1. Introduction
More than 90% of excess heat of the Earth system associated with anthropogenic greenhouse gas warming is stored in the oceans, manifesting as the increase of ocean heat content (OHC) (e.g., Fox-Kemper et al. 2021; Rhein et al. 2013; Cheng et al. 2019; Trenberth et al. 2014). Although relatively small in total area, the Indian Ocean has served as a considerable heat reservoir for anthropogenic heat since the mid-twentieth century (e.g., Levitus 2000; Levitus et al. 2012; Desbruyères et al. 2017). The remarkable upper-layer warming trend in the Indian Ocean played a role in the slowdown of global-mean surface warming during the 2000s (e.g., Lee et al. 2015; Nieves et al. 2015; Liu et al. 2016; Cheng et al. 2017; Gao et al. 2018; Li et al. 2018; Rathore et al. 2020). The increase of the Indian OHC (IOHC) also has notable regional impacts, including the rapid sea level rise in low-lying coastal areas (e.g., Han et al. 2010; Nicholls and Cazenave 2010; Jyoti et al. 2019), increased occurrence and severity of marine heatwaves and coral bleaching events (e.g., Wernberg et al. 2013; Feng et al. 2015; Maneesha et al. 2015; Zinke et al. 2015; Oliver et al. 2018), and modulated behaviors of tropical cyclones (e.g., Lin et al. 2013; Maneesha et al. 2015; Albert and Bhaskaran 2020; Vidya et al. 2020; Rathore et al. 2022). In this regard, investigating the IOHC trend is helpful for understanding and predicting climate change at both regional and global levels.
While existing studies have extensively investigated the prominent decadal-to-interdecadal variability of the IOHC (e.g., Han et al. 2014; Lee et al. 2015; Beal et al. 2019; Volkov et al. 2020; Ummenhofer et al. 2017, 2020), the long-term trend of the IOHC was less appreciated. Sea level studies (e.g., Schwarzkopf and Böning 2011; Trenary and Han 2013; Zhuang et al. 2013; Srinivasu et al. 2017; Han et al. 2018; Jyoti et al. 2019) also provide useful insights into the spatial pattern and underlying mechanisms of the IOHC change, given that the thermosteric component is a major contributor to regional sea level change (Rebert et al. 1985; Köhl 2014). Multiple drivers have been raised to explain the observed IOHC variability. Some studies stressed local wind-forcing effects in causing heat redistribution over the upper Indian Ocean (e.g., Han et al. 2006, 2014; Alory et al. 2007; Schwarzkopf and Böning 2011; Li and Han 2015; Volkov et al. 2020; Mohapatra and Gnanaseelan 2021), and others underscored the key role of the Indonesian Throughflow (ITF) heat transport from the Pacific in modulating the total IOHC (Lee et al. 2015; Li et al. 2017, 2018, 2020b; Jyoti et al. 2019; McMonigal et al. 2022). In addition to wind forcing, there is a considerable forcing effect by surface heat fluxes, particularly in recent decades (e.g., Li and Han 2015; Li et al. 2018, 2020b; Zhang et al. 2018). Zhang et al. (2018) attributed the remarkable southern Indian Ocean heat content increase during 1998–2015 to both the suppressed upward heat release and the increased heat transport of the ITF. Li et al. (2018) suggested that shortwave radiation contributed to the 1965–79 increase and 1980–96 decrease of the IOHC, whereas longwave radiation played a role in driving the post-2000 IOHC increase. Moreover, interdecadal variability in surface winds and heat fluxes of the Indian Ocean may contain signatures of the Atlantic Ocean climate (e.g., Li et al. 2016; Cai et al. 2019; Wang 2019; Xie et al. 2021), which also potentially modulate the IOHC variability.
Despite emerging results, our knowledge of the long-term OHC trend (ΔOHC) over the upper Indian Ocean in response to anthropogenic forcing remains fragmental. The 0–700 m IOHC showed a persistent increasing trend since the mid-twentieth century (e.g., Levitus et al. 2009; Han et al. 2014; Cheng et al. 2017; Beal et al. 2019; Ma et al. 2019, 2020; Roxy et al. 2020) and an evident acceleration since the end of the twentieth century (Lee et al. 2015; Li et al. 2018; Ummenhofer et al. 2020). The IOHC increase was not spatially homogeneous (e.g., Alory et al. 2007; Levitus et al. 2009; Han et al. 2014; Roxy et al. 2020). The regionally enhanced OHC increase in the southern Indian Ocean (SIO) over the past several decades have been widely reported (e.g., Zhang et al. 2018; Li et al. 2018; Yang et al. 2020; Nagura and McPhaden 2021). By contrast, the tropical Indian Ocean (TIO) did not show an evident warming trend until the 2000s (Srinivasu et al. 2017; Ummenhofer et al. 2020). The weak TIO warming trend before 2000 is likely attributable to wind forcing (Han et al. 2010; Li et al. 2018; Ummenhofer et al. 2020). Roxy et al. (2020) revealed a ΔOHC pattern for 1955–2015 over the TIO with a cooling band of 20°–5°S and warming elsewhere. A similar trend pattern emerges in sea level or thermosteric sea level estimates based on ocean historical observations and model hindcasts (Han et al. 2010; Durack et al. 2014; Jyoti et al. 2019; Duan et al. 2021; Lu et al. 2022). Existing studies also mentioned considerable uncertainties surrounding these trends, as indicated by discrepancies among different datasets and models (e.g., Li et al. 2018; Duan et al. 2021; Lu et al. 2022). It remains uncertain as to whether the ΔOHC pattern can stand out from interdata uncertainties and whether climate models can faithfully simulate it.
Mechanisms shaping the ΔOHC pattern are also worthy of in-depth investigation. Whether the key processes controlling decadal-to-interdecadal variability—the ITF and local wind forcing—also operate in long-term change is difficult to fathom. In the Southern Ocean adjacent to the Indian Ocean, the ΔOHC pattern can be explained by the surface winds through heaving of the thermocline (e.g., Shi et al. 2020; Lyu et al. 2020) or the heat advection by ocean circulation (e.g., Cai et al. 2010; Liu et al. 2018). It is also interesting to explore the mechanisms shaping the ΔOHC pattern in the Indian Ocean from the perspective of heaving and spicing. For the long-term trend, the external forcing by human activity is expected to play a pivotal role. The trend patterns of sea level and sea surface temperature (SST) over the Indian Ocean are primarily shaped by processes arising from anthropogenic forcing (e.g., Du and Xie 2008; Han et al. 2010; Timmermann et al. 2010; Dong et al. 2014). Han et al. (2010) attributed the formation of the sea level trend pattern to the enhanced Walker and Hadley circulations of the Indian Ocean in response to anthropogenic greenhouse warming. Dong et al. (2014) found that more than 90% of the SST warming over the TIO can be attributed to external forcing. In the adjacent Southern Ocean, external forcing, including greenhouse gases (Liu et al. 2018; Shi et al. 2020) and ozone (Li et al. 2021; Liu et al. 2022), has been suggested to play a critical role in shaping the OHC trends by modifying surface winds.
As reviewed above, our understanding of the ΔOHC pattern of the Indian Ocean since the mid-twentieth century, in spatial characteristics, uncertainties, mechanisms, and climate model simulation, is far from complete; the roles of external forcing, natural climate variability, and specific physical processes are to be clarified. These understandings provide insights into how the Indian Ocean responds and feedbacks to the ongoing climate change. Based on a variety of datasets, this study aims to provide a more reliable description of the ΔOHC pattern in the Indian Ocean with an estimate of uncertainties associated with interdataset discrepancies. With the aid of a simplified ocean model, the effect of wind-driven adiabatic ocean redistribution is examined. Furthermore, simulations and projections of state-of-the-art climate models are analyzed to understand the role of external forcing and the possible evolution in the upcoming future. The rest of the paper is organized as follows. Section 2 outlines the datasets and methodologies. Section 3 describes the observed and simulated ΔOHC pattern in the Indian Ocean, clarifies the controlling processes, and explores the projected future changes. Section 4 provides the summary and discussions.
2. Data and methods
a. Datasets
To estimate the 0–700 m ΔOHC since the mid-twentieth century, ocean temperature data for 1958–2014 from four observational analyses, one reanalysis product, and one ocean general circulation model (OGCM) hindcast are utilized. The four observational analyses are the Institute of Atmospheric Physics (IAP) ocean temperature analysis provided by Chinese Academy of Sciences (CAS) (Cheng et al. 2017), the World Ocean Atlas (WOA) ocean temperature analysis provided by National Oceanic and Atmospheric Administration’s (NOAA’s) National Centers for Environmental Information of the United States (Levitus et al. 2012); the Enhanced Ocean Data Assimilation and Climate prediction, version 4.2.1 (EN4), ocean temperature analysis (Good et al. 2013); and the Ishii ocean temperature analysis (Ishii et al. 2017). The ocean reanalysis product is the Ocean Reanalysis System 4 (ORA-S4) (Balmaseda et al. 2013) from the European Centre for Medium-Range Weather Forecasts (ECMWF). The OGCM hindcast is the Parallel Ocean Program version 2 (POP2) forced by the Japanese Meteorological Agency 55-year-do (JRA55-do) dataset (Tsujino et al. 2018) that has full-depth output as a member of phase 2 of the Ocean Model Intercomparison Project (OMIP2) (herein, OMIP2-POP2; Griffies et al. 2016). All these datasets provide ocean temperature fields on 1° × 1° grids at standard depth levels. The average of IAP, WOA, EN4, Ishii, and ORA-S4 is taken as the “observation.”
To explore atmospheric forcing effects, we employ four atmospheric reanalysis products, including the 1.25° × 1.25° JRA55 product for 1958–2014, the 1° × 1° ORA-S4 product as a combination of the 40-year ECMWF Re-Analysis (ERA-40) of 1958–88 (Uppala et al. 2005) and the ECMWF Interim Re-Analysis (ERA-Interim) of 1989–2014 (Dee et al. 2011), the ∼2° × 1.75° NOAA Twentieth Century Reanalysis, version 2 (NOAA20CR), for 1958–2012 (Compo et al. 2011), and the 1° × 1° ECMWF Twentieth Century Re-Analysis (ERA-20C) for 1958–2010 (Poli et al. 2016). The monthly surface winds and net heat flux of these datasets are all interpolated onto 1° × 1° grids to facilitate analysis.
We also analyzed sea surface height (SSH) fields from the 1° × 1° sea level estimate of Frederikse et al. (2020, hereinafter F20) for 1958–2014, the eddy-resolving (0.1° × 0.1°) simulation of Laboratory of Atmospheric Sciences and Geophysical Fluid Dynamics (LASG)/IAP Climate Ocean Model, version 3 (LICOM3-0.1°) for 1958–2010 (Duan et al. 2021; Li et al. 2020a; Ding et al. 2022), IAP, WOA, ORA-S4, and OMIP2-POP2 datasets and ocean current fields of ORA-S4, OMIP2-POP2, and LICOM3-0.1°. In addition, the 1° × 1° Met Office’s Hadley Centre Sea Ice and Sea Surface Temperature (HadISST) dataset (Rayner et al. 2003) for 1958–2014 is used to calculate indices of natural climate modes.
b. CMIP6 models
To examine the simulated and projected ΔOHC by climate models, we analyze 40 models’ historical simulations for 1958–2014 and 20 models’ shared socioeconomic pathway 5–8.5 (SSP5–8.5) simulations for 2015–2100 of the Coupled Model Intercomparison Project Phase 6 (CMIP6) project (Eyring et al. 2016). The historical simulations are forced by the observed greenhouse gases, ozone, aerosols, and solar cycle. SSP5–8.5 represents an economic vision in the future with relatively optimistic human development trends but assumes an energy-intensive and fossil-fuel economy and the radiative forcing reaches 8.5 W m−2 by 2100 (O’Neill et al. 2016). Table S1 in the online supplemental material provides more information on these models. Wind stress, SST, and ocean temperature fields from all models are interpolated onto 1° × 1° grids to facilitate analysis.
c. Reduced-gravity ocean model
List of the RGOM experiments.
d. Definitions
(a) ΔOHC (109 J m−2) of 0–700 m for 1958–2014 based on the ensemble mean of the observation-based datasets including IAP, WOA, Ishii, EN4, and ORA-S4. The black lines define the Indian Ocean (5°S–26°N, 30°–105°E plus 45°–5°S, 30°–120°E). ΔOHC is computed as the difference between the periods of 2000–14 and 1958–72. Stippling indicates significant changes exceeding the 95% confidence level based on a t test. (b) The annual mean temperature of 0–700 m of TIO (15°S–26°N, 40°–100°E), SIO (45°–22°S, 30°–115°E), their difference (TIO minus SIO), and the global ocean (gray bars), shown as deviations from their 1958–72 mean values. The differences of 2000–14 minus 1958–72 (ΔT) are specified. For the temperatures of the TIO, SIO, and their difference, the thick curves and shadings denote the ensemble mean and one standard deviation range of the five datasets, respectively, while the straight lines indicate the linear trends.
Citation: Journal of Climate 36, 7; 10.1175/JCLI-D-22-0534.1
Four natural climate variability modes are also considered: the interdecadal Pacific oscillation (IPO), the Atlantic multidecadal oscillation (AMO), the Indian Ocean dipole (IOD), and El Niño–Southern Oscillation (ENSO). Following Meehl et al. (2021), we use “natural” SST variability to compute indices of these modes with the externally forced component removed. To estimate the externally forced SST, the empirical orthogonal functions are applied to the CMIP6 ensemble-mean SST for 1958–2014. The first principal component (PC1) is considered as the time series of the externally forced SST. The externally forced SST in observation is obtained by regressing the observed SST (HadISST) onto this PC1 time series. After removing these externally forced SSTs from the observed SST, the residual is considered as the “natural” SST component. Following Henley et al. (2015), the IPO index is computed as the SST anomaly difference between the equatorial Pacific (10°S–10°N, 170°E–90°W) and the northwest plus the southwest Pacific Ocean (25°–45°N, 140°E–145°W plus 50°–15°S, 150°E–160°W) from December to February (DJF) of the next year. The AMO index is calculated as the SST anomalies in the North Atlantic (0°–60°N, 0°–80°W) (Trenberth and Shea 2006). The IOD mode index (DMI) is measured by an SST anomaly difference between the western (10°S–10°N, 50°–70°E,) and eastern (10°S–0°, 90°–110°E) equatorial Indian Ocean during September–November (SON), and the Niño-3.4 index of ENSO is defined as the SST anomaly averaged over the Niño-3.4 region (5°S–5°N, 170°–120°W) during DJF.
3. Results
a. Observed ΔOHC pattern in the Indian Ocean
The ensemble-mean 0–700 m ΔOHC of observation-based datasets for 1958–2014 shows a pattern with evident meridional differences (Fig. 1a). Strong and coherent warming (positive ΔOHC) emerges in the SIO, with the largest ΔOHC concentrating at the 45°–35°S (e.g., Cai et al. 2010; Liu et al. 2018; Lyu et al. 2020), particularly along the Agulhas Return Current (e.g., Rouault et al. 2009; Wu et al. 2012). Weaker warming trends show up in the Arabian Sea and Bay of Bengal of the northern Indian Ocean. Sandwiched by the two warming regions, the tropical south Indian Ocean between 15° and 5°S shows a cooling trend (negative ΔOHC). The cooling is insignificant in most areas of this band, except for the thermocline ridge in the southwestern tropical Indian Ocean (Hermes and Reason 2008) centered at ∼10°S. Collectively, these changes constitute a heat storage pattern over the Indian Ocean with more substantial warming in the SIO (45°–22°S, 30°–115°E) than in the TIO (15°S–26°N, 40°–100°E). This pattern is in line with the sea level trend over the Indian Ocean (Han et al. 2010; Durack et al. 2014; Jyoti et al. 2019; Duan et al. 2021; Lu et al. 2022). Similarly, the Pacific and Atlantic Oceans also show more heat storage in their subtropical bands (45°–20°S and 20°–45°N) than in the tropical parts (20°S–20°N).
Albeit with substantial interannual and decadal fluctuations, the 0–700 m average temperature of the SIO shows a persistent warming trend since 1958 (Fig. 1b), with a temperature change of ΔT = 0.25° ± 0.03°C (error denoting interdataset standard deviation), larger than the global average by 78.6% ± 23.1%. The TIO exhibits feeble and insignificant warming of ΔT = 0.04° ± 0.07°C. As a result, the meridional temperature gradient between the TIO and the SIO (TIO minus SIO) has decreased by −0.20° ± 0.09°C, which accounts for −16.4% ± 7.5% of the 1958–72 average value. These results show limited sensitivity to the choice of the southern boundary of the SIO. For example, retreating to 40°S leads to ΔT = 0.21° ± 0.03°C for the SIO and a decrease of −0.17° ± 0.09°C in the TIO-minus-SIO temperature gradient.
We then examine the ΔOHC patterns in different datasets. Despite differences in detailed spatial structure and magnitude, the four observational datasets (IAP, WOA, Ishii, and EN4), one reanalysis product (ORA-S4), and one OGCM hindcast (OMIP2-POP2) reached a consensus in showing stronger ΔOHC increase in the SIO than in the TIO (Fig. 2). Among the major features, the cooling of the thermocline ridge is of the largest uncertainty. It is notable and significant in ORA-S4 (Fig. 2e) but much weaker in IAP and Ishii (Figs. 2a,c). OMIP2-POP2 can broadly capture the ΔOHC pattern over the entire Indian Ocean that resembles those in observation, although it underestimates the warming of the northern Indian Ocean and fails to replicate the warming maximum between 45° and 40°S (Fig. 2f). Note that the hindcast of OMIP2-POP2 under JRA55-do forcing represents an independent ΔOHC estimate that is not influenced by ocean data sampling. Its overall agreement with observational datasets confirms the robustness of the ΔOHC pattern.
ΔOHC (109 J m−2) of 0–700 m during 1958–2014 derived from (a) IAP, (b) WOA, (c) Ishii, (d) EN4, (e) ORA-S4, and (f) OMIP2-POP2, computed as the difference of 2000–14 minus 1958–72. Stippling indicates significant changes exceeding the 95% confidence level based on a t test.
Citation: Journal of Climate 36, 7; 10.1175/JCLI-D-22-0534.1
b. Role of wind-driven redistribution
Figure 3 shows the results of the decomposition based on IAP data. While their sum (Fig. 3a) agrees well with the original ΔOHC of IAP (Fig. 2a), the two modes dramatically differ from each other (Figs. 3b,c). The strong warming of the SIO primarily arises from heaving, while the spicing mode shows a negative ΔOHC there. Both heaving and spicing contribute to the ΔOHC in the TIO. The spicing mode is responsible for the Arabian Sea warming, while heaving drives those in the Bay of Bengal and the eastern equatorial Indian Ocean. The cooling of the thermocline ridge arises from heaving, while that of the southeastern tropical Indian Ocean is attributed to spicing.
The 0–700 m ΔOHC (109 J m−2) for 1958–2014 derived from (a) the sum of the heaving and spicing modes, (b) the heaving mode, and (c) the spicing mode, based on IAP. (d)–(f) The corresponding zonal-mean temperature change ΔT of the Indian Ocean in the upper 1500 m. Stippling indicates significant changes exceeding the 95% confidence level. The black curves denote climatological isopycnal surfaces (kg m−3).
Citation: Journal of Climate 36, 7; 10.1175/JCLI-D-22-0534.1
Further insights are gained from the depth–longitude sections of the zonal-mean (30°–120°E) temperature change (Figs. 3d–f). The heaving mode drives deep-reaching warming in the SIO between 45° and 20°S extending to at least 1000 m. The maximum warming of ΔT > 0.6°C occurs at the depths of 300–600 m, characterizing the deepening of the main thermocline. There are discernible heaving-induced cooling trends between 100 and 300 m in the 20°S–10°N band, indicating the shoaling of the tropical thermocline. Existing studies have attributed this feature to surface wind changes (Han et al. 2010, 2014; Ummenhofer et al. 2020). The spicing-induced cooling in the SIO occurs in the subsurface layer of the 200–1200 m, reflecting the cooling and freshening trends on the isopycnal surfaces of σ = 26.6–27.4 kg m−3 (corresponding to the Subantarctic Mode Water and the upper portion of the Antarctic Intermediate Water) as reported by existing literature (Durack and Wijffels 2010; Swart et al. 2018; Lu et al. 2022). These cooling trends extend to the subsurface equatorial Indian Ocean and suppress the warming of the TIO. These comparisons overall suggest the dominance of heaving in causing stronger warming of the SIO than the TIO and highlight the essential role of adiabatic water redistribution in shaping the observed 0–700 m ΔOHC pattern. In the upper Indian Ocean, large-scale heaving primarily manifests the vertical displacement of the thermocline in response to wind forcing, which ΔULT of the RGOM can represent. With this regard, the 1.5-layer RGOM is useful in understanding the wind-driven dynamics associated with heaving mode in forming the observed ΔOHC pattern.
CTRL runs of the RGOM forced by the JRA55 and ORA-S4 winds can reasonably capture the elements of the heaving-induced ΔOHC pattern (Figs. 4a,b). The ULT increase in the SIO represents upper-ocean convergence (the increase of warm water volume) and depression of the thermocline, while the ULT decline in the TIO indicates upper-ocean divergence (the decrease of warm water volume) and shoaling of the thermocline. The two CTRL runs also replicate the evolution of the TIO–SIO gradient (Fig. 4c). Their temporal correlation coefficients with the heaving-induced OHC of the TIO-minus-SIO are 0.67 and 0.60, respectively. The similarity between the ΔULT in CTRL runs and the heaving-induced ΔOHC pattern suggests the essential role of wind-driven convergence/divergence in shaping the observed ΔOHC pattern. To better evaluate the ability of the RGOM in replicating observed ΔOHC, we estimate the OHC corresponding to the ULT by multiplying the regression coefficient of the main thermocline depth (depth of σ = 26.0 kg m−3 isopycnal surface) and OHC by the normalized ULT. The results (Fig. S1) show that, in addition to the spatial pattern, the RGOM can also predict the correct order of magnitude of the observed ΔOHC.
ΔULT (in m) for 1958–2014 based on the CTRL runs of the 1.5-layer RGOM forced by (a) JRA55 and (b) ORA-S4 winds (JRA55 CTRL and ORA-S4 CTRL, respectively). Stippling indicates significant changes at the 95% confidence level. (c) Annual ULT difference of the TIO minus SIO derived from JRA55 CTRL and ORA-S4 CTRL runs, along with OHC of the TIO minus SIO of the Heaving mode from IAP data. All variables are normalized by their standard deviations.
Citation: Journal of Climate 36, 7; 10.1175/JCLI-D-22-0534.1
Next, we explore the wind forcing effects in different regions. To pursue this, three sensitivity experiments (Table 1) are performed: the IND, SIO, and TIO runs. The IND run retains realistic winds in the Indian Ocean (same as CTRL) but climatological winds in the Pacific and Atlantic Oceans. It can isolate the local wind forcing effect within the Indian Ocean. The SIO run uses realistic monthly winds only in the Southern Hemisphere subtropical oceans (45°–18°S), and climatological winds are used elsewhere. In the TIO run, realistic monthly winds are retained only north of 18°S. The SIO and TIO runs measure the forcing effects of the SIO winds and TIO winds, respectively.
The JRA55 IND run produces a ΔULT pattern (Fig. 5a) close to that of JRA55 CTRL (Fig. 4a), confirming the dominance of local wind forcing within the Indian Ocean. This also indicates that the ITF change dictated by the Pacific winds plays a minor role. The JRA55 SIO run suggests that the SIO winds can cause both the positive ΔULT in the SIO and the negative ΔULT in the TIO (Fig. 5b). This implies a remote forcing effect of SIO winds on the TIO. Since the 1.5-layer RGOM is volume conservative, the wind-driven convergence in the SIO is compensated by the divergence of other regions. In our case, this compensation is provided by the TIO, as the ΔULT outside the Indian Ocean is negligible in the IND and SIO runs. As suggested by the JRA55 TIO run, the TIO winds can also drive negative ΔULT in the TIO but have weak impacts on the SIO (Fig. 5c). In the TIO run, the divergence in the TIO is not compensated by the convergence of the SIO but by that of the Pacific (figures not shown).
The ΔULT (m) derived from (a) IND run, (b) SIO run, and (c) TIO run forced by JRA55 winds for 1958–2014. Black lines denote the area with time-varying wind forcing. Stippling indicates significant changes at the 95% confidence level. (d) The ΔULT for 1958–2014 averaged over the TIO and SIO regions derived from the CTRL, IND, SIO, and TIO runs forced by JRA55 winds, with the error bars indicating the 95% confidence level intervals. (e)–(h) As in (a)–(d), but for the RGOM runs forced by ORA-S4 winds.
Citation: Journal of Climate 36, 7; 10.1175/JCLI-D-22-0534.1
Further quantification confirms that the SIO winds can wholly explain the ΔULT in the SIO plus a portion (∼30%) of that in the TIO (Fig. 5d). The TIO winds explain ∼70% of the TIO ΔULT but have a negligible effect on the SIO. These results imply that the SIO winds are more important than the TIO winds in shaping the ΔULT (also ΔOHC) pattern of the entire Indian Ocean. Note that artificial wind shears generated near the 18°S owing to the abrupt shift in wind fields in the TIO and SIO runs may impact the ΔULT distribution in the Indian Ocean. To examine this effect on our results, we repeated the TIO and SIO runs using wind forcing with 5° transition bands (22°–18°S for the TIO run and 18°–14°S for the SIO run) installed on the boundaries between realistic and climatological winds to damp possible artificial shears. The results of these runs (Fig. S2) are quite similar to those shown in Fig. 5. There are still some limits to the experiment design. For example, our experiments are relatively linear compared with the complex OGCM, and the nonlinear processes such as eddy are not simulated enough; the SIO and TIO runs also include the effects of wind forcing in the Atlantic and Pacific oceans, except for the Indian Ocean. Therefore, these RGOM experiments only provide a crude assessment for the relative importance of the SIO and TIO winds. We also repeated these experiments using ORA-S4 winds (Figs. 5e–h), and the results are consistent with those forced by JRA55 winds, suggesting the robustness of the conclusions regarding the choice of winds.
The warm-water redistribution between the SIO and TIO is mainly regulated by the shallow-layer meridional overturning circulation (SMOC) consisting of two overturning cells (Lee 2004). These two cells have a shared downwelling branch in the SIO. The overturning between the SIO downwelling zone and the upwelling zone in the tropical south Indian Ocean (mainly the thermocline ridge) is the “southern cell,” while that linking the SIO downwelling to the upwelling zones in the north Indian Ocean (mainly the western Arabian Sea) is dubbed the “cross-equatorial cell” (Lee 2004; Schott et al. 2002, 2009). In climatology, the SMOC transports warm surface water from the TIO to the SIO and carries cold thermocline water in the opposite direction. A strengthening of the SMOC can therefore lead to warm water convergence in the SIO and divergence in the TIO—the pattern seen in heaving mode and RGOM runs (Figs. 3–5). Existing studies have shown that changes in the SMOC strength can cause pronounced upper-ocean temperature variations on decadal-to-interdecadal time scales through heat redistribution (Lee 2004; Trenary and Han 2008; Meng et al. 2020). Whether the SMOC shows a significant multidecadal trend in strength and contributes to the ΔOHC pattern is still unknown.
(a) Climatological wind stress (arrows; Pa) and WSC (color shading; Pa m−1) of 1958–2014 derived from JRA55. The blue rectangles denote the Arabian Sea (3°–26°N, 46°–68°E), thermocline ridge (15°–5°S, 45°–100°E), and SIO regions adopted to compute the “wind proxies” of the SMOC. (b) Climatological SSH (m) of 1958–2014 derived from ORA-S4. The black rectangles denote the Arabian Sea, thermocline ridge, and SIO regions adopted to compute the “SSH proxies” of the SMOC. The (c) boreal summer (JJA) and (d) annual mean climatological overturning streamfunctions (Sv, 1 Sv ≡ 106 m3 s−1) in the Indian Ocean for 1958–2014 derived from ORA-S4. The positive and negative values indicate anticlockwise and clockwise overturning circulations. The black rectangle in (c) denotes the region of 0°–20°N, 0–400 m. Black rectangles in (d) indicate the regions of 40°–30°S, 0–400 m and 20°–5°S, 0–400 m. These regions are adopted to compute the “streamfunction proxies” of the SMOC.
Citation: Journal of Climate 36, 7; 10.1175/JCLI-D-22-0534.1
Figure 7 shows the percentage changes of the intensity proxies for the SMOC relative to the 1958–72 baseline, based on four surface wind datasets (JRA55, ORA-S4, NOAA20CR, and ERA20C), three ocean current products (ORA-S4, LICOM3-0.1°, and OMIP2-POP2), and six sea level estimates (F20, IAP, WOA, ORA-S4, LICOM3-0.1°, and OMIP2-POP2). The positive changes of all the three proxies have exceeded the interdataset standard deviation, suggesting the overall strengthening of the SMOC. The enhanced SMOC drives the migration of warm water from the TIO to the SIO and favors the formation of the observed ΔOHC pattern. Quantitatively, wind, streamfunction, and SSH proxies have increased by 15.5% ± 9.8%, 10.3% ± 2.0%, and 3.2% ± 2.1% in the Arabian Sea upwelling branch, 21.6% ± 6.5%, 11.9% ± 5.1%, and 5.7% ± 2.5% in the thermocline ridge upwelling branch, and 13.3% ± 5.1%, 15.0% ± 10.6%, and 31.9% ± 11.2% in the SIO downwelling branch, respectively (see also Table 2 for estimates based on individual datasets). These uncertainties are calculated as the interdataset standard deviations. Among them, the streamfunction proxy in the SIO downwelling branch is subjected to the largest uncertainty regarding the interdata discrepancies, which accounts for ∼70% of the ensemble-mean percent change. This mainly stems from difficulties in simulating the complicated structure of the SMOC by OGCMs and the sink and source associated with interbasin exchanges, that is, the ITF and the Agulhas leakage (e.g., Meng et al. 2020; Han and Huang 2020; Zhu et al. 2021).
(a) Percentage changes of wind, streamfunction, and SSH proxies of the SMOC for 1958–2014. Wind proxies are calculated using the JRA55, ORA-S4, NOAA20CR, and ERA20C; streamfunction proxies are calculated using the ORA-S4, LICOM3-0.1°, and OMIP2-POP2; SSH proxies are calculated using the F20, IAP, WOA, ORA-S4, LICOM3-0.1°, and OMIP2-POP2. For all proxies, a positive percent indicates the strengthening of SMOC. Error bars indicate the interdata uncertainties, calculated as standard deviations of different datasets. Note that the changes in ERA20C and LICOM3-0.1° are defined as the differences of 2000–10 minus 1958–72, while those in NOAA20CR are the differences of 2000–12 minus 1958–72. All percentages are calculated relative to the 1958–2014 baseline.
Citation: Journal of Climate 36, 7; 10.1175/JCLI-D-22-0534.1
Percentage changes in the SMOC proxies during 1958–2014 derived from observation-based datasets and OGCM hindcasts. The percentage is based on proxy value divided by its climatological mean. To obtain the regional SSH, the global-mean sea level time series has been removed. Note that the percentage values of SSH proxies are calculated by dividing the ORA-S4’s climatology as the reference surfaces of SSH datasets are different. An asterisk indicates insignificance at the 90% confidence level based on a t test. Note that changes from NOAA20CR and LICOM3-0.1° are for the 1958–2012 and 1958–2010, respectively.
To validate the reliability of these SMOC proxies, we also quantify the SMOC intensity based on wind-driven meridional Ekman transport (Lee 2004) and annual-mean meridional streamfunction (Meng et al. 2020). The changes of meridional Ekman transport from four surface wind datasets (JRA55, ORA-S4, NOAA20CR, and ERA20C) indicate the SMOC strengthening, with stronger downwelling in the SIO south of ∼12°S and upwelling in the north Indian Ocean between ∼2° and 10°N and tropical south Indian Ocean between ∼10° and 2°S during 2000–14 than during 1958–72 (Fig. S3). The annual-mean meridional streamfunction calculated following Meng et al. (2020) suggest an overall intensification of the southern cell, although the change is insignificant at 90% confidence level; meanwhile, there is no robust trend in the cross-equatorial cell (Table S2), as indicated by the large spread of three ocean current products (ORA-S4, LICOM3-0.1°, and OMIP2-POP2). This is in line with our RGOM experiments, which highlight the importance of the SIO wind forcing. Taken together, the southern cell is likely strengthened during 1958–2014 and acts to drive a redistribution of upper-layer warm water between the TIO and the SIO, thereby contributing to the formation of the ΔOHC pattern.
We next explore the changes in surface winds. As the forcing fields for OMIP2-POP2 and our RGOM runs, the JRA55 exhibits anticyclonic wind (positive WSC) trends in the SIO for 1958–2014, with southeasterlies between 30° and 10°S and westerlies in the midlatitude Southern Ocean (Fig. 8a). Anticyclonic winds favor the convergence of the upper-ocean warm water by driving Ekman downwelling and thereby causing the positive ΔOHC in the SIO. By contrast, cyclonic wind (negative WSC) trends dominate the tropical southern Indian Ocean, involving equatorial westerlies and southeasterlies between 20° and 10°S. By driving Ekman upwelling, the cyclonic winds cause the upper-ocean divergence and the negative ΔOHC in the thermocline ridge (Fig. 8a). Southerly alongshore wind trends emerge in the western Arabian Sea and cause enhanced coastal upwelling and negative ΔOHC there. The surface wind changes described above are the main drivers for the strengthened SMOC and play a central role in shaping the observed ΔOHC pattern. Results from four atmospheric reanalysis products (JRA55, ORA-S4, NOAA20CR, and ERA20C) are consistent in these key features (Fig. 8), suggesting the robustness of the wind trend pattern.
Changes of wind stress (arrows; Pa) and WSC (color shading; Pa m−1) derived from (a) JRA55 for 1958–2014, (b) ORA-S4 for 1958–2014, (c) NOAA20CR for 1958–2012, and (d) ERA20C for 1958–2010.
Citation: Journal of Climate 36, 7; 10.1175/JCLI-D-22-0534.1
c. Simulation and projection by CMIP6 models
Climate predictions and projections rely largely on climate models. The simulation of the ΔOHC pattern in state-of-the-art climate models is worthy of investigation, which also provides insights into the impact of anthropogenic forcing on ΔOHC. The multimodel mean (MMM) of the CMIP6 models can capture the basic features of the observed ΔOHC pattern in the Indian Ocean, with more substantial warming in the SIO than in the TIO (Fig. 9a). This suggests that anthropogenic forcing favors the generation of the observed ΔOHC pattern because natural climate variability has been precluded from MMM. The midlatitude oceans tend to store much more anthropogenic heat than the tropical oceans in climate models. There are also notable discrepancies between the simulated and observed ΔOHC patterns. Although the TIO warming in climate models is overall weak, models fail to capture the observed lack of warming in the thermocline ridge. Models have underestimated the heat storage in the subtropical SIO between 40° and 20°S, with the magnitude being about one-third of the observed value (Fig. 9b), and showed a sharp ΔOHC maximum between 50° and 40°S. This implies that heat tends to concentrate in the midlatitude Southern Ocean in climate models rather than stored more uniformly over the Southern Hemisphere oceans as in observation. These discrepancies may represent systematic biases in CMIP6 simulations or result from natural variability that modulates the observed ΔOHC but not the CMIP6 MMM ΔOHC. We discuss the two possibilities below.
(a) ΔOHC (109 J m−2) of 0–700 m for 1958–2014 from the ensemble mean of CMIP6 historical simulations. Stippling indicates significant changes at the 95% confidence level. (b) The corresponding zonal-mean ΔOHC for the Indian Ocean, compared with observational datasets. (c) Changes of wind stress (arrows; Pa) and WSC (color shading; Pa m−1) for 1958–2014 from the ensemble mean of CMIP6 historical simulations. (d) The corresponding zonal-mean ΔWSC for the Indian Ocean, compared with four reanalysis products (JRA55, ORAS4, NOAA20CR, and ERA20C). In (b) and (d), the shading denotes the one standard deviation range of different models/datasets.
Citation: Journal of Climate 36, 7; 10.1175/JCLI-D-22-0534.1
Given the essence of wind forcing, we examine the surface wind trends in CMIP6 simulations. Surface wind changes in MMM (Fig. 9c) bear some similarities to those in reanalysis products (Fig. 8), showing westerly trends at midlatitudes and southerly wind trends in the western Arabian Sea. These wind changes favor the heat convergence at the midlatitude and divergence in the western Arabian Sea, contributing to enhanced heat storage in the Southern Ocean and weaker warming of the northern Indian Ocean. However, the SIO in climate models is not covered by anticyclonic winds (positive WSCs) but instead shows cyclonic winds (negative WSCs). This is associated with the southward-shifted westerly trends in models: these westerly trends occur in a wide range of 65°–40°S in observation (Fig. 8), but they are confined south of 50°S in MMM. As a result, the anticyclonic winds and the resultant ΔOHC maximum shrink to the Southern Ocean in climate models, leaving the 40°–20°S subtropical band to cyclonic winds (negative WSCs in Fig. 9c) and weak ΔOHC. As quantified by the zonal-mean ΔWSC, the anticyclonic wind change is shifted southward by ∼8° in MMM than in reanalysis datasets (Fig. 9d). This bias also likely affects the TIO. The cyclonic winds are shifted to the subtropical SIO and cannot drive the thermocline ridge cooling as in observation. Figure 9d confirms that the climate models fail to reproduce the negative WSC trends between 18° and 5°S. This may explain the omitted thermocline ridge cooling in models. These model biases in wind changes have been reported and discussed by Duan et al. (2021). They suggested that CMIP6 and CMIP5 models have prevailingly underestimated the magnitude and meridional extent of surface atmospheric changes associated with the positive trend of Southern Annular Mode driven by greenhouse warming and Antarctic ozone depletion (e.g., Cai and Cowan 2007; Thompson et al. 2011). To summarize, the surface wind trends in models can well explain the simulated ΔOHC, including both the similarities to the observed pattern and the discrepancies, which further confirms the dominance of wind forcing in shaping the ΔOHC pattern.
The alternative possibility for the data–model discrepancy, that is, the impact of natural climate variability, is also examined. We consider the IPO, AMO, and IOD, which are the dominant natural variability modes in the tropical Pacific, Atlantic, and Indian Oceans, respectively. Figure 10 shows the OHC changes regressed onto the SST indices of these modes. These changes are one order smaller in magnitude than the total ΔOHC. In terms of spatial distribution, the ΔOHC induced by the IPO or the IOD shows a zonal dipole pattern in the TIO with insignificant trends in the SIO (Figs. 10a,c). The AMO induces negative ΔOHC in the central SIO and insignificant trends in most other areas (Fig. 10b). Therefore, natural climate variability modes can explain neither the observed ΔOHC pattern nor data–model discrepancies.
OHC changes (109 J m−2) regressed onto (a) DJF IPO, (b) annual AMO, and (c) SON DMI during 1958–2014 based on the ensemble mean of observational datasets. The change is computed as the difference of 2000–14 minus 1958–72. The IPO, AMO, and DMI indices are calculated using SST anomalies with the external forcing effect removed. Stippling indicates significant regressions at the 95% confidence level, based on an F test.
Citation: Journal of Climate 36, 7; 10.1175/JCLI-D-22-0534.1
Given the dominance of external forcing, the observed ΔOHC pattern is expected to persist into the future along with the increasing greenhouse gas concentration. In the SSP5-8.5 scenario projection of CMIP6 models, more substantial warming will occur over the entire Indian Ocean through 2100 (Fig. 11). This projected pattern greatly resembles the historical ΔOHC except for a much larger magnitude. This indicates that the observed ΔOHC pattern will persist and amplify in the upcoming future. The TIO and SIO is projected to warm up by +1.16° ± 0.25°C and +1.32° ± 0.36°C during 2015–2100, respectively (Fig. 11b). Correspondingly, the TIO-minus-SIO temperature gradient is projected to decrease by −0.16° ± 0.27°C, twofold larger than that in the historical simulation.
(a) ΔOHC (109 J m−2) of the 0–700 m Indian Ocean for 2015–2100 based on the ensemble mean of CMIP6 SSP5-8.5 projections. Stippling indicates significant changes at the 95% confidence level. (b) Annual mean temperature in the upper 700 m of the TIO, SIO (55°–22°S, 30°–115°E), and TIO minus SIO derived from CMIP6 simulations. The long-term changes for 2015–2100 are also shown. Each variable is shown as the anomaly relative to the 1958–72 baseline. The shading denotes the one standard deviation range.
Citation: Journal of Climate 36, 7; 10.1175/JCLI-D-22-0534.1
4. Summary and discussion
The upper-layer OHC changes in the Indian Ocean have widespread impacts on the regional and global climates, marine ecosystems, and human activities. The present study explores the characteristics and underlying mechanisms of the 0–700 m ΔOHC pattern over the Indian Ocean for 1958–2014 by analyzing various observation- and model-based datasets and performing RGOM experiments. The findings of our analysis are summarized below.
-
Multiple datasets have shown a robust spatial pattern of the 0–700 m ΔOHC since the 1950s, with enhanced warming in the SIO and weak to minimal warming in the TIO. The SIO warming is strong and spatially coherent, with the maximal ΔOHC at 45°–35°S, while cooling trends show up in the tropical south Indian Ocean between 15° and 5°S. The meridional temperature difference between the TIO and SIO has been reduced by 16.4% ± 7.5%.
-
Further analysis suggests the dominance of the heaving mode–adiabatic water redistribution–in causing the stronger warming of the SIO than the TIO. Experiments of a 1.5-layer RGOM demonstrate that changes in surface winds, particularly those in the SIO, give rise to a heat convergence in the upper SIO and a heat divergence in the upper TIO. These wind changes primarily include the strengthening of the midlatitude westerlies in the Southern Ocean and the subtropical anticyclone in the SIO.
-
The ΔOHC pattern and wind changes simulated by the MMM of CMIP6 models resemble the observation, suggesting the essence of anthropogenic forcing. Models underestimate the heat storage in the subtropical SIO between 40° and 20°S and fail to capture the observed cooling trend in the thermocline ridge. These data–model discrepancies are associated with the systematic model biases in surface wind trends. According to the future projection of CMIP6 models, the ΔOHC pattern will likely persist and amplify.
OHC changes induced by surface net heat flux (ΔOHCQ; 1010 J m−2) derived from (a) JRA55 for 1958–2014, (b) ORA-S4 for 1958–2014, (c) NOAA20CR for 1958–2012, and (d) ERA20C for 1958–2010. Stippling indicates significant changes at 95% confidence level.
Citation: Journal of Climate 36, 7; 10.1175/JCLI-D-22-0534.1
Using the simulations and projections by CMIP6 models, we preliminarily explore the potential influences of the ΔOHC pattern on the regional climate. We compute the partial correlations of the interannual SST anomaly with ENSO and IOD—the two most influential interannual climate modes in the Indian Ocean. In the historical period (1958–2014), significant positive partial correlation coefficients are seen in most areas of the Indian Ocean between the SST and ENSO during the mature period of the ENSO (DJF) (Fig. 13a). In the future projection (2015–2100), one can see a significant enhancement and southward expansion of the correlations in the SIO (Fig. 13b), whereas those correlations near the equatorial are reduced. The correlation between the SST and IOD shows significant positive values in most areas of the TIO with lower correlations in the SIO (Fig. 13c). In the future, positive correlations in the central SIO are amplified, and those near the equator are reduced (Fig. 13d). These results imply that owing to the rapid upper-ocean warming of the SIO, this region may more actively participate in the evolution of ENSO and IOD events. We also explore the potential influence of the ΔOHC pattern on sea level change. The projected pattern of sea level change shows more rapid rising rates in the SIO than in the TIO (figures not shown), as reported by previous studies (e.g., Slangen et al. 2014; Perrette et al. 2013; Wang et al. 2021).
Partial correlation coefficients between interannual SST anomalies and Niño-3.4 index during in DJF derived from (a) CMIP6 historical simulations for 1958–2014 and (b) the difference of 2015–2100 from SSP5–8.5 simulations minus 1958–2014 from historical simulations. (c),(d) As in (a) and (b), but for partial correlation coefficients between interannual SST anomalies and DMI in SON. An 8-yr high-pass filter is applied to anomaly fields and indices to highlight interannual variability. Stippling in (a) and (c) indicates coefficients of 70% models with the same signs and exceeding the 90% confidence level based on a two-tailed Student’s t test. Stippling in (b) and (d) indicates significant changes exceeding the 90% confidence level based on a t test.
Citation: Journal of Climate 36, 7; 10.1175/JCLI-D-22-0534.1
This work explains the strong subtropical SIO warming using wind-driven heat redistribution, but whether this mechanism operates in other ocean basins is unknown. The processes underlying the enhanced subtropical ocean warming (Wu et al. 2012) are worthy of investigation, which requires the aid of more complex ocean models. Here, the external forcing is suggested as the primary driver of the observed ΔOHC in the Indian Ocean, which contains various components such as greenhouse gases, anthropogenic and volcanic aerosols, and stratospheric ozone. Their individual contribution and relative importance await investigation. The Detection and Attribution Model Intercomparison Project (DAMIP) experiments (Gillett et al. 2016) and the NCAR Community Earth System Model, version 1 (CESM1), “Single Forcing” Large Ensemble Project (Deser et al. 2020) may provide valuable insights into this problem. Our analysis focused on the upper ocean and did not explore the deep portion of the Indian Ocean that is subjected to much larger uncertainties associated with sparse historical observational sampling. The issues mentioned above are interesting themes for future study. The role of the Southern Ocean climate by poleward shifting and intensifying westerly winds in the long-term changes of subtropical ocean circulation, heat storage, and sea level has been robustly identified (e.g., Cai et al. 2010; Liu et al. 2018; Qu et al. 2019; Lyu et al. 2020; Duan et al. 2021). It is worth investigating whether the Southern Ocean climate significantly affects the ΔOHC pattern of the Indian Ocean.
Acknowledgments.
Three anonymous reviewers provided insightful comments. This research is jointly supported by the National Natural Science Foundation of China (NSFC, Grant 42176007), the Strategic Priority Research Program of Chinese Academy of Sciences under Grant XDB40000000, the Shandong Provincial Natural Science Foundation under Grant ZR2020JQ17, and the Oceanographic Data Center, IOCAS. P. Lin is supported by the NSFC under Grant 41976026 and the National Key Research and Development Program under Grant 2020YFA0608902.
Data availability statement.
Temperature data of IAP, WOA, EN4, and Ishii are obtained from http://www.ocean.iap.ac.cn, https://www.ncei.noaa.gov/products/ocean-climate-laboratory, https://www.metoffice.gov.uk/hadobs/en4/download-en4-2-1.html, and https://www.data.jma.go.jp/gmd/kaiyou/english/ohc/ohc_global_en.html. Temperature, SSH, meridional velocity, wind, and net heat flux data of ORA-S4 are downloaded from http://apdrc.soest.hawaii.edu/datadoc/ecmwf_oras4.php and https://apps.ecmwf.int/datasets/. Temperature, SSH, and meridional velocity data of OMIP2-POP2 are obtained from https://esgf-node.llnl.gov/search/cmip6/. SSH and meridional velocity data of LICOM3-0.1° are available upon request (Pengfei Lin; linpf@mail.iap.ac.cn). Wind and net heat flux data of JRA55, ERA-20C, and NOAA20CR are downloaded from https://rda.ucar.edu/datasets/ds628.1/, https://apps.ecmwf.int/datasets/, and https://www.esrl.noaa.gov/psd/data/gridded/. F20 steric sea level anomalies can be obtained from https://zenodo.org/record/3862995#.Y1ji1NpBx3j. CMIP6 datasets are available at https://esgf-node.llnl.gov/projects/cmip6/. All codes for analysis and figure creation are available upon request.
REFERENCES
Albert, J., and P. K. Bhaskaran, 2020: Ocean heat content and its role in tropical cyclogenesis for the Bay of Bengal basin. Climate Dyn., 55, 3343–3362, https://doi.org/10.1007/s00382-020-05450-9.
Alory, G., S. Wijffels, and G. Meyers, 2007: Observed temperature trends in the Indian Ocean over 1960–1999 and associated mechanisms. Geophys. Res. Lett., 34, L02606, https://doi.org/10.1029/2006GL028044.
Balmaseda, M. A., K. Mogensen, and A. T. Weaver, 2013: Evaluation of the ECMWF ocean reanalysis system ORAS4. Quart. J. Roy. Meteor. Soc., 139, 1132–1161, https://doi.org/10.1002/qj.2063.
Beal, L. M., and Coauthors, 2019: IndOOS-2: A roadmap to sustained observations of the Indian Ocean for 2020–2030. CLIVAR-4/2019, GOOS-237, 206 pp., https://doi.org/10.36071/clivar.rp.4.2019.
Bindoff, N. L., and T. J. Mcdougall, 1994: Diagnosing climate change and ocean ventilation using hydrographic data. J. Phys. Oceanogr., 24, 1137–1152, https://doi.org/10.1175/1520-0485(1994)024<1137:DCCAOV>2.0.CO;2.
Cai, W., and T. Cowan, 2007: Trends in Southern Hemisphere circulation in IPCC AR4 models over 1950–99: Ozone depletion versus greenhouse forcing. J. Climate, 20, 681–693, https://doi.org/10.1175/JCLI4028.1.
Cai, W., T. Cowan, S. Godfrey, and S. Wijffels, 2010: Simulations of processes associated with the fast warming rate of the southern midlatitude ocean. J. Climate, 23, 197–206, https://doi.org/10.1175/2009JCLI3081.1.
Cai, W., and Coauthors, 2019: Pantropical climate interactions. Science, 363, eaav4236, https://doi.org/10.1126/science.aav4236.
Cheng, L., K. E. Trenberth, J. Fasullo, T. Boyer, J. Abraham, and J. Zhu, 2017: Improved estimates of ocean heat content from 1960 to 2015. Sci. Adv., 3, e1601545, https://doi.org/10.1126/sciadv.1601545.
Cheng, L., J. Abraham, Z. Hausfather, and K. E. Trenberth, 2019: How fast are the oceans warming? Science, 363, 128–129, https://doi.org/10.1126/science.aav7619.
Clément, L., E. L. McDonagh, A. Marzocchi, and A. J. G. Nurser, 2020: Signature of ocean warming at the mixed layer base. Geophys. Res. Lett., 47, e2019GL086269, https://doi.org/10.1029/2019GL086269.
Compo, G. P., and Coauthors, 2011: The Twentieth Century Reanalysis Project. Quart. J. Roy. Meteor. Soc., 137, 1–28, https://doi.org/10.1002/qj.776.
Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553–597, https://doi.org/10.1002/qj.828.
Desbruyères, D., E. L. McDonagh, B. A. King, and V. Thierry, 2017: Global and full-depth ocean temperature trends during the early twenty-first century from Argo and repeat hydrography. J. Climate, 30, 1985–1997, https://doi.org/10.1175/JCLI-D-16-0396.1.
Deser, C., and Coauthors, 2020: Isolating the evolving contributions of anthropogenic aerosols and greenhouse gases: A new CESM1 large ensemble community resource. J. Climate, 33, 7835–7858, https://doi.org/10.1175/JCLI-D-20-0123.1.
Ding, M., and Coauthors, 2022: A century-long eddy-resolving simulation of global oceanic large- and mesoscale state. Sci. Data, 9, 691, https://doi.org/10.1038/s41597-022-01766-9.
Dong, L., T. Zhou, and B. Wu, 2014: Indian Ocean warming during 1958–2004 simulated by a climate system model and its mechanism. Climate Dyn., 42, 203–217, https://doi.org/10.1007/s00382-013-1722-z.
Du, Y., and S.-P. Xie, 2008: Role of atmospheric adjustments in the tropical Indian Ocean warming during the 20th century in climate models. Geophys. Res. Lett., 35, L08712, https://doi.org/10.1029/2008GL033631.
Duan, J., and Coauthors, 2021: Rapid sea level rise in the Southern Hemisphere subtropical oceans. J. Climate, 34, 9401–9423, https://doi.org/10.1175/JCLI-D-21-0248.1.
Durack, P. J., and S. E. Wijffels, 2010: Fifty-year trends in global ocean salinities and their relationship to broad-scale warming. J. Climate, 23, 4342–4362, https://doi.org/10.1175/2010JCLI3377.1.
Durack, P. J., S. E. Wijffels, and P. J. Gleckler, 2014: Long-term sea-level change revisited: The role of salinity. Environ. Res. Lett., 9, 114017, https://doi.org/10.1088/1748-9326/9/11/114017.
Eyring, V., S. Bony, G. A. Meehl, C. A. Senior, B. Stevens, R. J. Stouffer, and K. E. Taylor, 2016: Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization. Geosci. Model Dev., 9, 1937–1958, https://doi.org/10.5194/gmd-9-1937-2016.
Feng, M., H. H. Hendon, S. P. Xie, A. G. Marshall, A. Schiller, Y. Kosaka, N. Caputi, and A. Pearce, 2015: Decadal increase in Ningaloo Niño since the late 1990s. Geophys. Res. Lett., 42, 104–112, https://doi.org/10.1002/2014GL062509.
Fox-Kemper, B. and Coauthors, 2021: Ocean, cryosphere and sea level change. Climate Change 2021: The Physical Science Basis, V. Masson-Delmotte et al., Eds., Cambridge University Press, 1211–1362.
Frederikse, T., and Coauthors, 2020: The causes of sea-level rise since 1900. Nature, 584, 393–397, https://doi.org/10.1038/s41586-020-2591-3.
Gao, L., S. R. Rintoul, and W. Yu, 2018: Recent wind-driven change in Subantarctic Mode Water and its impact on ocean heat storage. Nat. Climate Change, 8, 58–63, https://doi.org/10.1038/s41558-017-0022-8.
Gillett, N. P., and Coauthors, 2016: The Detection and Attribution Model Intercomparison project (DAMIP v1.0) contribution to CMIP6. Geosci. Model Dev., 9, 3685–3697, https://doi.org/10.5194/gmd-9-3685-2016.
Good, S. A., M. J. Martin, and N. A. Rayner, 2013: EN4: Quality controlled ocean temperature and salinity profiles and monthly objective analyses with uncertainty estimates. J. Geophys. Res. Oceans, 118, 6704–6716, https://doi.org/10.1002/2013JC009067.
Griffies, S. M., and Coauthors, 2016: OMIP contribution to CMIP6: Experimental and diagnostic protocol for the physical component of the Ocean Model Intercomparison Project. Geosci. Model Dev., 9, 3231–3296, https://doi.org/10.5194/gmd-9-3231-2016.
Häkkinen, S., P. B. Rhines, and D. L. Worthen, 2016: Warming of the global ocean: Spatial structure and water-mass trends. J. Climate, 29, 4949–4963, https://doi.org/10.1175/JCLI-D-15-0607.1.
Han, L., and R. X. Huang, 2020: Using the Helmholtz decomposition to define the Indian ocean meridional overturning streamfunction. J. Phys. Oceanogr., 50, 679–694, https://doi.org/10.1175/JPO-D-19-0218.1.
Han, W., G. A. Meehl, and A. Hu, 2006: Interpretation of tropical thermocline cooling in the Indian and Pacific oceans during recent decades. Geophys. Res. Lett., 33, L23615, https://doi.org/10.1029/2006GL027982.
Han, W., and Coauthors, 2010: Patterns of Indian Ocean sea-level change in a warming climate. Nat. Geosci., 3, 546–550, https://doi.org/10.1038/ngeo901.
Han, W., J. Vialard, M. J. McPhaden, T. Lee, Y. Masumoto, M. Feng, and W. P. M. De Ruijter, 2014: Indian Ocean decadal variability: A review. Bull. Amer. Meteor. Soc., 95, 1679–1703, https://doi.org/10.1175/BAMS-D-13-00028.1.
Han, W., D. Stammer, G. A. Meehl, A. Hu, F. Sienz, and L. Zhang, 2018: Multi-decadal trend and decadal variability of the regional sea level over the Indian Ocean since the 1960s: Roles of climate modes and external forcing. Climate, 6, 51, https://doi.org/10.3390/cli6020051.
Henley, B. J., J. Gergis, D. J. Karoly, S. Power, J. Kennedy, and C. K. Folland, 2015: A tripole index for the Interdecadal Pacific Oscillation. Climate Dyn., 45, 3077–3090, https://doi.org/10.1007/s00382-015-2525-1.
Hermes, J. C., and C. J. C. Reason, 2008: Annual cycle of the south Indian Ocean (Seychelles-Chagos) thermocline ridge in a regional ocean model. J. Geophys. Res., 113, C04035, https://doi.org/10.1029/2007JC004363.
Huang, R. X., 2020: Heaving, Stretching and Spicing Modes: Climate Variability in the Ocean. Springer, 391 pp.
Ishii, M., Y. Fukuda, S. Hirahara, S. Yasui, T. Suzuki, and K. Sato, 2017: Accuracy of global upper ocean heat content estimation expected from present observational data sets. SOLA, 13, 163–167, https://doi.org/10.2151/sola.2017-030.
Jyoti, J., P. Swapna, R. Krishnan, and C. V. Naidu, 2019: Pacific modulation of accelerated south Indian Ocean sea level rise during the early 21st century. Climate Dyn., 53, 4413–4432, https://doi.org/10.1007/s00382-019-04795-0.
Köhl, A., 2014: Detecting processes contributing to interannual halosteric and thermosteric sea level variability. J. Climate, 27, 2417–2426, https://doi.org/10.1175/JCLI-D-13-00412.1.
Lee, S.-K., W. Park, M. O. Baringer, A. L. Gordon, B. Huber, and Y. Liu, 2015: Pacific origin of the abrupt increase in Indian Ocean heat content during the warming hiatus. Nat. Geosci., 8, 445–449, https://doi.org/10.1038/ngeo2438.
Lee, T., 2004: Decadal weakening of the shallow overturning circulation in the south Indian Ocean. Geophys. Res. Lett., 31, L18305, https://doi.org/10.1029/2004GL020884.
Levitus, S., 2000: Warming of the world ocean. Science, 287, 2225–2229, https://doi.org/10.1126/science.287.5461.2225.
Levitus, S., J. I. Antonov, T. P. Boyer, R. A. Locarnini, H. E. Garcia, and A. V. Mishonov, 2009 : Global ocean heat content 1955–2008 in light of recently revealed instrumentation problems. Geophys. Res. Lett., 36, L07608, https://doi.org/10.1029/2008GL037155.
Levitus, S., and Coauthors, 2012: World ocean heat content and thermosteric sea level change (0–2000 m), 1955–2010. Geophys. Res. Lett., 39, L10603, https://doi.org/10.1029/2012GL051106.
Li, S., W. Liu, K. Lyu, and X. Zhang, 2021: The effects of historical ozone changes on Southern Ocean heat uptake and storage. Climate Dyn., 57, 2269–2285, https://doi.org/10.1007/s00382-021-05803-y.
Li, X., S. P. Xie, S. T. Gille, and C. Yoo, 2016: Atlantic-induced pan-tropical climate change over the past three decades. Nat. Climate Change, 6, 275–279, https://doi.org/10.1038/nclimate2840.
Li, Y., and W. Han, 2015: Decadal sea level variations in the Indian Ocean investigated with HYCOM: Roles of climate modes, ocean internal variability, and stochastic wind forcing. J. Climate, 28, 9143–9165, https://doi.org/10.1175/JCLI-D-15-0252.1.
Li, Y., W. Han, and L. Zhang, 2017: Enhanced decadal warming of the southeast Indian Ocean during the recent global surface warming slowdown. Geophys. Res. Lett., 44, 9876–9884, https://doi.org/10.1002/2017GL075050.
Li, Y., W. Han, A. Hu, G. A. Meehl, and F. Wang, 2018: Multidecadal changes of the upper Indian Ocean heat content during 1965–2016. J. Climate, 31, 7863–7884, https://doi.org/10.1175/JCLI-D-18-0116.1.
Li, Y., and Coauthors, 2020a: Eddy-resolving simulation of CAS-LICOM3 for phase 2 of the ocean model intercomparison project. Adv. Atmos. Sci., 37, 1067–1080, https://doi.org/10.1007/s00376-020-0057-z.
Li, Y., W. Han, F. Wang, L. Zhang, and J. Duan, 2020b: Vertical structure of the upper–Indian Ocean thermal variability. J. Climate, 33, 7233–7253, https://doi.org/10.1175/JCLI-D-19-0851.1.
Lin, I.-I., G. J. Goni, J. A. Knaff, C. Forbes, and M. M. Ali, 2013: Ocean heat content for tropical cyclone intensity forecasting and its impact on storm surge. Nat. Hazards, 66, 1481–1500, https://doi.org/10.1007/s11069-012-0214-5.
Liu, W., S.-P. Xie, and J. Lu, 2016: Tracking ocean heat uptake during the surface warming hiatus. Nat. Commun., 7, 10926–, https://doi.org/10.1038/ncomms10926.
Liu, W., J. Lu, S. P. Xie, and A. Fedorov, 2018: Southern Ocean heat uptake, redistribution, and storage in a warming climate: The role of meridional overturning circulation. J. Climate, 31, 4727–4743, https://doi.org/10.1175/JCLI-D-17-0761.1.
Liu, W., M. I. Hegglin, R. Checa-Garcia, S. Li, N. P. Gillett, K. Lyu, X. Zhang, and N. C. Swart, 2022: Stratospheric ozone depletion and tropospheric ozone increases drive Southern Ocean interior warming. Nat. Climate Change, 12, 365–372, https://doi.org/10.1038/s41558-022-01320-w.
Lu, Y., Y. Li, J. Duan, P. Lin, and F. Wang, 2022: Multidecadal sea level rise in the southeast Indian Ocean: The role of ocean salinity change. J. Climate, 35, 1479–1496, https://doi.org/10.1175/JCLI-D-21-0288.1.
Lyu, K., X. Zhang, J. A. Church