........................8~.......I -- 1
..
140 °
160 °
180"
160 °
140 °
120°W
FIG. 3e. Geostrophic u'ansport (Sv.) in layer 5 from model Ia.
/i,i
'"
-
"
....
SOLUTION Io
"r1111TOTA L •
'\
""
GEOSTROPHIC
i
40"
120°E
140"
160 °
180 °
160 °
140 °
Fla. 3f. Total geostrophic translxx't (Sv.) from model Ia.
120°W
190
D. ROEMMICHand T. McCALLISTER
is no water in the given density range. The numbers indicate Sverdrups of transport in the solution
of model Ia. Corresponding transports in models Ib and II are listed in Table 4.
In layer 5, (Fig. 3e), 10 Sv of the densest water found in the North Pacific flow northward
across 24 ° N and then westward across 152 ° E. The existence of such a flow is supported by a
number of pieces of independent information. First, since there is no source of deep water in the
North Pacific, the densest waters in the basin must be renewed from the south. Second, the abyssal
tracer distributions (MnrrrvI.n and REIO, 1983) indicate spreading along the path from the western
boundary of the Southwest Pacific Basin northward into the Central Pacific Basin and then into
the Northwest Pacific Basin. Third, the geostrophic transports calculated separately for the South
Pacific sections along 28 ° S and 43 ° S (WunscH, Hu and GRANT, 1983) and for the North Pacific
section at 24 ° N ( R o E ~ c H , 1986) both indicate northward flow of order 10 Sv in the abyssal
1ayers. Finally, the deep current meter data from 152 o E ( N m .ER,S~
and LEE, 1985 ) and from
165 ° E ( S ~ ,
1987) show strong westward mean flow at 4000 m. In combination, the
evidence appears quite strong.
Model Ia indicates about 10 Sv flowing westward in layer 5 into the region bounded on the
west by Japan and the Izu Ridge (areas 2 and 6 combined in Fig. 1a), but less than 2 Sv leaving
the area in this layer. In the deep layers, we believe one should disregard the transport estimates
across 35°N because of the small number of deep stations there. The divergence corresponding
to the layer 5 convergence is found principally in the overlying layer 4 (Fig. 3d). The implication
is that within this area there is conversion of layer 5 water into layer 4 water by diffusion and
mixing across the isopycnal boundary between these layers. In layer 4, as in layer 5, there is
northward flow across the deepest part of 24 ° N and about 21 Sv flows westward across 152 ° E,
but additionally there is strong flow out of the area, about 20 Sv northward across 47 ° N and 11
Sv southward across 24 ° N between the Izu Ridge and 152 ° E. The northward flow retums
southward on the west side of the Emperor Seamount chain, completing an anticyclonic
circulation in the deep Northwest Pacific Basin, beneath the cyclonic subpolar gyre. Layer 3 also
has northward flow across the deepest segment of 24 ° N and westward flow across 152 ° E. Again
there is substantial northward and southward flow out to the western boundary region. But there
is not a large divergence in this layer, as the water flowing into the western boundary area nearly
balances that flowing out. In this layer, one also begins to see the northward flow of the
subtropical gyre in the Philippine Basin, turning and crossing the Izu Ridge toward the east in the
Kuroshio.
It is interesting to examine the plausibility and implications of the deep water mass conversion
described in the previous paragraph, and this issue will be continued in the discussion section.
First, accepting the strong evidence that there is a net supply of abyssal water into the North
Pacific, an equal amount of water must return southward at a lower density. Results from the
South Pacific (WLrNSOt,HU and GRANT, 1983) indicate that the southward return flow is of middepth and deep water and not surface water. Given that the conversion occurs in the tropical and/
or northern Pacific and that the diffusive process is probably turbulent, it seems likely to be
concentrated near the western boundary at middle latitudes because this is where the kinetic
energy level is the highest. Here there are strong energy sources in the upper level (and possibly
deep) western boundary currents of the subtropical and subpolar gyres to drive eddy and boundary
mixing processes. Eddy kinetic energy at 4000 m is a factor of 10 to 50 greater at 152 ° E than at
152 ° W (ScriMn~, 1988). West of the Izu Ridge, there is still a high energy level, but the ridge
does not permit the very dense water of layer 5 to enter the Philippine Basin.
We are not singling out the northwestern boundary region as the only site of cross-isopycnal
transport. Indeed, there is a density decrease of the waters all along the path of abyssal flow from
Large scale circulationof the North Pacific
191
the South to the North Pacific ( ~
and R~m, 1983) so that any flow along this path must
cross isopycnals. To illustrate further, Fig. 4a shows the net transport of water across 24* N in
model Ia. Here, transport is shown in 14 layers. This should be compared to the flow across 28*
S and 43* S as shown by Wtrsscn, Hu and Gv,hrcr (1983). The maximum in the abyssal transport
across 43 ° S and 28* S occurs at a density of approximately sigma-4 equals 45.96 ° and a depth
of about 3700 m. At 24 ° N the northward abyssal flow is less dense but at greater depths. The
maximum occurs deeper than 5000 m with a density of about sigma-4 equals 45.90. But in
flowing into the western boundary region in layer 5 and out in layer 4, the water must flow upward
as well as across density surfaces. The deep southward transport across 24* N is at about
4500 m with a density slightly greater than sigma-4 equals 45.87 (Fig. 4a). In crossing 28* S and
43 ° S, the deep southward flow is spread between about 1000 and 2500 m. If the deep transports
of the North and South Pacific are part of a single ocean-scale system, as seems reasonable to
assume, then our picture of the thermohaline circulation of the deep Pacific is of abyssal flow
which is downward with decreasing density from the western South Pacific to the western North
Pacific, then upward with decreasing density along the return path. In this view, the distinguishing feature of the western North Pacific is that here is where the vertical velocity becomes upward.
The computed conversion of water from layer 5 to layer 4 in the western boundary area could
be sensitive to errors in the initial velocity field. Model lb, with a 50% reduction in the initial
estimate of westward flow across 152 ° E at 4000m, was constructed to test this issue. The results
(Table 4) do show a reduction in the transports but the basic pattern is unchanged. About 8 Sv
flow northward across the deep segment of 24 ° N in layer 5, with about 5 Sv then flowing
westward across 152 ° E. In layer 4, 11 Sv flow westward across 152°E with 20 Sv exiting the
area to the north and south. The deep anticyclonic circulation across 47 ° N remains, but is reduced
by about 25% from model Ia. Model Ib has some inconsistent aspects not found in model Ia. The
layer 4 divergence noted above is nearly twice the size of the corresponding layer 5 convergence.
Also, in this model there is a net southward flow of 9 Sv across 47 ° N in layer 4, a physically
implausible result since there is no way of producing this water north of 47 ° N. Thus this change
in the initial field does not alter the finding of a large conversion of water from layer 5 to layer
4 in the western ocean. Model Ia is preferred to model Ib because it is in better agreement with
the current meter data (Nm~R, S ~
and L ~ , 1985) and because it describes a more consistent
flow field when the whole western basin is considered.
In the westem Philippine Basin (area I in Fig. 1), model Ia shows a convergence of 7 Sv in
layer 4. Layer 4 is the densest water in that basin and, in analogy to layer 5 east of the Izu Ridge,
this is where one would anticipate enhanced cross-isopycnal transport. The corresponding
divergence is found in layers 3 and 2.
We will also argue that model Ia is preferred to model II, the model which included mass
conservation in layers 1 and 2. The two solutions are similar, but in areas where they differ, the
model Ia solution appears generally more plausible. For example, as in model Ib, model II has
a large net flow southward in layer 4 across 47°N, about 13 Sv, with no possible source for this
water. To understand the problem in this model, note that the initial field has upper layer
transports that are already fairly well balanced in most areas (Table 4). This is because the shear
across the thermocline dominates the transport. In areas where initial upper layer transport is not
nearly balanced, i.e. areas 2, 3, 6, and 7, the imbalances are likely to result from the undersampling
problems along 35 ° N discussed earlier, and perhaps in part from the fact that the axis of the
Kuroshio Extension is very close to 35°N. Note for example that the upper layers of combined
"WoNscn, Hu and GRANT(1983) used the pre-1980 equation of state for seawater whereas we have used the newex
equation (UNESCO, 1981) in all calculations. In refen'ing to their results an approximate conversion from the old to the
new equation has beon applied. This consists of subWacting .06 from the old values of sigma-4.
192
D. ROBMMlCHand T. McCALLISTER
o
I
2000
"1-
40O0
[] = I sv
6OOO
--0.08
--0.04
0
0.04
0.08
TRANSPORT PER UNIT DEPTH {lOemZs - I )
Flo. 4a. Transport per unit depth across 24°N in model Ia.
J
ZOO0
!
w
O
4000
ir, t'y3t3
--
I
I
I
I
I
I
0
I
I
I
I
I
=ISV
I
I
I
08
-0.04
0
0.04
0.08
TRANSPORT PER UNIT DEPTH (106m2s-I)
FIG. 4b. Transport per unit depth across 152°E in model Ia.
Large scale circulation of the North Pacific
193
Oi
J
2ooo
"-r
I--
o-
4OOO
6000
I-I -- I SV
l
I
I
I
I
I
I
I
I
I
I
I
I
i
-0.08
-0.04.
0
0.04
C~08
TRANSPORT PER UNIT DEPTH (106m2s-J)
Fro. 4c. Transportper unit depth ata'oss 175°Win model Ia.
areas 2 and 6 are near balance in the initial field, as are the combination of 3 and 7. In both cases,
it would be best to simply eliminate the 35°N section and thus to form larger areas. The effect
of reqniring a closer mass balance in upper layers in the areas including 35 °N is to introduce noise
into the solution.
Figures 4b and c show layer transport per unit depth from model Ia, for the 152 ° E and 175 °
W sections. At 152 ° E, it can be seen that the westward flow is nearly depth independent from
the ocean bottom up to about 1000 m. This highlights the fact that the westward flow has a very
large volume transport, principally at depths below the top of the Izu Ridge. Therefore, between
152 ° E and the ridge, which is at about 140 ° E, this large flow must turn to the north and/or south.
In model Ia, about 2/3 turns to the north and 1/3 to the south (Figs 3c-e). Very little net deep
transport is found across 175 ° E in the Northeast Pacific Basin (Fig. 4c). Flow is far more sluggish
in the Northeast Basin than in the Northwest Basin.
It is interesting to study the velocity field on a somewhat finer scale than is offered by the
transport maps, Fig.3. Figure 5 displays horizontally smoothed geostxophic velocity profiles
from each of the sections from Model Ia. The smoothing is done with a horizontal Gaassian filter
which has an e-folding length ofS00 km and is normalized to unit area. It is intended to suppress
mesoscale features which would otherwise dominate the display.
In Figs 5a, 5c, and 5e one can see the result of balancing total mass in the western boundary
area. The maximum westward flow at 4000 m across 152 ° E is decreased in the solution to about
2 cms 4. In order to move water out of the area, deep boundary flows appear in the zonal sections
- - southward across 24 ° N along the eastern slope of the Izu Ridge and northward across 47 ° N
along the continental slope. The only other large deep flow across 24 ° N is the broad area of
northward flow close to the bottom, with speed averaging more than 0.5 crn s-1 between 160 ° E
and the dateline. This is the densest water in the section (see Fig. 2a) alluded to earlier. The
northward flow across 47 ° N near the western boundary is balanced by southward flow that is
predominantly west of the Emperor Seamounts. Because there are deep gaps in the Emperor
194
D. ROEMMICtt and T. MCCALLISTER
Seamounts, they were not treated as a bamer in the model, but nevertheless the solutions show
the deep recirculation to be largely confined by this feature. Note that although our initial field
contained a guess at this pattern of flow, it has been greatly reinforced in the solution. This anticyclonic deep recirculation carries more than 30 Sv between 2000 m and the ocean bottom. In
the eastern basin at 47 ° N, our initial guess was of weak northward flow between 175 ° W and
160°W. Interestingly, in the solution this flow is shifted to the west and narrowed, occurring
mainly between 1750 W and the dateline.
In the 175 ° W section (Fig. 5f), the deep eastward flow of the initial field remains, centered
on 35 ° N. This is likely a pathway for flow from the northwest Pacific basin to the Northeast
Pacific Basin in layers 3 and 4. It is supported both by the current meter array along 175°W and
by the abyssal property distributions (MArCrYLAand REID, 1983). Farther to the north, a broad
weak westward flow is found below the subpolar gyre. At the northern boundary, details of the
field are obscured by the broad Gaussian filter, but the net eastward transport of the initial field
at 3000 m remains. At 152 ° W (Fig. 5g), the solution is little changed from the initial field. There
is a slight weakening of the broad westward flow and of the eastward flow near the northern
boundary which were specified in the initial field. The current meter data demonstrate that the
mean flows in the deep eastern ocean are weak and for purposes of the inversion there is little
signal.
5.2. Transport and circulation in upper ocean and thermocline layers
The flow in the upper layers is dominated by the subtropical and subpolar gyres, with large
transport in the western boundary currents balanced by broad return flows in the interior. Solution
velocities and transports in the top layers are relatively insensitive to the uncertainty in the initial
field of velocity because the shear in velocity across the thermocline is large compared to the
unknown deep component.
Total transport of the Kuroshio in the East China Sea is estimated to be about 32 Sv in model
Ia (Figs 3a-b and Table 4). This is slightly larger than the value of 28 Sv obtained by BRYom~,
ROmaMICHand CHURCh (1989) using 24 ° N data alone, with reference velocities provided by the
dopppler current profiler. The larger value in the present solutions results from matching to the
large Kuroshio transport downstream in the 137 ° E section. It is likely that differences of this order
can be resolved only by time-series measurements of the current. Additional northward transport
is found east of the Ryukyu Is in the Philippine Sea (Fig. 5a). But this region has a high degree
of variability and with the single snapshot we cannot distinguish between a tight permanent
recirculation and a strong transient eddy. The Kuroshio follows the Japanese coastline, being
found right against the coast in the 137 ° E section (Fig. 5d) and in the 35 ° N section. It subsequently
separates from the coast and flows eastward along a meandering path near 35 ° N. In Fig. 5b, the
meanders appear in the 35 ° N section as far east as about 170 ° E. The velocity at 152 ° E (Fig. 5e)
shows the Kuroshio centered at about 36 ° N, penetrating to a depth of about 1600 m. The shallow
westward recirculation seen in this figure is split into two maxima, one just south of the Kuroshio
and the second at least as far south as 24 ° N.
The Kuroshio can no longer be distinguished farther to the east, in the sections at 175 ° W and
152 ° W. The velocity at 152 ° W (Fig. 5g) shows a classical interior circulation - - t h e westward
limb of the subpolar gyre in the north, then broad eastward flow all the way to the center of the
surface subtropical gyre near the latitude of Hawaii. Below the broad eastward flow is a weaker
westward flow, with the zero surface sloping upward to the south. Thus the gyre center tends
poleward with depth, to about 40°N at 1000 m depth. In the zonal sections at 24 ° N and 35 ° N,
Large scale circulation of the North Pacific
195
Om --
i
("k
..
6000--
130*E
140"
150"
160 =
170"E
180"
170=W
160 °
150 °
140 =
130 =
120"W
1~o. 5 ¢ Horizontally smoothed geostrophic velocity (era s "1) across the 24*N transect. The filter is
a Gaussian which decays to e "1 at a distance of 500 kin. from its center. Positive velocity in this and
the subsequent figures indicates northward cf eastward flow. Negative velocity (shaded) indicates
southward or westward flow.
0mr-
2000
4000
6000
........
t41=E
I .........
150 °
I .........
I .........
160 °
170=E
I .........
180 °
I .........
1700W
I .........
160 •
I .........
150 °
I .........
140 =
I .........
150 °
l~o. 5b. Smoothed geostrol~ic velocity (cea s "t) across the 35"N transect.
0m
............................
Ji ..............
, .........................................
2000
¢q
4000
'~
. . . .
146°E
r . . . . . . . . .
150 °
i .........
160 °
i .........
170=E
i .........
180 °
i .........
170*W
i .........
160 °
i .........
150 °
| . . . . . . . . .
140 °
i .....
130 = 125°W
t~o. 5C. S m o o t h e d ~.x)slropbic v e l o c i t y ( c m s -~) across the 4 7 * N transect.
122*w
196
D. ROEMMICHand T. MCCALLISTER
Om
"
''
'
'
'
'
'
'
'
Om~'~'-"
'
2000 ---.
2000
E-
4000
~
4000
6000
l
6000 ~
I
I
r
I
33°N
I
I
I
~
30"
I
J
25°N
I
I
i
i
I
I
i
I
I
I
[
I
i
I
I
30"
I
25'~N
Fro. 5e. Smoothed geostrophic velocity (cm s 4) across the
152°E Iransect.
, . . . . . . . . . . . . .
Om
I
42°N 4 0 "
F~G. 5d. Smoothed geosl~ophic
velocity (cm s t ) across the 137°E
transect.
........
--4
,
,
r
,
,
,
,
,
,
,
,,,,,,,
,
.
.
.
.
.
.
,
\
_~
2O0O
-~
50*N
40 °
30 °
25°N
FIG. 5f. Smoothed geos~rophic velocity (cm s 4) across the 175°W transect.
_
~
~
:
A
-
~
4000 --
6000
"--q
II l l
--
_
i
r
56°N
i
i
i
i
I
50*
i
--4
::1
..2
i
i
i
i
¢
i
t
i
I
40*
I
i
~
t
i
i
i
i
i
I
i_~_-t__~
i
1
i
:30 °
FIG. 5g. Smoothed geostrophic velocity (cm s 4) across the 152"W transect.
~L~
22°N
Large scale circulationof the North Pacific
197
the southward flow of the gyre interior is spread across the basin with, in each case, intensification
of southward flow marking the California Current near the eastern boundary.
The structure of the geostrophic surface flow is best seen by examining geostrophically
balanced sea level. Figure 6 shows the sea level obtained from model Ia by integrating
geostrophic velocity westward from the eastern boundary, assuming no discontinuity at the
Hawaii Ridge, the Izu-Ogasawara Ridge, and at Okinawa:
f d ~ = 1/gl fu.n d x
From west to east, sea level rises by about 60 cm across the Kuroshio in the East China Sea.
Then, in the Philippine Basin, there is a series of large fluctuations, about 40 cm from peak to
trough, corresponding to the flow reversals found there (Fig. 5a). Farther to the east, there is a
downward trend in sea level marking the southward interior flow, with the magnitude of the
eddies diminishing to the east. In the Northwest Pacific Basin the sea level fluctuations are down
to about 20 cm. This level of eddy activity persists to about 1000 km east of the Hawaii Ridge
where the eddies diminish further, to about 5-10 cm in sea level height over the easternmost 4000
kin of the ship track. Finally, within 500 km of San Diego there is a sharper downward slope
amounting to about 20 cm across the California Current. Overall, San Diego sea level is about
25 cm below that of the East China Sea, indicating that the net surface geostrophic flow is
southward.
The zonal transect of the subpolar gyre center shows weaker large scale velocities (Fig. 5c)
than the subtropical gyre. The southward Oyashio is found at about 148 ° E with a band of
northward flow inshore and broad weak northward flow extending across the ocean interior.
Although the surface velocities are stronger than the deep velocities, the deep gyre at 47 ° N carries
greater transport since it is spread over a much greater depth range. Thus the map of total
geostrophic transport, Fig. 3f, shows an anticyclonic recirculation of about 30 Sv west of the
Emperor Seamounts in the northern ocean. Elsewhere, the total flow tends to follow that of the
surface layers, though again at 152 ° E the deep flow contributes a large fraction of the total.
5.3. Fluxes of heat and freshwater
For purposes of computing oceanic heat and freshwater transports, we consider the fluxes
across 24 ° N, across 35 ° N, and across 470 N, plus an eastern boundary area defined as east of
152 ° W and north of the 24 ° N section (areas 5, 9, and 12 in Fig. 1), and a western boundary area
defined as west of 152* E between the 24 ° N and 47 ° N sections (areas 1, 2, and 6 in Fig. 1). The
total mass transport into each of these regions is zero, and the heat flux is calculated by integrating
along the oceanic boundary of each region the product:
H.F. = ~S pCpOff.d dxdz
The velocity is composed o f a geostrophic part and an ageostrophic Ekman component. The
Ekman transport is estimated from wind stress tabulations by I-lm.,.rz~taN and ROSmCSTmN(1983),
with the ocean surface temperature used for the flux calculation. The flux calculations are
discussed in greater detail by BRYom~,ROEMMICHand C-MtmCH(1989).
The flux of sait into each region is also zero since there are no significant sources or sinks. If
there is a net excess of evaporation or precipitation plus runoffin a given region, the effect on the
198
D. ROEMMICHand T. MCCALLISTER
mass budget is negligible and is considered as part of the noise in balancing mass. However, since
the salinity contrast between oceanic salinities and water vapor is so much greater than it is
between water masses, a small net evaporation or precipitation has a greatly magnified effect on
the salinity budget. The mass balance equation, including the small residual, is:
F.W. = ~S P ti.h dxdz
The salt budget is:
0 = ~ pS u.n dxdz
Multiplying the first of these by a mean salinity, So, and subtracting the second gives:
F.W. = 1/So~ pS' t~.d dxdz
This equation is used to calculate the net excess of evaporation or precipitation plus runoff,
also referred to as the oceanic freshwater flux.
The heat and freshwater fluxes across each of the three parallels of latitude and into the eastern
and western ocean areas are listed in Table 5. BRYI)~N,ROEMMJCrIand CmraCH (1989) calculated
the heat flux across 24°N using data from that section alone and obtained a value of 0.85 x 10~sW
with an error estimated to be 0.25 x 10 ~5W. Our value, 0.75 x 10 ~5W, obtained from the complete
North Pacific data set, is not significantly different. Heat and freshwater fluxes in the Pacific are
dominated by transport in the upper kilometer of the ocean. In the upper kilometer, the total
velocity is due mainly to the shear across the thermocline, with the uncertainty in deep absolute
velocity making a small contribution. BRYDEN,R O A C H and C'm~cu (1989) used the bottom
as a reference surface in calculating heat transport across 24°N. As mentioned previously, this
produces unrealistic southward transport in the deepest layers. However, the deep shear is
sufficiently weak relative to shear in the upper kilometer, that the warm water transport which
dominates the heat flux is not greatly different in the two calculations. As long as one takes care
to balance the total mass, the flux calculations are not very sensitive to circulation in the deep
ocean.
The heat transports, Table 5, are consistent with local patterns of air-sea heat exchange from
bulk formula calculations (TaT.Lrcy, 1984), showing large heat loss by the western subtropical
ocean and modest heat gain in the east. Heat loss in the western boundary area is greater than the
heat transport across 24 ° N, indicating that the remainder of the ocean outside the western
boundary area has a net heat gain. As discussed previously, the northward Kuroshio transport
across 35 ° N is probably underestimated due to undersampling near the coast. An additional 10
Sv of transport at 15 ° C, balanced by depth independent flow in the interior, would change the heat
transport across 35 ° N to about 2 x 10 ~4W northward.
Errors in bulk formula calculations increase with the area of the region being considered.
Errors in oceanic heat transports do not depend on area but rather are affected by factors such as
the uncertainty in warm western boundary transport and Ekman transport. It follows that the bulk
formula calculations are the preferred means of establishing small-scale patterns or spatial
gradients of heat gain and loss while the oceanic heat transport calculations, applied over large
areas, may be used to remove systematic biases in the bulk formulae.
Large scale circulation of the North Pacific
199
TABLE 5. Fluxes of heat and freshwater in model Ia. A positive sign indicates northward transImrt
across 24"N, 35"N, or 47"N, or transport into the western or eastern boundmT areas. Thus, transport
of heat into the western boundary area indicates that the ocean loses heat to the auneaphere in that are&
T r a n s p ~ of freshwater out of the western boundary area indicates a net excess of precipitation and
OVer evaporation in that area
Heat flux (1014 W)
Fresllwater flux (10 s m 3 s "~)
24"N
7.5
-2.9
35"N
-1.6
-5.6
47"N
-0.9
-2.7
Western boundary area
(west of 152"E, 24"N - 47"N)
9.6
-1.5
Eastern boundary area
(east of 152"W, north of 24-N)
-0.5
-0.2
i00
8o
~60
_J
>
i.d
..J
<4o
lad
tD
2O
o
-12,ooo
-IO,OOO - 8 0 0 0
-6000
-4000
-2000
o
DISTANCE (km)
Fro. 6. Geostophically balanced sea level, relative to sea level at the eastern boundary, along
the 240N transect.
The freshwater fluxes, Table 5, are consistent with tabulations of net evaporation minus precipitation and runoff by BAtn~GARrSF.aand R~CrmL (1975). Here it is necessary to correct for the 1.7
Sv flowing northward through the Bering Straits (COACHMAN,A A G ~ and Tma,P, 1975) because
of the low salinity of that flow, about 32.6 psu compared to an average of about 34.6 psu at 24°N.
Our value of 0.29 x 106 m3s-1 for the freshwater divergence between 24 ° N and the Bering Sea is
not significantly different from the value of 0.26 x 10 ~obtained by BRYDEN,Rom~lr~acnand Or~Crl
(1989) and by BhtrMGARrt~-~ and RmcrmL (1975). Since the subtropical gyre has net excess
evaporation while the subpolar gyre has net precipitation, the largest freshwater divergence
200
D. ROEMMICHand T. McCALLISTER
is found in the area between the Bering Sea and 35 ° N, that section being near the gyre boundary
in the west. An interesting contrast between the North Atlantic and North Pacific is in the
freshwater balance near the western boundary. In the Atlantic the large net evaporation from the
warm Gulf Stream dominates. The western Pacific has much greater precipitation because of the
monsoon circulation. We calculate a net excess of precipitation and runoff in the western
boundary area amounting to 0.15 x l0 s m3s-~(Table 5).
6. DISCUSSION
We have applied a least squares inverse method to estimate the large scale circulation of the
North Pacific Ocean. First, an initial field of velocity was constructed by assimilating information
from the geostrophic shear field, from measured currents in the deep ocean, from the abyssal
distribution of properties, and from the basin topography. A solution was obtained which
conserves mass and does not allow flow through solid barriers and which is closest, in a least
squares sense, to the initial estimate. The sensitivity of the solution was tested with respect to
changes in the initial velocity field and to the addition of mass conservation constraints in layers.
The data which were used in constructing the initial field of velocity are incomplete. The
spatial coverage of the hydrographic database is limited and the combination of data from
different cruises and years leads to questions about the overall consistency of the data. The mass
conservation constraints have limited information content for driving a poor initial guess toward
the correct solution. Of what use, then, are such models? We offer three answers to this question.
First, there are aspects of the solution which stand above the likely errors in the model. For
example, it is probable that there is a net transport of order 10 Sv of abyssal water northward
across the 24 ° N in the Northwest Pacific Basin. The flow turns westward toward the IzuOgasawara Ridge and the Japanese coast. Near the ridge, most of the flow again turns northward,
becoming part of a deep anticyclonic recirculation in the northern Northwest Pacific Basin. The
area east of Japan and the hu-Ogasawara Ridge appears to be one of enhanced cross-isopycnal
transport and relatively strong upward velocity in the abyss.
Second, the model allows us to identify regions of particular interest and to specify the data
which are required in order to improve the circulation estimate in those areas. Again the western
boundary is an important example. Here, three aspects of the solution are closely coupled and
additional information on any of them would improve the total solution. These are:
- - the westward abyssal transport across 152 ° E. The moored current meter array did not have
sufficient spatial resolution or duration in time to permit a good estimate of the mean transport.
Model 1a obtains lower westward velocities (Fig. 5e) than those specified in the initial field, and
model lb shows that the solution is quantitatively different when a lower initial transport is
specified. To go beyond this and place proper error bounds on the solution, we require an error
estimate for this part of the initial field.
- - the existence of deep western boundary currents in the North Pacific. An elementary
deduction is that if there is a substantial deep westward transport at 152°E and 165°E, as indicated
by the moored arrays, then the flow must either turn northward near the Japan coast or southward
near the Izu-Ogasawara Ridge. Model la has about 2/3 of the flow turning northward and 1/3
southward. Additional information on the location, depth range, and strength of these possible
flows would greatly help to constrain the solution in this area and elsewhere.
- - flow across isopycnal surfaces. Solution 1a shows a conversion of 8 Sv from layer 5 to layer
4 in areas 2 and 6, west of 152°E (recall that these areas are considered together because the
transport estimates across 35°N are poor). The implied vertical velocity and diapycnal mixing
Large scale circulationof the North Pacific
201
coefficient for this region are about 100 meters per year and 6 x 10 -3m2s-1respectively in a depth
range around 5000 m. The values seem unacceptably large. On the one hand, if 10 Sv, or even
3 or 5 Sv of abyssal water flows across 24°N into the Northwest Pacific Basin, and even if a
considerably larger area of that basin is involved than the portion west of 152°E, still the diapycnal
transport must be elevated with respect to accepted mid-ocean values. One could place upper
bounds on the diapycnal mixing coefficient as a model constraint, as was done by Scrmn-ze.~
(1988) in a model of the North Atlantic. An effect of such a constraint would be to limit the
northward transport of abyssal water. However, it is not clear what the upper bound on the mixing
coefficient should be, nor is it necessarily valid to extrapolate thermocline values of the mixing
coefficient down to the deepest layer of the ocean and into the western boundary. While we would
agree that the implicit vertical velocity and diapycnal mixing of the present model may indeed
be too large, some relevant measurements are required in order to proceed further. Observations
are needed either of mixing in the abyss near the western boundary, or of the density and transport
of inflow and outflow waters, including measurements of the time variations in absolute flow and
geostrophic shear. We have chosen the moored array data as the most relevant at hand and the
resulting circulation depicts the consequences of the westward abyssal flow found in those data.
Third, although the model solution may be poor in some areas, it is a starting point. We have
combined a large body of hydrographic measurements and current measurements to estimate the
basin-scale circulation in a consistent manner. It is the best estimate we can make with the present
data set and a very simple model. The solution is highly underdetermined because of the small
number of constraints. We believe that the correct hierarchy of models begins conservatively in
this respect rather than with a unique solution based on dubious constraints. The present solution
is subjective in the sense that it is strongly dependent on our interpretation of the data used to
construct the initial velocity field. Once unrealistic aspects of the basic model are identified,
iterations are easily made in order to include additional data or added constraints.
The present study of the North Pacific completes a preliminary survey of the global
thermohaline circulation. Previous studies of the North Atlantic (e.g. Romv~flcn and WtrNscn,
1985), the South Atlantic (Fu, 1981; RecrotrL, 1988), the South Pacific (WtrNscrI, Hu and GRANT,
1983), and the Indian Ocean (Fu, 1986) have estimated the net meridional transport of water
masses in those oceans. Considered together, the global picture is a consistent one and is
relatively simple below 1000 m depth. The principal path of deep water renewal and transport is
shown schematically in Fig. 7. In the South Atlantic and North Atlantic, about 15 Sv of surface
and intermediate waters flow northward in the upper kilometer. Deep convection in the northern
North Atlantic produces North Atlantic Deep Water which flows southward out of the Atlantic,
balancing the northward transport of surface and intermediate layers (Fu, 1981; Rom~ICH and
Wtmscn, 1985; RlrcrooL, 1988). On reaching the Southern Ocean, deep water parcels are likely
to make multiple circuits of the globe before leaving that sector. This inference is made by noting
that circumpolar transport of deep water is several times the magnitude of the added contribution
of deep water from the North Atlantic (Rn,rrouL, 1988). While in the southern sector, deep water
is modified by deep exchange in the Weddell Sea. About 12 Sv of deep water between 3200 and
3800 m is removed from the Southern Ocean, flowing northward through the South Pacific
(WouscH, Hu and GRANT, 1983). Additionally, some bottom water flows northward into the
Atlantic, but the volume is only a few Sverdrups and is omitted from Fig. 7. The Indian Ocean
appears to have little net meridional flow below 3000 m (Fu, 1986). The North Pacific shows
almost as much deep northward flow as the South Pacific, though at greater depth and slightly
lower density. The southward return flow in the Pacific is in deep layers overlying the northward
flow. In the South Pacific the southward flows are in the depth range 1000 to 3000 m (WuNscrI,
202
D. ROEMMICH a n d T. McCALLISTER
Southern
Ocean
Atlantic
24oN
PAoN ~
'
0
Southern
Ocean
Pacific
I~8o8 24.oN
24o14 28o~
Indian
~
I
1000
P
I
t
~-~
o
L
~orthward flow ]
zo0o
T
I
I
I
3000
~
,
Southward flow
~' r,lom,,,,,,rd no.
\
/I
,~
Northward flow
4000
e~
Southw.rdnow
5000
~ /
n 0 sv
6000
i
i
[
Fro. 7. Scbemeticrepresentation of the major global pathway of deep water renewal based on
ROI~MMICI-Iand WUNSCH(1985), Fu (1981), Rn~TOUL(1988), WUSSCH,HU, and GRANT(1983), Fu
(1986), and the present work. The line acrossthe figure is drawn at the mean depth of flow across the
subtropicsin each ocean, with the verticalbarsretzesentingthe approximate rangein depth. Numbers
indicatethe transportin Sverdrups. Note that the flow reverses directions five times alongthis pathway
(in theNorth Adantic, Southern Ocean, North Pacific, Southern Ocean, and Northern Indian Ocean).
Hu and GRAter, 1983). The flow again reaches the Southern Ocean and again may make multiple
circuits of the globe before leaving that sector. The principal northward flow out of the Southern
Ocean into the Indian Ocean amounts to about 10 Sv and is in the depth range 1500-2500 m (Fu,
1986). At these depths, there may be contributions directly from the Atlantic as well as via the
Pacific. Finally, waters exit the Indian Ocean in surface and intermediate layers. At this level,
because of the multiplicity of sources of intermediate water, the global pathways intertwine and
become ambiguous. To summarize, we believe the principal pathway for global deep flow and
renewal is from the North Atlantic to the South Atlantic to the Southern Ocean to the South Pacific
to the North Pacific to the South Pacific to the Southern Ocean to the Indian Ocean (Fig. 7). Each
of the oceans has a meridional cell of order 10 Sv. The role of the Pacific and Indian Oceans is
to decrease the density of the deep layers in order to complete the loop started by the deep
convection in the northern North Atlantic. The Southern Ocean acts as a conduit to exchange deep
layers from one ocean to another and also to modify the deep water masses in transit.
The total volume of the oceans is about 1.3 x 1018m 3 (WoRa34_moxor¢, 1981), with the bulk of
this being at depths greater than 1000 m. If we take 20 Sv to be the total production of deep water,
from the contribution of the North Atlantic plus a smaller contribution of deep water from the
Southern Ocean, then the mean residence time for the deep waters of the globe is about 1000 years.
Since water in the deep western boundary currents may traverse an ocean basin in as little as a few
years, parcels which follow the fastest pathways around the globe can obviously travel through
the system in much less than 1000 years. Parcels which find their way to regions far from the
principal pathway, such as the eastern North Pacific, may have a much greater age.
It should also be pointed out that the results presented here, and implicit in the moored array
data from 152°E and 165°E, suggest a different view of the deep North Pacific from that predicted
by simple analytical models. The dynamical framework of S r o r , ~ L and ARor~s (1960) was
applied to the North Pacific by WARREN and OWF~S (1985), with a correction for bottom
Large scale ci~niafion of the North Pacific
203
topography. The model has uniform upwelling in the deep ocean, with a northward interior flow
drawing fluid out of a westem boundary current which is meridionally convergent. In contrast,
the current meter data show deep flow toward the western boundary over a wide latitude band.
In order to conserve mass with a westward zonal flow into the boundary, that boundary being the
Japanese coastline and its extension, the Izu-Ogasawara Ridge, our model requires the meridional
transport in the vicinity of the boundary to be divergent. The anticyclonic recirculation extending
from the western boundary to the Emperor Seamount Chain is also not a feature of the StommelArons type solutions. It appears that the effects of the western boundary penetrate quite far east
and also that the presence of the deep vertical-meridional cell is an important factor in establishing
the deep circulation in the western half of the Pacific.
JoYCs, WAg,raN and TAUmY(I 986) found that variations in potential temperature on surfaces
of constant salinity in the deep subarctic North Pacific are consistent with geothermal heating
rates and long (order 100 years) residence times of water in the bottom kilometer. This, in turn
seemed consistent with a sluggish Stommel-Arons type northward interior flow, slightly
modified by buoyancy driving from below. All except one of the stations displayed by JoYcs,
WARRENand TAH~F.y (1986) were in the Northeast Pacific Basin and the single station from the
Northwest Basin showed no evidence of heating. WARRENand Owm~s (1985) alSOconsidered data
from the Northeast Basin in finding a boundary current structure along the Aleutians that is
consistent with a modified Stommel-Arons interior circulation. The North Pacific current meter
dataset, on the whole, and the hydrographic observations, suggest that the Northwest Basin and
the Northeast Basin are markedly different regimes both in the levels of kinetic energy which are
found there and in the residence times of the deep water. We have found very rapid exchange and
modification of abyssal water in the western basin as part of the global thermohaline circulation.
Residence times, based on net transport and volume, are years to decades, depending on how
much of the Northwest Basin is included. On the other hand, the eastern basin appears to be away
from the main pathway, and accessible only through much slower processes.
7. A C K N O W L F B ' ) G E ~
The extensive dataset used here was collected through the efforts of many individuals who, in some cases, have made
the data available in advance of publication. For the hydrographic data, the list of credits includes: H. BRYDBN,IL
DESZOBKH,]~ HALL,T. JoYcE, K. KENYON,J. SWIFT,L. TALLEY,B. WARRSN,the Woods Hole CTD Group, the Scripps
Physical and Chemical Oceanographic Facility, and the Japan Meteorological Agency. P. Nm.ea and W. SCHMITZ
provided an early look at some oftbe corrent meterdataand some advice on its inteal~xetation. We thank Joe RmD, Lyrme
TALLeY, Csrl WUNSCHand anonymous reviewers for their careful reading and helpful suggestions. This work w a s
supportedbythe NmiunalScienceFoundationthroughGrantOCE-8317389. Computationswereperfc~medonthe San
Diego Supercomputer Center's Cray-XMP computer.
8. REFERENCES
BAUMOARTNEIt,A. and E. RmCHBL(1975). The World Water Balance, Elsevier, New Yotk, 179 pp.
BROBCKER,W., T. TAKAHASHIand T. TAKAHASm(1985) Sources and flow patterns of deep-ocean waters as deduced
from potential temperature, salinity, and irfifial phosphate concenwation. Journal of GeophysicaIResearch,
90, 6925-6939.
BRYDEN,H., D. ROI~MMICHarid J. CHug CH(1989) Ocean heat u'ansport across 24 ° N in the Pacific. Deep-Sea Research,
accepted for publication.
COACHMAN,L., K. AAOAARD8rid 1~ TRIpp(1975) BeringStrait: The RegionalPhysical Oceanography, University of
Washington Press, Seattle, 172 pp.
Fu., L. (1981) The general circulation and meridioual heat wansport of the subtropical South Atlantic determined by
inverse methods. Journal of Physical Oceanography, 11, 1171-1193.
Fu, L. (1986) Mass, heat, and freshwater fluxes in the South Indian Ocean. Journal of Physical Oceanography, 16,
1683-1693.
~.IKASAWA,M. alid T. TBRAMOTO(1986) Characteristics of deep currents off Cape Shiono-misaki before and after
204
D. ROEMMICHand T. McCALLISTER
formation of the large meander of the Kuroshio in (1981 ). Journal of the Oceanographical Society of Japan,
42, 53-68.
FUKASAWA,M., T. TERAMOTOand K. TAmA (1986) Abyssal current along northexn periphery of Shikoku Basin. Journal
of the Oceanographical Society of Japan, 42, 459-472.
I-IELLERMAN,S. and M. RosENsrem (1983) Normal monthly wind stress over the world ocean with error estimates.
Journal of Physical Oceanography, 13, 1093-1104.
Hu, J. and P. NIILER (1987) NEPAC current meter and XBT data for the circulation in northeast Pacific thermocline
42°N and 28"N, 152" W, July 1982 - October 1985. Scripps Institution of Oceanography Ref. No. 87-4
(unpublished data report).
JoxcE, T. (1987) Hydrographic sections across the Kuroshio extension at 165 ° E and 175° W. Deep-Sea Research, 34,
1331-1352.
JOYCE, T., B. WARRBNand L. TALLEY(1986) The geothermal heating of the abyssal subarctic Pacific Ocean. DeepSea Research, 33, 1003-1015.
KBNYON,K. (1983) Sections along 35 ° N in the Pacific. Deep-Sea Research, 30, 349-369.
MANrYLA,A. and J. REID(1983) Abyssal characteristics of the world ocean waters. Deep-Sea Research, 30, 805-833.
MARTIN,M~,L. TALLEYand R. D13SZOEKE(1987) Physical, Chemical, and CTD Data from the Marathon II Expedition.
Oregon State University Reference No. 87-15, (unpublished data report).
MASUZAWA,J. and K. NAGASAKA(1975) The 137" E oceanographic section. Journal of Marine Research, 33 supp.,
109-116.
NULER, P., W. SCHMITZand D. LEE (1985) Geostrophic volume transport in high eddy-energy areas of the Kuroshio
Extension and Gulf Stream. Journal of Physical Oceanography, 15, 825-843.
REID, J. (1961) On the geostrophic flow at the surface of the Pacific Ocean with respect to the 1000-decibar surface.
Tellus, 13, 489-502.
REIn, J. (1986) On the total geostrophic circulation of the South Pacific Ocean: flow patterns, tracers, and transports.
Progress in Oceanography, 16, 1-61.
REXD,J. and R. ARMOR(1975) Interpretation of maps of geopotential anomaly for the deep Pacific Ocean. Journal of
Marine Research, 33 supp., 37-52.
REID, J. and R. LYNN(1971) On the influence of the Norwegian-Greenland and Weddell Seas upon the bottom waters
of the Indian and Pacific Oceans. Deep-Sea Research, 18, 1063-1088.
ROEMMICH,D. (1981) Cirenlation of the Caribbean Sea: a weB-resolved inverse problem. Journal of Geophysical
Research, 86, 7993-8005.
ROEMMICH, D. (1983) Optimal estimation of hydrographic station data and derived fields. Journal of Physical
Oceanography, 13, 1544-1549.
ROEMMICH, D. (1986) Water mass and heat transport in the North Pacific Ocean. Eos, Transactions, American
Geophysical Union, 67, 1013 (abstract only).
ROnMMICn, D. and T. McCALLISr~R(1988) A transpacific hydrographic section along 24 ° N. in preparation.
ROEMIdlCH,D. and C. WUNSCH(1985) Two transatlantic sections: meridional circulation and heat flux in the subtropical
North Atlantic Ocean. Deep-Sea Research, 32,619-664.
SCHLITZErt,IL (1988) Modeling the nutrient and carbon cycles of the North Atlantic, 1, circulation, mixing coefficients,
and heat fluxes. Journal of Geophysical Research, 93, 10699-10723.
SCHMITZ,W. (1987) Observations of new, large, stable abyssal currents at midlatitudes along 165 ° E. Journal of Physical
Oceanography, 17, 1309-1315.
SCHMITZ,W. (1988) Exploration of the eddy field in the mid-latitude North Pacific. Journal of Physical Oceanography,
(in press).
SCRIPPSINSTITUTIONOFOCEANOGRAPHY,Univorsity of California (1978) Physical, Chemical, and Biological Data,
INDOPAC Expedition, Legs I, II, III, VII, VHI, XV, XVI. SIO Reference 78,21 (unpublished data report).
STOMMEL,H. and A. ARONS(1960) On the abyssal circulation of the world ocean - I. Stationary flow patterns on a sphere.
Deep-Sea Research, 6, 140-154.
TALLEY, L. (1984) Meridional heat transport in the Pacific Ocean. Journal of Physical Oceanography, 14, 231-241.
TALLEY,L. and T. JOYCE(1988) A transpacific section along 47 ° N. (in preparation).
UNESCO (1981) Tenth report of the joint panel on oceanographic tables and standards. UNESCO Technical Papers
in Marine Science No. 36, UNESCO, Paris.
WARREN,B. (1981) Deep circulation of the World Ocean. In: Evolution of Physical Oceanography B. WARRENand C.
WUNSCH,editors. The MIT Press, Cambridge, Mass. and London, England, 6-41.
WARREN,B. and W. B. OVENS(1985) Some preliminary results concerning deep northern-boundary currents in the
North Pacific. Progress in Oceanography, 14, 537-551.
WORrmNGTON, L. (1981) Water Masses of the World Ocean: a fine-scale census. In: Evolution of Physical
Oceanography, B. WARRENand C. WUNSCH,editors. The MIT Press, Cambridge, Mass., and London,
England, 42-69.
WONSCH, C. (1978) The North Atlantic general circulation west of 50°W determined by inverse methods. Reviews of
Geophysics and Space Physics, 16,583-620.
WUNSCH, C. and B. GRANT (1982) Towards the general circulation of the North Atlantic Ocean. Progress in
Oceanography, 11, 1-59.
WUNSCH,C., D. Hu and B. GRANT(1983) Mass, heat, salt, and nutrient fluxes in the South Pacific Ocean. Journal of
Physical Oceanography, 13, 725-753.