SEARCH

SEARCH BY CITATION

Keywords:

  • Argo;
  • Indian Ocean;
  • barrier layer;
  • interannual variability

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method
  5. 3. Seasonal Cycle
  6. 4. Interannual Variability
  7. 5. Conclusions
  8. Acknowledgments
  9. References

[1] Interannual variability of the barrier layer (BL) in the southeastern tropical Indian Ocean (SETIO) is examined using temperature and salinity profiles derived from Argo floats since 2004. We show that a quasi-permanent BL exists off Sumatra with a semi-annual cycle and a maximum in November. Further, interannual variability of the BL is closely related to the Indian Ocean Dipole (IOD) with the IOD leading the BL by one month. During the 2006 positive IOD (pIOD) season, equatorial easterly-induced upwelling Kelvin waves raise the isothermal layer (IL) off Sumatra; a salinity-stratified mixed layer (ML) shoals due to a reduced eastward salty water transport by a weaker Wyrtki Jet, despite an offset by a reduced freshwater flux. Consequently, thinning of the BL is dominated by thinning of the IL. During the 2010 negative IOD (nIOD), similar processes operate but in an opposite direction. As thinning of the BL during a pIOD enhances the thermocline-ML coupling, our results reveal that an IOD-induced co-varying BL in turn enhances the IOD positive feedbacks.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method
  5. 3. Seasonal Cycle
  6. 4. Interannual Variability
  7. 5. Conclusions
  8. Acknowledgments
  9. References

[2] The SETIO (60°E-100°E, 6°S-Eq) receives a considerable amount of local rainfall and a large quantity of freshwater from the Bay of Bengal and the Indonesian Throughflow [Sprintall and Tomczak, 1992; Jensen, 2003; Sengupta et al., 2006]. The large freshwater flux into the region makes waters in the near-surface layer less salty than the west, and maintains a strong haline stratification, resulting in a ML shallower than the surface IL [Eriksen, 1979; Sprintall and Tomczak, 1992]. This tends to favor a salinity-stratified intermediate layer between the base of the ML and the top of the thermocline (i.e., IL), a depth range referred to as a BL [Godfrey and Lindstrom, 1989].

[3] Because the BL is a blockade to turbulent entrainment of the cold thermocline water into the ML [Lukas and Lindstrom, 1991], its presence has a significant impact on sea surface temperature (SST) [Masson et al., 2005]. Further, the BL in the eastern Indian Ocean is tightly related to the development and evolution of the IOD, which reinforces the oceanic anomalies favoring a strengthening of air-sea interactions [Masson et al., 2004]. Therefore, understanding the evolution of the BL is essential to fully understand the development of the IOD.

[4] Seasonal evolution of the BL in the eastern tropical Indian Ocean has been investigated [Wyrtki, 1971; Eriksen, 1979; Sprintall and Tomczak, 1992; Godfrey et al., 1999; Masson et al., 2002; Qu and Meyers, 2005; de Boyer Montégut et al., 2007]. It is believed that the BL formation in the eastern basin is linked to the freshwater input in the near-surface layer from rainfall and river runoff [Sprintall and Tomczak, 1992], and is also forced by the equatorial waves and the Wyrtki Jet [Masson et al., 2002].

[5] Recent model studies [Murtugudde et al., 2000; Annamalai, et al., 2003] find an unusual absence of a BL off Sumatra during September and November (SON) of pIOD years, which favors strong air-sea interactions, suggesting that the evolution of the BL leads the IOD. In agreement,Masson et al. [2004] highlight the effect of salinity and a thinner BL on the 1997 pIOD. However, due to a lack of observations, an observational depiction of variability of the BL has been lacking and therefore its dynamical relationship with the IOD remains unclear.

[6] With an unprecedented collection of temperature and salinity profiles from Argo floats over the past decade, it is now possible to compile an observational image of interannual variability of the BL. Already Argo profiles have been used to examine the 2006–2008 three-consecutive pIOD events [Cai et al., 2009]. Here we examine the seasonal cycle of the SETIO BL and its interannual variability using Argo profiles. In particular, we focus on formation mechanisms of the BL during positive and negative IOD events.

2. Data and Method

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method
  5. 3. Seasonal Cycle
  6. 4. Interannual Variability
  7. 5. Conclusions
  8. Acknowledgments
  9. References

[7] Quality controlled Argo profiles for the period between January 2004 and September 2011 are obtained from the China Global Argo Data Assembly Center. Only profiles with data coverage both in the upper 10m and depths below are used. Depths of ML and IL are calculated from individual profiles and then interpolated onto a regular gird of 1° latitude × 1° longitude for each month using the Kriging algorithm [Oliver and Webster, 1990]. Following Bosc et al. [2009], the IL depth is defined as the depth where temperature is 0.2°C lower than that at the 10 m depth. The ML depth is defined in terms of a depth with a density equal to that at the 10 m depth plus an increment in density equivalent to −0.2°C. The BL thickness is then defined as the difference between the depths of the IL and the ML.

[8] Monthly 1° × 1° gridded temperature and salinity anomalies with a vertical resolution ranging from 7.5 m to 10 m for the upper 150 m, derived from Argo profiles [Roemmich and Gilson, 2009] are used to explore variability of temperature and salinity. Surface winds and currents, obtained from NCEP/NCAR reanalysis [Kalnay et al., 1996] and from monthly currents from Ocean surface current analyses – real time (OSCAR), which represent the average current from the surface to the 15 m depth, computed from satellite surface topography and vector winds [Bonjean and Lagerloef, 2002], respectively, are used to examine the associated circulation anomalies. Also used are precipitation from the Tropical Rainfall Measuring Mission (TRMM) 3B43-V6 product [Huffman et al., 2007], and evaporation from the Objective Analysis Flux (OAFlux) [Yu and Weller, 2007]. Monthly anomalies are computed by subtracting the corresponding monthly climatology. Below we examine the seasonal evolution of the BL, the associated dynamics, and how the process varies with the IOD leading to interannual variability.

3. Seasonal Cycle

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method
  5. 3. Seasonal Cycle
  6. 4. Interannual Variability
  7. 5. Conclusions
  8. Acknowledgments
  9. References

[9] Climatological seasonal cycle of the BL thickness (Figure 1a) and the IL and ML depths (Figures 1b and 1c) reveals several features. Consistent with the result of Sprintall and Tomczak [1992], the BL is present in the SETIO all months but with large semiannual fluctuations peaking in terms of an east-west extent and thickness during June and during November [Masson et al., 2002]. The November peak marks its annual maximum with a thickness (∼20 m) doubles that of the June peak (∼10 m), and extends further to the west.

image

Figure 1. Annual evolution of (a) barrier layer (BL) thickness, (b) isothermal layer (IL) depth, and (c) mixed layer (ML) depth in meters averaged between the equator and 6°S as a function of longitude and months.

Download figure to PowerPoint

[10] A semiannual cycle in the IL depth is also observed (Figure 1b) and is almost in phase with the BL. The peaks around June and November are largely due to equatorial downwelling Kelvin waves generated during the monsoon transition seasons [Han et al., 1999; Qu and Meyers, 2005; Qiu et al., 2009]. In addition to semiannual, the IL also exhibits well-defined annual variability, deepening during austral winter (June–September) and shoaling in summer (January–April) months. The equatorial convergence of water in response to annual component of zonal westerlies in winter months (June–September), and vice-versa in summer is primarily responsible for this annual variation [Qu and Meyers, 2005; Qiu et al., 2009].

[11] In contrast to the western Indian Ocean, where the IL and ML seasonal variations are almost identical (Figures 1b and 1c), the ML depth is significantly shallower than the IL all year round in the east, because of local rainfall and a freshwater input from the Bay of Bengal and the Indonesian Throughflow [Sengupta et al., 2006]. This gives rise to the BL. Over the eastern Indian Ocean, the ML displays a predominant annual cycle, almost in phase with the IL, deepening during winter and thinning during the rest of months. The fresh water flux into the ocean, with a greater rate of precipitation than evaporation, and a stronger surface heating during summer than those in winter are key factors influencing the annual variation [Qu and Meyers, 2005].

[12] Thus, over the SETIO, it is the difference in terms of both the phase and amplitude between the IL and the ML that explains the large semiannual cycle of the BL. Our results support the notion that seasonal variations of the BL thickness, occurring as a consequence of a Kelvin wave response of the thermocline to remote wind variability dominates the seasonal variation of the BL, as discussed by Masson et al. [2002]. Regressing BL monthly variations of the IL and the ML east of 85°E yields coefficients of 0.33 (p = 0.01) and −0.02 (not significant), respectively, suggesting the dominant role of the IL in the seasonal cycle of the BL.

4. Interannual Variability

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method
  5. 3. Seasonal Cycle
  6. 4. Interannual Variability
  7. 5. Conclusions
  8. Acknowledgments
  9. References

[13] An absence of a BL off Sumatra during SON of dipole years has been found to favor strong air-sea interactions and help to trigger pIODs [Murtugudde et al., 2000; Annamalai, et al., 2003; Masson et al., 2004]. These studies further suggest that the evolution of the BL leads the IOD. Among elements involved in the process is the westward South Equatorial Current that replaces an eastward transport by the Wyrtki Jet, spreading surface fresh water from Sumatra, and creating a shallow salinity stratification. A lagged correlation analysis using all monthly data averaged over the eastern Indian Ocean region (85°E–100°E, 6°S–0°N) finds that the correlation is highest when the IOD leads the BL by one month (Figure 2a). This suggests that, though reinforcing the associated positive feedbacks, thinning of the BL is a direct consequence of the IOD, rather than a trigger of the IOD. Figure 2bplots the relationship between the IOD and BL thickness anomalies shifted by one month. A higher sensitivity of BL thickness anomalies to the dipole mode index (DMI) is present during May–November (IOD prevalent months) than that in other months. A pIOD is associated with a thinner BL while a nIOD is with a thicker BL, with a negative correlation (−0.54) significant at the 99% confidence level. In contrast, the correlation (−0.26) for the non-IOD months is not significant.

image

Figure 2. (a) Lead/lag correlation between a dipole mode index (DMI) and BL thickness anomalies averaged over the southeastern Indian Ocean region (85°E–100°E, 6°S–0°N) using data from January 2004 to September 2011. A positive value in x axis means that the IOD leads. (b) Scatter diagram of BL thickness anomalies against DMI values with the DMI leading by one month. Scatter points in IOD months (May–November) and no-IOD months are shown as red solid and blue hollow circles, respectively. Linear regression lines and correlations are plotted. All data are normalized by their standard deviation.

Download figure to PowerPoint

[14] To elucidate the dynamic linkage between the IOD and BL, including the role of the Wyrtki Jet transport, we select a pIOD (2006) and a nIOD (2010) year to examine variations of the BL thickness in conjunction with the IOD events during their mature phase (September–November, or SON).

4.1. The 2006 pIOD Event

[15] During this pIOD event, the BL extends further westward to around 68°E and thickens in the western SETIO (68°E–85°E) but thins off Sumatra (east of 85°E) as indicated by the large positive and negative anomalies respectively (Figure 3a). The anomaly pattern of IL depths (Figure 3b) somewhat resembles that of the BL thickness anomalies in the SETIO, confirming the predominant role of the IL on the variation of the BL, as in the seasonal cycle. The ML anomaly pattern differs from that of the IL in that there is virtually little thinning off Sumatra, and that north of 2°S, a shoaling occurs (Figure 3c). The features favor a thinner BL off Sumatra, and a relatively thicker BL to the west (Figure 3a).

image

Figure 3. Spatial pattern of (a) BL thickness, (b) IL depth, and (c) ML depth anomalies in meters for SON 2006 with seasonal mean removed. (d–f) Same as Figures 3a–3c, respectively, but for SON 2010.

Download figure to PowerPoint

[16] Previous studies [e.g., Masson et al., 2002] and results described in Section 3 have suggested that local monsoon, equatorial Kelvin waves, precipitation and the Wyrtki Jet all play a role in the seasonal variation of the SON BL. These factors are therefore considered in Figure 4 to examine their relative importance. Firstly, over the eastern SETIO, salinity decreases throughout the ML despite a reduction in precipitation (Figures 4a and 4d), suggesting that rainfall is not a forcing factor. Secondly, the strong equatorial easterly anomalies associated with the pIOD (Figure 4b) excite eastward-propagating upwelling Kelvin waves lifting the IL off Sumatra (Figures 3b and 4a). This is in contrast with the western SETIO, where the IL deepens (Figures 3b and 4a), resulting from a downwelling Rossby wave excited in the 5°S–15°S latitude band by the equatorial zonal easterly anomalies [Cai et al., 2009; Chowdary et al., 2009], conducive to a thickening of the BL. Finally, a weaker Wyrtki Jet in response to the same equatorial easterly anomalies (Figure 4c) reduces the eastward transport of salty water, lowering salinity in the upper layer (upper ∼50 m) east to 65°E (Figures 4a and 4c). As a results, despite a reduced precipitation in the east (red, Figure 4d), a shallower haline stratification is seen, resulting in a thinner ML (Figure 4a).

image

Figure 4. (a) Longitude-depth distribution of salinity (S, shaded) and temperature (T, contour in °C) anomaly averaged between the equator and 2°S for SON 2006. The IL (dark solid) and ML (white solid) depths are superimposed, and climatological IL (dark dashed) and ML (white dashed) depths are also shown. (b) NCEP wind anomalies (m s−1), (c) sea surface salinity (SSS, shaded) and OSCAR surface current anomalies (vectors, in m s−1), and (d) evaporation minus precipitation (E-P) anomaly (mm day−1, positive values denoting divergence) averaged over SON 2006. (e–h) Same as Figures 1a–1d, respectively, but for SON 2010.

Download figure to PowerPoint

4.2. The 2010 nIOD Event

[17] The respective role of these factors in the 2010 nIOD is generally opposite in sign to that in the 2006 pIOD (Figures 3d–3f). Off Sumatra, the IL extends to a depth twice as deep as the ML, resulting in a thicker BL. The thinning of IL and ML in the western SETIO is of comparable amplitude and as a direct consequence, a BL anomaly is almost absent there.

[18] An increase in salinity is seen off Sumatra (Figure 4e) despite an increase in local rainfall (Figure 4h). Downwelling Kelvin waves generated by strong westerly wind anomalies (Figure 4f) deepen the IL in the eastern part (Figure 4e). The westerly wind anomalies drive a stronger Wyrtki Jet (Figure 4g), carrying more salty water eastward. Thus, despite a negative E-P (Figure 4h), there is a dramatic increase of salinity in the upper layer in the east (Figure 4e). In addition, a positive E-P anomaly in western tropical Indian Ocean helps salinize local water in the upper layer (Figure 4h). As a consequence, a large positive salinity anomaly occurs in the upper layer deepening the salinity-stratified ML (Figure 4e).

5. Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method
  5. 3. Seasonal Cycle
  6. 4. Interannual Variability
  7. 5. Conclusions
  8. Acknowledgments
  9. References

[19] The unprecedented temporal and spatial coverage of temperature and salinity observations by Argo allows an observational analysis of interannual variations of the SETIO BL. Our analysis reveals a robust interannual variation of BL thickness in the mature phase (September–November) of IOD events, which is predominantly resulted from variations of the IL. The IOD-related upwelling/dowelling Kelvin waves off Sumatra lifting/lowering the local thermocline and the IL is an essential process linking the IOD with the BL. We show that off Sumatra such thermocline-dominated variability in turn controls a coherent variation of the BL with the IOD. This is because the ML off Sumatra displays only weak IOD-covariance, due to two opposing processes: during a pIOD, a tendency for a ML shoaling by a salinity decrease from an equatorial easterly-induced weakening in the Wyrtki Jet is in part offset by a ML deepening tendency due to a reduced rainfall.

[20] A thickening of the BL tends to enhance the decoupling of the ML and the thermocline, hence influence the evolution of the IOD. Our observation analysis shows that the evolution of the BL is highly congruent with the evolution of the IOD with little lead or lag. If anything, the IOD leads by one month, suggesting that BL variability is a consequence of the IOD and its covariance with the IOD will enhance positive feedbacks crucial for the IOD development. For example, during a pIOD, the erosion of the BL will enhance the associated thermocline-SST positive feedback. The results will provide a valuable benchmark for assessing performance of climate models. We note however, our results are based on limited observations, and the robustness awaits a further confirmation.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method
  5. 3. Seasonal Cycle
  6. 4. Interannual Variability
  7. 5. Conclusions
  8. Acknowledgments
  9. References

[21] The work was done while Y. Qiu was visiting CSIRO. We thank Prof. Michael J. McPhaden and Dr. Won Moo Kim for fruitful discussions. Y. Qiu, L. Li, and X. Guo are supported by the National Natural Science Foundation (grant 40806014), the Ocean Public Welfare Scientific Research Project (grant 201005033-4), and the Chinese MoST Program (grant 2009CB421205, 2011CB403502). W. Cai is supported by The Goyder Institute and the Australian Climate Change Science Programme.

[22] The Editor thanks an anonymous reviewer for assisting with the evaluation of this paper.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method
  5. 3. Seasonal Cycle
  6. 4. Interannual Variability
  7. 5. Conclusions
  8. Acknowledgments
  9. References