SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. The Data
  5. 3. Results
  6. 4. Implications
  7. 5. Conclusions
  8. Acknowledgments
  9. References

[1] Satellite observations reveal a substantial weakening of the southeasterly trade wind over the South Indian Ocean during 1992–2000 and a resultant slowdown of the shallow meridional overturning circulation in this Ocean. The estimated rate of the slowdown of this circulation, 6.8 ± 1.4 × 106 m3/s over the 9-year period, is nearly 70% of the mean strength of this circulation. Such a change has important implications to upper-ocean heat content and decadal climate variability of the Indian Ocean, as well as to the ecosystem and air-sea exchange of CO2 in the region.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. The Data
  5. 3. Results
  6. 4. Implications
  7. 5. Conclusions
  8. Acknowledgments
  9. References

[2] The heat content in the upper few hundred meters of the Indian Ocean exhibits significant decadal to multi-decadal variability with an overall warming tendency since the mid-1960s [Levitus et al., 2000]. These variations not only have climatic impact on the Indian Ocean region, but on African rainfall and North Atlantic climate as well [Reason and Mulenga, 1999; Bader and Latif, 2003; Kerr, 2003]. Changes in ocean circulation, if any, that can potentially affect these variability of Indian-Ocean heat content have not been documented.

[3] The annual mean net surface heat flux is directed into the tropical Indian Ocean [Oberhuber, 1988]. This ocean exports the annual-mean net heat gain towards the subtropical South Indian Ocean (SIO, south of the equator) primarily through a wind-driven meridional overturning circulation in the upper several hundred meters [Wagcongne and Pacanowski, 1996; Lee and Marotzke, 1997, 1998; Garternicht and Schott, 1997; Miyama et al., 2003]. The circulation consists of a southern overturning cell and a cross-equatorial overturning cell, as conceptually shown in Figure 1.

image

Figure 1. Conceptual illustration of the time-mean meridional overturning circulation of the upper Indian Ocean that consists of a southern and a cross-equatorial cell. The time-mean zonal wind and surface heat flux are also shown schematically. The Indonesian throughflow that carries water from the Pacific Ocean into the Indian Ocean near the latitudes of 10°–15°S [Gordon, 1986] is not shown. This flow is believed to partially supply the cross-equatorial thermocline flow [Schott et al., 2002].

Download figure to PowerPoint

[4] Both the southern and the cross-equatorial cells transport warm water in the surface Ekman layer southward and colder thermocline water northward. These upper and lower branches of the overturning cells are connected by the downwelling (upwelling) south (north) of about 10°S [Lee and Marotzke, 1997]. For the downwelling, much of the surface water is subducted into the thermocline in the subtropical SIO [Zhang and Talley, 1998; Karstensen and Quadfasel, 2002]. For the upwelling, there is a southern branch between 10°S and the equator and a northern branch in the North Indian Ocean (NIO) [Lee and Marotzke, 1997; Susanto et al., 2001; Schott et al., 2002]. The southeasterly trade over the SIO controls the strength of the southern cell and thus affects the heat content of the upper SIO. The annual-mean southwesterly wind over the NIO regulates the strength of the cross-equatorial cell and thus influences the heat content of the upper NIO. The mean strengths of the southern and cross-equatorial cells are approximately 10 and 6 Sv (1 Sv = 1 × 106 m3/s), respectively [Lee and Marotzke, 1997; Schott et al., 2002].

[5] Not shown in Figure 1 is a pair of “equatorial rolls”. They are very shallow meridional overturning cells that lie primarily the upper 50 m and within a few degrees of the equator. On annual average, they are cyclonic to the north and anti-cyclonic to the south of the equator [Wagcongne and Pacanowski, 1996; Lee and Marotzke, 1997]. Due to their large intensities, these “rolls” contribute to the surface Ekman transport in the equatorial zone substantially. However, they have little effect on the meridional heat transport because of the small stratification in the upper 50 m [Schott et al., 2002].

[6] Decadal and longer variability of the heat content of the upper Indian Ocean, including the overall warming tendency since the mid-1960s, are mostly contributed by the SIO [Levitus et al., 2000]. A possible oceanic mechanism for such variability is the decadal-to-multi-decadal oscillations of the southern overturning cell. The larger volume of the SIO than that of the NIO may also contribute to the larger variability of heat content of the SIO. In this study, satellite observations of ocean surface wind stress and sea surface height (SSH) from the early 1990s to 2000 are used to analyze the potential changes of the two overturning cells. The data used for the analysis are described in section 2. The findings are presented in section 3. The implications to upper-ocean heat content, decadal climate variability, and biogeochemical cycles are discussed in section 4. Section 5 provides a summary.

2. The Data

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. The Data
  5. 3. Results
  6. 4. Implications
  7. 5. Conclusions
  8. Acknowledgments
  9. References

[7] The wind data include those derived from the ERS-1 and –2 scatterometers of the European Space Agency for the period of 1992–2000 [Bentamy et al., 2002] (http://www.ifremer.fr/cersat/). The two scatterometer missions span the period of late-1991 to early 2001. The QuickSCAT sactterometer mission of the National Aeronautics and Space Administration (NASA) has provided wind measurements since late-1999 [Liu, 2002]. However, non-negligible differences between the QuikSCAT and ERS-2 wind measurements are found during their overlapping period, probably due to the differences in sensor characteristics and data retrieval algorithm. To ensure the consistency, only ERS-1 and –2 measurements are used in this study because of their near decade-long record.

[8] The SSH data are obtained from the TOPEX/Poseidon altimeter that was launched in late 1992, a joint mission between NASA and the French National Center for Space Studies (CNES). The data are available through http://podaac.jpl.nasa.gov. Standard corrections, including those for the effects of tides and inverse barometer, have been applied [Fu et al., 1994].

3. Results

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. The Data
  5. 3. Results
  6. 4. Implications
  7. 5. Conclusions
  8. Acknowledgments
  9. References

[9] The scatterometer data indicate that the southeasterly trade over the SIO north of approximately 20°S experiences a near-decadal (1992–2000) weakening, dominated by the change of the zonal component (Figures 2a and 2b). The smaller changes of the meridional component are not shown. The largest weakening of the zonal wind occurs near 10°S, the approximate center latitude of the southern overturning cell. In contrast, the annual-mean southwesterly wind over the NIO shows little decadal change during this period (Figure 2a). These observations suggest that the strength of the southern cell has been slowing down during the study period while the cross-equatorial cell remains relatively steady.

image

Figure 2. (a) 1992–2000 linear trend of zonal wind stress in dyn/cm2/year (positive values indicate weaker easterly wind). The arrows show the directions of the annual-mean wind schematically. (b) Time series of zonal wind stress averaged over 20°–0°S. (c) Time series of meridional Ekman transport at 10°S.

Download figure to PowerPoint

[10] The total meridional volume transport of warm water in the surface layer (i.e., Ekman transport, the upper branch of the overturning circulation) at a certain latitude is given by the Ekman theory [cf. Pond and Pickard, 1978]: VE(y) = ∫−τx(y)/(ρ0f(y))dx where ∫dx is the zonal integration across the basin at latitude y, τx is zonal wind stress, ρ0 is a reference density of sea water, and f is the Coriolis parameter. The time series of VE at 10°S is shown in Figure 2c. Its linear reduction over the 1992–2000 period is 6.8 ± 1.4 Sv (1 Sv = 106 m3/s). The 1.4-Sv uncertainty is the standard deviation of the residual of the linear fit, which mostly reflects interannual variability.

[11] To further illustrate the slowdown of the southern cell and the relatively steady cross-equatorial cell, the latitudinal distribution of VE(y) averaged over the 1992–1995 and 1997–2000 periods are shown in Figure 3 (red and blue curves, respectively). The average of the two curves is qualitatively similar to the zonally and vertically integrated Ekman volume transport of the Indian Ocean estimated by Levitus [1988, Figure 5]. Let ΔVE(y) denotes the difference between these two periods (dashed curve). Its meridional variation, ∂ΔVE(y)/∂y, describes the temporal change in the divergence (convergence) of meridional Ekman transport that reflects the change of upwelling (downwelling). The positive values of ∂ΔVE(y)/∂y between about 10° and 2°S indicate that the upwelling in this region is stronger during 1992–1995 than during 1997–2000. The negative values of ∂ΔVE(y)/∂y south of about 10°S suggest that the downwelling in the tropical-subtropical SIO is stronger in the earlier than in the latter period. These evidences are consistent with the reduction of southward Ekman transport near 10°S, all of which reflect a slowdown of the southern overturning cell. The overall near-zero values of ∂ΔVE(y)/∂y in the NIO again highlight the lack of decadal change of the cross-equatorial cell. Since the cross-equatorial cell shows little decadal change, the 6.8-Sv reduction of southward Ekman transport at 10°S is mostly associated with the slowdown of the southern overturning cell. The near-decadal weakening of this cell is thus nearly 70% of its mean strength of 10 Sv (as mentioned earlier).

image

Figure 3. Latitudinal distribution of meridional Ekman transports averaged over the 1992–1995 and 1997–2000 periods (red and blue curves, respectively) and their difference (dashed curve). The values near the equator are masked out because Ekman transport is not well defined there.

Download figure to PowerPoint

[12] Satellite measurements of SSH provide additional evidence for the slowdown of the southern overturning cell. In tropical-subtropical ocean, zonal SSH difference across the basin, being coherent with zonal slope of the thermocline, is indicative of the zonal pressure gradient that drives the meridional geostrophic flow in the thermocline [Lee and Fukumori, 2003]. On interannual-to-decadal time scales, a larger east-west SSH difference reflects a weaker (stronger) northward thermocline flow in the southern (northern) hemisphere, vice versa. The linear trends of SSH estimated from the SSH data (Figure 4) show an increasing east-west SSH difference (Δh) across the SIO, suggesting a decrease of the net northward thermocline flow in this region. The linear change of Δh across the SIO is significantly larger than the interannual variability of Δh. The trends of east-west SSH difference across the NIO is much smaller and is statistically insignificant, consistent with a relatively stable cross-equatorial cell inferred from the wind observation.

image

Figure 4. 1993–2000 linear trend of SSH estimated from the TOPEX/Poseidon data. The data in year 1992 are not used as they do not cover the entire year.

Download figure to PowerPoint

[13] Different from the large increase of east-west SSH difference across the SIO (approximately 15 cm over the 1993–2000 period), the trend of sea level averaged over the SIO is close to zero. However, this does not necessarily mean that the total upper-ocean heat content of the SIO is not changing. The relative contribution of the variations in salinity and temperature to spatially averaged sea level need to be investigated, a subject beyond the scope of the present analysis.

4. Implications

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. The Data
  5. 3. Results
  6. 4. Implications
  7. 5. Conclusions
  8. Acknowledgments
  9. References

[14] A slowdown of the meridional overturning circulation connecting the upper tropical and subtropical Pacific Ocean since the mid-1970s has been reported along with discussions on its relation to Pacific decadal variability [McPhaden and Zhang, 2002; Lee and Fukumori, 2003]. The near decadal weakening of the meridional overturning cell in the SIO, 6.8 Sv over 9 years, is somewhat larger than the averaged rate of the slowdown of the tropical-subtropical overturning cell in the Pacific, about 5 Sv/decade (inferred from the analysis by McPhaden and Zhang [2002]). A prominent theory [Kleeman et al., 1999] suggests that the oscillation of the tropical-subtropical overturning cell in the Pacific Ocean plays an active role in Pacific decadal climate variability. Decadal change of the overturning cell in the SIO might also be important to decadal climate variability of the Indian Ocean for the following reason. The changes in the divergence of Ekman flow and upwelling in the tropical SIO would affect sea surface temperature (SST). A slight change in SST in this region could induce ocean-atmosphere coupling [Xie et al., 2002]. Evidence of such a coupling in the tropical SIO has been reported [Xie et al., 2002].

[15] The substantial weakening of the southern overturning cell reduces the rate of southward export of warm surface water and the rate of northward intrusion of colder thermocline water. This favors an increase of heat content in the upper SIO. The lack of decadal change of the cross-equatorial cell helps maintain a relatively stable heat content of the upper NIO. The relatively large changes of the wind and meridional overturning circulation in the SIO and the lack of them in the NIO may be ubiquitous features in the previous decades as well. Significant decadal-to-multi-decadal variability of Austral-summer wind during the period of 1900s to 1980s is found over the SIO, but not over the NIO [Allan et al., 1995]. The larger decadal variability of the wind over the SIO and the resultant change in the meridional overturning circulation appear to be consistent with the fact that heat content variability in the upper SIO is larger than that of the upper NIO [Levitus et al., 2000].

[16] The weakening of the shallow overturning circulation in the SIO also has implications to biogeochemical cycles. The upwelling in the tropical SIO is associated with an enhanced chlorophyll concentration [Murtugudde et al., 1999; Xie et al., 2002]. It reflects the supply of the nutrient-rich subsurface water. Chlorophyll is a measure of the phytoplankton biomass, which is the base of the marine food web. The reduction of the upwelling in the tropical SIO would decrease the nutrient supply to the euphotic zone. This influences the ecosystem by limiting biological productivity. Moreover, a change in biological production (e.g., photosynthesis) affects SST by modifying the ocean's absorption of short-wave radiation [Sathyendranath et al., 1991; Nakamoto et al., 2000]. The Indian Ocean is a net sink of atmospheric CO2 (about 20% of global oceanic uptake) because the uptake of atmospheric CO2 by the subtropical SIO exceeds the tropical outgassing of dissolved CO2 [Takahashi et al., 1997; Carr et al., 2002]. The reduced downwelling and upwelling in the subtropical and tropical SIO, respectively, may affect the air-sea exchanges of CO2 in these regions.

[17] The decadal change of the southeasterly wind over the SIO not only affects the meridional overturning circulation, but the horizontal circulation as well. The time-mean horizontal circulation of the SIO features an anti-cyclonic tropical gyre and a cyclonic subtropical gyre, separated by the South Equatorial Current (near 10°–15°S) [Schott et al., 2002]. The weakening of the southeasterly wind results in a positive (negative) wind stress curl north (south) of 11°S. As implied by the Sverdrup theory, the changes of the horizontal circulation are characterized by a cyclonic gyre between the equator and 11°S that weakens the time-mean anti-cyclonic tropical gyre, and by an anti-cyclonic gyre between 11°S and 30°S that weakens the time-mean cyclonic subtropical gyre (not shown). The strengths of these anomalous gyres estimated from the Sverdrup theory are 6.6 and 11.4 Sv, respectively. Based on the observed speeds of the dominant Rossby waves [Polito and Liu, 2003], it takes about 1.3, 2.7, and 4.7 years for these waves to cross the SIO at 10°, 20°, and 30°S, respectively. For the decadal change being discussed, the steadiness assumption of the Sverdrup theory is reasonable north of 20°S. Towards 30°S, however, the interpretation of the Sverdrup calculation becomes questionable. On time mean, the horizontal circulation has little contribution to the meridional heat transport because of the relatively weak zonal gradient of temperature [Lee and Marotzke, 1997]. Therefore, the decadal changes of the horizontal circulation are expected to play a minor role in the variation of meridional heat transport.

5. Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. The Data
  5. 3. Results
  6. 4. Implications
  7. 5. Conclusions
  8. Acknowledgments
  9. References

[18] Satellite observations of ocean surface wind stress and sea surface height anomaly are used to analyze the decadal change of the meridional overturning circulation of the upper Indian Ocean from 1992 to 2000. It is found that the strength of the overturning cell in the SIO decreases by 6.8 ± 1.4 Sv (about 70% of its mean strength) over this period in response to a substantial weakening of the southeasterly wind over the SIO. However, the cross-equatorial overturning cell that connects the subtropical SIO and the NIO shows little decadal change. The rate of the slowdown of the southern overturning cell is somewhat larger than the averaged rate of the slowdown of the shallow meridional overturning circulation in the tropical-subtropical Pacific Oceans from the mid-1970s to the 1990s reported by McPhaden and Zhang [2002]. The large decadal change of the southern cell and the lack of it for the cross-equatorial cell have important implications to upper-ocean heat content, decadal climate variability, ecosystem, and air-sea exchange of CO2 in the region.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. The Data
  5. 3. Results
  6. 4. Implications
  7. 5. Conclusions
  8. Acknowledgments
  9. References

[19] The author would like to thank Drs. W. T. Liu and W. Tang for providing the gridded QuikSCAT data and Dr. L.-L. Fu and Ms. Akiko Hayashi for providing the gridded TOPEX/Poseidon and JASON-1 data. Comments from Drs. L.-L. Fu, I. Fukumori, W. Patzert, and M.-E. Carr on earlier versions of the manuscript are appreciated. The research described in this paper was carried out at JPL, California Institute of Technology, under a contract with NASA.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. The Data
  5. 3. Results
  6. 4. Implications
  7. 5. Conclusions
  8. Acknowledgments
  9. References