Title: Impact of Indian Ocean Dipole on the salinity budget in the equatorial Indian Ocean
Abstract: Journal of Geophysical Research: OceansVolume 118, Issue 10 p. 4911-4923 Regular ArticleFree Access Impact of Indian Ocean Dipole on the salinity budget in the equatorial Indian Ocean Zhang Yuhong, Zhang Yuhong State Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, China University of Chinese Academy of Sciences, Beijing, ChinaSearch for more papers by this authorDu Yan, Corresponding Author Du Yan State Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, ChinaCorresponding author: Y. Du, State Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology, Chinese Academy of Sciences, 164 West Xingang Rd., Guangzhou 510301, China. ([email protected])Search for more papers by this authorZheng Shaojun, Zheng Shaojun State Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, ChinaSearch for more papers by this authorYang Yali, Yang Yali State Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, China University of Chinese Academy of Sciences, Beijing, ChinaSearch for more papers by this authorCheng Xuhua, Cheng Xuhua State Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, ChinaSearch for more papers by this author Zhang Yuhong, Zhang Yuhong State Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, China University of Chinese Academy of Sciences, Beijing, ChinaSearch for more papers by this authorDu Yan, Corresponding Author Du Yan State Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, ChinaCorresponding author: Y. Du, State Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology, Chinese Academy of Sciences, 164 West Xingang Rd., Guangzhou 510301, China. ([email protected])Search for more papers by this authorZheng Shaojun, Zheng Shaojun State Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, ChinaSearch for more papers by this authorYang Yali, Yang Yali State Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, China University of Chinese Academy of Sciences, Beijing, ChinaSearch for more papers by this authorCheng Xuhua, Cheng Xuhua State Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, ChinaSearch for more papers by this author First published: 16 September 2013 https://doi.org/10.1002/jgrc.20392Citations: 34AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Abstract [1] Based on observations and ocean reanalysis, this study analyzes the variability of salinity and its related ocean dynamics in the equatorial Indian Ocean (IO). The results show significant interannual variability of salinity associated with the Indian Ocean Dipole (IOD) mode in the boreal fall. During the positive phase of IOD (pIOD), when anomalous easterly winds prevail, westward advection along the equator strengthens in summer, while the eastward advection associated with the Yoshida-Wyrtki Jet weakens in fall. Analysis of salinity budget indicates that salinity anomalies are mainly due to advection, of which zonal component is dominant. As zonal current anomalies are symmetric with respect to the equator, the equatorward large northern IO zonal salinity gradient is more important than the current anomalies in determining the asymmetric distribution of low-salinity advection. During the mature phase of pIOD, low-salinity water is advected westward, which in turn shoals the surface mixed layer, thereby providing a favorable condition for warmer sea-surface temperature in the western equatorial IO. During the decay phase of pIOD, low-salinity water is advected across the equator to the southwestern IO. When pIOD concurs along with El Niño, the strengthened off-equatorial anticyclonic circulations, which is associated with El Niño, advect low-salinity water poleward after the decay phase. Key Points Salinity can record the significant climate variability in the EqIO IOD influences the variability of salinity and associated processes Mean zonal salinity gradients result in the asymmetry of salinity variation 1. Introduction [2] Sea-surface salinity (SSS) is an essential climate variable [Global Climate Observing System, 2004], which has been shown to play an important role in world's climate variability [Lagerloef, 2002]. Apart from river discharge, evaporation (E) and precipitation (P) are the major forcing of SSS variability over the global ocean [Yu, 2011]. The SSS in the tropical Indian Ocean (IO) shows significant spatial variability [Donguy and Meyers, 1996], featuring a dramatic change from the Arabian Sea to Bay of Bengal (BOB). High SSS in the Arabian Sea can be attributed to positive (E-P) [Kumar and Prasad, 1999], while low SSS in the BOB is due to negative (E-P) and large river runoffs [Rao and Sanil Kumar, 1991]. Abundant rainfall in the eastern equatorial IO is likely responsible for the low-salinity surface water in the region [Qu and Meyers, 2005b]. [3] Ocean dynamics contribute equally as (E-P) forcing to the SSS variability in the North Atlantic [Qu et al., 2011, 2013]. In the northern IO, to maintain long-term salinity balance, zonal water mass exchanges driven by ocean currents are important [Jensen, 2003; Rao and Sivakumar, 2003; Zhang and Du, 2012]. Northeast and Southwest Monsoon Currents in winter and summer, respectively, dominate the water mass exchanges north of 4°N [Schott et al., 1994; Jensen, 2003; Schott and McCreary, 2001; Zhang and Du, 2012]. The equatorial currents, especially the Yoshida-Wyrtki Jet [Yoshida, 1959; Wyrtki, 1973], control the zonal water transports between 2°S and 2°N [Wyrtki, 1973; Reppin et al., 1999; Yuan and Han, 2006]. In the northern part of the basin, the Southwest Monsoon Current [Murty et al., 1992; Suryanarayana et al., 1993] advects high-SSS water eastward into the BOB passing south of India and Sri Lanka during the summer monsoon season [Vinayachandran et al., 1999; Han and McCreary, 2001]. This high-SSS advection reaches its seasonal maximum in August and September. During the transition season of monsoon and in particular in October–November, the Yoshida-Wyrtki Jet advects saline and warm surface water to the east, exerting dramatic impacts on eastern IO [Wyrtki, 1973; Reppin et al., 1999]. [4] The Indian Ocean Dipole (IOD) [Saji et al., 1999] or Indian Ocean "Zonal Mode" (IOZM) [Webster et al., 1999; Clark et al., 2003], is an important climate mode on interannual time scale [Saji and Yamagata, 2003a, 2003b]. The dipole mode of sea-surface temperature (SST), with anomalously low SST off Sumatra and anomalously high SST in the western IO, corresponds well with local wind and precipitation anomalies from late summer to fall [Webster et al., 1999; Yu and Rienecker, 1999; Murtugudde et al., 2000; Saji and Yamagata, 2003a]. Anomalous SSS is also found in the tropical IO during IOD events [Thompson et al., 2006; Vinayachandran and Nanjundiah, 2009; Subrahmanyam et al., 2011; Grunseich et al., 2011]. Negative SSS anomalies appear in the equatorial IO and positive SSS anomalies are located south of the equator during the positive phase of IOD (pIOD). Southwestward current anomalies in the southern BOB, together with westward current anomalies on the equator, advect low-SSS water to the central and western IO [Thompson et al., 2006; Vinayachandran and Nanjundiah, 2009]. At the same time, intense northwestward currents off Sumatra strengthen the upwelling and westward high-SSS advection [Thompson et al., 2006]. Grunseich et al. [2011] investigated the SSS anomaly patterns during IOD and ENSO events, and aimed at designing a suitable salinity index for IOD events. [5] The IOD can occur with ENSO [Baquero-Bernal et al., 2002] or without ENSO [Behera et al., 2003]. Statistical analysis indicated that about half of the IOD events are independent of ENSO [Saji and Yamagata, 2003a; Meyers et al., 2007]. Although the relationship between IOD and ENSO is controversial, wind anomalies in the tropical IO show different patterns during those two events [Jensen, 2007; Drbohlav et al., 2007]. Wind anomalies appear near the equator during IOD, but off the equatorial IO during ENSO [Yu et al., 2005]. In addition, thermocline variability associated with the IOD is confined north of 10°S, while that associated with the ENSO is confined to the south of 10°S [Xie et al., 2002; Yu et al., 2005; Rao and Behera, 2005]. It is clear that both wind forcing and ocean dynamics are different between two types of IOD events, namely, with and without ENSO. [6] In this study, we focus on the interannual SSS variability and its associated ocean dynamics in the equatorial IO. Significant SSS variance is found to relate with the IOD, while the influence of ENSO plays a secondary role. Without a co-occurrence of El Niño, a pIOD modulates the zonal high-salinity water exchange, leading to anomalously low-salinity water advection westward along the equator, which then may accelerate the SST warming in the western IO. With a co-occurrence of El Niño, the off-equatorial low-salinity advection is strengthened. [7] The remainder of the paper proceeds as follows. Section 2 describes data and analysis methods. Section 3 presents general features of SSS along the equatorial IO and validates the model output using observations. Section 4 analyzes the mixed-layer salinity budget during pure IOD events, that is, without co-occurrence of ENSO. Section 5 discusses the influence of ENSO on the low-salinity advection and conducts a case study based on observations to confirm the salinity budget analysis of SODA composite. Section 6 is a summary. 2. Data and Methods Data [8] A recent release of Simple Ocean Data Assimilation reanalysis (SODA 2.1.6) for the period 1958–2008 is used for this study [Carton et al., 2005]. The model is based on the Parallel Ocean Program (POP) version 2.1 [Smith et al., 1992], with an original 0.25° × 0.4°, but outputs conservatively remapped onto a uniform 0.5° × 0.5° horizontal resolution and 40 vertical levels. Surface wind forcing is from the European Centre for Medium Range Forecasts (ECMWF) atmospheric reanalysis (ERA-40) [Uppala et al., 2005] for the period from 1958 to 2001 and from National Aeronautics and Space Administration (NASA) Quick Scatterometer (QuikSCAT) for the period from 2002 to 2008. Precipitation is the Global Precipitation Climatology Project (GPCP) monthly satellite-gauge merged product [Adler et al., 2003], while evaporation is from the bulk formula for the period from 1979 to present. Temperature and salinity data are from the World Ocean Database 2009 [Johnson et al., 2009]. Data used for assimilation include hydrographic profiles, ocean station data, moored temperature and salinity time series, surface temperature and salinity observations, and satellite SST data. For more details about SODA product, readers are referred to Carton and Giese [2008]. [9] The monthly climatology of temperature and salinity from Argo [Roemmich and Gilson, 2009] is used to validate the SODA product in our study area of (55°E–110°E, 15°S–10°N). We use the data set from 2004 to 2012, with a resolution of 1° × 1° and 58 levels ranging from 2.5 to 1975 m in the vertical. [10] The CPC Merged Analysis of Precipitation (CMAP) [Xie and Arkin, 1997] with a resolution of 2.5° × 2.5° and evaporation with a resolution of 1° × 1° from the Objectively Analyzed air-sea Heat Fluxes (OAFlux) [Yu and Weller, 2007] are used to calculate freshwater flux. The freshwater flux is then put on a 0.5° × 0.5° grid using linear interpolation for the period from 1979 to 2008. The monthly mean surface currents are from the Ocean Surface Currents Analyses Real time (OSCAR) data, which is available on a 1° × 1° grid spanning from 1992 to present [Bonjean and Lagerloef, 2002]. The most recent version of NOAA Extended Reconstructed SeaSurface Temperature (ERSST, v3b) analysis data with a resolution of 2° × 2° from the National Climatic Data Center of the U.S. is used to construct the dipole mode index (DMI) for the IOD, following the definition in Saji et al. [1999], and the Niño3.4 index (170°W–120°W, 5°S–5°N) for the period from 1958 to 2012. Methods [11] Following Feng et al. [1998], the salt conservation equation for the mixed layer can be expressed as (1)where S is the mixed-layer salinity, u and v are the zonal and meridional components of mixed-layer velocity, S0 is SSS, S−h is the salinity right below the mixed layer, P and E are the precipitation and evaporation rates, is the entrainment velocity, and h is the depth of the mixed layer, which is calculated from density based on a temperature criterion of 0.8°C decrease from the SST [Kara et al., 2000]. [12] To separate the interannual variability from the seasonal cycle, each variable is divided into two parts: the climatological mean seasonal cycle and interannual variability (e.g., ). In the rest of the paper, we focus our discussion on the interannual variability. By neglecting the higher order nonlinear terms, equation 1 can be rewritten as (2) [13] In this equation, the advection term contains two parts: One is due to the variability of ocean current (e.g., ), and the other is due to the variability of salinity gradient (e.g., ). 3. General Features of SSS Along the Equatorial IO and SODA Validation Mean State [14] To validate the SODA salinity, we first compare its long-term (1958–2008) mean with the Argo data and WOA09 (Figures 1a–1c). The three data sets show similar patterns in the tropical IO, with high SSS in the Arabian Sea and low SSS in the BOB and eastern IO. High SSS extends eastward along the equator and low SSS extends westward south of 10°S. On the seasonal time scale, the equatorial SSS variability from all three data sets is dominated by a semiannual oscillation. In the central equatorial IO (Figure 1d), the SSS is the lowest during March–April, due to the advection of low-SSS water from the eastern IO. The SSS increases quickly after April, when the spring Yoshida-Wyrtki Jet transports high-SSS water from the western IO [Wyrtki, 1973; Reppin et al., 1999]. The second lowest SSS appears in late August when westward equatorial surface currents prevail. During September–December, the highest SSS appears in the central equatorial IO, presumably due to the eastward advection of high-SSS water by the Yoshida-Wyrtki Jet [Wyrtki, 1973; Reppin et al., 1999]. The high SSS extends to its easternmost position along the equator during September–December (Figures 1a–1c). The seasonal cycle of SODA SSS shows a good agreement with Argo data and WOA09 for the period 2004–2008, but shows significant discrepancies for the period 1958–2008. The discrepancies are mainly due to long-term trend and decadal variability in the SODA salinity reanalysis; meanwhile, Argo and WOA09 have more scattered samplings in recent decade compared to SODA (figure not shown). The root mean square (RMS) of SODA and Argo SSS fields is higher from October to December and lower during the rest of the year (Figure 1e). The RMS of SODA SSS is about 0.15 higher than that of Argo SSS, which ranges from 0.1 to 0.3, probably due to the high resolution of SODA and data assimilation skill. SODA is quasimesoscale resolved model, with an original 0.25° × 0.4°, but outputs conservatively remapped onto a uniform 0.5° × 0.5° horizontal resolution, while Argo has samplings scattered with a reconstruction resolution of 1° × 1°. In essence, SODA reanalysis well represents the seasonal variation and mean spatial distribution of salinity. Figure 1Open in figure viewerPowerPoint (left) Annual (shaded) and September–December (contour) mean SSS in (a) SODA, (b) Argo, and (c) WOA09. (right) Comparison of SODA SSS with Argo and WOA09 SSS in the central equatorial IO (5°S–5°N, 70°–90°E), as indicated by the black box in Figure 1a. (d) The seasonal SSS values of SODA (dashed curve, averaged from 1958 to 2008; black curve, averaged from 2004 to 2008), Argo (orange curve), and WOA09 (blue curve), and (e) the seasonal RMS of SSS in SODA and Argo. Spatial Distribution of SSS Interannual Variability [15] In boreal fall (September–October), both SODA and Argo show large SSS variance near the equator, extending from the eastern to the central equatorial IO (Figures 2a and 2b). The RMS of SODA SSS shows similar structures as that from Argo in the eastern equatorial IO and east of Sri Lanka. Large discrepancies exist in the region where observations are relatively sparse (Figures 2a, 2b, and 3a). The uneven time-space distribution of the data causes differences in SSS RMS between SODA and Argo, e.g., in the central equatorial Indian Ocean. Even so, the data coverage in the equatorial IO seems to be enough to characterize the SSS pattern (Figure 3b), in particular the north shift of the salinity variance off the equator. Figure 2Open in figure viewerPowerPoint RMS of SSS, zonal velocity, and precipitation during September–October. (a) Argo SSS; (b) SODA SSS; (c) SODA zonal velocity (shaded in m s−1) and CMAP precipitation (contours in mm month−1). The rectangle in Figures 2b and 2c shows the area used in Figures 5 and 7 for averaging. Figure 3Open in figure viewerPowerPoint (a) Number of Argo temperature and salinity profiles in each 1° × 1° bin during September–December for the period of 2004–2012. (b) Monthly census of profiles during September–December in the equatorial IO (15°S–10°N, 55°E–110°E). [16] Precipitation is a major freshwater source in the open ocean, which can change ocean salinity in the surface layer, particularly in the equatorial region [Yu, 2011]. The RMS of precipitation shows a significant variability south of the equator, with its strongest signature located in the southeastern tropical IO (Figure 2c). Interestingly, there is a displacement between the precipitation and SSS variability. Previous studies suggested that advection dominates the salinity variability in the northern IO on the seasonal time scale [e.g., Rao and Sivakumar, 2003; Zhang and Du, 2012], and the advection is controlled by zonal surface currents [Zhang and Du, 2012]. Careful examination indicates that the RMS of zonal surface currents is maximum along the equator, with a symmetric pattern with respect to the equator, in contrast to the asymmetric pattern of SSS variability (Figure 2c). Then, two questions arise. Which climate processes result in the SSS variability in boreal fall? Which factors account for the asymmetry of SSS variability with respect to the equator? Interannual Variability [17] During a short period year 2004–2012, five pIOD and three negative IOD (nIOD) events occurred (Figure 4a). According to their peak times and durations, the IOD events can be classified as canonical IODs and unseasonable IODs [Du et al., 2013]. The former IODs are phase locked with the boreal fall [Saji et al., 1999], and the later peak in June–August (JJA) and quickly disappear in the following month [Du et al., 2013]. The canonical IODs can occur with or without ENSO. Hereafter, the later IODs are referred to as pure IODs, which are the focus of our study. Significant negative (positive) anomalies of Argo SSS appear in the central equatorial IO in September–December during pure pIOD (nIOD) events in 2011 and 2012 (2005 and 2009) (Figure 4b). In 2006 (2010), when pIOD (nIOD) event concurred with El Niño (La Niña), negative (positive) SSS anomalies were stronger than that of pure IOD events and persisted to the following February, which indicate the influence of ENSO. SSS anomalies were also found in JJA during unseasonable IOD events in 2007 and 2008. Meanwhile, SSS anomalies are barely seen during the 2009/2010 pure El Niño event and the 2008/2009 pure La Niña event. Figure 4Open in figure viewerPowerPoint SSS anomalies during IOD events in the central equatorial IO. (a) Dipole mode index (DMI; black curve in °C) and Nino3.4 index (orange curve in °C). The two thin horizontal lines mark one standard deviation of Nino3.4 index, and the magnitudes of DMI larger than the one standard deviation are indicated by red and blue color. ENSO and IOD events are defined as those with magnitudes larger than the one standard deviation, and the phase locked in boreal winter and late summer to fall, respectively. (b) Argo and (c) SODA SSS anomalies, superimposed with the DMI index (contours in °C, values less than the one standard deviation are omitted) in the central equatorial IO (5°S–5°N, 70°E–90°E), as indicated by the black box in Figure 1a. [18] SODA SSS has a good agreement with Argo SSS, clearly showing negative (positive) anomalies in pIOD (nIOD) events during the period from 2004 to 2008. Based on the above validation, we analyze SODA over a longer period in terms of interannual variability. Hereafter, anomalies of the SODA variables are referenced to the period from 1958 to 2008, and all filtered with a 4–84 months band-pass filter with a long-term linear trend are removed. [19] Figure 5 shows a time series of SSS and zonal surface-current anomalies, superimposed with the DMI SON and Nino3.4 NDJ Indices. Note that the two boxes shown in Figures 2b and 2c (see the bold rectangles) represent the areas where we average these anomalies. The two dashed lines in Figures 5a and 5b are one standard deviation of the two indices. Once the value goes over the dashed line, it is considered as an IOD or ENSO event. There are nine pIODs (1961, 1963, 1967, 1972, 1977, 1982, 1994, 1997, 2006) and six nIODs (1958, 1960, 1975, 1996 1998, 2005) during the period from 1958 to 2008. The bold font indicates pure IOD events to distinguish from IOD events that concurred with ENSO. This classification of IOD and ENSO events is similar to Meyers et al. [2007]. The SSS and zonal surface-current anomalies show opposite phases with the DMI SON index in the IOD years. Particularly, large negative SSS and westward surface-current anomalies along the equator tend to appear during pIOD. Figure 5Open in figure viewerPowerPoint Time series of (a) SODA SSS anomalies (black bar) in the region (74°E–88°E, 1°N–5°N) and Nino3.4 NDJ index (orange curve in 2°C) and (b) SODA zonal velocity anomalies (blue bar in m s−1) in the region (74°E–88°E, 2°S–2°N) in September–October. Superimposed with the DMI in SON (gray bar in °C). Two dashed horizontal lines in Figures 5a and 5b mark the one standard deviation of Nino3.4 index in NDJ and DMI in SON, respectively; ENSO and IOD events are defined as those with magnitudes larger than the one standard deviation. All anomalies are obtained by removing climatological mean and long-term linear trend, and then filtered the decadal variability with a 4–84 months band-pass filter. Partial Correlation [20] The partial correlation analysis can separate the influence between IOD and ENSO [Saji and Yamagata, 2003a, 2003b; Yamagata et al., 2004; Yu et al., 2005]. The significance level for partial correlation analysis is performed by the standard two-tailed Student's t test. The correlation coefficient at significance level 95% is 0.29. Figure 6 shows the partial correlation of wind-stress, surface-current, and SSS anomalies with the DMI SON (Figures 6a–6d) and Nino3.4 NDJ indices (Figures 6e–6h), in which the coefficient <0.29 are omitted. The results indicate that strong easterly wind-stress anomalies are related to the IOD. In September–October, northeasterly anomalies in the southern BOB and southeasterly anomalies along Java and Sumatra converge north of the equator, while in the western equatorial IO, the easterly anomalies diverge off the equator (Figure 6a). The converged easterly wind-stress anomalies north of the equator drive strong westward surface-current anomalies and play an important role in the low-SSS advection (Figure 6b). Meanwhile, the diverged wind-stress anomalies in the western IO force off-equatorial surface-current anomalies, which favor the low-SSS advection off the equator. The SSS anomalies associated with the IOD show a negative area north of the equator and a positive area off Java and Sumatra (Figure 6b), which is similar to previous results [Thompson et al., 2006; Grunseich et al., 2011]. In the west coast of Java and Sumatra, strong southeasterly wind anomalies and northwestward surface-current anomalies provide a favorable condition for upwelling [Vinayachandran and Nanjundiah, 2009]. The upwelling then brings high-salinity water from the thermocline to the surface, inducing positive SSS anomalies in the region. The shoaling thermocline and upwelling in turn contribute to the development of IOD [Behera et al., 1999; Vinayachandran et al., 1999]. The negative SSS anomalies north of the equator match the westward surface-current anomalies, reflecting strong SSS advection associated with the IOD. In November–December, when the IOD decays, the equatorial wind-stress anomalies decrease (Figure 6c), and the surface-current anomalies weaken in the central and eastern IO. In the western IO, westward surface-current anomalies remain south of equator and keep advecting low-SSS water to the region (Figure 6d). Figure 6Open in figure viewerPowerPoint Partial correlation of surface wind stress (vectors in N m−2), SSS (shaded), and currents (vectors in m s−1) with (a–d) DMI SON or (e–h) Nino3.4 NDJ index, with the influence of ENSO (Nino3.4 NDJ index) and IOD removed in Figures 6a–6d or Figures 6e–6h, respectively. Values higher than 0.29 are statistically significant at 95% level using a standard two-tailed Student's t test, and that <0.29 are omitted. (left) September–October; (right) November–December. [21] Compared to the processes associated with the IOD, the wind-stress, surface-current, and SSS anomalies during ENSO are not obvious in September–October (Figures 6e and 6f). Even during the mature phase of ENSO in November–December, wind-stress anomalies only appear in the eastern equatorial IO (Figure 6g), driving weak westward surface-current anomalies (Figure 6h). Meanwhile, SSS anomalies are weak and coincident with the westward current anomalies (Figure 6h). This result further indicates that the IOD rather than the ENSO dominates the SSS variability in the equatorial IO. For this reason, we mainly analyze the SSS anomaly and its related ocean dynamics during the pure IODs to distinguish the influence of IOD from ENSO in the next section. 4. Salinity Budget During Pure IOD Events Composite Analysis [22] We conduct a composite analysis to investigate the mixed-layer salinity variability and its associated processes during the pure IODs. Anomalies during each selected four pure pIOD years (1961, 1967, 1977, and 1994) and three pure nIOD years (1958, 1960, and 2005) together with the each following years are used to conduct the composite. During pIOD events, negative salinity anomalies start to appear in July, strengthen in the following months, and peak in October, which is out of phase with the seasonal cycle of salinity (Figure 7). The largest salinity anomalies exceed the amplitude of its seasonal variability (Figure 7a). Meanwhile, the westward surface-current anomalies along the equator occur during the developing phase of pIOD and peak in October (Figure 7b). These westward surface-current anomalies, being of similar strength as their monthly mean value, strengthen the westward equatorial currents in July–September and weaken the eastward Yoshida-Wyrtki Jet in October–December. Associated with these westward surface-current anomalies are negative salinity anomalies during the developing and mature phases of pIOD. This result indicates that the IOD is an important process for strengthening the westward advection of low-salinity water along the equator, which is against the zonal high-salinity water exchanges in boreal fall. Figure 7Open in figure viewerPowerPoint Composites of pIOD (red curve) and nIOD (blue curve): (a) SSS anomalies and (b) zonal velocity anomalies (m s−1) averaged in (74°E–88°E, 1°N–5°N) and (74°E–88°E, 2°S–2°N), respectively. Dashed curve is the climatological monthly salinity with the annual mean removed in Figure 7a and the zonal velocities in Figure 7b. [23] The conditions during nIOD are opposite to those during pIOD, with weaker amplitude and shorter duration of salinity anomalies. The salinity and zonal-current anomalies weaken the climatological mean values during pIOD, but strengthen them during nIOD. In the following part, we focus our discussion on the pIOD conditions, since the ocean dynamic processes are more clear and robust. [24] The evolution of salinity anomalies and their related processes during pIOD are shown in Figure 8. Negative salinity anomalies, together with converged westward surface-current anomalies, appear north of the equator during the developing phase of pIOD (July–August) (Figure 8a). Meanwhile, strong positive salinity anomalies occur along with the we