thermocline circulation and ventilation in the indian ...klinck/reprints/pdf/youdsr1993.pdfthe...
TRANSCRIPT
Deep-Sea Re.~eurrh I. V,.)I .-~). No. 1. pp I.!-50, lta,J3 0(~7-,(~37/t)3 S6.IIO .*. O.(t() Pnnted m Grea t Brttaln. ~ t t~ " Pergamon Pres6 Ltd
Thermocline circulation and ventilation in the Indian Ocean derived from water mass analysis
Y. You** and M. TOMCZAKt.~
(Received 23 July 1991; in revised[otto 27 January 1992; accepted 18 February 1992)
Abstract--A mixing model, which combines cluster analysis with optimum muitiparameter (OMP) analysis, is used to determine the spreading and mixing of water masses in the thermocline of the Indian Ocean. focusing on the ventilation process for the thermocline in the northern hemisphere. Mixing ratios are quantified and plotted on five isopycnal surfaces covedng the depth range 150- 800 m. two meridiona[ sections along 60°E and 90°E, and one zonal section along I0*S.
Three water masses are identified in the thermocline by cluster analysis. Indian Central Water (ICW) is subducted at the Subtropical Front in the southern Indian Ocean and advected with the southern subtropical gyre. Australasian Mediterranean Water (AAMW) enters from the Indone- sian seas as the result of throughflow from the Pacific Ocean. Red Sea Water (RSW) combines with water from the Persian Gulf to provide a minor source. A fourth water mass identified by cluster analysis occurs in the Bay of Bengal; it is labelled North Indian Central Water (NICW) and interpreted :is aged ICW. Cluster analysis did not produce evidence for the existence of Equatorial Water. a water mass often referred to in the literature.
Mixing ratios attd pathways of the thermoelinc water masses are established using OMP analysis. The input of RSW is insufficient to renew the thcrmoclinc witters of the northern Indian Ocean, which therefore has to be ventilated by ;tdvection front the south. The jet-like inflow of AAMW produces one of the strongest frontal systems of the world (~ce;.ln°s thcrmoclinc, suppressing mcridional motion across 10--15°S cast of 51)°E. This leaves the western boundary currents as the only rcgitm for advectivc transfer of thcrmoclinc water between the hemispheres. ICW is shown to advect along this path on the isopyenal surface ~r u = 26.7 (depth range 3(10-4(10 m). Above and below this surface ICW movement into the northern hemisphere is accompanied by significant diapycnal mixing. The annual mean distribution of ICW shows that it ages rapidly as it crosses the equator. This is interpreted as the result of seasonally varying advcction, producing very little mean net transport across the equator. As a consequence. ICW in the northern hemisphere is extremely low in oxygen and high in nutrients.
AAMW can be traced to 70"E, but is mixed into the ICW background and no longer recognizable as AAMW by the time it leaves the Indian Ocean with Agulhas Current eddies. It is argued that the so-called Equatorial Water is the result of the mixing process, i.e, a mixture of two well-defined water masses that should not be considered a water mass in its own right.
I. I N T R O D U C T I O N
THIS p a p e r s t u d i e s t h e p a t h s o f w a t e r m a s s e s in t h e I n d i a n O c e a n n o r t h o f 40°S b e t w e e n
a b o u t 150 a n d 800 m d e p t h ( t h e p e r m a n e n t t h e r m o c l i n e ) . SVERDRUP et al. (1942) i d e n t i f i e d
"Marine Studies Center, The Univcrsity of Sydney. NSW 2006, Australia. *Ocean Sciences Institute, The University of Sydney, NSW 2006, Australia. *Present address: School of Earth Sciences, The Flindcrs University of South Australia, G.P.O. Box 21(10.
Adelaide. S.A. 5(101, Australia.
13
14 Y. You and M. TOmCZAK
three water masses in this depth range and discussed their formation mechanisms. According to their description. Red Sea Water (RSW) forms in the mediterranean basin of the Red Sea, Indian Central Water (ICW) is subducted in the Subtropical Convergence of the southern hemisphere, and Indian Equatorial Water (IEW) is formed in the western equatorial region through some unspecified mechanism. In temperature-salinity (T -S ) diagrams ICW and IEW were displayed as nearly linear relationships over the entire thermocline range between these two conservative properties. MAMAYEV (1975) added two more water masses to the Indian Ocean north of 40°S: Bengal Bay Water, which is only found in the surface layer, and Timor Sea Water--in this paper referred to as Australasian Mediterranean Water (AAMW)---which originates from the deep basins of the Indonesian archipelago (the Australasian Mediterranean Sea). With respect to the processes of formation, Mamayev put strong emphasis on vertical mixing and did not consider subduction and spreading on isopycnal surfaces a possibility, so all his water mass definitions are based on single T-S points. A recent review by EMERY and MEINCKE (1986) returns to the ideas expressed in SVERDRUP et al. (1942) and presents the water masses of the Indian Ocean thermocline as T-S relationships rather than points. It also includes Arabian Sea Water [water influenced by outflow of Persian Gulf Water (PGW) through the Strait of Oman] as another water mass distinct from RSW, thus giving a total of six water masses for the upper kilometre of the northern Indian Ocean.
Modern methods of water mass analysis enable us to expand the pioneering work of SVERDRUP et al. (1942) beyond the definition of water mass properties and derive the water mass distribution in space through objective methods. In the Indian Ocean this is particularly true, since the International Indian Ocean Expedition (IIOE) of 1960-1965, documented in the atlas by Wvrr~l (1971), created an invaluable database, allowing the application of inverse modelling techniques. This paper uses the IlOE dataset and additional data collected at the World Oceanographic Data Centre for an application of optimum multiparamctcr (OMP) analysis. The mcthod, first proposcd by TOMCZA~ (198 i) and further developed by MACKAS et al. (1987) and TOmCZAR and LARGE (1989), USCS nutrients and dissolved oxygen along with temperature and salinity to derive a distribution of water masses that matches the observed distribution of properties in an objectively definable best tit.
The aim of the present study is to establish pathways for water masses in the permanent thermocline and contribute to our understanding of the upper circulation and thermocline ventilation in an ocean that has received surprisingly little attention, despite improve- ments in data density and its undeniably important role in climate variability. One example of the outstanding problems awaiting explanation is the origin of the so-called Indian Equatorial Water. SvErt)ruP et al. (1942) introduced this water mass but did not offer a formation mechanism, stating instead that IEW is not as well-defined as Indian Central Water. SHCIIEratNIN (1969) distinguished three main water masses in the equatorial area of the Indian Ocean; the central Indian; the equatorial transitional; and the north Indian; using maximum horizontal gradients of oceanographic parameters to define water mass boundaries, and again without reference to processes of formation. Using temperature- salinity frequency analysis, S~ARmA (1976) argued that IEW is a mixture of water masses from the northern and southern hemispheres and from the Pacific Ocean. This point of view is supported by Warren in a personal communication to QUADFASEL and ScHorr (1982), which suggests that IEW off the Somali Coast originates from AAMW (referred to as Banda Intermediate Water by Warren), and Antarctic Intermediate Water, which
Thcrmodine circulation and ventilation in the Indian Ocean 15
enters the equatorial Indian Ocean from the Pacific Ocean and is carried westward by the South Equatorial Current. If IEW is indeed merely a mixture of other water masses (i.e. formed without a contribution of air-sea interaction), the Red Sea and Persian Gulf Waters are the only sources for thermocline ventilation north of the equator. But the volumes of both sources are much too small to have a significant impact on nutrient and oxygen levels in the thermocline. It seems fair to say, therefore, that the ventilation of the Indian Ocean thermocline in the northern hemisphere is not well understood.
2. D A T A A N D M E T H O D S
The historical hydrographic data archived by the World Oceanographic Data Center in Washington (obtained through the Australian Oceanographic Data Centre in Sydney) were used for the study. The data contain physical-chemical observations recorded at discrete depth levels. Most of the observations were made using multi-bottle Nansen casts or other types of water samplers. A small amount of the data was obtained using CTD or STD instrumentation with rosette samplers. In this type of data set, the parameters potentially useful for OMP analysis include temperature, T, salinity, S, oxygen, O2, phosphate, PO4-P, nitrate, NO3-N and silicate, Si. Unfortunately, few of the historical stations contain nitrate data, and inclusion of nitrate in the analysis reduces the numbers of useful stations significantly. We therefore decided to use only five variables (T, S, O,, PO~-P, Si) in return for an acceptable station coverage, and interpolated all data linearly for five selected density surfaces. Figure I shows the station distribution for the oo = 25.7
40 ° E 60 ° 80 ° 100 ° 120 ° 140 ° E
I / I I 1 I I I I I
' ~ ~ , ~ ~ stations / , "
oo_ I _
- ~ , ~ ; ! . : : : ! .~" " 20 o_ ~20 °
" • ' ' . " • . ' . • - ' " • • •
40*S ( 40 S
I .I I I I I I I '1 I I
40 ° E 60 ° 80 ° 100 ° 120 ° 140 ° E
Fig. l. The distribution of hydrographic stations on the o, = 25.7 isopycn;d surface. Each station has five parameters: temperature, salinity, oxygen, phosphate and silicate.
16 Y. You and M. TOMeZAK
isopycnal surface that has the largest number of stations. We did not attempt to perform OMP analysis separately for the two monsoon seasons, which would have reduced the data density for each analysis beyond acceptable levels. All results presented here represent long-term climatological mean distributions.
We define a water mass as a body of water with a common formation history of all its elements. The permanent thermocline is shared between several such water bodies, which can penetrate each other as a result of mixing. Water masses are physical entities of finite volume; mathematically they can be described by functional relationships between their characteristic properties and a set of standard deviations. A water type is defined as a point in parameter space; it is a mathematical construct and does not occupy any volume in space. The functional relationship that is part of the definition of a water mass can then be described by an infinite set of water types, which we call source water types. In practice, the number of source water types for each water mass is finite and small. For example, if the fuctional relationship between all parameters of a water mass is linear, the number of source water types needed to describe the water mass is reduced to two.
As in previous applications of OMP analysis, all parameters are considered conserva- tive. This is not an unreasonable assumption if the analysis is restricted to mixing in the upper kilometre of the ocean over horizontal distances of order of hundreds of kilometres, since effects of advection and turbulent diffusion clearly outweigh any biochemical effects on that scale. The present analysis is concerned with circulation features on an oceanic scale where biochemical oxygen consumption and nutrient gain cannot be ignored. To cope with these effects in the framework of OMP analysis, we introduce a virtual water mass, i.e. a "water mass" for thc purpose of OMP analysis but not in the sense of our above definition, which we call North Indian Central Water (NICW). As will be seen later, NICW can be considered to represent aged Indian Central Water. (An alternative way would be to convert some of the nonconscrvative parameters into conservative ones, e.g. phosphate into initial phosphate. This way shows much promise but is fraught with problems at present. A preliminary discussion is given in the Appendix.)
We apply cluster analysis to determine the number of source water types required for adequate representation of all water masses in the Indian Ocean thermocline and their parameter values. Su et al. (1983) successfully applied this statistical method to define "modified water masses" in shallow seas from temperature and salinity data. The method generally can be applied to lind natural groupings, or clusters, of data in a situation where no previous knowledge is available about the number and nature of water masses present in a given region. Such clusters are usually identified by a parameter combination or point in parameter space (the cluster centre), which in the context of water mass analysis can be interpreted as a source water type. By applying cluster analysis on a number of isopycnal surfaces in succession it is possible to derive a series of source water types across the density range of the thermocline and approximate the parameter relationships of the unknown water masses to any required accuracy.
Detailed descriptions of cluster analysis can be found in ANDERBERG (1973), EVERITr (1980), ROMESaUra (1984), and others. Very briefly, application of the method in the present study proceeds as follows. On each isopycnal surface the hydrological data set is considered a matrix X of dimension n x p, where n is the number of stations and p the number of variables. The elements xii of the matrix are the normalized observational data (normalization is essential to convert all variables into comparable entities). Cluster analysis applies a distance algorithm to the data matrix to convert it into a matrix Yof inter-
Thcrmocline circulation and ventilation in the Indian Ocean 17
individual distances dq. The choice of algorithm is arbitrary to some degree and depends on the problem. We used the squared Euclidean distance
P
dq = Z (xa' - x/k)2 (1) kffil
to calculate Y. Clusters are identified by data with small distances between them. The next step is the use of hierarchical cluster analysis to establish a "clustering tree", which depicts the degree of relationship between clusters. An average linkage method is used to evaluate the potential merger of clusters i and j in terms of the average similarity for links between two clusters
N N
= N- N, (2)
where Ni and N i are the number of members in cluster i and/. respectively. The cluster tree shows the successive fusions of individuals, which culminate at the stage where all individuals are in one group. Which level of generalization (between a maximum of n x p clusters containing one data point each and a single cluster containing all data points) is appropriate is determined through interpretation of clusters and cluster groups as water masses, by association with recognisable formation processes.
Having established the water masses from cluster analysis, we proceed to calculate their distribution through OMP analysis. The method solves a linear system of mixing equations for each data point in which all water masses are represented through water types (see TOMCZAg and LARGE, 1989). Briefly, the linear system can be written for any water sample as
Gxg - xv = R (3)
where G is a matrix containing the parameter valucs which define the source water types, x,, is a vector containing the parameter values for the sample (the observations), xr, is a vector containing the relative contributions, or mixing ratios, of the source water types to the sample, and R is a vector containing residuals. Solving equation (3) by minimizing the sum of the squared residuals leads to the determination of the minimum of
RTR = (Gxg - Xv)TWTW(Gxg -- Xv) = W~ G/ixgi - x,j (4) j ~ l ffi
where W are weights attributed to the various parameters to reflect differences in measurement accuracy, degree of conservativeness, and other processes which may render some parameters less reliable than others. The weights are derived from
w~ = o./ i m,. (5)
where oj is the standard deviation of parameter/over the entire dataset (a measure for the ability of parameter/ to resolve differences in water mass content) defined as
ai= J l ~ (G, i - G/) 2 (6)
18 Y. You and M. ToMcz^K
and b/mi, is the smallest of the variances for the water masses. G/is a mean given by
cj = ! cj,. (7) /l
i= l
The largest of the weights W i (usually that of temperature) is also allocated to the mass conservation equation, since the above method of weight calculation is not applicable to mass conservation.
As mentioned above, five parameters were found to provide useful information for the present analysis. This restricts the number of source water types that can be handled by OMP analysis to a maximum of six and raises the problem whether six water types are sufficient to represent all water masses in the region under consideration. Our earlier discussion established the presence of six water masses in the Indian Ocean thermocline (ICW, IEW, AAMW, RSW, PGW, Bengal Bay Water). We also anticipated the use of an additional virtual water mass (NICW). A closer look at the definitions of all water masses is therefore required before the analysis can proceed.
Bengal Bay Water is the result of excess precipitation and river runoff over evaporation and characterized by very low salinity (between 32 and 33). This low salinity causes it to be isolated from the water in the thermocline by a sharp halocline located between 50 and 100 m depth. Bengal Bay Water is therefore clearly restricted to the surface layer and can be excluded front further discussion here.
Red Sea Water and Persian Gulf Water both result from excess evaporation over precipitation, and both have similarly high temperatures and salinities. The effect of the Persian Gulf Water on thcrmocline waters in the Arabian Sea is small even in the north (SilARMA, 1976), and the Arabian Sea Water identified by EMERY and MEINCKE (1986) is already a mixture of PGW and RSW. From thc point of view of large-scale circulation and ventilation, both sources can bc combined into a single source of high-salinity water if details of the water mass distribution in the northern Arabian Sea arc not addressed, and can then bc considered a point source located in the northwest of the Indian Ocean. Because the Red Sea is by far the more important of the two, we refer to water flowing from the combined source as RSW and represent it through a single source water type, detincd by the parameter valucs of Red Sea Water on its entry into the Gulf of Aden.
Australasian Mediterranean Water - - the Timor Sea Water of MAMAYEV (1975) or the Banda Intcrmcdiatc Water of RocH~:oro (1966)--is a major water mass of the world ocean. It is characterized by uniform salinity of about 34.8 or less throughout the thermoclinc range. The IIOE atlas (Wvr rKI, 1971 ) shows it as a tongue of low salinity and high silicate extending westward. AAMW plays a vital role in GorDoN'S (1986) global model of the thermoclinc circulation as part of the warm water route of North Atlantic Dccp Water return flow. Accurate representation of A A M W thus is a key element of any water mass analysis in the Indian Ocean. From the work of TOMCZAK and LarGe (1989) it can bc anticipated that two source water types are sufficient to represent AAMW.
Indian Central Water is also a well-defined water mass. SPrINTALL AND TOMCZAr (1992b) show that it is formed along the Subtropical Convergence in the southern Indian Ocean and contributes to thcrmocline ventilation both in the Indian and~ th rough advection in Agulhas Current eddies-- the South Atlantic Ocean. As a result of the subduction process and subsequent mixing, property relationships in ICW are nearly
Thermocline circulation and ventilation in the Indian Ocean 19
linear, and two source water types are again sufficient to represent the water mass (ToMCZAK and LARGE, 1989).
Indian Equatorial Water is not associated with a clearly established formation mechan- ism in the literature. It also does not appear in our cluster analysis (to be discussed below). We therefore agree with the arguments of SHARMA (1976) and Worthington, and exclude IEW from our compendium of water masses. Instead. we consider it a mixture of AAMW and ICW (and in its lower range possibly Antarctic Intermediate Water). In our results IEW therefore will be present as a mixture of AAMW and ICW and not be identified by name.
We are thus left with three water masses ( tCW. AAMW and RSW) that contribute to the ventilation of the Indian thermocline and can be represented through five water types. However. as will be seen in the next two sections, most observations in the Bay of Bengal cannot be explained on the basis of mixing these three water masses alone. The virtual water mass NICW is needed to produce meaningful results, and it requires two additional water types for its representation. This gives a total of seven source water types for the present problem, more than can be handled by OMP analysis on the basis of five parameters. Fortunately it turns out that the influence of RSW is negligible everywhere but in the upper thermocline west of 70°E, where AAMW in turn happens to contribute to the mixture in only negligible quantities. This allows us to proceed with OMP analysis in two parts: west of 70°E and above 25(1 m the analysis includes the five source water types of ICW, NICW and RSW; in the remainder of the region it is based on the six source water types of ICW, NICW and AAMW. Note that with five parameters and six source water types, OMP analysis does not degenerate into a uniquely determined system of equations (the multiparanlctcr analysis of TomczAK, 1981 ) because the solutions arc still subject to the non-negativity constraint and therefore have to be determined through rcsidual minimization.
3. TIlE DIS'FRIBU'FION OF PARAMETERS ON ISOPYCNAL SURFACES
Five isopycnal surfaces, defined by oo = 25.7, 26.7, 26.9, 27.1 and 27.3, were mapped in the thermoclinc of the Indian Ocean north of 40°S. The distance between surfaces is about 2(1(I m, being somewhat less in the north anti larger in the south. Pressure, potential temperature, salinity, oxygen, phosphate and silicate are shown respectively in panels (a)- (f) of Figs 2--6.
A remarkable feature of the Indian Ocean thermocline is that only one of the five isopycnal surfaces shown, the oo = 26.7 isopycnal surface found between 300 and 400 m (Fig. 3), offers some evidence for isopycnal mixing. This is obvious from the figures if it is recalled that on an isopycnal surface temperature and salinity remain constant in a water mass (as long as it does not mix with another water mass). It is then noted that the distributions of potcntial temperature (Fig. 3b) and salinity (Fig. 3c) show little change of both parameters along a path that follows the South Indian Ocean Current (Stramma, 19921 from the Subtropical Convergence (30°-40°S) west of 70°E eastward and northward around the subtropical gyre, then west along 20°S, across the equator west of 50°E, and eventually into the Bay of Bengal: potential temperature remains close to 12.0°C and salinity near 35.1 along the path. This corresponds to the O-S values of ICW on this density surface and indicates that this water mass spreads into the Bay of Bengal without mixing
20 Y. You and M. TOMCZAK
20°N
0 ° _
20 °
o 40 S -
Fig. 2,
40 ° E 60 ° 80 ° 100 °
~, ' ~,~.~,' ~0,oo'~so,.c'. o . ' 2 , , ' ) ,.'
•
o 120 140°E
_ 2 0 ° N
m
m O °
" 2 0 °
- - 4 0 ° S
I=" I I I I I I I I I 1
40 ° E 60 ° 80 ° 100 ° 120 ° 140°E
Dist r ibut ion o f hydrographic propert ies on the % = 25.7 isopycnal surface: (a) pressure, (h) potent ial temperature. (c) sal inity, (d) dissolved oxygen, (e) phosphate and (f) silicate.
20 ON --
0 ° _
20 ° _
o 4O S -
40°E 60 ° 80 ° 100 °
• ~ ' ~,~ ' l,oo,oo'=,u.a~, o :~., ' / ,-'
18.o / l s . o \ lS.O"-~,~'~ j / " ,~ / /<o1"~ , ,,~ .., ~ I
o o 120 140 E
- 20 ° N
m O °
- 2 C °
- 40 ° S
I I I ~1 I I ~ I I I 0 0
40 E 60 ° 80 ° 100 ° 120 140°E
Fig. 2(b)
Thcrmocl inc circulation and venti lation in the Indian Ocean 2 ]
20°N --
0 ° _
20 ° _
40 ° S -
40 ° E 60 ° 80 ° 1 O00 120 ° 140°E
I I I ~ I I I I I I
40°E 60 ° 80 ° 100 ° 120 ° 140 E
_20°N
_ 0 °
- 2 0 °
4 0 ° S
Fig. 2(c)
20°N --
M
0 ° _
20 ° _
40°S t
40 ° E 60 ° 80 ° 100 ° 120 ° 140 ° E
I / I I 1 I 1 I / I ~ isopycnal surface Cr . 25.7 / ' / "
, ~ " - - 0 °
- 2 0 °
l I l II~ l I l l I I I o o
40 E 60 ° 80 ° 100 ° 120 ° 140 E
Fig. 2(d)
22 Y. You and M. TOMCZAK
40 ° E 60 ° 80 • 1 O0 ° 120 o 140 ° E
20 N
20 °
o o2 © ¢ 0. 40 ° S' l
I I I I ~ ] I I I I I
40 ° E 60 ° 80 ° 100 ° 120 ° 140 ° E
_20°N
_0 °
-20 °
o -40 S
Fig. 2(c)
20°N _
0 ° _
20 ° _
40°S t
o o o 40 ° E 60 ° 80 100 120 140 ° E
I / I I I I I 1 I I
I I 1 ~ ' I I I I I I • 0 0
40 E 60 ° 80 100 120 ° 140 E
_20°N
_0 °
- 2 0 °
,,tOe S
Fig. 2(f)
Thcrmoc l in¢ c i rculat ion and vent i la t ion in the Indian Ocean
20°N -
0 ° _ _
20 ° _
40 ° S -
40 ° E 60 ° 80 ° 100 °
~,,' ,,,L...,,' 'o0,cn',,ur,,,c', ,, :2, , ' ~ ,.'
&
-;,
120 o 140 ° E
_20°N
L
_ 0 °
o - 2 0
. 4 0 ° S
Ir I .J I I I I ! I I I
40 ° E 60 ° 80 ° 100 ° 1 2 0 ° 140 E
Fig. 3. Di~,triht, t i lm ~t h~'dr~graphic pr~lpcrtic~ ~n the t l , = 2h.7 i',~.~p.vcm,I '~urfacc: (a) prcY.,urc. (b) p~tcn¢ial temperature . (c) ,,~dinity. (d) di,,,,~lvcd ~,xy~cn. (c) ph~,~ph;ntc and (f) ,;il icatc.
40 ° E 60 ° 80 ° 100 °
2o"N.--I ~ \ 2",,',s.x--'l '..' c°cl.¢"X ~ ~
0 12 .
I .I I I I I i i I i
40 ° E 60 ° 80 ° 100 °
120 ° 140°E
- -20°N
E
_ 0 °
o - 2 0
,40°S
120 ° 140 E
l"i~. 3(h)
24 Y. YOU and M. TOMCZAK
20°N - -
0 °
20 °_
O 4O S -
40 ° E 600 80° 100 ° 120 ° 140 ° E
I / I I I I I I I I ~ x x ~ _ ,~o,~= ~ , ,~ . o . 26.7 / ..
_20 °
> / ~ J ' ~ :3"5 2 35"2'~ ~ - V v - E] ~ , ,
- ~, J ~" ~35 .1 ~ , ' f ' " % - I .40 S
/
40 ° E 60 ° 80 ° 100 ° 120 ° 140 ° E
Fig. ~(c)
20°N --
0 ° _
20 ° _
O 4O S -
40 ° E 60 ° 80 ° 100 ° 120 ° 140 ° E
I / [ I I 1 I I I 1 ~ X ~ x isopycnal surface O . 26.7 ] d " •
_0 ° {2
_ - 2 0 °
6.0 ~ - 4 0 ° S
I .I I I I I i i i I
40 ° E 60 ° 80 ° 100 ° 120 ° 140 E
Fi~. 3(d)
Thermocline circulation and ventilation in the Indian Ocean
2 0 ° N _
0 ° _
m
20 ° _
40 ° S -
40 ° E 60 ° 80 ° 100 ° 120 ° 140 ° E
~' , , , ~ Isoo,or: ~. ' . o ' 26,' ) ..'
.I I ~ I ) I I I
40 ° E 60 ° 80 ° 100 ° 120 °
Fig 3(c)
- 2 0 ° N
m
m0 °
- - 20 °
O - -40 S
140
20°N _
0 ° _
20 ° _
40° S
40 ° E 60 ° 80 ° 100 ° 120 ° 140°E
I ,I I ~ I I I I 1
40 ° E 60 ° 80 ° 100 ° 120 °
- 2 0 ° N
m
_ 0 °
m
= 2 0 °
- - 4 0 ° S
140 ° E
Fig. 3(0
26 Y. Y o u and M . TOMCZAK
20°N - -
0 ° __
20 ° _
40 ° S -
40 ° E 60 ° 80 o 100 ° 120 ° 140 ° E
7oo <
I=, I I I
o
- -20 N
_ 0 °
_ 20 °
- 4 0 ° S
I 1 i I t I t o o o
40 E 60 ° 80 100 ° 120 140°E
Fig. 4. D i , ; t r ihu t ion c,f hydrogr~iphic pn~pcrti,,:,~ on lhc o , = 2h.~J i,;opyL'n;,l ,~url;,~.'~:: (a) pressure. (h) po l cn t iM tL'ml',,..'rlm.,r~:. (c) ',;~dinity. (d) di,.,,ol','~:,.l ~xygcn . (c) pho'~ph~It,.: ;,nd ( f ) ~;ilicah:.
20 ON - -
0 ° _
20 ° _
40 ° S
40 ° E 60 ° 80 ° 100 ° 120 o 140 ° E
l I I l I I I l I I ~'s°pycnalsurface (~V " 26'9 ~n "" '~ ' ) "CJ 0 ( ° C ) ~ , ~ " ~ _ 2 0 ° N
% v , . ' "
i I l ~ I i l I I w
40°E 60 ° 80 ° 100 ° 120 ° 140 E
Fig. 4(h)
Thcrmoctinc circulation and venti lation in the Indian Ocean 27
20°N _
0 ° _ _
20 ° _
40° S -
40 ° E 60 ° 80 ° 100 °
I I I I I I I I I I ~ , ~ _ = _ ,~opyc,=,u~c. o . ~.9 / ,-
'r ' ? ° - ,
120° 140 ° E
I I I ~ I I I I I I
40°E 60 ° 80 ° 100 ° 120 ° 140°E
_20°N
_ 0 °
_ 20 °
40° S
Fig. 4(c)
m
20°N --
0 ° _
20 ° _
40°S
40o E 60 ° 80 o 100 °
l I I I I I I l I l ~'~ . .~_= _ isopycna, surface a . 26.9 / , °
5.0 ~
/ A _ ~ ..---_
120 ° 140 ° E
_20°N
_ 0 °
_20 °
o 40 S
I I I I I I I I I
40 ° E 60 ° 80 ° 1 0 0 ° 1 2 0 ° 1 4 0 ° E
Fil. 4(d)
28 Y. You and M. TOMCZAK
40 ° E 60 ° 80 ° 100 °
I I I l I I I I I I I I ~ , , \ ~ . . , , + o p ~ ~,~-',, o . 269 / ,"
40 ° E 60 ° 80 ° 100 °
o 120 1400E
_ 2 0 ° N
_ 0 °
_ 2 0 °
- 40 ° S
120 ° 140°E
Fig. 4(c)
20°N _
0 ° __
20 ° _
o 40 S -
40 ° E 60 ° 80 ° 100 ° 120° 140 ° E
i I i ~ I J ' I , i
40°E 60 ° 80 ° 100 ° 120 ° 140 E
_ 2 0 ° N
_ 0 °
o -20
o 40 S
Fig. 4 ( 0
Thermocline circulation and ventilation in the Indian Ocean 29
20°N --
0 ° _
20 °
40 • S - -
40 ° E 6 0 ° 80 ° 1 O0 ° o
120 1 4 0 ° E
_20°N
_0 °
_ 20 °
_40°S
li- t I I I I I I I I I
40 ° E 60 ° 80 ° 1 O0 ° 120 ° 140 ° E Fig. 5. Distribution of hydrographic properties on the ~ru = 27, I isopycnal surface: (a) pressure,
(b) potential temperature, (c) salinity, (d) dissolved oxygen, (c) phosphate and (f) silicate.
40 ° E 60 ° 80 ° 1 O0 ° 120 ° 140 ° E
I I [ I I I r f l I I I I l ~J~ ~ ~ 's°pycnalsu ace 00" 27'1 ~ '" I
O° I~ j ~ ~ ' ~ - , q j ~ °°
20 ° 20 °
I l i r "' i - - i I i i i ~ i
40 ° E 60 ° 800 1 O0 ° 120 ° 140 ° E
Fig. 5(b)
30 Y. Yot" and M. T.MCZAK
40 ° E 60 ° 80 ° 100 °
I / I I I I I I I I
• L~ ~ ~ 3 4 . 8 - ~ .~'7-.-~, ,,-¢.=--=.-',~
4 0 O S t / " .._..~.34.4 ~
' I I I i , ; I I I I I I
40 ~ E
120 ° 140 ° E
_20°N
_ _ 0 °
- 2 0 °
_40°S
600 80 ° 100 ° 120 ° 140 ° E
Fig. 5(c)
40 ° E 60 ° 80 ° 100 ° 120 ° 140 ° E
I I / I I I I I I 1 l, I ~ J ~ " ~ . ~ _ _ ~ _ _ isopycna, surface G - 27.1 ~ r ,"
/ \(o,?<o, ..½ \ -
4 0 ° S ~ - 4 0 ° S
40 ° E 60 ° 80 ° 100 ° 120 ° 140 E
Fig. 5(d)
Thcrm¢~: l inc c i rcu la t ion and ~¢nt i la t ion in the Ind ian Ocean .3 ]
40 ° E 60 ° 80 o 100 ° 120 ° 140 ° E
I I I I I I I I I I \'~..~ =p~,~ s.,,=¢, o = 27., / ,"
ON_ \~,,~ " ~ ' ~ ~ , ~ . ~ ) 0 ~ 0 _2oON
13 2.4
. [ ~ " . ' . u ~ (J ' \ . 2 ~ / " - " . - - % - , = , , , , ~ - , , = 1 , ,0 ,-
2.2 2.4
40 S 40 S
I I I I ~ ] I I I I I o
40 E 60 ° 80 '~ 100 ° 120 ° 1 ..z0 ° E
l.i~..~ (~..)
20°N --
0 ° _
20 ° _
40 o S
40 ° E 60 ° 80 ° 100 °
o ' ; ,.'
• 45 ~
o 120 140°E
Ii. i i I I I I l I I I
40 ° E 60 ° 80 ° 100 ° 120 ° 140 E
- 2 0 ° N
_ 0 °
- 2 0 °
' 40 ° S
Fig. 5 ( 0
32 Y. You and M. ToMCZAK
20°N --
0 ° _
20 ° _
o 40 S'
40 ° E 600 80 ° 1 O0 ° 120 ° 140 ° E
I I I i I I I I I I ~ ' ~ ~%.~,-~'L isopycnal surface G - 27.3 / ,,"
< o
. . r - ~ v u'5o [ ) ~ u o lo.~o looo '~,~,-- ,~-- '~"vL t 200 [ . . . . V , '~_'" . _
i i i ~ I J i I i J
40°E 60 ° 80 ° 100 ° 120 ° 140 E
Fig. 6. Distribution of hydrographic properties on the rJ, = 27.3 isopycmd surface: (a) pressure. (b) potential tcmpcr;={ure. (c) salinity. (d) dissolved oxygen. (c) phosphate and (f) silicate.
o o o 40 ° E 60 ° 80 1 O0 120 140°E
I . . l ~ . l I i I , I I I I ~ ~'t,,,.,.:~._ ~pycna l surface O - 27.3 / , "
.~ • (b)
0 [:2 |"
4 0 0 8 1 / f 4 . 0 ~ O' _4008
/ l e i i i i I i i 1 i i o o
4 0 E 6 0 ° 8 0 ° 1 0 0 ° 1 2 0 1 4 0 E
Fig. 6(b)
Thermoc l inc c i rculat ion and vent i la t ion in the Indian Ocean ,3.3
40 = E 60 ° 80 ° I OO ° 120 = 140 ° E
' ' 'L
/ I I I I I I ' ~ ~ '~ .~ __ =~,¢~.., ,=. o - 27~ / ,-
20ON _ 2 0 ° N
o,_ F,
/ / ~ 4 ~ 3 + . 4 , . ( ' N O , 40OS - ~" - 4 0 S
Ie I I I I I I I I I I
40 ° E 60 ° 80 ° 1 O0 ° 120 ° 140 E
Fi~. 6(c)
40 ° E 60 ° 80 ° 1 O0 °
~1~ ' ~ , ~ Isopycn[alsurfacle O / 27,3 I ; . '
~oO,_ ,~\ '-'~~O~<m,,~.,...~' ~.--"~
- : :, Tz A+
I--~--~ ~ . o ~ ~ ~ .-,
20 ° _ 2.5
120° 140 ° E
_20°N
_ 0 °
- 2 0 °
. 4 0 0 5
i r i i i i i i i i i i
40 ° E 60 ° 80 ° | O0 ° 120 ° 140 ° E
..34 Y . YoL" and M . T()~cz.a,,K
20°N -
0 ° _
20 ° _
40°S -
O O 0 O
40 E 60 ° 80 100 ° 120 140 E
I / I I I I I 1 I I V , \ ' ~ .~ , iso0yc°al suaace a = 27.a / , "
.,e.. a
; II "I- ~ \\,.,~~_. ~E- I X :-. _.3 _ . . "~ 2.2,---C, 2-2 % - \
_ 2 0 ° N
mO °
_ 2 0 °
L 40°S
1 I I I I I I I I I
40°E 60 ° 80 ° 100 ° 120 ° 140 E
Fig. 6(c)
20°N .
0 ° _
20 ° _
40°S _
40 ° E 60 ° 800 100 °
, ~ J \ ~ Iso°yoo'a,'~o~o'. o '2,.~" ) ..' O
120 140°E
- 2 0 ° N
_ 0 °
- 2 0 °
O - 4 0 S
i i i =1 I i i i i I
40°E 60 ° 80 ° 100 ° 120 ° 140°E
Fig. 6( f )
Thcrm~:linc circulation and ventilation in the Indian Ocean 35
much with AAMW, crossing the equator near the western boundary. However. oxygen and nutrient levels (Fig. 3d-f) change significantly through biochemical nutrient release along the path. COLBOR.~ (1975) pointed out the usefulness of oxygen in the Indian Ocean as an indicator for the relative a~,e of ICW. i.e. the time since last contact with the sea surface. Since no thermocline water is formed in the eastern Indian Ocean north of the equator, the very low oxygen content in the northern Indian O c e a n I t h e lowest oxygen values at thermocline level in the world ocean--means that thermocline water in the Bay of Bengal can only be very old ICW.
The other two sources for thermoclinc ventilation are an intrusion of low temperature. low salinity water along 10°S from the east and an influx of high temperature, high salinity water from both the Persian Gulf and the Gulf of Aden. The sources are not prominently visible in the distributions of oxygen or nutrient content, due to strong meridional gradients and the fact that RSW by the time it enters the Indian Ocean is also low in oxygen and high in nutrients and does not contrast strongly with old ICW in these properties.
In contrast to the o , = 26.7 surface, the density surfaces above and below display temperature and salinity distributions that only can be explained if some degree of diapycnal mixing is invoked. On the o , = 25.7 isopycnal surface (Fig. 2). for example. salinity changes from about 35.6 to below 34.9 and potential temperature from above I S.I)°C to less than 16.I}°C along the path of ICW. This density surface is shallow, but with the exception of a small region in the south between 711°E and 90°E it is well below the mixed laver depth (Svrtnrat.t, and TomcZaK. 19t)2a). iligh salinities in the Bay of Bengal certify that it does not reach into the surface layer dominated by low-salinity Bengal Bay Water. "ieml~er~ltt, rc and salinity therefore should be conservative on this surface, which leaves diapycn,d tuixing or mixitlg with AAMW as the only possibilities to explain their decrease from the south into the Bay of l~cngal. Mixing with AAMW is indeed likely to ~otnc dcgrcc but camlot explain the low oxygen and high ntttrient levels in the Bay of I~ctlgal. Thus. sotnc dial~ycnal tuixing tntts| occur in the northern Indian ()ccan, a fact idrcatly i~itllctl out by ('~)I I~}rn (It)75).
Strong mcridional gradients ¢~1 all properties at abottt 13'~S indicate the presence of a hydrological fr(mt south of the re, ion of AAMW inllo~v. The front indicates that throughottt much of the Indiatl ()cean there is little commttnication between water in the tlt)rthcrtl and in the sottthcrn hemisl~hcre, and that the region of the western boundary cttrrcnts along the coast of Africa is the only p~th for water to cross the equatorial current system (WYrtKt, 1973). This adds additional weight to the argument that mixing in the Bay of Bengal thcrmoclinc tnust bc diapycnal in n:tture, since ICW entering the Bay from the west has no exit route avail:tblc to the south, and Malakka Strait is too shallow to provide a p a s s a g e .
Further down on the or,, = 26.9 isopycnal surface potential temperature and salinity change again along the path of ICW. but this time both properties increase rather than decrease. This cottld bc interpreted as indicating a contribution from RSW. and oxygen and nutrient levels do not contradict that possibility. Howevcr, it is noticeable that most of the property changes occur during the passage of ICW from Madagascar north across the cquator, while little change is ohscrvcd from there on into the Bay of Bengal. Also, the oxygen distribution in particular supports the notion of northward advection in the west followed by further aging of the water on its way east. Thus. the RSW inllucnce can only bc very minor, and at least some diapycnal mixing must be invoked to explain the O-S distributions.
36 Y. You and M. TOMCZAK
The lowest two density surfaces (Figs 5 and 6) show distributions quite similar to the ones just discussed. The lowest surface is found at depths where some influence of Antarctic Intermediate Water might reasonably be expected; results for that surface may have to be interpreted with caution.
4. THE DEFINITION OF SOURCE WATER TYPES AND THE MIXING MODEL
Application of cluster analysis to the data fields displayed in Figs 2-6 produced four well defined clusters, which were linked with equally well defined geographic regions. The first cluster contained all data points south of 15°S--on the lowest surface (oo = 27.3) it was restricted to data south of 20°S---and could clearly be identified with ICW. Another cluster was found in the Red Sea west of the Strait of Bab El Mandeb and represents RSW. The third cluster was located in the region between Australia and Java, protruding eastward into the Indian Ocean beyond 100°E on the highest (oo = 25.7) density surface but not much beyond 110°E on all other surfaces: it is seen to be associated with AAMW. The fourth cluster occupied the Bay of Bengal north of about 8°N on all but the lowest density surface. As discussed before, it was taken to represent the virtual water mass NICW. The remainder of the Indian Ocean was filled with data of no particular cluster association. In particular, no cluster that could have been associated with Equatorial Water was seen in the equatorial western Indian Ocean.
It is of course important to verify that clusters are not simply the result of a non-uniform data distribution in space. Ideally, objective analysis should be used to map all obser- vations into a regular grid and avoid artificial clustering. This step was not deemed necessary in the present case, since the four clusters found by the method are clearly not related to data density in space. As an inspection of Fig. I will show, regions of high data density that could produce a bias in the cluster analysis are found along the west Australian coast and in several longitude bands in the western tropical Indian Ocean, None of these regions overlaps with the cluster regions described above.
The data in the respective regions were used to determine cluster centres, i.e. points in parameter space which give the smallest sum of the distances to all other points in their respective cluster. The calculation was done for all rive density surhtccs; consequently, five cluster centrcs, or source water types, were obtained for every water mass. Figure 7 shows the parameter values obtained in this way. It is seen that in all cases but one (the 0-Si relationship for ICW) the property-property relationships defined by the source water types are close to linear (this is true even for RSW, which indicates that RSW does not act entirely as a point source, despite its well defined release point in the Strait of Bab El Mandeb). The closest linear representations are determined through regression analysis, using the reduced major axis representation where appropriate, and are shown in Fig. 7 by the straight lines. The procedure also produces a variance for each water mass used to determine C~jmi, in equation (5). Because the 0-Si relationship of ICW cannot be represented by a single linear relationship, the procedure was applied separately for the density surfaces above and including the oo = 26.9 density surface, and below and including that surface.
The source water types needed for OMP analysis were taken from the regression lines as the values found on the highest (oo = 25.7) and lowest (oo = 27.3) density surfaces and on the density surface (oo = 26.9) at the break point of the 0-Si relationship of ICW. They are tabulated, together with the weights derived from the data and water mass variances, in
Thcrmoclinc circulation and vcntila¢ion in the indiun Occan ~7
~ l l ~ l i l i l i l i l , l , l i l i l ~ l , l , l i
" ' ~ V ; o
i | i i i I i I . I | , i . I | I .
( M c ~ ( M ( M 0 4 ~
(0o) e
, i , l i l , l i l , l i l , l , l ~ l ~ l ~ l , l ,
• < + D O
, I , I , ! , I , I • I , J , I ~ ! , ! • ! J # i I i
b.,
T,~ .
.:- ~
v
(0o) e
38 Y. You and M. Tomcz.~
- - ~ o
(0o) O
0
0 0
0
0
4
0
0
0
• ,~ ~ c> o !
I
q
/ (0o) 0
Thcrmocl inc circulation and ventilation in the Indian Ocean 39
Table !. Source water ~'pe definitions and parameter weights
Water type Pot. temp. (°C) Salinity Oxygen (ml I - t ) Phosphate (.ug-at 1 - I ) Silicate (ug-at I - I )
ICW on o , = 25.7 18.160 35.624 6.241 0.145 2.921 ICW on o , = 26.9 9.070 34.8(15 4.746 1.327 10.089 ICW on o , = 27.3 4.39l 34.384 3.977 1.936 43.879 NICW on o~ = 25.7 15.720 34.840 0.1)60 2.099 25.594 NICW on oo = 27.3 7.800 35.098 0.726 2.796 60.179 A A M W on as = 25.7 14.890 34.774 2.527 1.615 28.638 A A M W on oo = 27.3 5.554 34.521 2.2118 2.662 79.103 RSW on o~ = 25.7 25.900 38.082 3.625 0.372 0.701 Weights 69.551 69.551 13.811 6.189 7.271
Table 1. This procedure is somewhat different from other determinations of source water types, which often select property extrema ("outliers") to represent source water types, on the argument that newly formed Intermediate Water for example should have the lowest salinity and the highest oxygen content. This is, however, only true if the data are collected in synoptic fashion. Historical data collected over many years inevitably contain seasonal and interannual variability, and to select the lowest observed salinity as a source water type descriptor for Intermediate Water would not represent the climatological average. Wc determine the cluster centres in the formation regions of the water masses, i.e. where the water masses are freshly formed and unadulterated. We therefore consider the cluster centres to give us the long-term mean parameter values fi~r the source water types and the variances to represent seasonal and interannual variability.
A linear representation of all but one of the property-property relationships is, of course, a simplification of the real situation imposed by the limitations of the method. Rcsolution of the oxygen minima and maxima described by Warrt:.n (1981) for the 18°S latitude band is beyond OMP analysis, at least until the matrix can be increased by two ntorc parameters. In its present form, OMP analysis exploits the fortunate fact that most property-property relationships of Central Water are very close to a linear connection between Intermediate Water and the water of the shallow subtropical salinity maximum, even though Central Water is no t a mixture of the two but subducted at the Subtropical Convergence Zone. The fact that this subduction may not occur uniformly across the entire parameter range does not interfere with the analysis as long as the linearity of the property-property relationships is maintained. Mode Water, for example, constitutes one point in parameter space, and where it contributes to the formation of Central Water, this point is situated on the T-S-curve of Central Water. A problem arises where Mode Water, being produced in large quantities and sinking relatively fast, distorts the linear temperature--oxygen relationship by introducing higher oxygen levels. The oxygen distri- bution in the present data suggests that the effect is minor and the inability of OMP to resolve it is no serious drawback.
Mixing between six source water types, appropriately chosen above and below the a , = 26.9 density surface, constitutes the mixing model used to derive the results in the fi~llowing section.
4 0 Y. You and M. TOMCZAK
O O
40 E 60 ° 80 ° 100 ° 120 ° 140 E
x l I I I I I I I
20°N - - . I C W : (a)
_
- 5 0 " "~, d ="
80 90
40° S - / i I !
i,.' I 1 I I ] I I I 1 I
Fig. 8.
- -20°N
_ 0 °
_20 °
0 ,40 S
40°E 60 ° 80 ° 100 ° 120 ° 140 E
Water mass contrihutions on the aH= 25.7 isopycnal surfacc. (a) the contr ibut ion o f ICW. (h) the contribution (,f A A M W ;.,nd RSW and (c) the contribution ()f NICW.
20°N -
0 ° _
20 ° _
40 ° S t
40 ° E 60 ° 80 ° 100 ° 120 ° 140 ° E
1 / I I I 1 I I I I
I I I I I I I I I i
40 ° S
40 ° E 60 ° 80 ° 100 ° 120 ° 140 ° E
Fig. ~(h)
Th~zrmoclim: circulation and vcntilation in the Indian Ocean J, ]
40 ° E 600 80 ° 1 O0 ° 120 ° 140 ° E
I / I I I I o ~, ~ ~ ~°~'c~''~- ~o''' '~~.'o "
~ ~ o ' ~.o
~J2 ~"'\ 40 ° S ~ . '
I I 4 i I I ! I I
40 ° E 60 ° 80 ° 100 ° 120 ° 140 E
_20°N
E
_ 0 °
- -20 °
_ 4 0 ° S
Fig. 8(c)
40 ° E 60 ° 80 ° 100 ° 120 ° 1400 E
20 N
40 ° S 100 90
Fig. 9.
- 2 0 ° N
_ 0 °
_20 °
' 40 ° S
40° I= 60 ° 80 ° 1 O0 ° 120 ° 140 F:
Water mass ct)ntrihutions on the ao = 26.7 is(~pycnal surface. (a) the contribution of ICW, (b) the contribution of A A M W and (¢) the contribution of NICW.
4 2 ¥ . Y o u and M. TOMCZAK
2 0 ° N ~
0 ° _
20 ° _
o
4 0 S - -
40 ° E 60 ° 80 ° 100 ° 120 ° 140 ° E
I I I I I I I I I l
' , ~ ~ \ ~ ~ ~ L . ~ ~ ' "
I I I I | I I I l I
40°E 60 ° 80 ° 100 ° 120 ° 140 E
Fig. 9(h)
o
4 0 S
20 °N - -
m
0 °
20 °
40 ° S --
40 ° E 60 ° 80° 100 ° 120° 140 ° E
I I I I I I I I I o
40 E 60 ° 80 ° 100 ° 120 °
- 2 0 ° N
_ 0 °
_20 °
- 40 ° S
140 E
Fig. 9(c~
Thcrmoclinc circulation and ventilation in thc Indian Ocean 43
2 0 ° N _
0 ° _
o 2 0 -
4 0 ° S -
Fig. I0 .
• • • 4 0 E 6 0 ° 8 0 1 0 0 ° 1 2 0 1 4 0 ° E
I l I I I I I I / I
oo
o - 4 0 S
b" I I I I I I I I I I
4 0 ° E 60 ° 80 ° 100 ° 120 ° 140 E
Water m~,ss contributions on thc o~ = 26.9 isopycnal surface, (a) the contribution of ICW. (b) the contribution of A A M W and (c) the contribution of NICW.
2 0 ° N _
0 ° _
20 ° _
40 ° S - -
40 o E 6 0 ° 80 ° 100 °
\ ' 2 .'
b" I I I I I I I I I
40 ° E 6 0 ° 8 0 ° 100 ° 120 °
F ig . I l l ( b )
o 120 140 ° E
- 2 0 ° N
_ 0 °
_ 2 0 °
o 40 S
140 E
4.4 Y. You and M. TOMCZAK
20°N _
0 ° =
20 ° . .
40°S 1
40 ° E 60 ° 80 • 1 O0 °
I / I I I I I I I I
"t\ / ~,.,cw ,-'-'k ~ 0 ,o o%.h - g .
i ,00
I I I l i I I I I l
400 E 60 ° 120 ° 14
o 120 140°E
.20°N
w
_ 0 °
- 20 °
- 40 ° S
80 ° 100 °
Fig. lO(c)
1 4 0 E
o 2 0 N - -
w
O ° j
m
20 ° _
o 40 S -
Fig. I 1.
40 ° E 60 ° 80 ° 100 °
I I I I I I I I I I ~'~.,,.,.~ tsopycnal surface O . 27.1 / / ,"
W
120 ° 140°E
_20°N
_ 0 °
_20 °
o 40 S
I I I I~ J I I I I I
40 ° E 60 ° 80 ° 100 ° 120 ° 140 E
Wazcr mass contributions on the o , = 27.1 isopycnal surface. (a) zhc contribution of ICW, (b) the contribution of A A M W and (c) the contribution of NICW.
Thcrml~:linc circulation and t-cnzilalion in the Indian Ocean 45
O
20 N -
0 ° _
20 °
O
40 S -
40 ° E 60 ° 80 ° 100 ° 120 ° 140 ° E
I I I I I I I I I I I %'~_ isopyc, surface %=27.1 .~ .° I
V':" %.Y 7 .MA -, /_ ,, , "-.X% ¢- .~¢" ~,...._ _0
.'¢. - -,,,' ~..',,..~ (z,~ f".~ .~ <." ~ ~ _ - j, : , • .,_
I I I I I | I I I I
40 ° E 60 ° 80 ° 100 ° 120 ° 140 ° E
Fig. 1 l(b)
40 ° E 60 ° 800 100 °
I / I I I I I I I I
~10 ~ 4 0 ° S t ' 1 0 1 Ir
I I I l I I I I I
40 ° E 60 ° 80 ° 1 O0 °
O
120 140 ° E
_20°N
B
_ 0 °
_20 °
0
40 S
120 ° 140 E
Fig. ll(c)
46 Y. You and M. TOMCZAK
20°N--
0 ° _
20 ° _
u
40 ° S -
Fig. 12.
40 ° E 60 ° 800 1 O0 ° 120 ° 140 ° E
I I I I I I I I I I ~ ~ ~ i s o p y c n i d surface a O = 27.3 , " _--20°N
• M O °
_20 °
/% ~ / X _40°S
I I ! ~ I i 1 I I I
40 ° E 60 ° 80 ° 1 O0 ° 120 ° 140
Water mass contributi tms on the o. = 27.3 isopycnal surface, (a) the contribution of [CW, (h) the contribution of A A M W and (c) the contribution of NICW,
0 20 N--
0 ° _
20 °_
40 ° S -
0 0 0 0 40 E 60 ° 80 100 ° 120 140 E
I / I I t I t t I I ~ is°pycnal surface G8 " 273 ~ ""
_ oO
! I I ~ I ! i ! ! ! - - I 40
S
40 ° E 60 ° 80 ° 1 O0 ° 120 ° 140 ° E
Fig. 12(b)
Thcrmocline circulation and ventilation in the Indian Ocean 47
20°N _
0 ° _
I
20 ° _
40 ° S -
40 ° E 60 ° 800 100 ° 120 °
I I I I I I I I I isopycnal ,~dace 0 = 27.3 , / / , "
e
12 l
o
140 E
40 ° F: 60 ° 80 ° I O0 ° 120 ° 140 I::
o
- -20 N
- 0 "
m
_ 2 0 °
.40°S
Fig. 12(c)
5. R E S U L T S
Figures 8--12 show the contributions of the water masses to the waters of the thermocline derived from the mixing model. The contribution of RSW is shown only for the uppermost density sttrhlcc (Fig. 8) where the model uses ICW, NICW and RSW west of 7I)°E and ICW, AAMW and NICW east of 70°E. The cro = 25.7 surface is the only surhtce where exclusion of RSW leads to entirely unrealistic results in the Arabian Sea. RSW influence is seen to be restricted to the Arabian Sea, where it contributes up to 311% to the thermocline water. South of 5% the RSW contribution is negligible. 70°E is a convenient meridian to match the two mixing models since both RSW and A A M W contribute very little to the thermocline at that meridian; this is evident from the good match between percentage contours for ICW and NICW on either side of 70°E, particularly south of the equator.
Results for all other density surfaces are only shown for the mixing model based on ICW, AAMW and NICW. The distributions appear reasonable even in the Arabian Sea, but it is clear that RSW is present in some quantities on these surfaces in the Arabian Sea and along the African coast north of the equator and the percentages of ICW and NICW in the Arabian Sea are only a guide to the general circulation.
Indian Central Water dominates the thermocline and provides the largest contribution throughout the thermocline. On the uppermost surface (Fig. 8a) it does not penetrate the hydrological front at 13°S, where AAMW enters with the westward flowing South Equatorial Current. ICW therefore enters the northern Indian Ocean in the western boundary current from where it again moves eastward with the Equatorial Countercurrent or Southwest Monsoon Current (depending on season). Its aging during the process is reflected by the gradual decrease of the ICW percentage contribution and simultaneous
48 Y. You and M. TomczaK
increase of the NICW contribution along the path. With the exception of the region of AAMW inflow and the Arabian Sea, the sum of the ICW and NICW contributions is always close to 100%. indicating that ICW at different stages of aging occupies most of the thermocline.
An interesting feature to note is the indication of northward spreading of A A M W into the Bay of Bengal Bay along the eastern boundary at the uppermost surface (Fig. 8). SASAmaL (1989) found a strong narrow southerly current in the northern Bay of Bengal on the west and a broad northerly flow on the eastern side. The A A M W distribution (Fig. 8) supports northward annual mean flow along the eastern side of the Bay. suggesting that ICW cannot leave the area along the Sumatra coast.
Another feature worth mentioning is the A A MW percentage anomaly near 70°--80°E south of 40°S. It coincides with the area where the thermocline comes to within 50 m of the surface (Fig. 2a). Most likely the low ICW percentages and the unrealistic AAMW percentages are an artifact produced by non-conservative behaviour of temperature and salinity. East of 80°E this seems to effect data on the next density surface (oo = 26.7; Fig. 9) down to 200 m (compare Fig. 3a).
Figure 9 shows the results for the v , = 26.7 density surface. This is the surface that earlier was shown to be dominated by isopycnal mixing. The path of ICW is reflected well by the percentage contours, While on the a , = 25.7 surface the contribution from AAMW west of 70°E is well below 10%, this water mass is seen to spread further westward on this surface, restricting passage of ICW into the northern hemisphere to the region west of 50°E. It is also noticeable that most of the transition from ICW to NICW (i.e. aging) occurs in the west between 15°S a11d 5°N. In a steady state situation this would indicate redt, ced advcction velocities along the African coast. In reality the transition region is a region of intense boundary currents that reverse direction with the nlt~nsoons. Thus. the rapid drop in ICW content from Madagascar to Somalia and corresponding increase in NICW content has to bc seen as the annual mean signal of seasonally varying ICW penetration into the northern hemisphere.
As on the level above, some AAMW is seen moving north toward the Bay of Bengal. The movement appears less restricted to the area close to the Javanese shelf but extends west beyond 90°E.
The distribution of w~,tcr mass contributions changes signiticantly below the surhlce ch~,racterized by isopycmtl mixing. The most prominent feature on the ao = 26.9 density surface (Fig. 1(1) is the southward spread of AAMW that reduces the ICW presence even in the southern hemisphere to 80% and less. The high percentages of A A MW along the Australi:m coast may be the result of advection with the Leeuwin Current as described by TOmCZAK and Larcl~ (1989). Penetration of ICW into the northern Indian Ocean still occurs in the west. but transition from ICW to NICW is observed as far east as 70°E. On annual mean the northward movement is so slow that north of 5°N the entire thermocline contains only the very old NICW.
The distribution on the a , = 27. I density surface (Fig. I 1) shows a similar pattern with very much reduced AAMW influence. The hydrographic front at 130S does not reach down into this surface. As a consequence, northward movement of ICW across the equator now occurs all across the Indian Ocean. The same movement can be seen on the lowermost (tro= 27.3; Fig. 12) density surface but with much reduced speed, as is evident from the fact that transition from ICW to NICW sets in as far south as 300S.
The effect of the hydrological front and its weakening in the west is seen when the
Thermoclinc circulation and ventilation in the Indian Ocean 49
40°S
-0
2001,
400
soo
800
1000
20 ° 10 ° 0" 10 <' 20°N
I I I I I I I I I
=...-'- 80 40 _- .o . -
I I I I I I I I I
4 0 ° S 2 0 ° 1 0 ° 0 °
0 I I I I , ~ I
- . 0 0 --
600 ~ 20 - ' J J 120"~ )
8o0.O~.....~ / 1 0 ~ / . ~ 1000
10 ° 20°N
I I I
(b) I I I I I I I I I
40os 20 ° 10 ° 0 ° 10 ° 20°N
0 / I I I I I I I I I
I - '
I000 s I ~ •~'~r~'~S
Fig. 13. W:ztcr mass contributions ahmg f~rE. (a) the contribution of ICW, (h) the contribution of AAMW and (c) the contribution of NICW.
section at 60°E (Fig. 13) is compared with the section at 90°E (Fig. 14). Larger ICW percentages north of IO°S in the west in the upper thermocline indicate again that the major pathway for ICW into the northern Indian Ocean is in the west.
The contribution from AAMW intrusion remains high until the water reaches 80°E and than falls rapidly below 20% at 60°E (Fig. 15), making room for an increasingly strong ICW contribution. As discussed previously, exclusion of RSW makes the results west of 70°E somewhat unreliable; however the error involved is below 10% (Fig. 8), i.e. the various water mass contributions west of 70°E should be accurate to within _ 10%. The largest ICW contribution is found near the African coast, indicating the core of the northward flow of ICW at 2(X}-400 m depth.
An idea of the quality of lit is conveyed by Figs 16 and 17, which show the residuals of the conservatkm of mass equation for one of the isopycnal surfaces studied. For most observation points the calculated water mass contributions do not depart from the correct values by more than 1-2% (Fig. 16); nearly all solutions give the correct volume within 5%. The few residuals larger than 5% are all located in the source region of AAMW. Large environmental variability in this area (both seasonal and interannual) produces signiticant fluctuations in the source water type properties. As a result some observational data in the source region have property values outside the range of possible analysis (e.g. salinities lower than the lowest salinity in Table 1). Within the accuracy of the method, i.e.
5 0 Y. You and M. TOMCZAt¢
400 :3
600 == " 800
I000
40os
0
200
20" 10 ° 0 ° 1 O* 20°N
g,,--- _ oi
40°$ 20 ° 10 ° 0 o 10 ° 20°N
0 , I I I I I | I I I /
, 0 0 0 p ' " - T , , , , " , ,
40°S 20 ° 10 ° 0 ° 10 ° 20°N
0
~" 200
400 :3
eoo
o. 800
1000
Fig. 14,
L__._ ) x
W~ltcr mass contributions along 90°E. (a) 1he contr ibution of ICW, (h) the contr ibution of A A M W and (c) the contr ibut ion of NICW.
within one standard deviation, such observations represent unmixed AAMW. Multipara- meter analysis would guarantee mass conservation by producing an AAMW contribution in excess of 100%, balanced by a negative contribution from another water mass. OMP analysis forces all contributkms to be positive but conserves mass no better or worse than temperature or salinity. This produces the correct A A MW contribution of I(X)% at the expense of mass conservation.
As a summary of the results, water mass contributions were averaged over 10 ° squares and over all five density surfaces. The result, the relative contributions of all water masses to the water of the thermocline over the depth range studied, is shown in Fig. 18. The path of ICW and its gradual conversion into NICW come out clearly.
6. S U M M A R Y A N D D I S C U S S I O N
Application of cluster analysis as a precursor to OMP analysis reduces the amount of subjective oceanographic knowledge needed before OMP analysis gives meaningful results. Nevertheless, the method still requires substantial input in the form of decisions with respect to the number and kind of water masses relevant for the region, in the present case it required the introduction of a virtual water mass to represent the aging of Indian Central Water. Compared with other inverse techniques for tracer modelling, OMP
Thermo¢line circulation and ventilation in the Indian Ocean 51
(a) 42"E 50* 60* 70* 80* 90* 100" 110 ° 120* 130"E 0 I I I I I I I !
200
• , , , , , , ,
(b) 420E 50* 60* 70 ° 80* 90* 100* 110 ° 120* 130"E
°ii J*
200 10
400
3. 600
(c) 420E 50 °
I"ig. 15.
I I I I I I I ]
~ , ,ooq
I I I I I I I I
60* 70* 80* 90* 100" 110" 120 ° 130°E
0 I ~ I I I I I I I J
Water mass contributions along l()°S, (a) the contribution of ICW. (b) the contribution of A A M W and (c) the contribution of NICW.
0.6
0.5
0,4
O = 0.3 g
0.2
0.1
Fig. 16.
0 ~ -0.05 0 0.05 0.1 0.15 0.2
residual
Histogram of the rcsiduals for mass conscrvation on thc u, = 25.7 isopycnal surface.
52 Y. You and M. TomczAK
20°N --
0 ° _
20 ° _
4 0 ° S f I t
/ <>o.o~ !
! t I
40 ° E 60 ° 80 ° 100 ° 120 °
I ~ , ~ 1 lisopyclnalsurflace oel 257 l ; oJ residual "'" . . . ~ 0
Fig. 17.
I I I I I
0.01"~ 0.01
140~E !
O
--20 N
. . . 0 °
.20 °
! o - -40 S
400 E 6 0 ° 8 0 ° 1 0 0 ° 1 2 0 ° 1 4 0 ° E
The distribution of lhc mass conservation residual cl, = 25.7 i~'Jpycnal surface.
analysis relies more on skilled preconditioning and m~,y not reproduce observed tields a,s well as other techniques (though the residuals shown in Figs 16 and 17 are rather satisfactory). On the other hand, results from OMP analysis always can be interpreted in terms of existing physical entities and processes. A situation such as that reported for principal component .'malysis by FUKUMORI and WUNSCH (1991), that the modes which produce the best tit to the data c:mnot be interpreted in oceanographic terms, is impossible with OMP analysis. The fidlowing picture of the Indian Ocean thermocline emerges from our study.
Fig. 18.
20°N ~ 25%
ICW 2 0 °
O NICW 40oS
RSW 40=E 60 ° 80" 100 ° 120°E
Total contributions by ICW. NICW. A A M W and RSW. integrated over the thcrmoclin¢ bclween n , = 25.7 and a e = 27.3 and averaged over I(P squares.
Thcrmocline circulation and ventilation in the Indian Ocean 53
Being closed in the subtropics, the northern Indian Ocean does not have its own subtropical convergence; its thermocline water has to be replenished from the tropics and further south. The analysis established Indian Central Water and Australasian Mediterra- nean Water as the two sources for thermocline ventilation, with a small contribution from the Red Sea and Persian Gulf in the Arabian Sea.
The idea that the thermocline of the northern Indian Ocean may be ventilated from the south was first put forward by SWALLOW (1984). In the present analysis supply of ICW to the northern hemisphere is clearly seen on the o0 = 26.7 isopycnal surface (300-400 m depth) which is characterized by isopycnal mixing. ICW dominates the thermocline south of the front at 13°S and enters the northern Indian Ocean with the western boundary currents. Oxygen values are fairly uniform south of the hydrological front, suggesting reasonably swift recirculation of ICW in the subtropical gyre. Transition into the northern hemisphere is accompanied by a rapid fall in oxygen values, which is interpreted as the expression of a seasonally intermittent transition process in the annual mean. The decrease in oxygen values continues into the Bay of Bengal, which contains the oldest Central Water. The oxygen decrease in the northern Indian Ocean can be explained by noting that the annual mean transfer of ICW between the hemispheres in the western boundary region is small. Therefore, the circulation of ICW in the northern Indian Ocean is slow.
Movement of AAMW through the Indian Ocean is a key aspect of recircu[ation models of North Atlantic Deep Water. Outflow from the Indonesian seas into the Indian Ocean occurs in the whole thermocline, stronger in the upper thermocline and weaker in the lower thermocline. In the upper thermocline the path of AAMW is evident from the salinity distribution on isopycnal surfaces as a band of low salinity along 5-10°S, from the entry point in the east to Madagascar in the west. In the lower thermoclinc the same pattern can be seen in the silicate distribution, where the presence of AAMW down to the lowest isopycnal surface is indicated by a silicate maximum. The jet-like inflow of AAMW produces one of the strongest frontal systems of the world ocean's thermocline. This leaves the western boundary currents as the only region for advcctivc transfer of thcrmocline water between the southern and northern Indian Ocean. AAMW does not leave the Indian Ocean; all of it mixes with ICW; the mixture is carried into the northern lndi:m Ocean, and into the South Atlantic Ocean with Agulhas Current eddies.
The formation of Indian Equatorial Water, a water mass frequently mentioned in the literature, could not be established. It is argued that this water is the mixture of ICW and AAMW found in the present analysis.
R E F E R E N C E S
ANDF'RllERG M. J. (1973) Cluster analysis for applications. Academic Press, 359 pp. BROECKER W.. T. TA~HASm and T. TAKAIIASlII (1985) Sources and flow patterns of deep-ocean waters as
deduced from potential temperature, salinity, and initial phosphate concentration. Journal of Geophysical Research. 90, 6925-6939.
Col.aoR.~ J. G. (1975) The thermal structure of the hulian Ocean. University Press of Hawaii, Honolulu, 173 pp. EM~.RV W. J. and J. MEINCXE (1986) Global water masses: summary and review. Oceanologica Acta, 9.383-391. Ew.Rrrr B. (1980) Cluster analysis. Halstel Press, 2nd edn, 135 pp. Fut<uMokt 1. and C. Wussol (1991) Efficient representation of the North Atlantic hydrographic and chemical
distributions. Progress in Oceanography. 27. I I 1-195. GoRoos A. L. (1986) Inter-ocean exchange of thcrmt~:linc water. Journal of Geophysical Research. 91, 5037-
5(146.
54 Y. You and M. TOUCZAK
M,~CKAS D. L., K. L. DEm,,^~ and A. F. BENNETT (1987) Least-square multiple tracer analysis of water mass composition. Journal oi" Geophysical Research. 92, 2907-2918.
~IAMAHYEV O. I. (1975) Temperature-salinity analysis of world ocean waters. Elsevier, Amsterdam, 374 pp. QUADO, SEL D. R. and F. Scnorr (1982) Water-mass distributions at intermediate layers off the Somali Coast
during the onset of the southwest monsoon. Journal of Physical Oceanography, 12, 1358--1372. REDFtELO A. C.. B. H. KL='rcnuM and F. A. R,O-O~RDS (1963) The influence of organisms on the composition of
sea water. In: Thesea, Vol. 2, M. N. HILl., editor, Interscience, New York, pp. 26-77. ROCHFORO D. J. (1966) Distribution of Banda Intermediate water in the Indian Ocean. Australian Journal of
Marine and Freshwater Research, 17, 61-76. ROMESBUK<:; H. C. (1984) Cluster analysis for researchers. Lifetime Learning Publications, San Diego, 334 pp. SASAMA[. S. K. (1989) Hydrography of the northern Bay of Bengal during south west monsoon. Mahusagar, 22,
[05-112. SHARMA G. S. ([976) Transequatorial movement of water masses in the Indian Ocean. Journal o/Marine
Research, 34, 143-[54. SHCHERatN,S A. D. ([969) Water structure of the equatorial Indian Ocean. Oceanology, 9,487-495. SPR,N'rALL J. and M. TOMCZAK (1992a) Evidence of the barrier layer in the surface layer of the tropics. Journal o/
Geophysical Research. 97, 7305-7316. SPmN'rALL J. and M. TOMCZAK (1992b) On the formation of Central Water and thermocline ventilation in the
southern hemisphere. Deep-Sea Research, in press. SrRAMMA L. (1992) The South Indian Ocean Current. Journal of Physical Oceanography, 22, 42 i-430. Su Y. S.. Z. X. Yu and F. Q. Lt (1983) Application of the clustering method in analysing shallow water masses
and modified water masses in the Huanghai Sea and East China Sea. Chinese Journal of Oceanology and Limnology, 3. 272-284.
SvERORUP H. U., M. W. JOHNSON and R. H. Ft.EMXN(.; (1942) The oceans, their physics, chemistry and general biology. Prentice-Hall. Englcwood Cliffs. 1087 pp.
Sw^Lt.OW J. C. (198J,) Some aspects of the physical oceanography of the Indian Ocean. Deep-Sea Research. 31, 639-650.
TOMCZAK M. (1981) A multiparamcter extension of temperature/salinity diagram techniques for the analysis of non-isopycnal mixing. Progress in Oceanography, tO, 147-171.
TOMCZAK M. and D. LAa(;t:. (1989) Optimum multiparametcr analysis of mixing in the thermocline of the cast Indian Ocean. Journal of Geophy.~ical Research, 94, 16141-16[49.
WAKm.:N B. A. ( 1981 ) Transmdian hydrographic section al lat. 18°S: property distributions and circulation in the South Indian Ocean. Deep-Sea Research, 28, 759--788.
WvarKJ K. (1971) Oceanographic atlas of the haternational Indian Ocean Expedition. National Scicncc Foun- dation, 531 pp.
Wvar~t K. (1973) Physical occanogr;iphy of the Indian Ocean. In: The biology of the Indian Ocean, B. Zi.:rxzson~L, editor, Springcr-Verlag, Berlin, pp. 18-36.
A P P E N D I X
Elimination of NICW by the imrodaction of initial phosphate
REDI-'t [-:LD et al. (1963) noted that in and below the oceanic thermocline the concentration of certain nutrients is coupled with the concentration of oxygen through the chemical processes active during rcmineralization and pointed out that this fact can bc used to derive truely conservative parameters from oxygen and nutrient observations by correcting for biochemical uptake hnd release. Briefly, the method assumes water mass formation at the surface, where oxygen concentration is at saturation value and nutrient concentrations are at their " in i t ia l" levels. By comparing the oxygen concentration for a sample away from the formation region with thc oxygen saturation value and taking into account the ratio between oxygen loss and nutrient gain, it is possiblc to determine how much nutrient was rcmincralizcd since thc water parccl left the surface and to rccovcr the initial nutrient level. The oxygen saturation value for the sample can be calculated from its temperature and salinity. which are conservative and therefore unchanged. Initial nutrient levels thus calculated are also maintained along the path of the parcel; they arc conservative properties.
Thermocline circulation and ventilation in the Indian Ocean 55
4 0 " E 6 0 ° 8 0 " 1 0 0 ° 1 2 0 ° 140 ~ E
" 0 . 5
I I I I I I 1 ! ! I I
40 ° E 6 0 ° 8 0 ° 1 0 0 ° 1 2 0 ° 140 = E
20 ° N -
0 a - -
20 ° _
40 ° S -
40 ° E 6 0 ° 8 0 ° 1 0 0 = 1 2 0 ° 140 = E
,~o~C ~'~ ~ / ~ ,~ ~ - - . .
~ , / ~ o~ ~ r ° ' ~ ~ ~ ' ~ , ~ ' - ~ °' "* Jr,
I I I I I I I I I I
Fig. AI.
- 2 0 ° N
- - 0 °
- 2 0 °
• 4 0 ° S
- - 2 0 ° N
_ 0 o
- 2 0 °
.40 ° S
40 ° E 6 0 o 8 0 ° 1 0 0 = 1 2 0 ° 140 o E
Distribution of initial phosphate on the o= = 26.7 isopycnal surface: (a) calculated with Re = 175, (b) calculated with Re = 135.
56 Y. You and M. TOMCZAK
20 ° N -
0o~
20 ° _
40 ° S -
40 ° E 60" 80" 100"
I I I I I I I I ,. t = , , , 0 , o . 2s.r / , '
e
% ( a
I~' I l I I I l
120" 140" E
- 2 0 ° N
. 0 °
-20"
l
- - 4 0 ° S
i l I I
40 ° E 60 ° 80" 100 ° 120 ° 140 ° E
Fig. A2. Water mass contributions on the or, = 26.7 isopycnal surface derived from temperature. salinity and initial phosph;,tc. The figure shows the contribution of ICW; the contribution of
A A M W equals I00 minus the numbers shown.
PO 4 s such a conservative property, and the only nnc that can he derived from our d;=ta. It is Initial phosph:ttc ~ " calculated from
) t l I O 4 = PO4 - (O;" - ._,)/Re
where I'O4 and 02 arc the observed concentrations :rod O~'" is the oxygen saturation value at the temperature and salinity of the s:implc. The coefficient Rc is known as the Rcdlicld ratio; R~:DFa'LD eta/, (1963) give 13~ for phosph;ttc. More recently. BNo~.C'K~., el ,,I. (1985) suggested the um of 175 instead.
I:igurc A I shows initial phosphate on the u, = 26.7 isopycnal surface, using the values Rc = 135 and Rc = 175 h~r the Rcdficld ratios. Remembering that on this surface both temperature and salinity arc nearly constant ahmg the path of the Central Water, it is interesting to note that the s;tmc is truc for initial phosphate if Rc = 135 is used, while the distribution ba.~d on Rc = 175 shows significant variation. This suggests that in the upper thcrmt~:linc of the Indian Ocean the classical value Rc = 135 m;,y he more appropriate than the revised value. The situation is, however, rather complicated. Use of Rc = 135 on the dccpcr isopycnal levels does not result in much improvement and produces quite unrealistic distributions below 0(K) m depth. It appears that the Rcdficld ratio may not bca constant across the depth range of the thcrm(~:linc.
Avoiding the use of the virtual water mass NICW by the use of more conservative parameters is thus not a trivi:d matter, and more work is required before wc can report on the desirable alternative in full. However. assuming that Rc = 135 is the correct Rcdficld ratio for the % = 26.7 density surface, and noting that on this surface diapycnal mixing is extremely unlikely, wc can solve the mixing problem for that surface only by using three parameters---temperature, salinity, initial phosphate- -and two source water types---the parameter values for ICW and A A M W on the isopycnal surface. This eliminates the artificial water mass NICW and should result in a water mass distribution where the A A M W contribution remains unchanged and the contribution of ICW is equal to the combined contributions of ICW and NICW from Fig. 9. The result of this analysis is given in Fig. A2; it is indeed very similar to Fig. g. This shows that the use of NICW as a proxy of aged SICW is acceptable.