pdf reprint - Physical Oceanography Research Division

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 99, NO. C1, PAGES 963-979, JANUARY 15, 1994
Surface layer variations observed in multiyear time series
measurements from the western equatorial Pacific
Janet Sprintall and Michael J. McPhaden
NOAA Pacific Marine Environmental Laboratory, Seattle, Washington
Abstract. Mooring measurements at 0 ø, 165øE, for the period November 1988 to
August 1991 indicate that surface layer structure was characterized by two distinct
climatic regimes associated with dramatic differences in large-scale atmospheric and
oceanic conditions. La Nifia conditions existed from November 1988 to November
1989, during which time the easterly trades were strong, Ekman divergence and
upwelling were pronounced, surface velocity was strongly westward, and rainfall was
low. The surface layer was cold, salty and well mixed down to 100-m depth, with
density variations controlled primarily by temperature. In contrast, from November
1989 to August 1991, the zonal winds were on average westerly and punctuated by
frequent westerly wind bursts, the surface currents reversed and flowed eastward in
the upper 50 m, and rainfall was high. Compared to the La Nifia period, the surface
layer was warmer and fresher, and the density mixed layer was shallower than the
isothermal layer owing to the presence of a 30-m-thick mean halocline (or barrier layer)
between 55- and 85-m depth. Moreover, density variations in the mixed layer were
determined primarily by salinity. During the November 1988 to November 1989 La
Nifia period, variability in sea surface temperature was influenced by local upwelling
and zonal advection. However, during November 1989 to August 1991, the presence of
the barrier layer effectively prevented the entrainment of cooler, saltier water from the
thermocline into the surface layer. Local air-sea heat fluxes were therefore more likely
to be prominent in the surface layer temperature balance. The barrier layer thickness,
which varied with a dominant time scale of 12-25 days, appears to have been affected
by variations in zonal advection of low-salinity water past the mooring.
1. Introduction
The world's highest open ocean sea surface temperatures
(SST) are found in the western equatorial Pacific. Deep
atmospheric convection is associated with the "warm pool,"
producing a corresponding maximum in global oceanic rainfall.
The nature of the complex air-sea interactions that
occur in the region because of these features is important for
the understanding and prediction of global climate variability.
This has been a significant focus of attention during the
decade-long Tropical Ocean Global Atmosphere (TOGA)
program and in particular of the international TOGA Coupled
Ocean Atmosphere Response Experiment undertaken
during 1991-1994 [Webster and Lukas, 1992].
The purpose of this paper is to examine the variability in
surface layer hydrography of the western equatorial Pacific
on daily to interannual time scales for the period November
1988 to August 1991. The emphasis is on moored time series
measurements from 0 ø, 165øE. The mooring is part of the
TOGA observing array and is located in the warm-pool
region.
The upper ocean structure in the western equatorial
Pacific is particularly interesting, as previous observations
have revealed substantial salinity stratification resulting in a
shallower mixed layer occurring within a deeper, nearly
isothermal layer [Lukas and Lindstrom, 1991; Delcroix et
This paper is not subject to U.S. copyfight. Published in 1994 by the
American Geophysical Union.
Paper number 93JC02809.
al., 1992]. This may be a persistent feature of the general
hydrography of the region, as demonstrated in the analysis
of historical hydrographic data by Sprintall and Tomczak
[1992]. The difference between the mixed-layer depth and
the isothermal-layer depth, referred to as the barrier layer,
may be important in limiting communication between the
ocean surface and the thermocline. The generation of shallow
haloclines and associated barrier layers has been attributed
to a localized heavy rainfall in the warm-pool region
[Lukas and Lindstrom, 1991]. Those authors also suggested
that the barrier layer may be maintained in part by the
westward subduction of saltier water from the central Pa-
cific. Conversely, Roemmich et al. [1993] suggest that barrier
layers can be created as fresher surface water from the
west flows eastward over the central Pacific water in an
equatorial surface jet. Numerous authors have speculated
that the presence of the barrier layer in the western equatorial
Pacific may be important for the development of E1
Nifio-Southern Oscillation (ENSO). However, the precise
role of the barrier layer in affecting ocean-atmosphere interactions
over the warm pool has yet to be clearly determined.
The mooring at 0 ø, 165øE, is unique in that it has been
measuring salinity since November 1988. The availability of
the salinity time series data allows us to address the issues
regarding the role of salinity in controlling surface layer
buoyancy and the potential of the salt-stratified barrier layer
in affecting air-sea interaction. We are interested in the
following specific questions related to upper ocean dynamics
and thermodynamics of the warm pool region. (1) What are
963

964 SPRINTALL AND MCPHADEN: WESTERN EQUATORIAL PACIFIC SURFACE LAYER VARIATIONS
I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I
0 ø 165øE
- løS 167øE
• -10
c
o
::•: .
,i
..
' , ,i:_.'! . ::i,ll ,. ! ,
- i
i'
ili¾i? ....
I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I
NDJFHRHJJRSONDJFHRHJJRSONDJFHRHJJR
1989 1990 1991
Figure 1. Daily averages of the zonal wind component from the 0 ø, 165øE mooring (solid curve) and
Nauru Island at 0ø32'S, 166ø54'E (dotted curve) for the period November 1988 to August 1991.
the relative importances of temperature and salinity in
determining surface layer density at 0 ø, 165øE? (2) Do the
relative importances of temperature and salinity vary with
time? (3) How do upper ocean temperature and salinity vary
in response to local atmospheric forcing? (4) How prominent/persistent
is the barrier layer at 0 ø, 165øE? In particular,
what is its thickness, time scale, relation to surface forcing,
and effect on SST?
In section 2 we briefly describe the mooring measurements
and additional data set• used in the study. In section 3 we
show that our study period encompassed two distinct climatic
regimes in the western equatorial Pacific. The first
regime (November 1988 to November 1989) coincided with a
La Nifia and was characterized by unusually high Southern
Oscillation Index (SOI) values, cold SSTs, high surface layer
salinities, strong easterly trade winds, and low precipitation
in the western equatorial Pacific. The second regime (November
1989 to August 1991) was characterized by low SOI
values, high SSTs, low surface layer salinities, westerly
surface winds, and high precipitation. The surface layer
hydrography is discussed in section 4, first in regard to the
relationship between temperature and salinity and their
relative control of the surface layer density structure within
the two climatic regimes and second in regard to the time
evolution of the barrier layer. Variability in upper ocean
thermodynamics in response to surface wind forcing and
current fluctuations are analyzed in section 5. We conclude
in section 6 with a general discussion of the air-sea interaction
scenarios which are likely to have existed in the western
equatorial Pacific during the two different climatic regimes.
2. The Data
2.1. The Mooring Data
The primary data used in this study are moored time series
measurements of currents, temperature, salinity, and wind
collected at 0 ø, 165øE for the period November 1988 to
August 1991.
Wind velocity was measured on the surface toroid at a
height of 4 m above mean sea level by a vector averaging
wind recorder (VAWR) and also by an Argos meteorological
platform (AMP). The VAWR recorded vector-averaged
wind components, air temperature, and SST (1-m depth) at
15-min intervals, while the AMP recorded the same parameters
at 2-hour intervals and telemetered them to shore via
the Argos system. Additional wind data were available from
the Nauru Island monitoring station at 0ø32'S, 166ø54'E,
where an R.M. Young model 05103 propeller and vane
anemometer were mounted on a tower 10 m above the
ground. Data were vector averaged for 40 min of each hour,
and three individual hourly samples were transmitted to
shore via the GOES satellite. The zonal wind components
from both the mooring and the Nauru Island station are
shown for the entire record in Figure 1. There is a high
correlation (r = 0.87) between the daily time series of zonal
winds at the two locations for the 1988 to 1991 time period.
Current velocity was measured using E.G.&G. model 610
vector-averaging current meters (VACMs) at 50-m intervals
between a depth of 50 and 300 m. Additionally, an EG&G
model 630 vector measuring current meter (VMCM) recorded
velocity at a 10-m depth. The instruments internally
recorded vector-averaged velocity components at 15-min
intervals (VACMs) and 2-hour intervals (VMCM). The
depths of the current meters are indicated in the zonal
velocity time-depth section of Figure 2. Overall velocity data
return for the period November 1988 to August 1991 was
95%. Data return at individual depths is summarized by
McCarty and McPhaden [1993].
Subsurface temperature was measured by the current
meters and by SeaData temperature recorders (TDR-2s) at
30, 75, 125, 175, and 225 m. Measurements from TDR-2s
were spot samples taken every 30 min rather than the 15-min
averages of the thermistors associated with the VACMs.
In November 1988 the 0 ø, 165øE mooring was instrumented
with Seabird SEACATs (model SBE-16) located at
depths of 11, 51, 101, and 201 m. SEACATs record both
temperature and conductivity and, like the TDR-2s, take
spot samples every 30 min. Salinity data were obtained using
the standard algorithm which is a function of temperature
and conductivity. Additional SEACATs were deployed be-

_
_
_
_
_
_
_
_
_
_
_
_
_
_
1oo.,•
SPRINTALL AND MCPHADEN: WESTERN EQUATORIAL PACIFIC SURFACE LAYER VARIATIONS 965
i i i i i i i i i i i i [ i i i i i i i i i i i ! i i i i
3,00 , . , , . , , , , , , . , , , , . , . , .
DJFMAMJ JASONDJFMAM JASONDJFMA'M'J
1989 1990 1991
Pour le D6veloppement en Coop6ration) group based in
Figure 2. Contours of zonal current component in centi- Noumea, New Caledonia. In all cases, only those CTDs
meters per second from the mooring record at 0 ø, 165øE, for taken within 60 min in time and within 5 nautical miles (1 n.
the period November 1988 to August 1991. Data have been mi = 1852 m) of the mooring were included in the calibration
low passed using a 51-day Hanning filter. Contour interval is analysis. Differences between CTD and SEACAT measure-
20 cm s -], with the zero contour enhanced, and shading
ments may be expected because of internal wave variability
indicates westward (negative) flow. Instrument depths are
shown by solid squares on the right axis.
and the spot sampling nature of the SEACAT instrument
ginning November 1989 at 3-, 30-, and 75-m depths and also
beginning May 1991 at 151 m. The 30- and 75-m SEACATs
replaced the SeaData temperature recorders during this
period. Figure 3 shows the time series of temperature and
salinity in the upper 201 m at the 0 ø, 165øE mooring.
Further details of the instrumental accuracies and of
acquisition and processing of the 0 ø, 165øE wind, current,
and temperature data are dealt with by Feng et al. [1991].
The performance of moored SEACATs in the central equatorial
Pacific at 0 ø, 140øW is discussed by McPhaden et al.
[1990]. For our study the accuracy and calibration of the 0 ø,
165øE SEACAT data is evaluated through comparison with
in situ conductivity-temperature-depth (CTD) data. Generally,
CTDs are taken at the time of recovery and deployment
of the mooring. In addition, there are also periodic hydrographic
transects along 165øE as part of the U.S.-People's
Republic of China (PRC) bilateral air-sea interaction program
and semiannual SURTROPAC cruises along 165øE by
the ORSTOM (Institut Fran•ais de Recherche, Scientifique
[McPhaden et al., 1990]. We stratified CTD/SEACAT pairs
by time since mooring deployment. For nine pairs that fell in
the 6- to 9-month range (the longest mooring deployments),
the rms difference is 0.023 practical salinity units (psu). This
compares favorably with the manufacturer's specification of
about 0.060 psu over 6 months of deployment and is also
comparable to accuracies determined by McPhaden et al.
[1990] for 1 year of data at 0 ø, 140øW.
30-
28
30-
28-
30-
28-
11m
b
35.1 -
34.5 -
35.1 -
34.5 -
'--' 30-
o
r.._) 9.8-
[,1 30-
CK 28-
ß --) _
CIZ 30-
CK 28-
[,1 _
•- 28-
L,_I 26-
26-
22-
26-
22-
35.1 -
34.5 -
CO -
.--. 35.1 -
>-, 34.5 -
,_, 35.1 -
(]Z 34.5 -
35.1 -
34.5 -
24-
20-
35. I -
34.5 -
151 m
22-
18-
N' IJ ''''''''$' M M J "N' I J '''''"'''"" M M J 5 N I j' •M''''' M J
1989 t 990 t 99 t
35.1 -
34.5 -
201 m
' I''''M''' 'S'N' I'' ''''d'''''
I d'''''''
N J M J J M M $ N M M J
1989 1990 1991
Figure 3. (a) Daily averaged temperature from the SEACAT and SeaData instruments and (b) salinity
from the Seabird SEACAT instruments at the mooring at 0 ø, 165øE, for the period November 1988 to
August 1991. Depths of the SEACAT sensors are indicated.

966 SPRINTALL AND MCPHADEN' WESTERN EQUATORIAL PACIFIC SURFACE LAYER VARIATIONS
Table 1. Data Returned by SEACAT Instruments
at the 0 ¸, 165øE Mooring
Data Returned, %*
Depth,
m Period of Deployment Terriperature Salinity
3
11
30
51
75
101
151
201
Total
Nov. 18, 1989, to Apr. 8, 1991 79 79
Nov. 15, 1988, to Apr. 8, 1991 96 83
Nov. 18, 1989, to Apr. 8, 1991 100 100
Nov. 15, 1988, to Apr. 8, 1991 100 100
Nov. 18, 1989, to Apr. 8, 1991 79 79
Nov. 15, 1989, to Apr. 8, 1991 100 100
Mar. 29, 1991, to Apr. 8, 1991 100 100
Nov. 15, 1988, to Apr. 8, 1991• 100 100
94 92
*Percentage of daily averaged temperature and conductivity
(salinity) data available.
?SEACAT instrument at 201 m was removed on November
15, 1989, and redeployed on March 29, 1991.
Percentage data return for the temperature and salinity
associated with the depth of each SEACAT sensor is shown
in Table 1. For the most part the time series appear to be
complete; however, data gaps are found in the SEACAT
temperature and salinity records at 3 m for February 17,
1990, to June 30, 1990; at 11 m for May 22, 1990, to June 30,
1990; and at 75 m for November 19, 1990, to March 29, 1991.
A data gap also occurred in the SEACAT salinity record at
11 m for March 28, 1991, to August 4, 1991. For the nearly 3
year record, correlation between daily 3-m (1 l-m) temperature
data and those at 30 m is 0.89 (0.96), and for salinity it
is 0.93 (0.98). The correlation between temperature (salinity)
data at 3 and 11 m is 0.96 (0.98). Hence the data gaps were
filled for the 3-m record by a linear regression formula based
on the 11-m record when available, and when this record was
not available, then by linear regression with the 30-m record.
Similarly, correlation between the daily 75-m record of
temperature (salinity) with data at 51 m was 0.65 (0.90), so
data gaps at 75 m were filled using the resultant linear
regression formula. Gaps due to mooring replacement (generally
of only 1- or 2-day duration) were filled by linear
interpolation in time. In the following, only the calculations
of mixed-layer depth and isothermal depth required the use
of the interpolated time series. The interpolated time series
in this case provided data segments of maximum length and
to a depth sufficient for determination of these parameters.
For this analysis all variables were processed to daily
averages. In the case of density, mixed-layer and isothermallayer
depths, wind speed, wind stress, and wind work,
computations were made using the higher frequency data
before averaging to the daily means.
2.2. Additional Data Sets
Any study involving salinity variability of the surface layer
in the rain-drenched western equatorial Pacific should necessarily
include some estimate of the local precipitation.
Here we will use a pentad (5-day-averaged) oceanic rainfall
estimate at 0 ¸, 165øE, derived from the Global Precipitation
Climatology Project (GPCP) [Arkin and Ardunay, 1989;
Janowiak and Arkin, 1991]. These estimates are based on
infrared measurements of cloud top temperatures from polar
orbiting and geostationary satellites. The method of rainfall
rate estimation is based on a high correlation between
observed rainfall and the fractional coverage of "cold"
cloud tops with effective blackbody temperatures of
15
lO
5
o t"' .... ,.,,,.,,,.,,,/'•"""'-'""""'""V""
'.....
%"'",,,I'•,,,,,' "-' ' --5 •1 it
O JFM FIM J J FI S 0 N O JFM FI M J J FI S 0 N O JF'M FI M J J FI
1989 1990 1991
Figure 4. Time series of pentad GPCP rainfall estimates and evaporation rate. Evaporation rate was
calculated using the formula E = (pacœ/Po)lUIAq with P0 as the mean water density and using mooring
wind spee data, an air-sea specific humidity difference of Aq = 5 g kg -• , an exchange coefficient of cE
-- 1.2 x 10 -3 , and an air density of pa - 1 ß 2 kg m -3

SPRINTALL AND MCPHADEN: WESTERN EQUATORIAL PACIFIC SURFACE LAYER VARIATIONS 967
time series. It is evident that in magnitude, precipitation
generally exceeds evaporation, particularly from November
1989 to August 1991.
CTD data from four U.S.-PRC bilateral cruises and five
French SURTROPAC cruises along the 165øE meridian
collected from November 1988 to August 1991 will be used
to describe the hydrodynamic characteristics and circulation
around the mooring. For the U.S.-PRC cruises, casts were
made every 0.5 ø longitude equatorward of 3øN and 3øS, with
station spacing of about 1 o latitude poleward of 3 ø. The casts
were generally made to at least 1000 dbar and mostly to
within 100-200 m of the ocean bottom. Calibrated data were
processed to a 2-dbar depth interval. For the SURTROPAC
cruises, station spacings were approximately 0.5 ø latitude
between 5øN and 5øS and 1 ø poleward of these latitudes.
Again, most of these casts were made to ---1000 dbar,
although in some cases the equatorial SEACAT calibration
casts were shallower. Calibrated data were processed to a
5-dbar vertical resolution. Further details of the acquisition,
calibration, and processing of these data are found in work
by Lake et al. [1991] for the U.S.-PRC cruises and in work
by Henin et al. [ 1989], Dandonneau et al. [ 1990], du Penhoat
et al. [1990], Delcroix et al. [1991], and Rual et al. [1991] for
the SURTROPAC cruises.
Finally, to characterize the regional-scale wind field variability
in the western Pacific, we utilize the Florida State
University pseudostress analyses [Goldenberg and O'Brien,
1981].
3. Two Climatic States in the Western
Equatorial Pacific
Over the 3-year record of this study, there appear to be
two distinct regimes in the western equatorial Pacific distinguishable
by dramatic differences in both atmospheric and
oceanic conditions.
3.1. November 1988 to November 1989
For the period February 1988 to November 1989, climatic
records show that the SOI of surface pressure distribution in
the tropical Pacific was positive, with the highest values
since late 1975 registered in November 1988 (1.9) [Climate
Analysis Center, 1991]. These high values are indicative of
La Nifia conditions prevailing in the tropical Pacific throughout
much of 1988 and 1989. After November 1988, the SOI
declined, crossing zero around November 1989. Thus the
first year of our time series (November 1988 to November
1989) coincides approximately with La Nifia conditions in
the equatorial Pacific.
The Florida State University zonal wind stress (Figure 5a)
indicates the presence of strong easterly trade winds in the
equatorial Pacific from November 1988 to November 1989.
Averaged over the region shown in Figure 5a, the trades are
O (0.02 N m -2) stronger than the annual climatology given
by $tricherz et al. [1992] in their atlas. Also, easterly winds
at 0 ø, 165øE are 3.1 m s- • stronger than normal compared to
the annual mean mooring climatology of McCarty and
McPhaden [ 1993]. The GPCP precipitation estimates for the
coinciding period (Figure 5a) show that the Pacific equatorial
band east of 160øE received on average lOO.
20ON I ............ ]................
I .............. !............... I ............... !...............
I... I. ............ ! ....
...... '• '*-- "•"""'"'---••'•"•'-•-'••:-
- ------- .,,•ii•i:-•!•:'*-":-":.•i:: :•;•7:i"•- :":.•=:!•i•i•:--':':'/.."x•""'"""""
-"-
- ß: "'•""•"'•:'•:•:- ...,,•,•. :'••:.-:-i•::+.-:• , o •'""••'" ............ *'•'•••• •.-'.-.:i ø-.-': '--'•i!•=.•$•:':. :.:..' :•i•ii
•'"'"""• '•'••!iii!•..-;'."'•
- '-"'•-'
................. - •'
...................... ••.•---.-•'--•:-:.••••••• x:
•0o•-
'•-'"'"'•""'"'"'"•'"'•'•:•::•:'•'•:i;'-;."•::•i•'--"':'•:-•
*•'"'"'••••'••-•i•i:!'i•
."•...........
,............... .5•,••*'/.:•'•.'•..............
i•..:• •oo •
• ........................................................................ • !=
10øS •-%"•..':".•="" • '--- '-•"................................
•'•:'""'"•' '"':':':•'":' '""•• •
100 q•
20øs • i
:•.....................
"¾'
'•'"'•i:"'*•"•**i'•:'"'•'• ............... •?"'"' -'••••. "•':':''•'• •'*•'•"'
• :•4
145øE
155øE 165øE 175øE 175øW 165øW
LONGITUDE
DEC 89 - AUG 91. > lOO.
20ON I I I I I I I I I
:•:. '--•- '-..• "•-•!•'
:_:•:::.-• •iii:: :::'"• i ::i::•if:•'•::i::ii!i:::::-."'•!i::iiiii:::
'-'•:.•;i::::ii::•::':•:•:•!;::i::::!;:•-•::ii::i::i'-::'•
:-'
:•:-•!::ii::iiii!:: :-:-:- ======================================
•
- f•..•..•.•.•*.•.•`••:;•;•:•:•4•;4•.:.•.•.•.•::•:•::?:•F•:•::•f:•..j•::::•;•:..•::•:.:.•:.`•:•:•f.•.•.•:•P•:5:•::•:•.• ß
100N ...........................................
•'""'"'"'"'*'*'•""•'*'*•••••C•**:J:"•::•***:':':•:•*****'"'""*-'
'•'•'"•'"•':':•:••2•*::•yy• 300 •
10øS
o
.. ::efi•...-•---..:-•.•:•:•:5:•:•:•:5:•:•:•:•
lOO •
::,•:;:.. .-*-.: . .• •
20 oS ':•::•'• "•½•{•½:•:•
'•
'• ...........................
i
•
t '• o •
145•E 155•E 165•E 175•E 175•W 165•W
LONGITUDE
Figure 5. Pseudo wind stress (m 2 s -2) from Florida State
University (FSU) climatology and precipitation estimates
from the GPCP climatology for the tropical Pacific, averaged
for the periods (a) November 1988 to November 1989 and (b)
December 1989 to August 1991. Length of the 100 m 2 s -2
vector is shown in the top right corner; contour interval for
shaded rainfall estimates is 100 mm month -• .
composite study of a 109-year analysis period of precipitation
patterns associated with the high index phase of the SOI
[Ropelewski and Halpert, 1989], the equatorial region at
165øE is shown to lie in a rainfall deficit zone. Further to the
west, a high SOI is associated with wetter conditions.
Average meridional CTD sections of temperature, salinity,
and density along 165øE have been constructed between
10øN and 20øS from six cruise transects made between
November 1988 and November 1989 (Figures 6a-6c). The
salinity field (Figure 6a) shows strong meridional gradients
located in the vicinity of the North Equatorial Countercurrent
(NECC) at ---3øN to 5øN and again at 3ø-7øS. North and
south of these respective bands, relatively fresh (•34.6-psu)
water is found, probably in response to the rainfall in the
convergence zones (Figure 5a). The southern hemisphere
subtropical salinity-maximum water, here shown as salinity
of >35 psu, occupies a large volume of the southern latitudes
at depth, surfacing in the equatorial regions (3øN-3øS) between
the fresher water found north and south. On the
surface at the equator, salinity is ---35.3 psu. The meridional
temperature field (Figure 6b) shows only two small patches
of water of >29øC centered at 7øN and 7øS. The meridional
temperature, salinity, and density (Figure 6c) sections illustrate
the upward sloping isolines toward the equator that are
characteristic of Ekman divergence and equatorial upwelling.
The spreading of the equatorial pycnocline and
thermocline between 2øN and 2øS is a consequence of the
geostrophically balanced Equatorial Undercurrent (EUC).
Means of the 0 ø, 165øE mooring data for the period
November 1988 to November 1989 (Figure 7) show an
approximately isothermal (28.1øC) and isohaline (35.3 psu)

968 SPRINTALL AND MCPHADEN: WESTERN EQUATORIAL PACIFIC SURFACE LAYER VARIATIONS
8 Nov 88 - 15 Nov 89
D 5 Dec 89 - 10Aug 91
I , I...I / /. I /
0. E.• L '•, ,'-'-__--¾
1.1[,,• l•,t•,.• •,• 0
100.
100 m
200.
300.
400.
500.
20øS 15øS 10øS 5øS 0 ø 5øN 10øN
200 ua
300 u)
400 a_
I I I I I I I
500
20øS15øS 10øS 5øS 0 ø 5øN 10øN
B
I I I I
lOO
200
300
400
500
20øS 15øS 10øS 5øS 0 ø 5øN 10øN
100 m
200 •a
300 u)
400 a_
500
20øS 15øS 10øS 5øS O" 5øN 10øN
c
! I 1_ I 0 0
^ I - 100 100•
E •'"""'•o• 200 200 •
• - ' 300 300 •
< o _ - - 400 400 •
i i i 500 500
20øS 15øS 100S 5øS 0 ø 5ON 10øN
LATITUDE
20øS 15øS 100S 5øS 0 ø 5ON 10øN
LATITUDE
Figure 6. (a) Salinity, (b) temperature, and (c) density along 165øE calculated from CTD data during
November 1988 to November 1989 and (a) salinity, (b) temperature, and (c) density and during November
1989 to August 1991. For salinity the contour interval is 0.1 psu, with salinities of >35 psu shaded and
those of 29øC shaded. Contour
interval for density is 0.2 kg m -3, with values of

SPRINTALL AND MCPHADEN: WESTERN EQUATORIAL PACIFIC SURFACE LAYER VARIATIONS 969
o
SFIL I N I T¾ TEMPERFITURE S I GMFI-T
I __ I I I I I ,-... I • I
ZONAL
VELOC I T¾
I I I . I I
50
E
-i-
1oo
15o
b_] 200
250
300
3•4
(psu)
I I I
(øC)
•IERN
(kg m -a)
I I I
-50-25 0 25 50
(cm
-I
s )
I I I I I I
50-
• 1oo -
n- 15o -
bJ 200 -
250 -
300
0.1 0.3 0.8 1.2 1.8 •.1 •.4 0•.7
I
5 10 15 20 25 30
$TRNDRRD
DEV I RT I ON
Figure 7. Means and standard deviations of salinity, temperature, density, and zonal velocity from the
daily averaged mooring data at 0 ø, 165øE for the periods November 14, 1988, to November 15, 1989, (open
circles) and November 17, 1989, to August 5, 1991 (asterisks). The symbols indicate the depth of each
sensor used in the calculations.
ents found in the NECC region during La Nifia that separate
the fresh water from the salinity-maximum water are now
located just south of the equator between 1 ø and 2øS. Unlike
water during November 1988 to November 1989, no water at
the surface between 5øN and 5øS is more saline than 35.0
psu. Surface salinity at the equator is reduced to 34.5 psu,
and strong vertical salinity gradients occur between 50 and
100 m. In the CTD meridional section of temperature (Figure
6e), the 29øC warm pool spreads from 7øN to 16øS and is well
mixed to a depth of-100 m on the equator. There is no
suggestion of Ekman divergence and upwelling in the shallow
isopycnals near the equator (Figure 6f), which is
consistent with the occurrence of westerly winds in the
western equatorial Pacific during this period.
Zonal wind velocity from Nauru (Figure 1) for November
1989 to August 1991 is 0.4 m s -1 on average, and the record
is frequently punctuated by westerly wind bursts with
speeds of up to nearly 10 m s -1. In response to these
westerlies, the zonal velocity in the surface layer frequently
reverses and becomes eastward (Figure 2). The GPCPderived
rainfall for the period (Figure 5b) estimates that
between 200 and 300 mm of precipitation occurred on
average each month. The SOI was mostly negative (typically
between 0 and -1 for 5-month running means) during
November 1989 to August 1991 [Climate Analysis Center,
1991]. The low SOI-high rainfall relationship is consistent
with that found by Ropelewski and Halpert [1987] for the
equatorial region around 165øE during ENSO events.
For the period November 1989 to August 1991, the mooring
means (Figure 7) show a warm (29.6øC), fresh (34.2-psu)
surface layer that is isothermal and isohaline to different
depths. This discrepancy between the shallower isohaline
and the deeper isothermal layer will be dealt with in more
detail in section 4.2. Compared to the mean temperature
profile of the La Nifia period, the isothermal layer now
appears to be shallower. Once again, most of the temperature
variability occurs in the thermocline region. However,
the maximum standard deviation of 2.4øC is located at 125-m
depth, which is consistent with the shallower isothermal
layer. Salinity, too, shows much greater variability in the
surface layer than during the La Nifia period. A maximum
standard deviation of 0.4 psu occurs at 101 m before falling
to around the same means and the same variabilities at 151
and 201 m as were found in the preceding period. As a result,
most variability in the density profile (0.75 kg m -3) is also
found at 101 m.
Associated with westerly wind forcing, mean eastward
flow is found in the surface layer (for example, 8.6 cm s -• at
10-m depth), below which there is a mean westward counterflow
at 101 m. There does not seem to be much difference
in the strength of the EUC between the two periods (52 cm
s -• at 201 m during this second time period). Most of the
zonal velocity variability is found at the surface (standard
deviations of 34 and 31 cm s -• at 10 and 50 m, respectively),
also likely associated with the occurrence of westerly wind
bursts.

970 SPRINTALL AND MCPHADEN: WESTERN EQUATORIAL PACIFIC SURFACE LAYER VARIATIONS
a l,•-NOV-G8 'co IS-NOV-89
TœMPœRRTURœ V$ $1GHR-T V• $IGMR-T V•
SRL ! N I TY TœHPERRTURE SRL I N I TY BETR RLPHR
øt
t øJT I,' FøJ o,,,,,,, o , ,
I I I ! I I
i i i ! !
75 , i , 75 75 , , 75 75
IOO
i
,
i
, I00
t
, , 100 100
i ! i i I
125 ' ' 125 ' ' 125 ' ' 125 125
I I I
i i i
150 150 ' ' 150 ' ' 150
]i,l!xf '' ,,
I I ! I
175 175 , ' 175 175 175
i
i
2OO 200 ' ' 200 200 200
225. , ' , , . 225 ,' ', 225. , . , . 225. , • •', , .225
-1.0-0.5 0.0 0.5 1.0 -I.0-0.5 0.0 0.5 1.0 -I.0-0.5 0.0 0.5 1.0 -6 -4 -2 0 2 4 6 -1.0-0.5 0.0 0.5 1.0
XCORR XCORR XCORR REGRESSION SLOPE REGRESSION SLOPE
b 1S-NOV-09 to '•-RUG-91
50
TœMPERRTURE V$
SRLINITY
0 I I I. t
75
i
, ,
•oo
•;-5 , ,
2OO
,
SIGMR-T Ve
TEMPERRTURE
0 I • I ,• m 0
7s /il' ,, F 7s
-i½ ;1', !-
•øøJI,' I ', L •øø J ', I : JL
•=Sll ; I ',
•so-• : I ', I-
-II ; I ', 1-
/ •=s
200
225 , ', , ,' . 225 200
225. , , , .
-1.0-0.5 0.0 0.5 1.0 -I.0-0.5 0.0 0.5 1.0 -I.0-0.5 0.0 0.5 1.0
XCORR XCORR XCORR
5IGMR-T V.
5RL I N I TY BETR RLPHR
• ', ! ,' •d o ..... ' ' o ,
: I ', I!-
, , 25 25
, , 50 50
75 75
, , 100 : 100
'!
' 125 125
150
200 200
'
225. • , , , , , 225 ,
-6-4-2 0 2 a, 6 -1.0-0.5 0.0 0.5 l.O
REGRESSION SLOPE REGRESSION SLOPE
Figure 8. Cross correlation as a function of depth between daily averages of temperature and salinity,
temperature and density, and salinity and density for the periods (a) November 1988 to November 1989
and (b) November 1989 to August 1991. Estimates of/3 (the haline contraction coefficient) and a (the
thermal expansion coefficient) for both time periods are also shown. Open circles indicate depths of the
sensors used in the calculations. Dashed lines are 95% confidence intervals for the null hypothesis in the
cross-correlation plots and expected values of a and/3 in the regression slope plots.
Overall, conditions during December 1989 to August 1991
tended to be ENSO-like in the western equatorial Pacific.
However, ENSO conditions did not prevail during this time
across the entire basin. It was not until September 1991 that
substantial warming in the eastern Pacific began to develop,
marking the onset of the 1991-1993 ENSO [McPhaden,
1993].
4. Surface Layer Hydrography
4.1. The Density Equation
The temperature-salinity relationship during both climatic
regimes is quite variable in the upper 150 m. Cross correlations
between the two variables (Figure 8) are generally
either not significant or not consistent between the two time
periods. Only at 201-m depth is a consistent and significant
relationship manifested between temperature and salinity,
with a cross correlation of about 0.7 over the entire mooring
record.
We will examine the relative importance of temperature
and salinity in the density field by assuming a linearized
equation of state, that is,
P=P0 +/38S+aST+'" (1)
where/3 = 0 p/OS, a = 0 p/O T, and higher-order terms have
been neglected. In the limit where salinity controls the
density field, the cross correlation between/5S and/Sp (= p -
P0) would be close to unity (high salinity • high density).
The slope of the linear regression curve between the two
variables would be the haline contraction coefficient /3.
Conversely, in the limit where temperature controls the
density field, the cross correlation between/ST and/Sp would
be - 1 (low temperature • high density), and the slope of the
regression would be equal to the thermal expansion coefficient
a.
The orthogonal regressions for the detrended and demeaned
daily data of density and salinity and density and
temperature for the period November 1988 to November
1989 are shown as a function of depth in Figure 8a. While
there is significant correlation between density and salinity

SPRINTALL AND MCPHADEN: WESTERN EQUATORIAL PACIFIC SURFACE LAYER VARIATIONS 971
Table 2. Comparison of Mixed-Layer Depth, Isothermal-Layer Depth, and
Barrier Layer Thickness*
Thickness, m
rms
Difference,
m
Period Parameter CTD SEACAT Mean ms
Nov. 15, 1988, to Nov. 15, 1989
Nov. 16, 1989, to Aug. 4, 1991
isothermal layer 111.4 95.4 16.0 18.4
mixed layer 109.1 93.3 15.8 16.8
barrier layer 2.3 2.1 0.2 7.3
isothermal layer 90.1 92.1 -2.0 6.0
mixed layer 54.5 55.0 -0.5 4.5
barrier layer 35.7 37.1 - 1.4 7.2
*Results were calculated from nearly simultaneous CTD casts and data from SEACAT
sensors at 0 ø, 165øE, CTD casts were made within 5 nautical miles and 1 hour of the
SEACAT measurements for November 1988 to November 1989 (five casts) and November
1989 to August 1991 (six casts).
at all sensor depths in the water column, the slope of the
regression line is --•1.5-2.5 at the surface and -5.4 at 201-m
depth. The true value of/3 calculated from mean salinity and
temperature at the corresponding depths is --•0.75. For the
same time period, the correlation of density and temperature
at all depths in the water column (Figure 8a) is close to -1
and highly significant. In addition, the slope of the regression
coincides with the values of a (--• -0.32) calculated again
from the corresponding mean temperature and salinity values.
The conclusion to be drawn from this is that during the
cold La Nifia period, temperature is the controlling factor for
the density structure in the surface layer and in the thermocline
at 0 ø, 165øE.
For the period November 1989 to August 1991, Figure 8b
indicates correlation of density and salinity to be near unity
and highly significant for the upper layer to a depth of 101 m.
The slope of the regression closely follows the estimate of/3
to 75-m depth. On the other hand, correlation of density and
temperature (Figure 8b) is highly significant and near -1 for
depths between 101 and 201 m where the temperature/
density regression slope is similar to the expected value of a.
There is also a similarity of the regression slope and a at the
11- and 30-m depths, but the correlation of temperature and
density at these depths is not significant. Hence for the
period November 1989 to August 1991, salinity is the controlling
factor in the density structure of the upper layer,
while in the thermocline region the density structure is more
strongly influenced by temperature.
4.2. The Barrier Layer
During the period November 1989 to August 1991, a
systematic difference developed between the depth of the
isohaline and the isothermal layer, as is evident in the mean
profiles of temperature and salinity shown in Figure 7. This
implies the presence of a barrier layer whose existence in the
western Pacific warm pool has been demonstrated previously
in several publications mentioned in the introduction.
These studies suggest that the spatial extent of the barrier
layer in the equatorial Pacific may be quite significant.
However, little is known about the temporal persistence of
the barrier layer. By the nature of the high temporal resolution
of the salinity and temperature mooring data used in the
present study, it is possible to determine the temporal scales
of variability of the barrier layer phenomenon.
First, we must address the issue of whether the thickness
of the barrier layer can be accurately defined with the
vertical resolution of the mooring data at 0 ø, 165øE. This is
possible through a comparison of the SEACAT data with the
nearest available CTD casts in time and space. As with the
CTD-SEACAT calibration study outlined in section 2, the
differences in location and sampling of the CTD and
SEACAT data corresponded to generally

--
972 SPRINTALL AND MCPHADEN: WESTERN EQUATORIAL PACIFIC SURFACE LAYER VARIATIONS
I
I I I I I I I I I I I I I I I I I I I
I
• •o
• 12o
• HLxed La•e• .
I I I I ' I ! I I I I ! I I I I I
D J F H R H J J R S 0 N D J F H R H J J R
1989 1990 1991
I I I I I I I I ! I I I I I I I I I I I I
b_• •= 80
n-' r,1
n-' 0
1" '
I I I I I I I I I I I I I I I I I I I I I
D J F M R M J J R $ 0 N O J F M R M J J R
1989 1990 1991
Figure 9. Daily averaged values of the mixed-layer depth, isothermal-layer depth, and barrier layer
thickness calculated from the mooring data at 0 ø, 165øE for the period November 1989 to August 1991.
Contours are in meters. Corresponding values calculated from simultaneous CTD profiles are indicated by
diamonds for mixed-layer depth, asterisks for isothermal-layer depth, and crosses for barrier layer
thickness.
Nifia period (November 1988 to November 1989), the statistics
were calculated from simultaneous SEACAT-CTD data
for five different days; for the period November 1989 to
August 1991, the comparisons are based on data from six
different days.
For the La Nifia period, the mean mixed-layer depth
calculated using the CTD profiles is 109.1 m, and for the
corresponding SEACAT data it is 93.3 m, that is, a difference
of 15.8 m. The mean depth of the isothermal layer over
the same time period is 111.4 m for the CTD data and 95.4 m
for the SEACAT data, a difference of 16.0 m. These differences
arise because of relatively poor vertical resolution of
the SEACAT data during November 1988 to November
1989. However, both SEACAT and CTD measurements
indicate a barrier layer thickness (isothermal minus mixedlayer
thickness) of 2.1 and 2.3 m, respectively. These values
are close to the vertical resolution of the CTD data and are
not significant relative to the vertical resolution of the
SEACAT data. Hence it appears that under La Nifia conditions,
strong easterly trade winds and associated upwelling
are not conducive to the maintenance of the barrier layer at
0 ø, 165øE. Indeed, the result is not surprising, as it was
shown in section 4.1 that during November 1988 to November
1989, temperature is the controlling factor of the density
structure (and hence the mixed layer) at the mooring site.
For November 1989 to August 1991, the mean mixed-layer
depth for the CTD profiles is only 54.5 m, and for the
SEACAT data it is 55.0 m, a difference of only 0.5 m. During
the same period, the mean depth of the isothermal layer is
90.1 m for the CTD data and 92.1 m for the SEACAT data,
only a 2-m difference in calculations between the two data
sets. Hence for this time period, the mean thickness of the
barrier layer is 35.6 m using the CTD data and 37.1 m using
the SEACAT data. The CTD-SEACAT rms differences are
only 4.5 m for the mixed-layer depth, 6.0 m for the isothermal
depth, and 7.2 m for the barrier layer thickness. Since
the mean barrier layer thickness during this time period is
much greater than these rms differences, we are assured that
the existence of the barrier layer is not an artifact of the
vertical resolution of the SEACAT data. Hence calculations
of mixed-layer depth and isothermal depth using the
SEACAT data from the ENSO-like period produce a meaningful
measure of the barrier layer thickness.
The time series of mixed-layer depth, isothermal depth,
and the difference between the two (the barrier layer thickness)
calculated using the SEACAT data are shown for
November 1989 to August 1991 in Figure 9. For reference,
the corresponding CTD data for the same period used in the
comparison above are also shown in Figure 9. A scatterplot
(Figure 10) of the mixed-layer depth and isothermal layer
depth during the period succinctly illustrates the persistence
of the barrier layer. The mean mixed-layer depth is 54.5 m,
while the mean depth of the isothermal layer is 84.5 m. The
difference between the two depths (30 m) is the mean
thickness of the barrier layer, which can also be inferred
from mean profiles for November 1989 to August 1991

SPRINTALL AND MCPHADEN: WESTERN EQUATORIAL PACIFIC SURFACE LAYER VARIATIONS 973
160
os
AS
---I- 8t U' VS + w e •-= S(P - E)h -1 (4b)
where OT/Ot (OS/Ot) is the local time rate of change of
temperature (salinity), u is the horizontal velocity, VT (VS)
is the horizontal temperature (salinity) gradient, w e is the
vertical entrainment velocity through the base of the mixed
layer at depth z = -h, AT(AS) is the vertical temperature
(salinity) gradient across the base of the mixed layer, Qo is
surface heat flux, P and E denote precipitation and evaporation
rates, p is a constant water density, and C•, is the heat
capacity of water at constant pressure. In this form, the
conservation equations (4a) and (4b) express the local rate
of change of mixed-layer temperature and salinity (first term)
that result from variations in horizontal advection (second
term), vertical entrainment (third term), and surface fluxes
(fight-hand side). Terms that relate to such processes as
turbulent vertical diffusion through the base of the surface
layer have been neglected in (4a) and (4b), as they cannot
be adequately estimated from our data sets. The entrainment
velocity w e can be expressed as
Figure 10. Scatter diagram of daily averaged isothermal
depth versus daily average mixed-layer depth calculated
from mooring data at 0 ø, 165øE, for November 1989 to
August 1991.
(Figure 7). Standard deviations about the mean are 23.0 m
for the mixed-layer depth, 24.6 m for the isothermal layer,
and 20.9 m for barrier layer thickness.
From the autocorrelations of the detrended and demeaned
daily data, we can determine the integral time scales of the
mixed-layer depth, isothermal layer depth, and barrier layer
thickness. The integral time scale is a measure of the time
required to gain a new degree of freedom in the time series
[Davis, 1976]. It is effectively the same as the decorrelation
scale, which in turn can be thought of as one fourth of the
period of the dominant fluctuations. For the period November
1989 to August 1991, the integral time scales of the
mixed-layer depth, isothermal layer depth, and barrier layer
thickness were 15, 24, and 25 days, respectively. However,
calculation of the integral time scales was sensitive to
whether the large oscillation seen in Figure 9 at the beginning
of the time series (November 1989 to February 1990) was
included. When this part of the record was removed, estimates
of the integral time scale decreased to 14, 17, and 12
days for the mixed-layer depth, isothermal layer depth, and
barrier layer thickness, respectively. Thus, broadly speaking,
the dominant time scales in the time series shown in
Figure 9 range between 12 and 25 days. These scales are
consistent with a wide band of spectral energy at periods
between 10 and 100 days in variance-preserving spectra of
mixed-layer depth, isothermal-layer depth, and barrier layer
thickness. Some of this variability may be related to intraseasonal
wind oscillations associated with 30- to 60-day
Madden and Julian [1972] waves.
5. Evolution of Surface Layer Variability
The heat and salt balances of the surface layer are
OT
AT
--+ ot u. VT + We = Qo(pC•,h) -•
(4a)
Oh
We -- W_h q--- (5)
ot
where W-h is the vertical velocity at the base of the mixed
layer, and Oh/Ot is the change in mixed-layer depth
[McPhaden, 1982].
5.1. Surface Layer Response to Local Atmospheric Forcing
In this section we primarily examine the variations in
surface layer temperature and salinity in response to windforced
changes during the two climatic regimes. In section
5.2, variability due to lateral advection, represented by the
second term in (4), will be examined. McPhaden and Hayes
[1991] have discussed the parametric dependence of the heat
balance equation (4a) on surface wind forcing. In a general
sense the turbulent components of latent and sensible heat
flux are proportional to wind speed, Ekman-related vertical
velocity (W-h) is proportional to zonal pseudostress, and
wind work is related to entrainment mixing. Parallel relationships
between the wind constructs and terms in the surface
layer salt balance (4b) also apply. There can be salinity
variability related to the wind speed with its effect on
evaporation. Similarly, wind stress and wind work can result
in the vertical advection and entrainment of waters of
differing salinity into the surface layer. However, as with the
heat balance, direct wind-related variability is not the only
control on surface layer salinity. The salt balance is also
affected, for example, by precipitation, which can be intense
in the warm-pool region (for example, see Figure 5).
Figure 11 shows the cross correlations of the time rate of
change of temperature 0 T/O t and salinity 0 S/0 t (calculated
using centered differences) with the zonal wind pseudostress
> for November 1988 to November 1989 and November
1989 to August 1991. All the daily data used in the
cross-correlation analysis have been smoothed using an
l 1-day Harming filter. The filter removes periods shorter
than about 5 days and enhances the cross correlations with
the wind field. However, results do not qualitatively change
for calculations based on unsmoothed daily data or daily
data low passed with a 7-day Harming filter. Although not
presented, calculations of cross correlations between sur-

ß
974 SPRINTALL AND MCPHADEN: WESTERN EQUATORIAL PACIFIC SURFACE LAYER VARIATIONS
a
b
d•/dt Vs [UIU'"'
100
150
!
i
i
I
i
i
i
! i
200 ' '
250 •'
lOO
-I 0
15-N0V-88 to 14-N0V-89
i
i
i
i
I
dS/dr Ms lU[U'"'
50 ,
O0 ,
50 ' i I '
O0
i
i
I
i
I
i
i
I
i
i
i
'
50 •
-1 0 ]
XCOEE
18-NOV-89 to •,-RUG-91
dT/dt Vs ]LIp'"
0 I • I I
150 , ,
i
i
I
i
i
i
! i
! i
200 ' '
250
-! 0
XCORR
i
i
i
i
i
i
I
i
i
i
i
i
i
i
i
dS/dL Vs ILImU'"'
o ,,
i
I
50 '
i
'
i
-
oo
DO
50
i
I
i
i
i
i
i
i
! !
! i
-1 0
i
XCDRR
Figure 11. Cross correlations of time rate of change of
temperature and salinity at 0 ø, 165øE with zonal pseudostress
for the periods (a) November 1988 to November 1989 and
(b) November 1989 to August 1991. All data have been low
passed with an 11-day Hanning filter. Open circles indicate
depths of the instruments used in the calculations. Dashed
lines are 95% confidence intervals for the null hypothesis
(i.e., the correlation of zero).
face layer 0 T/O t and 0 S/0 t with meridional pseudostress
(U U(Y)), wind speed (IU), and a construct related to wind
work (U 3) were also computed. For the correlations with
meridional pseudostress, either the results were not significantly
nonzero at the 95% level of confidence, or, if they
were significant, then the magnitude was much less than for
correlations between OT/Ot, OS/Ot, and zonal pseudostress.
Hence subsequent discussion of pseudostress will focus only
on the zonal component. The statistical relationships between
OT/Ot and OS/Ot on the one hand and wind speed and
wind work on the other hand resembled those between
0 T/O t, OS/O t, and zonal pseudostress shown in Figure 11.
Indeed, the three wind constructs of wind speed, wind work,
and zonal wind pseudostress are highly correlated among
themselves, with correlation values ranging in magnitude
between 0.78 and 0.97. A similarly high correlation between
wind constructs was found by McPhaden and Hayes [1991]
in an earlier study of surface layer variability at 0 ø, 165øE.
Thus distinguishing the effects of turbulent air-sea exchanges,
vertical advection, and entrainment requires fur-
i
i
i
ther discussion of how these processes work in the upper
ocean.
During the November 1988 to November 1989 period,
cross correlations of OT/Ot with zonal pseudostress (Figure
11 a) are significantly nonzero at the 95% level of confidence
from the sea surface down to 150-m depth. The sign of these
correlations implies that easterly winds (relative to the
mean) lead to falling surface layer temperature. The deeper
significant correlations to 150 m in the temperature field
indicate that a shoaling thermocline is associated with falling
SST during easterly winds (relative to the mean), while a
deepening thermocline is associated with a warming surface
layer during westerly winds (relative to the mean). This
relationship is consistent with variations in wind-driven
vertical advection and entrainment (which in combination
we refer to as upwelling) producing temperature variability
in the surface layer. We can estimate the magnitude of this
vertical velocity at and below the base of the mixed layer by
assuming a purely vertical advection balance in the thermocline,
that is,
OT OT
--+ w --= 0 (6)
ot oz
For this case, the velocity is then simply the ratio of the time
rate of change of the temperature to the mean vertical
temperature gradient near the base of the isothermal layer.
The variability of OT/Ot at a depth of 125 m just below the
mixed layer is 0.83 x 10-6øC s -1. The product of this and
the magnitude of the cross correlation between 0 T/Ot and
U U (x) (r = 0.23) gives an estimate for OT/Ot due to
wind-driven upwelling of 1.8 x 10-7øC s -1. The average
temperature gradient is determined from the temperature
change (0.87øC) between the base of the mixed layer (100 m)
and just below the mixed layer (125 m). Hence the ratio
_
(-OT/Ot)/(OT/Oz) implies a standard deviation wind-driven
vertical velocity of magnitude 5.2 x 10 -6 m s -1 (0.45 m d -1)
in the upper thermocline for the period November 1988 to
November 1989. This vertical velocity, if representative of
the flow at the base of the mixed layer (W-h), could drive an
entrainment heat flux resulting in in-phase thermocline and
mixed-layer temperature variability.
By substituting (6) into (5) we can obtain an estimate of the
entrainment velocity We. From November 1988 to November
1989, the barrier layer is absent, and there is no difference
between the mixed-layer depth and the isothermal
depth; hence Oh/Ot in (5) is equivalent to the time rate of
change of the isothermal layer depth. The time series of
isothermal layer depth is estimated for November 1988 to
November 1989 using the combined set of SEACAT and
SeaData temperature recorders. The cross correlation be-
tween We and O(SST)/Ot is significant (r = 0.22) at the 95%
confidence level. The sign of the correlation implies that the
sea surface tends to cool during periods of entrainment
(We > 0) and tends to warm when entrainment is shut off
(We < 0). This relationship is consistent with that found
above between O T/Ot and the zonal pseudostress and supports
the importance of Ekman-induced upwelling in cooling
the surface layer during the La Nifia period.
For salinity, OS/Ot at 11 m is negatively correlated with
zonal pseudostress during November 1988 to November
1989 (Figure 1 l a), indicating that for periods of easterly
(westerly) winds relative to the mean, salinity rises (falls).

SPRINTALL AND MCPHADEN: WESTERN EQUATORIAL PACIFIC SURFACE LAYER VARIATIONS 975
This is consistent with variations in surface salinity responding
to variations in vertical advection and entrainment of
high-salinity water from the thermocline as suggested in
Figure 6a. Enhanced evaporation or reduced rainfall, associated
with periods of stronger than normal easterly wind
speeds, probably also contributes to increases in surface
salinity. There is no significant correlation between the zonal
pseudostress and salinity variation below the surface layer.
During the period November 1989 to August 1991, the
cross correlations between OT/Ot and the local zonal pseudostress
(Figure 11 b) again show significant positive correlations
between 100 and 125 m, as during the La Nifia period.
However, now the sign of the significant cross correlations
changes across the base of the isothermal layer. In contrast
to the relationship observed during the La Nifia period, this
implies that easterlies (westerlies) relative to the mean are
associated with rising (falling) surface layer temperatures.
Temperature variations in the thermocline are still likely to
be related to wind-driven vertical advection, but (as discussed
in more detail below) surface layer temperature
variability is probably not the result of simple entrainment of
thermocline water brought close to the surface by Ekmanrelated
vertical advection. As the mean surface wind at the
mooring during this period is to the west, the falling surface
temperatures would therefore be associated with the high
wind speeds of the westerly wind bursts. Surface cooling
may be caused by enhanced latent heat flux associated with
higher wind speeds and/or by decrease in insolation due to
the prevailing cloudiness associated with westerly windrelated
convective events [Nitta and Motoki, 1987]. Conversely,
the relationship between rising surface temperatures
and easterlies (relative to the mean) would correspond
to conditions of light winds, reduced evaporative cooling,
and clearer skies at 0 ø, 165øE.
Cross correlations between 0S/0 t and zonal pseudostress
during the November 1989 to August 1991 period are significantly
nonzero at the 95% confidence level between the sea
surface and 100-m depth. The correlations have the same
sign (negative) as during the La Nifia period and imply that
westerly (easterly) winds relative to the mean are associated
with falling (rising) surface layer salinity. This is consistent
with the enhanced deep convection and precipitation found
during periods of westerly wind bursts [Figure 4; Nitta and
Motoki, 1987; McPhaden and Milburn, 1992]. Ekman pumping
could account for the high correlation at 50-100 m found
between OS/Ot and local zonal pseudostress. Compared to
the La Nifia period, a pronounced mean halocline has
developed between 50 and 100 m where OS/Oz < O. This
halocline, when advected vertically, would produce significant
salinity variations over this depth range. We can again
estimate the magnitude of the wind-driven vertical velocity
near the base of the mixed layer by w_ h = (-OS/Ot)/(O•/
Oz), where OS/Ot is the time rate of change of that part of the
salinity variance correlated with zonal pseudostress (0.11 x
10 -6 psu s-l), and OS/Oz is the mean vertical salinity
gradient (0.01 psu m-I). For the period November 1989 to
August 1991, the standard deviation of vertical velocity is
estimated to be 9.8 x 10 -6 m s -1, or 0.85 m d -•.
As for the La Nifia period, we can again estimate the
entrainment velocity w e through (5) and (6). In this case we
estimated O h/Ot using density rather than temperature because
of the presence of the barrier layer. The cross correlation
between We and 0(SST)/0 t is not significant at the 95%
Table 3. Cross Correlations of Mixed-Layer Depth,
Isothermal-Layer Depth, and Barrier Layer
Thickness with SST, Wind Speed (IUI), Zonal
Pseudostress (IUl(X)), Wind Work (IUI 3) and
GPCP Rainfall Estimates
Variable
Correlation
Mixed-Layer Isothermal Barrier Layer
Depth Layer Depth Thickness
SST -0.36* (2) -0.60* (1) -0.31 (x)
Zonal pseudostress 0.47* (-3) 0.61' (-4) 0.14 (x)
Wind work 0.49* (-2) 0.55* (-3) 0.06 (x)
Wind speed 0.48* (-2) 0.54* (-3) 0.07 (x)
GPCP rainfall 0.27* (-5) 0.36* (-5) 0.10 (x)
All data have been low pass smoothed with an l 1-day
Hanning filter (except for the correlations with GPCP rainfall,
which are 5-day estimates) and are for the period
November 1989 to August 1991. Maximum correlation occurred
at lags in days shown in parentheses. Positive lags
mean that column variables (e.g., mixed-layer depth) lead
row variables (e.g., zonal pseudostress); an x indicates that
no lag was identified because the cross correlation was not
significant at the 95% level of confidence.
*Significantly different from zero at 95% confidence level.
confidence level. This suggests that from November 1989 to
August 1991, vertical advection and entrainment were not
responsible for cooling the sea surface. In contrast to the La
Nifia period, the barrier layer is preventing Ekman-related
entrainment of the cooler thermocline water from reaching
the mixed layer. Similar results were found during the
1986-1987 ENSO by McPhaden and Hayes [1991], who
suggested that entrainment may not have been effective in
the heat balance owing to the presence of a barrier layer
limiting communication between the surface layer and the
deeper thermocline region. However, McPhaden and Hayes
[1991] had no salinity measurements to verify this hypothesis.
We further explore the variability in mixed-layer depth,
isothermal-layer depth, and the barrier layer thickness by
computing cross correlations with SST, GPCP pentad rainfall
estimates, and various wind constructs for November
1989 to August 1991 (Table 3). Both SST and the wind
constructs of zonal pseudostress, wind work, and wind
speed are significantly correlated with changes in the mixedlayer
depth and the isothermal-layer depth. The signs and
lags of the correlations indicate that westerly (easterly)
winds are followed 3-4 days later by a relatively deep
(shallow) mixed-layer depth and isothermal depth. This is
consistent with zonal wind-driven Ekman convergence leading
to changes in the depth of these layers. Ignoring zonal
variations for simplicity, mass continuity would imply that
Oh/Ot - H(Ov/Oy), where the meridional Ekman flow (v)
over the mean mixed-layer depth (H) would be in phase with
the zonal wind stress. Thus it is expected that variations in
zonal pseudostress would lead those in mixed-layer depth,
and the results presented in Table 3 are consistent with this
relationship. In this context we view the high cross correlation
that exists between wind work and wind speed on the
one hand and mixed-layer depth and isothermal-layer depth

976 SPRINTALL AND MCPHADEN: WESTERN EQUATORIAL PACIFIC SURFACE LAYER VARIATIONS
on the other hand as not physically significant. Rather, we
interpret these high cross correlations as an artifact of the
high correlation between these wind constructs and zonal
pseudostress.
None of the cross correlations between the wind con-
structs and the barrier layer were significantly nonzero at the
95% confidence level. Of particular note in this regard is the
lack of significant correlation between wind work and barrier
layer thickness. This lack of correlation implies that there is
no erosion and no entrainment across the barrier layer in
response to the westerly wind burst forcing during this
period. The result is consistent with the vertical structure of
the correlation between temperature and zonal pseudostress
(Figure 11 b), which changed sign at the mean depth of the
barrier layer (50-80 m), and also the lack of correlation
between We and O(SST)/Ot for this time period, implying an
isolation of processes controlling temperature in the mixed
layer and the thermocline.
The mixed-layer depth and isothermal-layer depth are also
significantly correlated with GPCP rainfall estimates. The
sign of the correlation suggests that a deep mixed-layer is
associated with high rainfall 5 days (1 pentad) earlier. This is
not a relationship one would expect if buoyancy flux associated
with the rainfall controlled the mixed-layer depth. It is
more likely that wind-driven dynamics rather than buoyancy
flux associated with rainfall determines the mixed-layer
depth variability on the dominant time scales (10-25 days) of
our data set and that, coincidentally, westerly (easterly)
winds are correlated with high (low) rainfall rates. Indeed,
we found that zonal wind pseudostress and GPCP rainfall
estimates were significantly cross correlated for November
1989 to August 1991, with a coefficient of 0.45.
5.2. Lateral Advection
Results in the previous section suggest that onedimensional
vertical processes are likely to affect variations
in the hydrography of the surface layer at 0 ø, 165øE. We
examined the potential impact of various mechanisms using
physical reasoning guided by correlation analysis. However,
the significantly nonzero correlations we found between
surface layer water mass properties, wind constructs, and
GPCP rainfall estimates were generally in the range of
0.2--0.6. Thus not all the observed variability in surface layer
temperature and salinity can be accounted for by local windor
rain-related vertical fluxes. This leads us to consider the
possible role of lateral advection in the surface layer temperature
and salinity balance. In this section we explore the
advective balance obtained in (4) by equating the first two
terms on the left-hand side of the equation. We then use
correlation analysis to look for significant relationships between
velocity components and changes in temperature and
salinity (Figure 12).
For the period November 1988 to November 1989, the
zonal current is significantly correlated with O T/Ot at the
95% confidence level between 11 and 101 m. The sign of the
correlation implies that eastward (westward) velocity relative
to the mean would lead to rising (falling) temperatures in
the surface layer. An estimate of the time-averaged zonal
temperature gradient can be obtained from the slope of the
regression of O T/Ot against zonal velocity. For the La Nifia
time period, at 11 m, this is -0.81 x 10-6øC m -1, implying
a temperature gradient of 0.9øC for every 10 ø longitude from
east to west. This gradient is comparable to the SST gradient
that existed across the equatorial Pacific during this period
when the highest temperatures of the warm pool were
located further to the west of the mooring [Climate Analysis
Center, 1991].
For salinity during this period, meridional velocity is
usually better correlated with OS/Ot than with zonal velocity,
particularly at depths of 11, 101, and 201 m (Figure 12). The
positive correlation suggests that salinity increases (decreases)
with northward (southward) flow. The slope of the
regression of OS/Ot against meridional velocity implies a
meridional salinity gradient of-0.53 x 10 -6 psu m -1 at 11
m, -0.28 x 10 -6 psu m -1 at 101 m, and -0.7 x 10 -6 psu
m -• at 201 m. These gradients correspond to salinity decreases
per 1 ø latitude of 0.5, 0.3, and 0.7 psu at 11,101, and
201 m, respectively. The meridional salinity section that
coincides with this period (Figure 6a) illustrates the presence
of a salinity front near the equator; the magnitude and
sign of the salinity gradient associated with this front are in
general agreement with the values inferred from crosscorrelation
analyses of the moored time series data.
During the period November 1989 to August 1991, the
cross correlation between O T/Ot and the local zonal current
again show significant correlations at the 95% confidence
level in the surface layer at 11 and 51 m. However, at these
depths the sign of the correlation is negative, implying that
westward (eastward) current is associated with rising (falling)
temperatures. During this time period the slope of the
regression between O T/Ot and the zonal velocity component
at 11 m is 0.21 x 10-6øC m -1 implying a temperature
gradient with an increase of 0.2øC for every 10 ø longitude
from west to east. The maps of SST for the period show that
there are times when the warmest temperatures are located
to the east of the mooring [Climate Analysis Center, 1991];
however, for the period as a whole, there is no significant
gradient in the zonal temperature field at the surface. This
lack of a strong mean zonal temperature gradient across 0 ø,
165øE implies that zonal advection is probably negligible
during this period. The observed correlation between O T/Ot
and zonal velocity in the surface layer probably reflects a
coincident strong correlation between enhanced eastward
flow during periods of westerlies and a falling surface layer
temperature due to local wind-related air-sea heat exchange.
During the period November 1989 to August 1991, the
salinity is more significantly correlated with the zonal velocity
component than with the meridional velocity component.
The negative correlations of OS/Ot with zonal velocity in the
surface layer at 11 and 51 m imply that eastward (westward)
current is associated with falling (rising) salinities. At 11 m
the slope of the regression between O S/Ot and the zonal
velocity component is 0.31 x 10 -6 psu m -1 implying a
salinity gradient with an increase of 0.3 psu for every 10 ø
longitude from west to east. This is comparable to the
eastward increase of 0.2 psu for every 10 ø longitude found in
SURTROPAC sea surface salinities for the overlapping
period of data availability between November 1989 and July
1990.
6. Discussion and Conclusions
Using the 0 ø, 165øE mooring data for temperature, salinity,
currents, and winds, we have examined the processes that
influence the surface layer hydrography in the western
equatorial Pacific warm-pool region. For the approximate

SPRINTALL AND MCPHADEN' WESTERN EQUATORIAL PACIFIC SURFACE LAYER VARIATIONS 977
a 15-NOV-88 to l•-NOV-89
Current
Current
; ' J ;]1', l
•ø1 ',1;/" !- •ø-I ',•; r
/ ',1 •/ / / :1•; /
,001 il/ I ,oo-I : If,,' t-
! I
! I
! I
i i i i
150 , , 150
i !
i
! i
' '
200 200 ' '
50
de/dr
Ve Zonot
Current
O0
!
, ,
i
i
i
I
I
!
I
I
! i
i !
50
i
,
!
,
co
! i
i
0
.'oo
dS/dr
Ve NerLdtonot
Current
• ,
!
I
':
250. , , , . 250
-1 o I -1 o I
XCORR
XCORR
50. , __ , .
-I 0 !
XCORR
!SO. , __ ! .
-! 0 1
XCORR
b
18-NOV-89 to 4-RUG-91
O.
dT/dt
,.ol
20O
250. , •,
-! 0
Ve ZonoL
Current
i
,,
XCORR
dT/dt, V. MerLdtonoL
Current,
o
I
I
I
I
I !
50 , ,
O0 , ,
so
I
I
I
I
! !
150.
-I 0
XCORR
o
d$/dt,
Ve ZonoL
Current
50 , ,
!
!
!
!
i
oo
/ ;ll', /
J :11', L
Søl :'l : r
!
!00 ' '
!50
-! 0 I
XCORR
i
dS/dr
50 , ,
i
O0 ,
Ve NertdLonoL
Current
i
i
,,
i
i
i !
/ :1\',
•o• ,' I•',
i
O0 •,•!
,,
'50
-] 0
XCORR
i
i
i
i
i
i
Figure 12. Cross correlations of time rate of change of temperature and salinity at 0 ø, 165øE, with the
zonal and meridional current components for the periods (a) November 1988 to November 1989 and (b)
November 1989 to August 1991. All data have been low passed with an 11-day Hanning filter. Open circles
indicate depths of the instruments used in the calculations. Dashed lines are 95% confidence intervals for
the null hypothesis (i.e., true correlation of zero).
3-year record (November 1988 to August 1991), two climatic
regimes are evident in both atmospheric and oceanic conditions.
During the period November 1988 to November 1989, La
Nifia conditions existed in the equatorial Pacific. The Southern
Oscillation Index was high, unusually strong easterlies
were evident across the basin, and the equatorial cold tongue
penetrated into the western Pacific. Along 165øE, strong
easterlies were associated with equatorial divergence and
upwelling, a strong westward SEC, and unusually low rainfall
rates (< 100 mm month -1) near the equator. The mixed
layer at 0 ø, 165øE was relatively dense, cold, and salty and
penetrated to 100-m depth. Under the La Nifia conditions of
strong easterly trade winds and associated upwelling, no
barrier layer was observed in either the mooring record or
simultaneous CTD casts. Density variations in both the
mixed layer and the thermocline during this period were
controlled by temperature rather than salinity.
For the period November 1989 to August 1991, equatorial
winds at and to the west of 165øE were on average westerly
and punctuated by frequent westerly wind bursts. Rainfall
increased to 200-300 mm month -1, and near-surface zonal
velocity reversed, flowing eastward on average at 0 ø, 165øE
in the upper 50 m. The mooring record shows a shallower
thermocline and a warmer, fresher surface layer. During this
period the mixed layer was shallower than the isothermal
layer owing to the presence of a 30-m-thick barrier layer.
The standard deviation of barrier layer thickness was 20.9 m
with dominant time scales of 12-25 days. As during the La
Nifia period, density variations in the thermocline were
controlled by temperature; in contrast, however, density
variations in the mixed layer during November 1989 to
August 1991 were controlled by salinity.
With the limited data set available, we attempted to gain
some insight into the processes that might account for the
observed variability in surface layer temperature and salinity
within these two climatic regimes. Our basic strategy was to
use correlation analysis supported by physical reasoning.
The analyses are necessarily incomplete and inferential;
nonetheless, they provide some unique perspectives in the
differences between the two climatic states for which mean
conditions and variability are markedly different.
During the La Nifia period of November 1988 to November
1989, variability in the SST and sea surface salinity was
influenced by local Ekman-induced upwelling and horizontal
advection. We estimated a vertical velocity standard devia-

978 SPRINTALL AND MCPHADEN: WESTERN EQUATORIAL PACIFIC SURFACE LAYER VARIATIONS
tion of 5.2 x 10-6 m s- • in the upper thermocline associated
with zonal wind pseudostress forcing during this period. A
shoaling thermocline was correlated with decreasing SST
and rising surface salinity. Surface layer temperature appears
also to have been affected by zonal advection in the
cold tongue, and surface layer salinity appears to have been
affected by meridional advection of a salinity front situated
just north of the equator.
During November 1989 to August 1991 the presence of the
barrier layer effectively limited communication between the
surface layer and the deeper thermocline region. While
wind-induced Ekman pumping affected temperature and
salinity variations in the thermocline, the barrier layer prevented
entrainment of this cooler and saltier water into the
surface layer. Air-sea heat and freshwater fluxes were therefore
likely to play a more prominent role in determining the
surface layer hydrography. Surface cooling, for example,
was likely related to enhanced latent heat flux associated
with high westerly wind speeds that existed during this
period and to diminished insolation owing to the increased
cloudiness associated with the westerly wind-related deep
convection. Though we have no radiation data available to
substantiate the hypothesized role of insolation, we note that
our cloudiness-related GPCP rainfall estimates were positively
correlated with zonal winds; that is, cloudiness was
greater during episodes of westerlies. The weak horizontal
SST gradients that existed in the vicinity of 0 ø, 165øE during
this period imply a secondary role for horizontal temperature
advection.
During November 1989 to August 1991, westerly (easterly)
winds relative to the mean were associated with falling
(rising) surface layer salinity. This relationship may have
been in part the result of enhanced precipitation during wind
burst-related convective events. We also found that east-
ward flow was associated with advection of low-salinity
water past the 165øE mooring in the upper 50 m. Zonal
advection may have led to variations in barrier layer thickness
in the manner discussed by Roemmich et al. [1993].
They noted the presence of an eastward freshwater jet
overriding relatively salty water, resulting in the presence of
a barrier layer along 170øW during the 1991-1992 ENSO. The
vertical shear within this jet, acting on a zonal salinity
gradient, tilts the isohalines to the east to produce a vertical
salinity gradient in an isothermal surface layer. The hypothesized
process would operate on about 10-day time scales
and may be responsible for some of the observed variations
in barrier layer thickness. Eastward flow in the upper 50 m,
for example, would lead to lower surface layer salinity,
steepening the vertical salinity gradient across the mean
halocline located at depths between 50 and 80 m. However,
in contrast to the findings of a study by Roemmich et al.
[ 1993] in which zonal current variations were driven by zonal
salinity gradients in the mixed layer, observed zonal currents
at 0 ø, 165øE are predominantly wind driven.
While we have demonstrated the existence of a barrier
layer during the period November 1989 to August 1991 at 0 ø,
165øE, we have not specifically addressed why a barrier
exists during this part of the moored time series record but
not during the part of the record coincident with La Nifia
conditions. Nor have we attempted to define the role (if any)
of the barrier layer in the transition between the two different
climate regimes identified in this study. Our discussion of
mechanisms has concentrated on variations within rather
than between the two climatic states observed in the data.
However, we have been able to provide insight as to the
barrier layer' s influence in affecting ocean-atmosphere interaction
over the warm-pool region. Our analysis emphasizes
the reason for the term "barrier" with regard to the prevention
of water properties being transferred vertically through
mixing and entrainment. The barrier layer effectively decouples
the surface layer from the deeper water and also isolates
the physical processes that occur within each of these
regions.
Acknowledgments. We would like to thank Paul Freitag of
NOAA's Pacific Marine Environmental Laboratory for assistance
with processing and analyzing the moored time series data. We are
also grateful to Jim O'Brien of Florida State University for providing
pseudostress data; to Thierry Delcroix of ORSTOM, Noum6a, for
providing shipboard salinity data; and to John Janowiak of NOAA's
Climate Analysis Center for providing GPCP pentad rainfall estimates.
We would also like to acknowledge the very constructive
comments of two reviewers, Eric Lindstrom of the U.S. WOCE
Interagency Office, Washington, D.C., and Stuart Godfrey of
CSIRO, Hobart, Australia. Support for one of us (J.S.) was provided
by the National Research Council postdoctoral program.
Production of this manuscript was made possible by grants from
NOAA's U.S. TOGA project office and the Equatorial Pacific Ocean
Climate Studies program from NOAA Pacific Marine Environmental
Laboratory contribution 1449.
References
Arkin, P. A., and P. E. Ardunay, Estimating climatic-scale precipitation
from space: A review, J. Clim., 2, 1229-1238, 1989.
Climate Analysis Center, Near real time analysis, August 1991,
Clim. Diagnostics Bull., 63 pp., Natl. Oceanic and Atmos. Admin.,
U.S. Dept. of Commer., Washington, D.C., 1991.
Dandonneau, Y., J. Grelet, F. Masia, and P. Waigna, Rapport de la
campagne SURTROPAC 12 h bard du N.O. Le Suroft, 131 pp.,
Institut Franqais de Recherche Scientifique Pour le Developpement
en Coop6ration (ORSTOM), Noum6a, New Caledonia,
1990.
Davis, R. E., Predictability of sea surface temperature and sea level
pressure anomalies over the north Pacific Ocean, J. Phys. Oceanogr.,
6, 249-266, 1976.
Delcroix, T., F. Gallois, F. Masia, and P. Waigna, Rapport de la
campagne SURTROPAC 14 h bord du N.O. Le Noroit, 117 pp.,
Institut Franqais de Recherche Scientifique Pour le Developpement
en Coop6ration (ORSTOM), Noum6a, New Caledonia,
1991.
Delcroix, T., G. Eldin, M.-H. Radenac, J. Toole, and E. Firing,
Variation of the western equatorial Pacific Ocean, 1986-1988, J.
Geophys. Res., 97, 5423-5445, 1992.
du Penhoat, Y., F. Gallois, M.-J. Langlad, G. Reverdin, and H.
Walico, Rapport de la campagne SURTROPAC 13 h bord du N.O.
Le Suroft, 167 pp., Institut Franqais de Recherche Scientifique
Pour le Developpement en Coop6ration (ORSTOM), Noum6a,
New Caledonia, 1990.
Feng, Y., H. P. Freitag, M. J. McPhaden, and A. J. Shepherd,
Wind, current and temperature data at 0 ø, 165øE: January 1986 to
March 1991, NOAA Data Rep., ERL PMEL-36, 54 pp., Natl.
Oceanic and Atmos. Admin., Boulder, Colo., 1991.
Goldenberg, S. B., and J. J. O'Brien, Time and space variability of
tropical Pacific wind stress, Mon. Weather Rev., 109, 1190-1207.
H6nin, C., B. Camusat, F. Gallois, and P. Waigna, Rapport de la
campagne SURTROPAC 11 du N.O. Coriolis, 161 pp., Institut
Franqais de Recherche Scientifique Pour le Developpement en
Coop6ration (ORSTOM), Noum6a, New Caledonia, 1989.
Janowiak, J. E., and P. A. Arkin, Rainfall variations in the tropics
during 1986-1989 as estimated from observations of cloud top
temperature, J. Geophys. Res., 96, suppl., 3359-3374, 1991.
Lake, B. J., K. Yang, Z. Luizhi, R. C. Millard, S. Pu, J. M. Toole,
Z. Wang, and L. J. Mangum, Hydrographic observations from the
US/PRC Cooperative Program in the western equatorial Pacific

SPRINTALL AND MCPHADEN: WESTERN EQUATORIAL PACIFIC SURFACE LAYER VARIATIONS 979
Ocean: Cruises 5-8, Tech. Rep. WHOI-91-19, Woods Hole
Oceanogr. Inst., Woods Hole, Mass., 1991.
Levitus, S., Climatological atlas of the world ocean, NOAA Prof.
Pap., 13, 173 pp., U.S. Govt. Print. Off., Washington, D.C.,
1982.
Lukas, R., and E. Lindstrom, The mixed layer of the western
equatorial Pacific Ocean, J. Geophys. Res., 96, suppl., 3343-3357,
1991.
Madden, R. A., and P. Julian, Description of global scale circulation
cells in the tropics with a 40-50 day period, J. Atmos. Sci., 29,
1109-1123, 1972.
McCarty, M., and M. J. McPhaden, Mean seasonal cycles and
interannual variations at 0 ø, 165øE during 1986-1992, NOAA Tech.
Memo., ERL PMEL-98, 64 pp., Natl. Oceanic and Atmos. Admin.,
Boulder, Colo., 1993.
McPhaden, M. J., Variability in the central equatorial Indian Ocean,
II, Oceanic heat and turbulent energy balances, J. Mar. Res., 40,
403-419, 1982.
McPhaden, M. J., TOGA-TAO and the 1991-92 E1 Ni•o/Southern
Oscillation event, Oceanography, 6, 36-44, 1993.
McPhaden, M. J., and S. P. Hayes, On the variability of winds, sea
surface temperature, and surface heat content in the western
equatorial Pacific, J. Geophys. Res., 96, suppl., 3331-3342, 1991.
McPhaden, M. J., and H. B. Milburn, Moored precipitation measurements
for TOGA, Trop. Ocean-Global Atmos. Notes 7, pp.
1-5, Nova Univ., Fort Lauderdale, Fla., 1992.
McPhaden, M. J., H. P. Freitag, and A. J. Shepherd, Moored
salinity time series measurements at 0 ø, 140øW, J. Atmos. Oceanic
Technol., 7, 568-575, 1990.
McPhaden, M. J., F. Bahr, Y. du Penhoat, E. Firing, S. P. Hayes,
P. P. Niiler, P. L. Richardson, and J. M. Toole, The response of
the western equatorial Pacific Ocean to westerly wind bursts
during November 1989 to January 1990, J. Geophys. Res., 97,
14,289-14,303, 1992.
Nitta, T., and T. Motoki, Abrupt enhancement of convective
activity and low-level westerly burst during the onset phase of the
1986-87 E1 Nifio, J. Meteorol. $oc. Jpn., 65,497-506, 1987.
Roemmich, D., M. Morris, W. R. Yot•g, and J. R.q)onguy,, Fresh
equatorial jets, J. Phys. Oceanogr., in press, 1993.
Ropelewski, C. F., and M. S. Halpert, Global and regional scale
precipitation associated with E1 Nifio/Southern Oscillation, Mon.
Weather Rev., 115, 1606-1626, 1987.
Ropelewski, C. F., and M. S. Halpert, Precipitation patterns associated
with the high index phase of the Southern Oscillation, J.
Clim., 2, 268-284, 1989.
Rual, P., J. Grelet, M.-J. Langlade, H. Walico, and S. Bonnet,
Rapport de la campagne SUTROPAC 15 •t bord du N.O. Le
Noroit, 134 pp., Institut Franqais de Recherche Scientifique Pour
le Developpement en Coop6ration (ORSTOM), Noum6a, New
Caledonia, 1991.
Sprintall, J., and M. Tomczak, Evidence of the barrier layer in the
surface layer of the tropics, J. Geophys. Res., 97, 7305-7316,
1992.
Stricherz, J. N., J. J. O'Brien, and D. M. Legler, Atlas of Florida
State University tropical Pacific winds for TOGA, 1966-1985,
technical report, 261 pp., Mesoscale Air-Sea Interaction Group,
Fla. State Univ., Tallahassee, 1992.
Webster, P. J., and R. Lukas, TOGA COARE: The Coupled
Ocean-Atmosphere Response Experiment, Bull. Am. Meteorol.
$oc., 73(9), 1377-1416, 1992.
M. J. McPhaden and J. Sprintall, NOAA Pacific Marine Environmental
Laboratory, 7600 Sand Point Way, N.E., Seattle, WA 98115.
(Received March 29, 1993; revised September 20, 1993;
accepted September 24, 1993.)