Comparisons of Summer-Winter Carbon�te and Nutrient Data in the Southwest Indian Ocean

Chemical data were obtained in the southwestern Indian Ocean in the austral Winter (July) of 1984 and in the austral summer (Feb. March) of 1985 aboard the French research vessel, MARION DUFRESNE. The winter data represent initial chemical concentrations in an important source region of the Antarctic Intermediate Water at the time of its formation. For the first time, one can evaluate variations in the chemical cycles in the Antarctic Intermediate Water of the Indian Ocean with reference to the source water in winter. These winter data were compared with our summer data and with the summer data obtained in 1978 during the GEOSECS Expedition. Prelimi­ nary results indicate the following: the surface pH and normalized nitrate, alkalinity and total C02 values are found to correlate linearly with temperature; small deviations from the linearity are related to the Subantarctic and Subtropical Fronts and to the equatorial upwelling; large variations in nitrate and pH are found in surface waters collected at the same location but in different seasons; however, there is less variation between pH or normalized nitrate concentrations when compared at the same tempera­ ture; a seasonal difference in alkalinity and total C02 may exist, even when compared at the same salinity and temperature; the decrease in alkalinity and total C02 between the Antarctic Waters and the Indian Central Water found north of the Subtropical Front can perhaps be attributed to the decrease in nitrate and the increase in temperature; the remnant North Atlantic Deep Water (NADW), which has a very weak salinity signal, is identified clearly by pH and total C02 data; and nutrient and oxygen data also help in tracing NADW.


INTRODUCTION 59
Deep waters from the three oceans move to the Southern Ocean and mix there.The resultant, relatively homogeneous water becomes the major source of the Antarctic Bottom Water (AABW), which spreads back out into the deep world oceans.Consequently, the chemistry of Southern Ocean water is a baseline for the deep wotld oceans.One must therefore know the chemistry of the Southern Ocean water in order to understand global biogeochemical cycles (Wust, 1939;Bolin, 1983;Chen, 1984).Unfortunately, only a few high-precision chemical oceanographic programs have been conducted in the Indian Ocean College of Marine Sciences, National Sun Yat-Sen University, Kaohsiung, Taiwan, R.O.C. section of the Southern Ocean.For instance, I know of only 7 stations with good quality carbonate and nutrient data south of 30° S prior to our investigations (Weiss et al., 1983).
Not knowing the characteristic properties of the water near its origin, therefore, makes it difficult, if not impossible, to interpret variations in the carbon chemistry or to calculate excess C02 in the Indian Ocean.Furthermore, the scant data in the southern Indian Ocean were all collected in the summer; whether the summer data are representative of the mainly winterformed waters is entirely uncertain (Chen and Pytkowicz, 1979;Chen, 1982a,b;1987a,b;1988b;Chen and Poisson, 1984;Poisson and Chen, 1987).
Recognizing the importance of collecting more chemical data, especially seasonal data, in the southern Indian Ocean, I participated in two cruises, INDIVAT

OUTLINE OF THE EXPEDITIONS
The MARION DUFRESNE departed from la Reunion for the INDIVAT 1 Expedition on 3 July 1984, reoccupied GS 427 on 5 July, and reoccupied GS 429 on 19 July after a stop in Crozet.The vessel then proceeded to Kerguelen and Amsterdam Is. and returned to Ia Reunion on 4 August.The cruise track is shown in Figure 1.The Subtropical Front was near 40° S and the Antarctic Front was near 4T S. While underway, sea surface samples were collected hourly from a seawater intake located at the bow 4 m below the surface.Tumpera ture and salinity at the intake were recorded by a Thermosalinograph.During INDIVAT 1 seawater was pumped through a rubber tube to a van near the laboratory where samples were taken.Because of the delay in sampling, the recorded temperature and salinity may not minutes) required to collect all samples (salinity, oxygen, pH, alkalinity, total C02, p C02, nitrate, phosphate, and silicate) may also have caused some discrepancies in the data, as the water flowing out of the tube at the beginning of sampling may have been somewhat different from the water flowing out at the end of sampling.Subsurface samples at GS 427 and 429 were obtained using a Neil Brown CID-Rosette system (Chen et al., 1986).March.Chemical data were collected from 23 stations including four GEOSECS stations (427 -429, 454).Underway samples were also taken directly near the underwater intake.
The attempt to reach the Antarctic Continent, however, was thwarted by foul weather.The cruise track is given in Figure 1.The Subtropical Front was at approximately 43• Sand the Antarctic Front was near 52• S (Chen et al., 1986).

EXPERIMENTAL TECHNIQUE
The pH samples were all analyzed at 25 ±0.02•C with a Radiometer combination electrode within 30 minutes.NBS 4.004 and 7.4 15 buffers were used to calibrate the elec trode.The reproducibility of the pH measurements is better than ±0.003 units for replicate samples.The electrode drift (assumed to be linear) was determined at approximately 10-day intervals.The drift was approximately 0.002 pH unit/day, and the correction was made to the measured values (Byrne et al., 1988;Chen, 1988aChen, , 1989;;Chen et al., 1986).
The CID-Rosette used to obtain deep samples malfunctioned once during INDNAT 1, so all 11 bottles were closed at approximately 3400 m at GS 427.Four replicate samples were taken from each bottle.The standard deviation of the pH data (44 points) is 0.0027 pH units (1 a) which includes random error in both sampling and analysis.The standard devia tion corresponds to roughly 1 µ mol.kg-1 in total C02 (Chen et al., 1986).
Alkalinity and total C02 were determined at 25" ± 0.02" C with an Apple fl-controlled titration cell using a program similar to that of Almgren et al. (1977) and Bradshaw et al. (1981).These measurements have a precision of ±4 µ eq.kg- 1 for alkalinity and ±5 µ mol.kg-1 for total C02 and were performed within 12 hours of sampling ( Chen et al., 1986;Keeling et al., 1988).Duplicate samples from 3 Rosette bottles were taken at 3400 m at GS 427.The standard deviation of the alkalinity data is 1.3 µ eq.kg-The above mentioned pH, alkalinity, total C02 and nitrate data are listed in a technical report (Chen et al., 1986) and are not relisted here.

CHEMISTRY OF THE SURFACE WATERS
Many surface-water chemical properties, especially when normalized to a constant salinity to remove the effects of evaporation and precipitation, are known to correlate linearly with temperature (e.g., nitrate: Chen et al., 1982b; pH, phosphate and silicate: Chen, 1984;calcium: Chen et al ., 1982a;alkalinity: Edmond, 1974;Chen and Millero, 1979; and total C02: Chen and Millero, 1979).My normalized nitrate (NN03= N03 x 35/S) values for surface waters are also found to correlate generally linearly with surface tempera ture (Figure 2; Chen, 1988b) between 22• C and 1 r C, except for slight changes in slope near the Antarctic and Subpolar Fronts (approximately 4° C and 13° C, respectively).There is essentially no nitrate above 1 r C. The normalized phosphate shows the same trend (Chen et al., 1986).Le Corre and Minas (l983) also observed similar phenomena for nitrate and phosphate in the same general region in summer (March) 1977.
The pH also correlates linearly with temperature below 1 r C (Figure 3) except for a slight change in slope near the fronts.Note that above 1 r C the nitrate is essentially used up.There seemed to be a pH-stat between 17° C and 23° Cduring INDIGO 1/INDIVAT3, a phenomenon not observed during INDIVAT 1 (Figure 3).The pH seems to increase again with temperature above 23 • C.
Normalized alkalinity (NTA =TA x 35/S), and normalized total C02 (NTC02 = TC02 x 35/S) also correlate roughly linearly with temperature (Figures 4 and 5    ..  At the same time, the NTC02 must be reduced by the same amount.In addition, the warming of seawater drives out dissolved C02 and reduces the NTC02 by 8 µ mol kg-1 , assuming that the surface seawater remains at the same degree of saturation with respect to C02• Production of organic carbon as soft tissue is associated with the production of inor ganic carbon as hard tissue or shells at roughly a fout to one ratio (Broecker and Peng, 1982).Thus, the production of 185.  ).These results strengthen the suggestion that biological activities contribute to most of the reduction in nitrate, alkalinity and total C02 before the Antarctic waters reach the tropical zone.It should be pointed out that the above is the first attempt to make calculations along this line and it remains to be seen whether the assumptions are valid.

CHEMISTRY OF THE SUBSURFACE WATERS
During INDIVAT 1 only limited samples were collected at two stations, GS 427 and 429.The temperature, salinity, pH, NTA and NTC02 for these stations are plotted vs. depth in Figures 6 and 7, respectively (Chen et al., 1986).GS 427 is north of the Subtropical Front; Antarctic Intermediate Water (AAIW) is found here as a S-min layer at approxi mately 1000 m.A pH-min is located slightly below this.depth.NTA and NTC02 seem to increase with depth, but at a faster rate near the surface.Most of this increase is due to the increase in preformed values.
GS 429 is south of the Antarctic Front.AAIW is absent here (Figure 7).T, S, pH, NTA, A very pronounced pH-min occurs at 800 m, with a NTCOz-max immediately below it.They do not coincide, perhaps, because dissolution of CaC03 below 800 m, signaled by the increase of NTA, increases NTC02 further but does not aff ect pH.A very weak S-max exists at about 2000 m, reflecting the influence of the remnant North Atlantic Deep water (NADW) which is low in NTC02 (minimum; Figure 7) and low in nutrients but high in pH (maxi mum; Figure 7).The pH and NTC02 signals are much stronger than the salinity signal and are quite useful as additional tracers in the Southern Ocean (Chen, 1984;Chen andRodman, 1985, 1990).
The temperature cross-section for the stations occupied during the INDIGO 1 I INDIVAT 3 expedition is given in Figure 8. Upwelling is evident for the southernmost stations (G 10-16), and is also shown clearly in the salinity cross-section (Figure 9).In addition, Figure 9 shows the Subtropical Front near 43 • S and the low salinity tongue of AAIW.One particularly interesting feature was the core of S-max water found at about 2700 mat G 9 near 43• S.This is probably the core of remnant NADW (Redfield, 1960), The pH cross-section is given in Figure 10.Decomposition of soft tissue decreases the pH with increasing water depth.However, the water near the bottom is affected by AABW, which has a relatively high pH value.As a result, a pH-min layer is formed (Chen, 1984;Chen and Rodman, 1985;Chen et al., 1986).
The NTA cross-section is given in Figure 11.NTA increases with depth almost monotonically in regions north of the Subtropical Front because of the dissolution of calcar eous hard tissue and shells.South of the Subtropical Front, the water contains mainly  .... ..

�
siliceous organisms so that the increase in NTA with depth is small.Because Circumpolar upwelling brings deep water (and its high NTA) to i:he surface, the entire region south of the front is high in NTA throughout the water column.Biological consumption reduces surface NTA again when the water flows northward.Note the NTA-max at G 1 caused by the under cutting of AABW,Jhe data of Naqvi and Naik (1983) and Sen Gupta and Naqvi (1984) in the North Indian Ocean show this NTA maximum clearly.
Figure 12 gives the NTC02 cross-section.The .resolution is coarser than the pH plot because of the relatively poor precision of the NTC02 data.Nevertheless, the major features are preserved, i.e. the core of NT C02-min water at G 9 (also shown on S and pH plots), the NTC02-max at G 3 (also shown on NTA plot), and the effect of Circumpolar upwelling (also shown on T, S, pH, and NTA plots).For comparison, the apparent oxygen utilization cross section is presented in Figure 13 (Chen et al.,198 6).The similarity between Figures 12 and 13 is apparent.The AOU cross-section shows a maximum at the southernmost stations.This maximum corresponds to the pH minimum found on Figure 10.However, the AOU maxi mum for the northernmost stations fall slightly deeper,than the pH-min layer.higher than the INDIVAT 1 values, which are in turn 10 µ eq kg-1 higher than the GEOSECS values (Figure 4).The INDIGO 1/ INDIVAT 3 NTC02 values are approxi mately 20 µ mol kg-1 higher than the INDIVAT 1 values, which are in turn 10 µ mol kg-1 higher than GEOSECS (Figure 5).These differences may indicate either seasonal effects or systematic differences in analysis.More data are needed in order to make a better judge ment.) independently of each other, and during GEOSECS.These fo ur sets of data generally agree to within IOµ eq kg-1 for NTA and 10 µ mol kg-I for NTC02, variations only slightly larger tl lan th e combined experimental error.The agreement is equally good at GS 428 (Chen et al., 1986).
Significant differences, however, exist between different data sets a� GS 429.There is a large difference in NTC02 near surface, with the winter INDIVAT 1 values the highest (between 20 and 25 µ mol kg-I higher th an the IN DIGO 1 and GEOSECS data), obviously because of seasonal effect.The smaller systematic difference for subsurface waters, how ever, is probably due to analytical errors.Poisson's NTA values are approximately 10 µ eq kg-1 higher than the GEO SECS data with Chen's data in between the two.Poisson's NTC02 values agree with Chen's data and both are approximately IOµ mol kg-1 higher than the GEQSECS data (Chen et al., 1986;Poisson et al., 1985).

CONCLUSION
I have obtained the first contemporary winter nitrate and carbonate data in the south western Indian Ocean.Surface nutrients, pH, alkalinity and total C02 values differ signifi cantly from the summer data when compared at the same location.But the differences are less pronounced when compared at the same salinity and temperature.These winter data can now be used to estimate the initial concentrations of nutrients, pH, alkalinity and total C02 fo r waters formed in the southwestern Indian Ocean, such as the Antarctic Intermediate water in th e Indian Ocean sector.
1 and INDIGO 1 / INDIVAT 3. INDIVAT (INDIEN VALORISATION de TRANSIT) took advantage of the travel schedule of the French research/supply vessel, MARION DUFRESNE, which steams from Ia Reunion to Crozet, Kerguelen, Amsterdam, and back to la Reunion three to four times per year.The French TA AF (TERRES AUS TRALES et AN TARC TIQUES FRANCAISES), which operates the vessel, agreed that I could use the ship to collect and measure surface samples during transit.Because the vessel crosses the Subtropical and Antarctic Fronts many times a year, it provides an excellent opportunity to study the sea sonal variation of chemicals in the formation region of the Antarctic Intermediate Water (Chen and Poisson, 1986; Chen et al., 1986).TAAF also agreed to permit me to reoccupy two GEOSECS stations (GS 427 and 429) during each INDIVAT cruise.Deep samples provide me with a means to calibrate my results for comparison with data reported in the literature.In addition, I can determine how much the seasonal variation affects the water column.Such information is essential for compari sons from year to year.The second program, the INDIGO (INDIEN GAZ OCEAN), involves the use of the MARION DUFRESNE to collect deep samples in different regions of the Indian Ocean once a year for a minimum of four years.The first INDIVAT expedition took place in July 1984.INDIVAT 2, however, had to be cancelled owing to logistical problems.The combined INDIGO l / INDIVAT 3 expedition was carried out in February/March 1985.
The first INDIGO expedition (with emphasis on subsurface samples) and the third INDIVAT expedition (with emphasis on underway surface samples) were combined.R/V MARION DUFRESNE departed from la Reunion on 23 February 1985 and returned on 30
5 µ mol kg-1 in organic carbon should result in a further reduction of 46.4 µ mol kg-1 in NTC02 and 63 µ eq kg-1 in NTA •after gtldng into considera tion the effect of nitrate and phosphate on alkalinity (Brewer et al., 1975; Dyrssen, 1977; Chen et al., 1982a).Consequently, I expected a total reduction in TC02 of 240 µ mol kg-• and observed a reduction of 215 µ mol kg-1 • I also observed a NTA reduction of 70 µ mol kg-1 • These correlations indicate that biological activities contribute to most of the reduction in nitrate, alkalinity and total C02.

For
waters north of 35 • S (waters wanner than 1 T C ), nitrate concentration is so low that other sources of nitrogen, such as ammonia, or nitrogen fixers may be important in biological consumption.Thus, the Redfield ratio is no longer applicable.Further, the effect of equatorial upwelling becomes important, and simple linear relations do not exist.The decrease in NN03 of 28 µ mol kg-1 between4 and 17" C for INDIGO 1 I INDIVAT 3 corresponds to the decrease of 239 µ mol kg-1 in NTC02 and 63 µ eq kg-1 in NTA.I observed a similar decrease in NTC02 and NTA of 200 µ mol kg-1 and 75 µ eq kg-1 , , respec tively (Chen et al., 1986 NTC02 and NN03 (not shown on Figure..c7) remain constant above 150 m due to winter mixing.To my knowledge, this is the first time such winter data have been collected in the Indian ocean section of the Circumpolar Water (Chen et al., 1986).

Fig. 6 .Fig. 7 .
Fig. 6.Vertical profiles of temperature, salinity, pH, NTA, and NTC02 at GS 427 reoccupied during INDIV AT 1 (T, S data taken from Poisson et al., 1985).The lines for pH, NTA and NTC02 are drawn by eye.Note the changes in scales.

Fig. 8 .
Fig. 8.The temperature cross-section for stations occupied during the INDIGO l / INDIV AT 3 Expedition (data taken from Poisson et al., 1985).

6.
COMPARISON OF INDIVAT 1 AND INDIGO 1 / INDIVAT 3 DATA WITH GEOSECS DATA Even when data are collected at the same locations, it is difficult to compare those collected at different times because one can never be sure that the same waters;especially surface waters, have been sampled.For example, surface N03 values show significant varia tions over three cruises to the same location.At GS 42 9, occupied in summer (Feb.1978 ), the surface concentration was 20.2 µ mol kg-1 (Weiss et al., 1983).The INDIGO 1 / INDIVAT 3 value was 21 .2µ mol kg-1 (March 1985) and the INDIVAT 1 value was 24 .5 µ mol kg-1 (July 1984).Tue correlations between NN03 and temperature, however, only changed slightly, with the INDIGO 1/ INDIVAT 3 concentrations systematically higher (Figure 2) .For waters south of the Subtropical Front, I also do , not see much difference between the pH/temperature correlation for data collected in winter and that obtained in summer given the statistical accuracy of the data.There are, however, significant differences in surface NTA and NTC02 values among the three cruises (I subtracted 15 µ mol kg-1 from the reported TC02 values according to the recommendation given in Weiss et al., 1983).INDIGO l / INDIVAT 3 seems to have pro duced the highest NTA and NTC02 values.These NTA values are approximately 5 µ eq kg-1