Evidence of the Barrier Layer in the Surface Layer of the Tropics

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 97, NO. C5, PAGES 7305-7316, MAY 15, 1992

Evidence of the Barrier Layer in the Surface Layer of the Tropics

JANET SPRINTALL 1

Department of Geology and Geophysics, University of Sydney, Sydney, Australia

MA'ITH/AS TOMCZAK 2

Ocean Sciences Institute, University of Sydney, Sydney, Australia

Comparisons between isothermal depth to the top of the thermocline, and the mixed layer depth based on a ot criterion were undertaken for the tropical world oceans. In three equatorial regions, a shallower mixed layer than isothermal layer occurs, implying the presence of a strong halocline above the thermocline. This distance separating the top of the thermocline and the bottom of the mixed layer is referred to as the "barrier layer", in relation to its impediment to vertical heat flux out of the base of the mixed layer. Different mechanisms are responsible for maintaining the barrier layer in each of the three regions. In the western equatorial Pacific Ocean a salinity budget confirmed that heavy local precipitation most likely results in the isothermal but salt-stratified layer. In the northwest equatorial Atlantic, it is hypothesized that high salinity waters are subducted at the subtropics during winter and advected westward as a salinity maximum in the upper layers of the tropics, resulting in the barrier layer. In the eastern equatorial Indian Ocean, monsoonal related rainfall and river runoff contribute significantly to the freshwater flux, producing salt stratification in the surface. These results suggest the need to include the effects of salinity stratification when determining mixed layer depth.

1. INTRODUCTION The creation of the barrier layer exerts an important influence on mixed layer dynamics and, in particular, the interplay between the

The ocean surface mixed layer generally denotes a quasi- kinetic energy and potential energy processes mentioned above. The homogeneous layer with little variation in temperature, salinity, and term barrier layer is indicative of its effect on the mixed layer heat

density. The zone owes its high degree of vertical uniformity to budget. In a steady state situation, one would expect the formation mixing caused by turbulence. The turbulence, which may result in of a temperature gradient in response to the salinity gradient, in the mixed layer deepening or stratifying, requires an energy input

that may be generated by kinetic energy and/or by potential energy at the sea surface. Kinetic energy, from the transfer of the wind momentum to the sea, results in mixing processes such as wave action, entrainment, and horizontal advection by currents. Potential energy is measured by buoyancy flux that reflects change of density due to heat and freshwater fluxes.

order to balance the surface heating with the mixing from entrainment of cooler water from below the halocline. However, this does not seem to occur in the western equatorial Pacific, where the discrepancy in isothermal and isohaline depths has been observed. The presence of the barrier layer means that any water entrained from the isothermal layer into the mixed layer has the

It is the pycnocline, where the water is highly stable, that defines same temperature as the water in the mixed layer. There is therefore no heat transferred through the bottom of the mixed layer. Thus the the lower boundary of the mixed layer. That is to say, it represents a heat input at the surface would perpetually raise the temperature of vertical limit to air-sea interaction, and the associated turbulence is the upper ocean if it were not horizontally advected away from the less able to penetrate through this layer. Generally the pycnocline

coincides with both the halocline and the thermocline; however, region, or in fact if the net heat input through the surface of the recent investigations [Delcroix et al., 1987; Lukas and Lindstrom, mixed layer was not close to zero. As the sea surface temperature 1991] have noted the presence of a shallower halocline than (SST) gradients in the region are thought to be too small for

horizontal advective processes to be significant [Enfield, 1986], the thermocline in the upper layer of the western equatorial Pacific Ocean. The sharp gradient of the halocline was reflected in the case for the near zero heat flux is being strengthened [for instance, pycnocline (as illustrated, for example in Lukas and Lindstrom's see Godfrey and Lindstrom, 1989]. However, Lewis et al. [ 1990] [1991] Figure 3), and we refer to the distance separating the bottom suggest that there may be a net surface heat flux of the order -15-25

W m -2 in to the western Pacific, that may then be lost through of the mixed layer from the top of the thermocline (i.e., the bottom of the isothermal layer)as the "barrier layer". penetration of solar radiation through the transparent waters to

below the mixed layer. Further oceanographic studies under differing oceanic conditions that will directly measure vertical

1Now at NOAA/Pacific Marine Environmental Laboratory, Seattle, mixing of heat may be required to conf'um the interaction between Washington. the barrier layer, SST, and heat flux in the region.

2Now at School of Earth Sciences, Flinders University, Adelaide, Australia.

Copyright 1992 by the American Geophysical Union.

Paper number 92JC00407. 0148-0227/92192.1C-00407505.00

Differences between the isohaline and isothermal depths have been observed in the western equatorial Pacific [Delcroix et al., 1987; Lukas and Lindstrom, 1991]; however, little has been done to identify other regions of the world oceans where this discrepancy may exist. The major aim of this present analysis is to map the large-scale distribution of these differences in the tropical oceans

7305

7306 SPRINTALL AND TOMCZAK: EVIDENCE OF BARRIER LAYER IN THE TROPICS

and to determine the horizontal and vertical processes that are contributing to the mismatch in vertical scales between mixed layer and isothermal depth.

Historically mixed layer depth has been considered synonymous with thermocline depth. This assumption developed more out of necessity (though of course it can still be correctly physically based), as observations consisting of salinity and hence density data were much less common. The existence of the barrier layer indicates that determination of a mixed layer depth based on a temperature criterion alone is not sufficient to indicate the layer affected by surface mixing processes. The isothermal layer does not always correspond to a vertically uniform layer because salinity can govern the stratification of the water column. The mixed layer has to be der'reed relative to both temperature and salinity. The improved data base available today will enable us to confirm the isothermal/mixed layer assumption, thereby gaining a better understanding of characteristics of the mixed layerß Recently Sprintall and Tomczak [1990] undertook a comparison of mean depth and standard deviation for seven different criteria of mixed layer depth using the CTD cruise data of the Western Equatorial Pacific Ocean Circulation Study (WEPOCS) I and II expeditions in the western equatorial Pacific. The der'tuitions for mixed layer were based on differences in temperature, salinity, and density using a variety of critical gradient criteria and net decrease criteria from the surface. Their study suggests the suitability of most of these criteria in determining either mixed layer or isothermal layer depth. The two criteria used in this study are discussed in the analysis of section 2. One, based on a 0.5øC change from the sea surface temperature, gives an indication of the isothermal layer to the top of the thermocline and has been used extensively as a proxy to mixed layer depth. The other method is based on a variable ot criterion that accounts for the thermal expansion required to obtain a net temperature difference of 0.5øC. Hence, this technique accounts for both salinity and temperature effect in determining the depth of the mixed layer.

Three equatorial regions are found to display a significant 0 positive difference (10-50 m) between the top of the thermocline and the mixed layer depth calculated from the {It based criterion. 25 This phenomenon implies the presence of a strong halocline occurring above the thermocline. The dynamics of the three regions 50 that show a barrier layer, namely the western Pacific, the equatorial Atlantic, and the Bay of Bengal in the Indian Ocean, differ distinctly • 75 from each other. In section 3 we develop a salinity budget and invoke different surface forcing mechanisms for each of the three •. 100 regions in order to explain the salt-stratified isothermal layer. -o Section 4 gives a summary and conclusions. 125

2. DATA AND METHODOLOGY

Data

The Levitus [1982] world ocean data set will be used to determine the extent of salinity influence on the mixed layer depth. The data set consists of objectively analyzed temperature and salinity fields at standard oceanographic observation depth levels on a 1 o latitude-longitude grid for the world ocean. These standard depth levels are given at 0 m, 10 m, 20 m, 30 m, 50 m, then thereafter every 25 m to 150 m, every 50 m to 300 m, and every 100 m to a maximin depth of 1500 m. In the tropical ocean, mixed layer depths greater than 150 m are rare, and more often are substantially shallower than this. For instance, in the western equatorial Pacific Ocean, Lukas and Lindstrom [1991] determined the average mixed layer depth to be 29 m, with a range between 1 m

and 106 m over the two WEPOCS cruises, while the isothermal

layer was ~ 64 m deep, implying a mean barrier layer of 35 m in thickness. With this in mind, the resolution of the standard vertical

levels given above, for the surface layer of the Levitus [ 1982] data set, appears to be quite sufficient for computations of tropical mixed layer depth, and for determining barrier layers of greater than 10-m thickness.

The representativeness of the original data used in the objective analysis in terms of spatial and temporal biases has been discussed by Levitus [ 1982], who noted that the distribution of observations in the tropical regions was adequate for defining large-scale features that are representative of the real ocean. It is seasonal data grouped according to boreal seasons, although here we will refer only to the monthly groupings to avoid confusion when referring to cross- equatorial regions. These groupings are February, March, and April (FMA); May, June, and July (MIJ); August, September, and October (ASO) and November, December, and January (NDJ).

Definition of the Mixed Layer

The definition of the mixed layer depth (mld) is based on a variable sigma-t criterion, and determines the depth where ot is equal to the sea surface ot plus the increment in ot equivalent to a desired net decrease in temperature. This increment in ot uses the coefficient of thermal expansion, calculated as a function of surface temperature and salinity. Thus the technique accounts for both the salinity and temperature effect in determining the depth of the mixed layer. We wish to interpolate to the depth (z--mld) at which

Ot, ml d = ot, 0+ AT Dt DT (1) where ot,0 is the surface ot value, A T is the desired temperature difference and &Or/&T is the coefficient of thermal expansion

150

175

200

14

0

lO 20

30

50

-75

-100

-125

- 150

200 30

ß i ß i ß i - i ß i ß i ß

I

t,

ß

,, /

./// T

, I '"•'7 I I I I II I I ,'• I , I

18 22 26

temperature

. , , , , , salinity

L I I ' '

33 37

I I I , I ß I ' 2•½ ' i , i , i , i 2O 22 26 28

sigma-t

Fig. 1. Temperature, salinity, and sigma-t at 158.5øE, 2.5øS during FMA from the Levitus [1982] data set. (Standard depth levels of the Levims data are given on the fight axis.) The difference between mixed layer depth (solid line) and isothermal depth (dashed line) is referred to as the barrier layer.

SPRINTALL AND TOMCZAK: EVIDENCE OF BARRIER LAYER IN THE TROPICS 7307

evaluated with surface values of temperature and salinity. This definition allows for comparison of mixed layer depths on a world ocean basis, as it accounts for the difference in surface salinities and

temperatures and their effect on the seawater expansion coefficient. A vertical profile of temperature, salinity, and sigma-t from the Levitus data for the equatorial Pacific Ocean at 158.5øE, 2.5øS during FMA is shown in Figure 1. The mixed layer depth calculated from (1) is 37 m, assuming a temperature difference of A T = 0.5 øC. If salinity stratification within the surface layer was negligible, then equation (1) would give a mixed layer depth as calculated by the depth of this desired temperature difference (A T) from the surface isotherm. In Figure 1 the depth of the isotherm, which is 0.5øC colder than the surface isotherm, is 79 m, that is, 42 m deeper than the mixed layer depth. This 42-m difference between the c•t based mixed layer and the deeper isothermal layer represents the thickness of the barrier layer for this profile. Note that the depths of the isothermal and mixed layers do not coincide with the standard depth levels used by Levitus [1982], as we have linearly interpolated between these depths to satisfy the criteria used.

This paper considers the distribution of the barrier layer in the tropical (i.e., 30øS-30øN) regions of the world ocean, that is, the difference when isohaline layers were found to be shallower than isothermal layers. Seasonal charts of the absolute values for c•t based mixed layer, depth of the isothermal layer, and the differences between the two, are available in Sprintall and Tomczak [1990] for the world ocean between 60øS and 60øN. The reader is referred to

this report for further differences, such as in the subtropics, where a deeper isopycnal than isothermal layer exists due to the convergence of the Ekman layer transport and the subduction of surface waters along the isopycnals. This feature has been investigated by J. Sprintall and M. Tomczak (On the formation of Central Water and thermocline ventilation in the southern hemisphere, submitted to Deep Sea Research, 1992).

3. SEASONAL STRUCTURE IN THE THICKNESS

OF THE BARRIER LAYER

Three equatorial regions consistently displayed, throughout all seasons, a significant positive difference (10 - 50 m) between the isothermal layer and c•t based mixed layer criterion, implying the presence of a strong halocline occtmSng above the thermocline. As mentioned previously, the difference between the two depths is referred to as the barrier layer. Here, we will determine the different mechanisms of surface forcing responsible for the maintenance of the barrier layer in each of the three regions. In the following, a Peter's Projection [Peters, 1989; Tomczak and Krause, 1989] has been used for all figures of the equatorial oceans, as this projection combines fidelity of area with a rectangular latitude-longitude grid, both appropriate features for large-scale mapping of tropical oceanographic data.

The Western Equatorial Pacific Ocean

The western equatorial Pacific region (Figure 2) has in its core, generally centred at-•160øE, a 25-m difference in depth between isothermal and isohaline layers (i.e., a 25-m barrier layer thickness). It stretches from the northeastern coast of New Guinea, straddles the equator between 10øS and 10øN, and reaches its most extensive eastern limit at 170øW during MJJ (Figure 2b) of the southwest monsoon season. Barrier layers of 50-m thickness are evident in the southern hemisphere during NDJ (Figure 2d) and FMA (Figure 2a). There is a 10-m contour present during all seasons, enclosing the 25-m barrier layer thickness, running directly east of the Philippines to --150øW then westward to the Coral Sea.

The importance of salinity in the upper layers of this region has been recognized by many authors. The WEPOCS cruises of 1985 and 1986 [Lindstrom et al., 1987; Lukas and Lindstrom, 1991] determined differences between isothermal and isohaline layers

30øN

20øN

10øN

o

10øS

20øS

30øS (a)

,•,.25

• : 25

, ,

120øE 150øE 180 ø 150øW 120øW 90øW

Fig. 2. Differtraces (meters) between the isothermal and mixed layer depth for the Pacific Ocean during (a) February-April, (b) May- July, (c) August-October, and (d) November-January. Positive differences (solid lines) indicate a shallower isohaline layer than isothermal layer, and numbers give the thickness of the barrier layer.


30øN

20øN

10øN

o

10os

20os

30os (b) (• ', ',

I

120øE 150øE 180 ø 150øW 120øW 90øW

150øE 180 ø 150øW 120øW 90øW

Fig. 2. (continued)

which were of approximately the same magnitude as what has been determined here from the climatologic seasonal data set of Levitus [1982]. Lukas [1988] suggested that the isothermal but salt stratified layer was produced by a net flux of freshwater into the region, creating the distinct stable barrier layer separating the bottom of the mixed layer from the top of the thermocline.

The amount of freshwater required to induce surface freshening in order to maintain a barrier layer can be obtained by applying salinity and mass conservative principles to the region where the

barrier layer exists. In the western equatorial Pacific Ocean, the region bounded by 10øS - 10øN and 150øE - 170øE di•lays a barrier layer of at least 25-m thickness in all seasons. Using a simple box model for the upper surface layer, mass and salt conservation are given by

AlulSlPl + A3u3S3P3 = A2u2S2P2 +A4u4S4P4 (2) Alul +A3u3 +RB = A2u2+A4u4

where Ai is the cross-sectional area with a salinity Si, density Pi,

SPRINTALL AND TOMCZAK: EVIDF2qCE OF BARRIER LAYER IN THE TROPICS 7309

30øN

20øN

10øN

o

10øS

20øS

30øS

120øE

' ! ' hO ! ! ! 0 ...... ........ ,,, .......... ........

J"'"•,13'/ ,•'• k10i i 101 • ' )• '

'

..... ..... , ........ ................. ..... .........

0 0

150øE 180 ø 150øW 120øW 90oW

Fig. 2. (continued)

and velocity normal ui, for the upstream (i=1), downstream (i=2), and side boundaries (i=3,4). The freshwater gain (R) through the surface area (B) of the box represents the differences between precipitation (P) and evaporation (E).

Further refinement of the model (2) is carried out by making a number of assumptions which may be applied in general, and specifically in applying the model to the western equatorial Pacific region. First, we will assume that the differences in densities are negligible. Second, water may be transported in the model either through the sides or through the surface. Entrainment from below is neglected. This may at first appear a gross assumption; however, in view of the light mean easterly tradewinds experienced in the western equatorial Pacific [Wyrtki and Meyers, 1975], this closed bottom boundary condition may be justified. Certainly off the equator we can expect the wind to induce a surface convergence. The validity of the assumption on the equator where divergence may result is discussed further below. Further, if the box is orientated along the mean zonal equatorial current in the region, then this mass may be used to dilute the salinity from S1 at the eastern boundary to S2 at the western boundary, assuming no change in either the zonal flow (i.e., Ul=U2) or the inflowing and outflowing cross-sectional areas (i.e., A 1=A2). In the region of interest in the western equatorial Pacific an average mixed layer of 35 m was estimated [Sprintall and Tomczak, 1990] with an average salinity at 170øE of 34.9 (S1) and at 150øE of 34.6 (S2) according to the Levitus [1982] atlas. Zonal flow (Ul) in the upper 50 m of the region is toward the west in the South Equatorial Current (SEC) with a meridional extent between 10øS to 3øN. Delcroix et al. [1987] made direct current measurements along 165øE during six half-yearly cruises between 7ø-10øN and 20øS from January 1984 to June 1986 and observed a maximum speed in the SEC of -50 cm s-1. Similarly, Lindstrom et al. [1987] found a peak speed of -50 cm s-1 in the SEC from their mooring at 0% 150øœ from August 1985 through January 1986.

Thus in the calculations undertaken here we will assume an SEC

velocity with Ul = -50 cm s-1 with a meridional boundary between 10øS to 3øN. Finally, meridional movement is thought to play only a small part in the dynamics of the mixed layer of this region [McPhaden and Picaut, 1990]; hence we will consider only the total divergence through the meridional sides of the box (i.e., A5u5 = A3u3-A4u4) with no change in the meridional value for salinity of 34.8 (S5) [Levitus, 1982].

Are these reasonable assumptions for our model in an attempt to explain the existence of the barrier layer within the western equatorial Pacific to be due to ocean-atmosphere freshwater flux? The assumption that entrainment processes are negligible for this region is supported in the recent observations of Godfrey and Lindstrom [1989] and Lukas and Lindstrom [1991]. As discussed already, the very presence of the barrier layer implies that entrainment cooling, through heat budget implications, cannot be significant in the waters of the western Pacific region. Ekman transport divergence and upwelling at the equator are proportional to the local zonal wind stress [Gill, 1975]. As stated previously, the zonal wind stress for the western Pacific is close to zero along the equator in the annual mean, and is weak during most months of the year [Wyrtki and Meyers, 1975]. Indeed, Levitus [1982] shows little variation in sea surface salinity across the equatorial region here, perhaps implying upwelling of the saltier water of the barrier layer to be m'mimal. In accordance with this, our neglecting of entrainment processes on the equator in this region is not an unreasonable assumption.

In both studies by Delcroix et al. [ 1987] and Lindstrom et al. [1987] it was observed that if the zonal component of the wind changed to westerlies, within a week the zonal flow in the SEC at the surface reverses. This could change both the sign and the magnitude of our Ul value. However, with the arrival of strong westerly wind bursts, Lukas and Lindstrom [1991] also observed


that the halocline was eroded, the barrier layer broken down, entrainment processes initiated, and strong mixing occurred. Thus in a model where we are explaining the existence of the barrier layer per se in a large-scale climatologic data set, these conditions hold no relevance to us, and hence the assumptions seem reasonable. Correspondingly, after the proposed modifications, the model (2) may be written as

A5u5S5 + Alul(S1-S2) = 0 A5u5 + RB = 0 (3)

and solved for the above outlined values to give a net freshwater flux (R) of 2140 mm yr -1 (positive value denotes excess precipitation) and a meridional mass transport (A5u5) of 0.2 Sv which implies a mean north-south velocity of 2.8 x 10 -3 m s-1. As expected, this velocity is much smaller than the zonal flow of the SEC substantiating the hypothesis of small meridional influence in the mixed layer dynamics of the region. So, approximately 2140 mm yr -1 of freshwater is required for the formation of the freshwater lens at the upper surface of the western equatorial Pacific. Is this value in accord with the estimates of freshwater flux

across the sea surface based on the existing distribution maps? Weare et al. [1981] established an annual E-P balance by

comparing the annual estimation of rainfall in the tropical Pacific from Taylor [1973] to their computed latent heat transfer. The distribution of this computed E-P balance indicates two m'mima of over -1500 mm yr-1, one on the equator north of New Guinea, the other at 10øS, 170øE. This excess precipitation over evaporation coincides extremely well with the region where the mixed layer is shallower than the isothermal layer (Figure 2). Indeed, recent estimates by Oberhuber [1988] of the net downward freshwater flux in the two minima puts the value at more like -2400 mm yr-1. The difference lies in the precipitation data used. The rainfall analysis used to prepare the fluxes of Oberhuber [1988] were those of Shea [1986] derived from land or island station records and completed by satellite observations. Uncertainties of oceanic rainfall calculations

in both the Taylor [1973] and the Shea [1986] precipitation data sets lead to a mean estimate of approximately -2000 mm yr-1 in the E-P balance for the western equatorial Pacific. The value is in good agreement with the computed amount of freshwater required to maintain the shallow isohaline layer.

The result is not surprising considering the coincidence of the two negative E-P extremes and the maximum rainfall of over 5000 mm yr-1 in Taylor's [1973] annual rainfall map. These, in turn, correspond to the mean point of intersection between the Intertropical Convergence Zone 0TCZ) of the northern hemisphere, and the South Pacific Convergence Zone (SPCZ) of the southern hemisphere. Both zones are known tropical rainfall bands, and their influence in producing the barrier layer at their confluence and across the entire tropical Pacific can be followed in Figure 2.

The mean position of the ITCZ is near the equator in the western Pacific, rising to ~10øN in the central Pacific [Barnett, 1977]. The 10-m difference contour (Figure 2) between mixed layer depth and thermocline depth follows the ITCZ across the equatorial Pacific, emphasizing the importance of rainfall in salt stratifying the mixed layer. The seasonal variation in ITCZ movement is also reflected in the distribution of the barrier layer thickness (Figure 2). During February the ITCZ reaches a northern extreme at ~15øN, then retreats to ~10øN during August. The 10-m isopleth of barrier layer thickness follows this seasonality, as evidencexl in Figures 2a and 2c, respectively.

The average annual cycle of the surface wind divergence

indicates that the SPCZ is less pronounced than the ITCZ, although it still exhibits changes in both position and magnitude [Goldenburg and O'Brien, 1981]. During January, maximum convergence is observed near 15øS in the western equatorial Pacific when associated convective activity brings maximum rainfall [Taylor, 1973]. By September the SPCZ is less marked and has shifted more northerly, reaching about 10øS. The seasonality in zonal movement of the SPCZ is not as pronounced in the distribution of the barrier layer in the southwestern Pacific Ocean as it is for seasonal ITCZ movement. However, it is observed in Figure 2 that the maximum values of the barrier layer lie along a southeast/northwest axis, as is characteristic of the mean S PCZ location.

East of 160øW (Figure 2), the negative isopleths in the subtropics indicate that the equatorial barrier layer is maintained by seasonal subduction of the surface water. This process alternates between the northern hemisphere in FMA (Figure 2a) and the southern hemisphere in ASO (Figure 2c). It is described more fully in the following Atlantic Ocean section, where it is shown to be the main source in maintaining the barrier layer structure.

The Northwestern Equatorial Atlantic Ocean

The equatorial northwest Atlantic is also characterized by a shallower isohaline than isothermal layer (Figure 3). In NDJ (Figure 3d) a 25-m discrepancy in scales is found in the north from the West Indies (~20øN) out to ~50øW then along the northeastern South American coastline to Cabo de S•o Roque just below the equator. It contains a maximum of 50 m in the depth of the barrier layer at ~55øW, 15øN. In FMA (Figure 3a) the 25-m discrepancy reaches its most eastward extent to 30øW at ~15ø-20øN, and the 50-

m maximum has extended to lie between 12øN-20øN. By MJJ (Figure 3b) the 25-m isopleth has diminished in area, and the 50-m barrier layer has become only of the order of a few degrees square to lie directly adjacent to the South American coast at ~8øN. Finally in ASO (Figure 3c) three small 25-m maxima remain in the northern hemisphere, while a new region of more than 25-m barrier layer thickness develops in the south.

Updated annual and seasonal maps of E-P for the north Atlantic Ocean were produced recently by Schmitt et al. [1989]. The data sets used for this calculation are the heat flux estimates of Bunker

[1976] and the precipitation estimates of Dorman and Bourke [1981]. Both of these studies provide an increased spatial and temporal resolution for estimation of E-P over previous work. Their map of annual E-P shows that in the region west of 40øW, corresponding to maximum discrepancies in isothermal and isohaline depths (Figure 3), there is a net water loss to the atmosphere (E-P>0) of 1000-1250 mm yr-1. Local freshening of the surface layer by excess precipitation, the mechanism that produces the barrier layer in the western equatorial Pacific Ocean and modeled by equation (3), can therefore not be responsible here. Clearly, we must look to other processes, such as horizontal advection from adjoining regions or continental runoff, as a source for dilution of the surface layers.

The region of net evaporation lies immediately adjacent to a band of net precipitation in the tropics found between the equator and 10øN but east of 40øW [Schmitt et al., 1989]. The greatest seasonal variability in E-P in terms of both magnitude and location of gradients is found in this equatorial region. It is associated with the migration of the Atlantic Ocean ITCZ and its accompanying rainfall. There generally appear to be two modes of variability. During December-May a net precipitation band extends across the Atlantic between 0 ø and 10øN, and a net evaporation maximum lies between 10øN and 20øN with a strong zonal gradient between the two. Both


30øN

20øN

10øN

10øS

20øS

30øS

30øN

20øN

10øN

0 ø

10øS

20øS

30øS

25,

80øW 60øW

i

40øW 20øW 0 o

10

80øW 60øW 40øW 20øW 0 ø

i

80ow 60oW

-10 :•{ '•,, '" •-,'-- i',-- ,': ,---

0

40ow 20ow 0 ø

i

80øW 60øW 40øW 20øW 0 o

Fig. 3. Differences in mixed layer depth (meters) between the isothermal and mixed layer depth for the Atlantic Ocean during (a) Febmary-April, (b) May-July, (c) August-October, and (d) November-January. Positive differences (solid lines) indicate a shallower isohaline layer than isothermal layer, and numbers give the thickness of the barrier layer.

features have extrema in excess of 120001 mm yr-1. This period corresponds to when the barrier layer is thickest (Figures 3a and 3d); however, as its distribution lies in the region of evaporation excess, some northward advection of the freshwater gain at the equator is required. In fact, Schmitt et al. [1989] do estimate that

northward transport of freshwater flux, including river runoff, is maximum at 10øN; however, this represents the meridional divergence across the entire ocean basin. Furthermore the U.S. Navy Oceanographic Office [ 1965] atlas indicates that the Ariantic Ocean South Equatorial Current (SEC) strengthens and extends

7312 SPRINTALL AND TOMCZAK: EVIDF•CE OF BARRIER LAYER IN THE TROPICS

north of the equator to converge with the Atlantic Ocean North Equatorial Current (NEC) in the region of maximin barrier layer thickness, thus providing a mechanism of transport. By June to November, the adjacent bands of net precipitation and net evaporation have moved northward, associated with the northward intensification of the equatorial rainfall band. As a result, the E-P maximum in the region coinciding with the maximum thickness in the barrier layer suffers a reduction in magnitude to -•500 mm yr- 1. The barrier layer during this period (Figures 3b and 3c) is generally shallower and more diffuse and extends down to the equator, thus corresponding in part with the region of net precipitation-1500 mm yr-1 in the E-P balance. This indicates that during some times of the year at the equator there is an excess of precipitation over evaporation, which may be sufficient to allow for the formation of a fresh isohaline layer above a deeper isothermal layer. In all however, the seasonal east-west anisotropy in precipitation allows evaporation to dominate in the annual average in a narrow band extending to the equator from the South American coast to -•45øW. The major area of freshwater gain remains east of 40øW. This evaporative dominance in the western equatorial Ariantic implies that some other mechanism than rainfall must be maintaining the salt stratified isothermal layer.

It is likely that the low salinity outflow from the Amazon and Orinoco Rivers contributes to the freshwater flux in the surface

waters of the equatorial western Atlantic. The Orinoco River has its mouth at-•7øN and contributes 845 mm of freshwater per annum, and the Amazon, with its estuary located on the equator, has a freshwater outflow of 835 mm yr-1 [Baumgartner and Reichel, 1975]. This may be sufficient to account for some of the freshwater excess required to maintain the isohaline layer. These discharge values represent outflow at the river's estuary and have been calculated by Baumgartner and Reichel [1975] using the ratio of total water volume over the catchment area. Interestingly though, Baumgartner and Reichel [1975] assumed that with Coriolis force, one-third of the Amazon flow would be to the northern hemisphere, with two-thirds to the southern hemisphere. However, it is now recognized that the Guiana Current strongly deflects all Amazonian flow to the north and the considerable discharge possibly even strengthens the flow northward at its outlet [Landis, 1971]. Peak flow of the Amazon occurs during the months of May-July, and this is evident in the 50-m barrier layer during these months (Figure 3b) hugging the coast north of the river's mouth. There is no doubt that river runoff must have a contributory effect on the freshwater budget of the tropical western Atlantic; however, Neumann [ 1969] suggests that local precipitation along with advection of extremely low or high salinity water masses must also be considered.

The presence of a shallower isohaline than isothermal layer in the tropical Atlantic Ocean has also been observed by Defant [1961] in the Meteor cruises of 1936, and also during the Barbados Meteorological and Oceanographic Experiment (BOMEX) cruises in May, June, and July of 1969 [e.g., see Elliott, 1974]. During the Meteor cruises, almost all stations in the tropics and subtropics were characterized by a nearly isohaline layer with a thin layer containing a well-developed salinity maximum at its base, located above the top of the thermocline and a deeper layer of lower salinity. This saline layer originates at the surface in the subtropics and is formed midgyre in both hemispheres, partially in response to high evaporation [Luyten et al., 1983]. This water is then subducted as the shallow subtropic salinity maximum [Worthington, 1976] and spreads out southward and westward in the NEC of the northern hemisphere, and northward and westward in the SEC of the southern hemisphere. The SEC is the stronger and more constant of

these two westward equatorial currents, and usually extends north of the equator to converge with the NEC at -•7øN, 30øW [U.S. Navy Oceanographic Office, 1965]. From here, both currents proceed into the Caribbean as the Guiana Current, or during summer, some flow may deflect eastward to feed the Atlantic Equatorial Counter Current (ECC). The salinity of the maximum varies little and maintains about the same thickness over its long course within the NEC and SEC to well within the Gulf of Guinea. It thus represents an inlxusion of high-salinity water beneath the surface layers of lower salinity in the equatorial regions. The salinity maximum appears to be present everywhere in the tropical Atlantic except in two narrow bands; one between 10 ø and 15øN and extending from 40øW eastward to Africa, and a second, more narrow band between 2 ø and 3øS and extending from 30 ø to 10øW. The two bands without salinity maximum mark the southern and northern limit, respectively, of the subtropical Central Water masses as they extend toward the equator [Worthington, 1976]. Between the two bands the maximum appears again and may be very defined due to the presence of the ECC, being fed by the NEC and SEC from regions west of 35ø-40øW. The distribution of the salinity maximum is imitated in the distribution of the barrier layer evident in Figure 3. Negative isopleths indicate a shallower isothermal than isopycnal layer, and these regions are coincident with where maximum downward velocity in the Ekman layer transport occurs [Sprintall and Tomczak, 1990], and can therefore be interpreted as a region where Central Water mass is formed through subduction along isopycnals. J. Sprintall and M. Tomczak (submitted manuscript, 1992) show that the region 30ø-45øS is a formation region of South Atlantic Central Water. Figure 3c shows active subduction for ASO south of 15øS, coupled with formation of a barrier layer to the north. During FMA (Figure 3a) the same process occurs in the northern hemisphere in the vicinity of 20øN. The two sources alternate in renewing the barrier layer structure in the west equatorial Atlantic Ocean, where the barrier layer is found during all seasons.

The thickness of the barrier layer is generally maximin in the region of confluence between the NEC and SEC (Figure 3). The seasonality of the presence of the ECC is also strongly reflected in the distribution of the barrier layer thickness. During FMA the countercurrent dissipates and is least evident, while the SEC widens from the South American coast eastward to -•10øW.

Correspondingly, the barrier layer (Figure 3a) is almost nonexistent between the latitudes of the ECC during this period; however, it extends far to the south and west in the southern hemisphere relative to other seasons. During MJJ and ASO the countercurrent strengthens and develops into a continuous eastward flow, with Figures 3b and 3c showing a corresponding response in the barrier layer. By NDJ, the ECC has weakened considerably, and the presence of the barrier layer (Figure 3d) is once again confined to the northwestern side of the Atlantic.

The spreading of the water containing the salinity maximin is mainly by advection. The salinity maximin fixes the position of the pycnocline, and the corresponding stability may strongly suppress turbulence effects, so that there is almost horizontal spreading of the salinity surface. In this case, a simple model may be devised whereby the advection of the salinity maximum water balances the surface net heat flux (QN) such that

z

u•T.)dz QN= I P Cw ( '•'• o

(4)

where Z is the mixed layer depth of 30 m [Sprintall and Tomczak,


1990], u is the current speed along the direction x, and Cw is the specific heat of water at 4180 W s / kg øC. Recent estimates of net annual heat flux by Oberhuber [1988] in the equatorial northwest Atlantic are of the order of -5 W m-2, while the Levitus [ 1982] maps of surface temperature give a horizontal gradient of 1 x 10-6 øC m-1. This leads to an estimated westward current speed (u) of approximately -4.0 cm s-1. Robinson and Stornmel [1959] estimate the mean westward velocity of the layer containing the salinity which is in general agreement with our results. Of course this estimate must be treated with caution due to insufficient knowledge concerning typical heat fluxes and advective velocities. Still it appears likely that the presence of the shallower mixed than isothermal layer in the north equatorial Ariantic is maintained not so much by the dilution of salinity in the surface layers (although the outflow from the Amazon and Orinoco Rivers probably contribute to this) as by the presence of a well defined salinity maximum subducted from the surface in the subtropics below the mixed layer and advected by the major equatorial current system of the region.

The Eastern Equatorial Indian Ocean

The third region where the isohaline layer is shallower than the isothermal layer is located in the small area immediately to the west of Sumatra (Figure 4), extending along the whole western coastline

during MJJ (Figure 4b), ASO (Figure 4c), and NDJ (Figure 4d), west to around 85øE and bounded by the 25-m isopleth. During FMA (Figure da) there is a uniform 10-m barrier layer present over much of the Indian Ocean north of 10ø$. This 10-m isopleth is also present during other seasons, extending northward into the Bay of Bengal and south into the equatorial Indian Ocean to -10ø-20øS.

According to Oberhuber [1988] the region lying adjacent to the west coast of Sumatra receives an annual freshwater flux of-1800

mm yr-1. This region of local maximum in (P-E) is present during all months, indicating that rainfall likely contributes to the existence of the salinity stratified surface layer of this region.

To the north, in the Bay of Bengal, Oberhuber [ 1988] shows that the freshwater flux would be sufficient to produce a fresh lens at the surface only during the wetter months of the southwest monsoon (June-September), and even then only in the eastern Bay of Bengal. Frequently, the western Bay of Bengal experiences a regime of evaporative dominance in the freshwater flux. It is hypothesized that in this region the freshwater flux required to maintain the barrier layer is obtained from the substantial river runoff into the northern Bay of Bengal, and furthermore the seasonality which exists in this runoff is reflected in the seasonal distribution of the barrier layer. The Ganges-Bramaputra-Irrawaddy contribute an annual average ranoff of 455-1070-1020 ram, respectively [Baumgartner and Reichel, 1975], into the north and eastern sides of the Bay of

30øN

20øN

10ON

o

10øS

20øS

(a) 30øN

(b)

20øN

10ON ...........

0 ø .,

10øS

20øS

30øS 30oS 60øE 80øE 100øE 120øE 60øE

Fig. 4. Differences (meters) between isothermal and mixed layer depth for the Indian Ocean during (a) Febmary-April, (b) May- July, (c) August-October, and (d) November-January. Positive differences (solid lines) indicate a shallower isohaline layer than isothermal layer, and numbers give the thickness of the barrier layer.

12( )OE


30øN

(c) 30øN

(d)

20øN 20øN

10øN

o

10øS

20øS

30øS

60øE

•5

10

80øE 100øE 120øE

10øN

o

10øS

20øS

30øS

60øE 8(

----. ............. I..- ............ I. ........... 4 ...........

tOE 100øE 120øE

Fig. 4. (continued)

Bengal. This freshwater flux is more than adequate to sustain a shallow isohaline layer. Most of this river runoff occurs during the southwest monsoon and results in a strong horizontal surface salinity gradient southward across the Bay of Bengal from a low of 30 in the north to around 34 in the south (~5øN) [Pickard and Emery, 1990]. A strong pycnocline also forms as a result of this freshwater discharge, leading to stable stratification in the upper layers of the water column. Previously estimated mixed layer depths based on temperature data alone [e.g., see Colburn, 1975; Krishna et al., 1988] indicate horizontal homogeneity of temperature in the isothermal layer throughout the region during this period, and surface temperature maps of Pickard and Emery [ 1990] indicate that the entire Bay of Bengal region south to the equator is of the same 28øC temperature. This enables the warm, low-salinity surface waters from the fiver runoff to form a slab over the warm, more

saline waters of the eastern equatorial Indian Ocean, producing strong halodines and their resulting pycnoclines within the warm isothermal layer (Figure 4c). This supports the present hypothesis that stratification due to the considerable freshwater flux from river

runoff plays a dominant role in the surface layer dynamics. Gradually as the runoff decreases during the drier northeast

monsoon in February-April, the salinity in the Bay of Bengal increases and becomes more uniform, while the temperatures

decrease from 28øC at the equator to 25øC up in the very north. This removes the mechanism for advection of the low-salinity water which maintains the halocline. Hence active mixing again takes place until finally the halocline is eroded and the barrier layer is diminished (Figure 4a).

Another region of the Indian Ocean that displays a shallower isohaline than isothermal layer is on the southeastern side of the Arabian Sea. Here, 10-m barrier layers are present during FMA (Figure 4a) and ASO (Figure 4c) and have thickened to 25 m during the intervening NDJ (Figure 4d). The influence of the monsoonal system is again evident; however, it seems to be a season out of phase with the Bay of Bengal region. In this case, river runoff from the Indus River provides the most substantial contribution to the freshwater flux.

During MJJ (Figure 4b) and ASO (Figure 4c) the thick barrier layer centered at ~ 20øS, 80øE is probably associated with Central Water formation and subduction induced by Ekman pumping as shown by the negative isopleths in the subtropical region. This process is similar to that found and described in the equatorial Atlantic; however, in the Indian Ocean it is present mainly during the austral winter months, as obviously Central Water formation can only occur in the southern subtropics of the Indian Ocean basin due to the proximity of the northern Asian land mass.


4. SUMMARY AND CONCLUSIONS

In this analysis we determine the thickness of the barrier layer that separates the mixed layer from the thermocline in the tropical regions of the world oceans. The thermocline depth was estimated as the depth at which a net temperature change of 0.5øC occurs from the surface. This method was frequently employed by previous studies as being proxy to mixed layer depth. Here, the mixed layer depth is calculated as the depth where the potential density is the sea surface density plus an estimate of the density change required to obtain a temperature difference from the surface equal to 0.5øC if salinity is held constant. Hence any differences in depth produced by the two criteria must be caused by salinity effects.

Three equatorial regions consistently displayed a significant positive difference between the mixed layer depth and the isothermal layer. Different forcing mechanisms were proposed for the existence of the shallow haloclines in each of the three regions. In the westem equatorial Pacific a salinity budget confmned that heavy local precipitation most likely results in the formation of the layer which is isothermal but salt stratified. This is expected due to the coincidence of the distribution of the thick barrier layer with a global minimum in E-P. Understanding the thermodynamics that maintain the barrier layer, and also its role in the E1 Nifio-Southern Oscillation instability will necessarily be of high priority in the imminent Tropical Ocean Global Atmosphere-Coupled Ocean Atmosphere Response Experiment (TOGA-COARE) in the western Pacific warm pool region.

In the Guiana Basin of the north equatorial Atlantic, the region is characterized by net water loss to the atmosphere; hence precipitation cannot be responsible for the distribution of the barrier layer here. It appears that the high salinities found at the surface in the subtropics are subducted from both hemispheres toward the equator during the respective winter season, at the upper end of the temperature/salinity range of the Central Water. This creates a salinity maximum above the Central Water in the tropics. This may then be advected westward into regions of uniform temperature in the equatorial current system, forming a barrier layer in the tropical Atlantic Ocean.

In the eastern equatorial Indian Ocean, monsoonal related rainfall and river runoff produce salinity stratification in the upper layers. The Ganges-Bramaputra-Irrawaddy river systems emptying into the Bay of Bengal during the southwest monsoon season contribute to the freshwater flux, maintaining the shallower isohaline layer, while in the eastern Arabian Sea this task is undertaken by the Indus River.

The above findings stress the importance of including the effects of salinity stratification when determining mixed layer depth of the world oceans. Previously it was thought that the significance of haloclines in the surface layer need only be considered for higher latitudes. However, the results presented here indicate that the inclusion of salinity dynamics are also necessary in the mixed layers of the tropical regions. Often a depth produced by temperature criteria alone does not truly represent the depth of convective overturn which physically defines the mixed layer. The variable c•t criterion for mixed layer depth used here allows for a comparison of mixed layer depths on a world ocean basis, as it accounts for the differences in surface salinities and temperatures with their effect on the coefficient of expansion of seawater. In regions where salinity stratifications are not of consequence, mixed layer depths of the same order as derived from a temperature-based criterion will be found. However, where they differ, truly representative measures of mixed layer depths may be quantified.

REFERENCES

Bamett, B., The principal time and space scales of the Pacific trade wind fields, J. Atmos. Sci., 34, 221-236, 1977.

Baumgartner, A., and E. Reichel, The World Water Balance: Mean Annual Global, Continental and Maritime Precipitation, Evaporation and Runoff, 179 pp., Elsevier, New York, 1975.

Bunker, A.F., Computations of surface energy flux and annual air-sea interaction cycles of the North Atlantic Ocean, Mon. Weather Rev., 104, 1122-1140, 1976.

Colburn, J.G., The Thermal Structure of the Indian Ocean, 173 pp., University of Hawaii Press, Honolulu, 1975.

Defant, A., Physical Oceanography , Vol. 1, 729 pp., Pergamon, New York, 1961.

Delcroix, T., G. Eldin, and C. Henin, Upper ocean water masses and transports in the western tropical Pacific, J. Phys. Oceanogr., 17, 2248- 2262, 1987.

Dorman, C.E., and R.H. Bourke, Precipitation over the Atlantic Ocean, 30øS-70øN, Mon. Weather Rev., 109, 554-563, 1981.

Elliott, G.W., Precipitation signatures in the sea surface layer conditions during BOMEX, J. Phys. Oceanogr., 4, 498-501, 1974.

Enfield, D.B., Zonal and seasonal variations of near surface heat balance of the equatorial Pacific Ocean, J. Phys. Oceanogr., 16, 1038-1054, 1986.

Gill, A.E., Models of the equatorial currents, in Numerical Models of Ocean Circulation, pp. 181-203, National Academy of Sciences, Washington D.C., 1975.

Godfrey, J.S., and E.J. Lindstrom, The heat budget of the equatorial western Pacific surface mixed layer, J. Geophys. Res., 94, 8007-8017, 1989.

Goldenburg, S., and J.J. O'Brien, Time and space variability of tropical Pacific wind stress, Mon. Weather Rev., 109, 1190-1207, 1981.

Krishna, V.V.G., Y. Sadhuram, and V. Rameshbabu, Variability of mixed layer depth in the northern Indian Ocean during 1977 and 1979 summer monsoon seasons, Indian J. Mari. Sci., 17, 258-264, 1988.

Landis, R.C., Early BOMEX Results of sea surface salinity and Amazon river water, J. Phys. Oceanogr., 1,278-281, 1971.

Levitus, S., Climatological Atlas of the Word Ocean, NOAA Prof. Pap., 13, 173 pp., Natl. Oceanic and Atmos. Admin., Rockville, Md., 1982.

Lewis, M.R., M.E. Carr, G.C. Feldman, W. Esaias, and C. McClain, Influence of penetrating solar radiation on the heat budget of the equatorial Pacific Ocean, Nature, 347, 543-545, 1990.

Lindstrom, E., R. Lukas, R. F/ne, E. Firing, J.S. Godfrey, G. Meyers, and M. Tsuchiya, The Western Equatorial Pacific Ocean Circulation Study, Nature, 330, 533-537, 1987.

Lukas, R., On the role of the western Pacific air-sea interaction in the E1 Nino / Southern Oscillation phenomenon, Proceedings of the U.S. TOGA Western Pacific Air-Sea Interaction Workshop, edited by R. Lukas and P. Webster, pp. 43-69, UCAR Tech. Rep., US TOGA 8, Univ. Corp. for Atmos. Res., Boulder, Colo., 1988.

Lukas, R., and E. Lindstrom, The mixed layer of the western equatorial Pacific Ocean, J. Geophys. Res., 96 (Suppl.), 3343-3357, 1991.

Luyten, J.R., J. Pedlosky, and H. Stommel, The ventilated thermocline, J. Phys. Oceanogr., 13(2), 292-309, 1983.

McPhaden, M.J., and J. Picaut, E1 Nifio-Southern Oscillation displacements of the western equatorial Pacific warm pool, Science, 250, 1385-1388, 1990.

Neumann, G., Seasonal salinity variations in the upper strata of the western tropical Atlantic Ocean, 1. Sea surface salinities, Deep Sea Res., 16, 165-177, 1969.

Oberhuber, J.M., An atlas based on the COADS data set: The budgets of heat, buoyancy and turbulent kinetic energy at the surface of the global ocean, Rep. 15, Max-Planck-Institut ffir Meteorologie, Hamburg, 1988.

Peters, A., Peter's Atlas of the World, 231 pp., Longman Press, Harlow, 1989.

Pickard, G.L., and W.J. Emery, Descriptive Physical Oceanography, 5th ed., Pergamon, New York, 1990.

Robinson, W., and H. Stommel, The oceanic thermocline and the associated thermohaline circulation, Tellus, 11 (3), 295-308, 1959.

Schmitt, R.W., P.S. Bogdena, and C.E. Dorman, Evaporation minus precipitation and density fluxes for the north Atlantic, J. Phys. Oceanogr., 19, 1208-1221, 1989.

Shea, D.J., Climatological Atlas: 1950-1979, National Center for Atmospheric Research, Boulder, Colo., 1986.


Sprintall, J., and M. Tomczak, Salinity considerations in the oceanic surface mixed layer, Ocean Sciences Institute Rep. 36, 170 pp., University of Sydney, 1990.

Taylor, R.C., An atlas of Pacific islands rainfall, Data Rep. 25, HIG-73-9, 83 pp., Hawaii Inst. of Geophys., Honolulu, 1973.

Tomczak, M., and G. Krause, The Peters projection as a tool for oceanographers during WOCE, WOCE Newslett., 8, 9-11, 1989.

U.S. Navy Oceanographic Office, U.S. Navy Oceanographic Atlas of the North Atlantic, Washington, D.C., 1965.

Weare, B.C., P.T. Strubb, and M.D. Samuel, Annual mean surface heat fluxes in the tropical Pacific Ocean, J. Phys. Oceanogr., 11,705-717, 1981.

Worthington, L.V., On the North Atlantic Circulation, 110 pp., John Hopkins University Press, Baltimore, Md., 1976.

Wyrtki, K., and G. Meyers, The trade wind field over the Pacific Ocean, Part 1, The mean field and the mean annual variation, Publ. HIG-75-1, 26 pp., Univ. of Hawaii, Honolulu, 1975.

J. Sprintall, NOAA/PMEL, 7600 Sand Point Way, Seattle, WA 98115. M. Tomczak, School of Earth Sciences, Flinders University, GPO Box

2100, Adelaide, SA 5001, Australia.

(Received May 2, 1991; revised January 17, 1992;

accepted January 17, 1992.)

Documents

Evidence of the Barrier Layer in the Surface Layer of the Tropics