1. Introduction
Salinity influences seawater density, thereby influencing oceanic dynamics and thermodynamics through its impact on vertical stratification (and therefore vertical mixing) and density gradients (and therefore currents). The importance of salinity has long been acknowledged at high latitudes, for instance, because of its impact on deep water formation rates and the thermohaline circulation (e.g., Bryan 1986; Keigwin et al. 1991; Jungclaus et al. 2006; Smith and Gregory 2009; Jackson and Wood 2018; Wang et al. 2018; Haskins et al. 2020). Its potential importance for tropical climate, where temperature strongly controls the stratification, was first raised in the western Pacific, where a strong near-surface salinity stratification coincides with a deep isothermal layer (IL). The salinity-stratified layer between the surface mixed layer (ML) and the top of the thermocline was named the “barrier layer” (BL) by Lukas and Lindstrom (1991). Thick BLs act as an insulating salinity-stratified layer between the warm ML and cold thermocline waters, inhibiting turbulent entrainment cooling (Vialard and Delecluse 1998a). Salinity-sustained temperature inversions can even occur within the BL, leading to a rather unusual warming of the surface layer by entrainment (Vialard and Delecluse 1998a; de Boyer Montégut et al. 2007; Foltz and McPhaden 2009). Thick BLs are also associated with a shallow ML, inducing a strong current response to equatorial wind forcing (Vialard and Delecluse 1998b; Vialard et al. 2002). As a result of these processes, salinity BLs are believed to enhance air–sea coupling and affect climate (Ando and McPhaden 1997; Han et al. 2001; Maes et al. 2005; Masson et al. 2005; Cai et al. 2009; Foltz and McPhaden 2009; Seo et al. 2009; Wang et al. 2011; Balaguru et al. 2012; Drushka et al. 2014; Qiu et al. 2019). For instance, coupled general circulation model experiments have demonstrated that the suppression of the BL prior to an El Niño onset could reduce its peak amplitude (Maes et al. 2004).
One region where the effect of ocean salinity on climate has been studied a lot is the northern and equatorial Indian Ocean. This region indeed receives abundant monsoonal rainfall and river runoff, which induce a strong near-surface salinity stratification (e.g., Schott and McCreary 2001). Shenoi et al. (2002), for instance, hypothesized that the Bay of Bengal (BoB) strong salinity stratification inhibits entrainment of cold water, hence maintaining high sea surface temperature (SST) and climatological rainfall. Three studies later addressed the potential effect of salinity stratification on the regional climate in coupled models, with contrasted results. Seo et al. (2009) found that the effect of salinity stratification on regional climate was mostly visible during the winter monsoon. Masson et al. (2005) and Krishnamohan et al. (2019) found a relatively small effect of salinity stratification on both the summer and winter continental rainfall. Krishnamohan et al. (2019) explained this weak effect of salinity on SST by a compensation between the salinity-induced warming through the BL insulating effect, and cooling due to the trapping of the upward surface heat fluxes over a thinner ML. The potential effect of the north Indian BL on regional climate is thus still an open topic. But thick BLs have other important implications. They can promote stronger tropical cyclones by inhibiting the surface cooling they induce (e.g., Balaguru et al. 2012; Neetu et al. 2019) and inhibit the upward mixing of subsurface nutrients, keeping the productivity low in the BoB (Prasanna Kumar et al. 2002). Understanding the processes that control the barrier layer thickness (BLT) distribution in the north Indian Ocean (NIO; 5°S–30°N, 40°–100°E) is therefore important, due to its potential importance for regional climate, oceanic productivity, and high-impact events such as tropical cyclones.
Observations indicate that BLs thicker than 30 m (and up to 60 m) are present in the NIO during boreal winter (Fig. 1a; Thadathil et al. 2007; Kumari et al. 2018), mainly in the BoB (5°–25°N, 77.5°–95°E; green box on Fig. 1c), southeastern Arabian Sea (SEAS; 5°–18°N, 65°–77°E; orange box), and equatorial eastern Indian Ocean (EEIO; 5°S–4.5°N, 80°–100°E; black box). The BL first appears in the eastern BoB in boreal summer and is thickest and covers the entire northern bay in winter, associated with freshwater forcing and its horizontal redistribution by the ocean circulation, including Ekman transport (Thadathil et al. 2007; Kumari et al. 2018; Chaitanya et al. 2021). The BL peaks in winter in the SEAS, due to the two processes. First, the mixed layer depth (MLD) becomes shallow due to a surface inflow of low-salinity BoB water by the monsoonal currents (the East Indian Coastal and Northeast Monsoon Currents; e.g., Schott et al. 1994; McCreary et al. 1996; Shetye et al. 1996; Han and McCreary 2001; Jensen 2001; Durand et al. 2009). Second, the westward radiation of the downwelling Rossby wave from the Indian coast deepens the isothermal layer depth (ILD), leading to a thick BL during winter (Thadathil et al. 2008; Agarwal et al. 2012). The EEIO is thickest in November, also due to surface freshwater inputs through rain and advection, and the ILD deepening due to equatorial Kelvin waves and the zonal convergence of surface currents (Masson et al. 2002; Qu and Meyers 2005; Agarwal et al. 2012).
Observed and average CMIP6 November–February (NDJF) barrier layer thickness (BLT), isothermal layer depth (ILD), and mixed layer depth (MLD) climatologies. NDJF (a)–(c) BLT (m), (d)–(f) ILD (m), and (g)–(i) MLD (m) in the (left) observation (ISAS-15 analysis based on Argo data), (center) CMIP6 multimodel mean (MMM), and (right) bias (MMM minus observations). The green, orange, and black boxes in (c) denote the Bay of Bengal (BoB; 5°–25°N, 77.5°–95°E), southeastern Arabian Sea (SEAS; 5°–18°N, 65°–77°E), and equatorial eastern Indian Ocean (EEIO; 5°S–4.5°N, 80°–100°E), respectively.
Citation: Journal of Climate 37, 1; 10.1175/JCLI-D-23-0366.1
Coupled climate models of the Coupled Model Intercomparison Project (CMIP; Meehl et al. 1997, 2000, 2004, 2007; Eyring et al. 2016) database are precious tools not only for climate change projections but also to study the BL formation mechanism and its effect on the climate system. Reeves Eyre et al. (2019) examined three climate models and found that the BL in CMIP models is too thin in tropical regions and too thick at higher latitudes, and its insulating effect is clearest in the tropics. Breugem et al. (2008) documented a spurious (i.e., not observed) BL in the southeastern equatorial Atlantic in CMIP models, contributing to a warm SST bias there. Wei et al. (2021) documented that the thin BLT bias in the western Pacific can influence the development of El Niño in their climate model. There have been only a handful of studies of climate models BLT biases in the NIO (Parekh et al. 2016; Di Sante et al. 2019; Rahaman et al. 2020), documenting a shallow bias in the BoB, EEIO, and SEAS. However, none of these studies provided an in-depth analysis of the mechanisms that control these biases or documented biases in the latest CMIP database (CMIP6; Eyring et al. 2016). In particular, those studies did not attempt to link the BL biases to other common biases in CMIP models.
Most CMIP models since CMIP3 indeed suffer from common mean state biases. Those biases include a cold/dry bias in the eastern Indian Ocean and warm/wet bias in the western Indian Ocean, and an equatorial easterly wind bias that leads to an unrealistic upward thermocline tilt toward the eastern Indian Ocean (e.g., Cai and Cowan 2013; Li et al. 2015). Such biases over the equatorial Indian Ocean are tightly coupled through the Bjerknes positive feedback loop (Cai and Cowan 2013). Figure 2 demonstrates that this is still the case in CMIP6: there is a cold and dry bias in the BoB and a warm and wet bias in the western equatorial Indian Ocean (Figs. 2c,f), leading to an easterly wind bias along the equator (Fig. 2i). This easterly wind bias induces easterly current anomalies and a thermocline deepening in the west and shallowing in the east, which reinforce the original SST bias through zonal advection and thermocline feedback (Cai and Cowan 2013; Annamalai et al. 2017). The cold bias in the northern Arabian Sea (Fig. 2c) was already present in CMIP3 and CMIP5, and linked to overestimated latent heat uptake due to too cold and dry air blowing from the continent during the winter monsoon (e.g., Levine et al. 2013; Marathayil et al. 2013; Sandeep and Ajayamohan 2014; Fathrio et al. 2017): this is also the case in CMIP6 (not shown).
Observed and average CMIP6 NDJF sea surface temperature (SST), precipitation rate, and wind stress climatologies. NDJF (a)–(c) SST (ISAS-15 analysis based on Argo data; °C), (d)–(f) precipitation rate (GPCP data, mm day−1), and (g)–(i) wind stress (computed by 10-m wind from ERA5 reanalysis product; N m−2; vectors) and wind speed (computed by 10-m wind from ERA5 reanalysis product; m s−1; shadings) in the (left) observations, (center) CMIP6 MMM, and (right) bias.
Citation: Journal of Climate 37, 1; 10.1175/JCLI-D-23-0366.1
This study aims to examine the ability of CMIP6 models to simulate BL in the NIO, to link those biases with other previously described CMIP biases in the Indian Ocean, to investigate the physical mechanisms that explain those BLT biases, and to explore whether BLT biases contribute to climatological SST biases. We restrict our analysis to boreal winter season (November–February), the season when the NIO observed BLT is thickest (and also the season for which BLT biases are largest, as we will demonstrate). The rest of this paper is organized as follows. Section 2 briefly describes the CMIP simulations and observations used in this study. Section 3 provides an overall estimation of the NIO simulated BL in CMIP6. The mechanisms of the winter BLT biases and their potential contribution to the SST bias are, respectively, discussed in sections 4 and 5. A summary and discussion are provided in section 6.
2. Data and methodology
a. Observations and model datasets
We use CMIP6 historical runs, which will be more comparable with recent observations than preindustrial control runs. Table 1 shows a list of 25 CMIP6 models used in this study, selected on the basis of the availability of necessary variables when initiating this study. We use the first ensemble member for each model. Further information on each model is available online at https://esgf-node.llnl.gov/projects/cmip6/. We computed the CMIP ocean temperature, salinity, surface currents, surface winds, wind stress, precipitation, and evaporation climatologies based on the 2002–14 average, for which we have observations. The climatology is not sensitive to the length of period (Fig. S1 in the online supplemental material). All CMIP data are interpolated onto a common regular grid (0.5° × 0.5° grid). Throughout the paper, we refer to the multimodel mean of the 25 CMIP6 models that we consider as the MMM.
List of 25 CMIP6 coupled atmosphere–ocean climate models.
To analyze BLT biases in CMIP6 models, we use temperature and salinity monthly gridded fields provided by the In Situ Analysis System (ISAS-15). ISAS-15 is an optimal interpolation product designed for the synthesis of the Argo global dataset (Gaillard et al. 2016; Kolodziejczyk et al. 2021). Monthly fields are provided at a 0.5° × 0.5° horizontal grid resolution at the equator, with reduced latitudinal resolution toward the poles, and a vertical resolution of 2 m near the surface, 5 m at 100-m depth, and 10–20 m below. We have compared the BLT computed from ISAS-15 with the other observation-based climatology (de Boyer Montégut et al. 2007, hereafter DBM07). The spatial distribution of the NIO BLT in ISAS-15 is similar to that in DBM07, but it is thicker in the EEIO region (Fig. S2). The DBM07 climatology product is based on 1961–2008 in situ data. We preferred to use the ISAS-15 dataset, which is also based on in situ data but includes 2002–14, and hence contains many more Argo profiles.
Surface ocean currents were derived from the 1/12° resolution Global Ocean Physical Reanalysis (GLORYS12) based on the current real-time global forecasting Copernicus-Marine Environment Monitoring Service system (https://doi.org/10.48670/moi-00021), which assimilates altimetry data, hence ensuring the accuracy of the surface velocity geostrophic component. SST and sea surface salinity (SSS) were derived from the ISAS-15 (surface layer of temperature/salinity profiles). The 10-m horizontal wind components at 0.25° resolution were obtained from the latest European Centre for Medium-Range Weather Forecasts reanalysis product (ERA5; Hersbach et al. 2020), and they were also used to compute wind stress components. Evaporation at 1° resolution was obtained from the Woods Hole Oceanographic Institute OAFlux project (Yu and Weller 2007). We use precipitation data from the Global Precipitation Climatology Project (GPCP; Adler et al. 2003). All those observational datasets were obtained over 2002–14, all bilinearly interpolated to the common grid used for CMIP6 models. Throughout the article, “bias” refers to the model minus observations of 2002–14 climatology (in general shown for November to February, i.e., boreal winter).
b. Calculation of BLT
The BLT is defined as the difference between the ILD and the MLD (i.e., BLT = ILD − MLD). Following DBM07, the ILD is defined as the depth at which temperature falls below Tr − ΔT, where Tr is the temperature at the reference depth (10 m) and ΔT = 0.2°C. The MLD is defined as the depth where the density is equal to the reference depth density plus a density change equivalent to a 0.2°C cooling [Δσ = σ(Tr − 0.2, Sr, Pr) − σ(Tr, Sr, Pr)], where Sr is the salinity and Pr is the pressure at the reference layer. DBM07 did show that it was necessary to compute MLD or BLT on individual profiles and then average spatially or temporally, due to the nonlinear nature of this computation. Here, we hence estimate ILD/MLD/BLT from monthly mean ocean variables and then obtain the climatology (rather than computing those variables from the temperature/salinity climatologies).
c. Analysis methods
For convenience, we denote November–February as NDJF and October–December as OND. The equatorial zonal wind stress index is defined as the normalized OND zonal wind stress bias averaged over the central equatorial Indian Ocean (1°S–1°N, 70°–90°E; black box as shown in Fig. 6a). Note that we shifted the sign of this index so that a positive value designates an easterly bias. The SEAS MLD index is the normalized NDJF MLD bias averaged over the SEAS (orange box as shown in Fig. 1c). The buoyancy related to freshwater flux Bw is computed as Bw = gβ(P − E)S0, where β is the coefficient of saline expansion, g is the gravitational constant, P is the precipitation, E is the evaporation, and S0 is the surface salinity (Gill 1982). All terms were evaluated using the monthly data. Positive (negative) haline buoyancy flux here corresponds to a buoyancy gain (loss) by the ocean.
3. Basin-scale climatological BLT biases
a. Errors in NDJF climatological BLT
As discussed in the introduction, the thickest NIO BLT occurs in boreal winter (NDJF; Fig. 3) with the thickest BLT in the BoB, SEAS, and EEIO regions (Fig. 1a). The BL abruptly strengthens at the beginning of the winter monsoon in November, is thickest in February (∼17 m on average, Fig. 3), and abruptly decreases in March. Individual CMIP6 models (orange lines) and their MMM generally display a reasonable seasonal cycle of the NIO BLT (Fig. 3). It implies that the current climate models can capture the main physical processes associated with BL formation, and its seasonality. However, most models have a climatological BLT that is thinner than in observations (64%–88% of the models depending on the month; Fig. 3). The NDJF season has both the thickest observed BLT and the largest MMM bias (over 80% of the models underestimate the BLT during NDJF; see Fig. 3). In the rest of the paper, we will thus focus on the NDJF period.
Seasonal variations of BLT averaged in the north Indian Ocean (NIO; 5°S–30°N, 40°–100°E). Blue (cyan) bars denote the BLT in observations (CMIP6 MMM). Light orange bars denote the BLT in each CMIP6 model. The percentage of models that have a shallower than the observed average NIO BLT is indicated above each month (80% for the annual-mean climatology).
Citation: Journal of Climate 37, 1; 10.1175/JCLI-D-23-0366.1
Let us now investigate if CMIP6 models reproduce the climatological BLT spatial pattern in the NIO. Figure 1 compares the CMIP6 MMM spatial patterns of the climatological NDJF BLT, ILD, and MLD with that from observations. The CMIP6 MMM has maxima in the same three regions as observations (the EEIO, BoB, and SEAS), but with a clear shallow BLT bias in each region (Fig. 1c). Note that this shallow BLT bias in the northern BoB is robust: it appears relative to both the ISAS-15 and DBM07 climatologies (Fig. S3). On the other hand, the shallow BLT bias in the EEIO and SEAS is still present relative to DBM07, but much weaker than in ISAS-15 (Fig. S3).
b. Skill of models in simulating NDJF climatological BLT
Here, we use the skill score of Eq. (1) (Taylor 2001) to estimate the CMIP6 performance in reproducing the NDJF climatological-mean observed BLT distributions in the NIO (Fig. 1). We recall that those skill scores evaluate both the realism of the spatial pattern and amplitude of NDJF climatological BLT. The BLT skill scores in CMIP6 models range from 0.13 (bad) to 0.84 (quite good; see ordinate range in Fig. 1), revealing a strong diversity among models in simulating the climatological NDJF BLT. While the CMIP6 MMM is generally associated with deeper BLTs in the EEIO, BoB, and SEAS as in observations, the shallow biases in those regions result in a relatively mediocre skill score of 0.46. We will now discuss how the spatial resolution or ability to reproduce the climatological precipitation and surface circulation explains this diversity in CMIP6 skill scores.
We will first investigate the ILD and MLD respective influences on the NDJF BLT climatology skill. The diversity of BLT skill is primarily determined by that of the ILD (correlation coefficient r = 0.67, p < 0.05; Fig. 4a). The MLD contributes less to the BLT skill (r = 0.33, p = 0.1; Fig. 4b). As mentioned earlier, precipitation and the surface ocean circulation both play a crucial role in the NIO BL formation. Figures 4c and 4d display the relationship between the BLT skill, and those for precipitation and surface currents. The BLT skill is significantly related to that of surface currents (r = 0.45, p < 0.05), but the correlation with precipitation is relatively weak (r = 0.37, p = 0.06). The above results indicate that the ocean dynamical processes that control the thermocline position (ILD) and the surface circulation control more the BLT spatial pattern and amplitude than rainfall, which corroborate earlier findings about the importance of ocean circulation for the BL (Cronin and McPhaden 2002; Saha et al. 2021).
Controls of the NIO NDJF BLT pattern. Scatterplot between the skill scores [Eq. (1); Taylor 2001] of NDJF BLT and the simultaneous (a) ILD, (b) MLD, (c) precipitation rate, and (d) surface ocean currents in the NIO for CMIP6 models. Scatterplot between the skill scores of NDJF BLT averaged over the NIO and the (e) horizontal and (f) vertical resolutions of CMIP6 models. The black lines indicate the least-squares fit, with the associated correlation r and its confidence level p displayed on each panel.
Citation: Journal of Climate 37, 1; 10.1175/JCLI-D-23-0366.1
We would also expect higher-resolution models to better resolve the BLT. We define horizontal resolution as the square root of number of points in the zonal times number of points in the meridional directions, and vertical resolution as the number of layers between the surface and 80 m (the typical ILD; see Table 1). The BLT skill score exhibits a significant correlation with horizontal resolution (r = 0.46, p < 0.05; Fig. 4e), indicating that a higher horizontal resolution leads to a better representation of NDJF BLT climatology. Additional analyses indicate that the skill improvement at high horizontal resolution is due to a reduced shallow BLT bias in the BoB and SEAS (not shown). This improvement is linked to the better simulation of surface ocean currents with the increase of horizontal resolution (the correlation coefficient between skill score and resolution is 0.62, p < 0.05). In contrast, there is no relation between vertical resolution and the climatological BLT representation, probably because a lot of models have similar vertical resolutions (Fig. 4f).
In summary, the CMIP6 model skill in simulating NDJF BLT climatology can be significantly improved by increasing the horizontal resolution and achieving a more accurate simulation of the top of thermocline and surface ocean circulation. The BLT skill appears to have less dependence on the representation of climatological precipitation in CMIP6 models.
4. Mechanisms of the CMIP6 model biases in each region
a. Overall introduction
The previous section established qualitative links between the quality of the winter BLT climatology representation and that of surface currents, or spatial resolution. In this section, we will investigate more specifically the mechanisms that control the BLT biases in each of the three regions (EEIO, BoB, and SEAS) where there is a thick observed climatological BLT and large bias.
We first quantify the respective roles of the ILD and MLD in controlling the BLT biases in each region. We will both discuss the MMM bias and the intermodel diversity. In the EEIO, the shallow MMM BLT bias is due to a shallow MMM ILD bias (Figs. 5a,d; the MMM MLD bias is close to zero there). The diversity in the EEIO BLT bias is also controlled by that of the ILD (Fig. 5a), with little influence of MLD. The ILD hence both controls the BLT bias MMM and its intermodel diversity. In the SEAS, the shallow MMM BLT bias is largely due to a deep MLD bias (Figs. 5c,f), and the MLD bias also controls the intermodel BLT bias diversity (Figs. 5c,f). For the BoB, the MMM shallow BLT bias is due to the MMM deep MLD bias (Fig. 5e). However, the ILD and MLD both contribute almost equally to the intermodel diversity in the bias (regression coefficient of 0.49 and −0.51; Figs. 5b,e). In the following subsections, we will investigate the mechanisms that control the BLT biases in each region.
MLD/ILD controls of the CMIP6 BLT bias in key regions. Scatterplot of the NDJF BLT bias (m) with simultaneous (top) ILD (m) and (bottom) MLD biases (m) in the (a),(d) EEIO, (b),(e) BoB, and (c),(f) SEAS for CMIP6 models. The regression coefficient of the ILD bias to that of BLT bias is indicated in (a)–(c) and that of the −MLD bias is indicated in (d)–(f). Those two regression coefficients sum to 1 by construction and quantify the respective contributions of the ILD and MLD biases to the BLT bias diversity. The black solid lines indicate the least-squares fit, and the slope and its p value are indicated in each panel.
Citation: Journal of Climate 37, 1; 10.1175/JCLI-D-23-0366.1
b. EEIO
Above, we showed that the ILD bias is responsible for both the MMM EEIO BLT bias and its intermodel diversity. It is thus necessary to clarify the causes of this ILD bias. Along the equator, the zonal pressure gradient associated with the slope of the thermocline depth is largely in a Sverdrup balance with the zonal wind stress forcing at low frequency (e.g., Cane and Sarachik 1981; McPhaden and Taft 1988; Schneider et al. 1995). As discussed in the introduction, most CMIP models do exhibit a strong easterly wind stress bias along the equator, resulting in an unrealistic thermocline slope tilting toward the eastern Indian Ocean (e.g., Cai and Cowan 2013; Li et al. 2015; Annamalai et al. 2017; Wang et al. 2021). Figures 6a and 6b display regression maps of the ILD and BLT to the OND equatorial zonal wind stress index defined in section 2c (easterly bias is counted positive). This analysis confirms that the zonal slope of thermocline is strongly controlled by the easterly wind bias in the equatorial Indian Ocean. From the scatterplots in Figs. 6c–e, it appears that the equatorial easterly wind bias strongly controls the NDJF ILD and BLT bias diversity (slope of −455 and −567 m Pa−1, respectively, p < 0.01), while its influence on the MLD bias is insignificant. It confirms that a strong equatorial easterly wind bias increases the EEIO shallow ILD and BLT biases in CMIP6 models.
Controls of the CMIP6 BLT bias in the EEIO. Regressions of the NDJF (a) ILD (m) and (b) BLT (m) biases onto the October–December (OND) zonal surface wind stress index (normalized by its standard deviation, counted positive westward). The dots in (a) and (b) are at a significance level of p < 0.05. The red (black) boxes in (a) and (b) denote the region of the EEIO (central equatorial Indian Ocean; 1°S–1°N, 70°–90°E). Scatterplots of the OND zonal surface wind stress (N m−2) bias over the central equatorial Indian Ocean with the (c) NDJF BLT (m), (d) ILD (m), and (e) MLD (m) biases in the EEIO. The black solid lines in (c)–(e) indicate the least-squares fit, and the slope (100 m Pa−1) and its p value are indicated in each panel.
Citation: Journal of Climate 37, 1; 10.1175/JCLI-D-23-0366.1
The same dynamics as above explain the EEIO MMM BLT bias. The CMIP6 MMM has too-weak equatorial westerlies wind (Figs. 2g,h and 7a,b), which results in an easterly equatorial wind stress bias (Figs. 2i and 7c), a flat MMM thermocline (Fig. 7e), and hence a shallow ILD and BLT MMM bias in the EEIO.
Observed and average CMIP6 NDJF zonal surface wind stress and temperature climatologies. NDJF (a)–(c) zonal surface wind stress (10−2 N m−2; counted positive westward; shading) and wind stress (N m−2; vectors), and (d)–(f) depth–longitude diagram of temperature (°C) averaged over 5°S–5°N for (left) observations, (center) CMIP6 MMM, and (right) bias. The green boxes in (a)–(c) denote the region of the central equatorial Indian Ocean. The black solid lines in (d) and (e) denote the 20°C contour. The black and blue solid lines in (f) denote the zero contour and the D20 bias between CMIP6 MMM and observation, respectively.
Citation: Journal of Climate 37, 1; 10.1175/JCLI-D-23-0366.1
c. BoB
In section 4a, we showed that the BoB BLT bias intermodel spread is jointly due to the ILD and MLD biases (Figs. 5b,e). Figure 6a indicates that the EEIO shallow ILD biases associated with the equatorial easterly wind bias propagate into the BoB as expected from the “leaky equatorial waveguide” framework proposed by McCreary et al. (1993) and Shetye (1998): the upwelling signal at the eastern equatorial boundary travels northward around the BoB as coastal Kelvin waves, with upwelling signal also radiating westward as Rossby waves from the eastern boundary to fill the BoB. As a result, increased easterly wind biases result in a shallowing of the BoB-average ILD (−368 m Pa−1, p < 0.05; Fig. 8b). Figures 6a, 6b, and 8a indicate that this influence of the ILD on the BLT translates into a more pronounced shallow BLT bias as the equatorial easterly wind bias increases. The BLT response to the easterly wind bias is actually larger than that of the ILD (cf. Figs. 6a,b, and the slopes of Figs. 8a,b: −703 m Pa−1 vs −368 m Pa−1). This is due to the influence of the equatorial easterly wind bias on MLD: a stronger equatorial easterly wind bias is associated with a deeper MLD in the BoB (although the relation is only statistically significant at the 90% level). While this may sound surprising at first, we will show below that the equatorial easterly wind bias and BoB dry bias are linked, which explains this relation.
Controls of the CMIP6 BLT bias in the BoB. Scatterplots of the OND zonal surface wind stress (N m−2; counted positive westward) bias over the central equatorial Indian Ocean with the (a) NDJF BLT (m), (b) ILD (m), and (c) MLD (m) biases in the BoB. Scatterplots of the NDJF buoyancy due to freshwater flux (N m−2 day−1; counted positive upward) bias with the NDJF (d) BLT, (e) ILD, and (f) MLD biases in the BoB. The black solid lines in (a)–(f) indicate the least-squares fit, and the slope [100 m Pa−1 for (a)–(c); m (N m−2 day−1)−1 for (d)–(f)] and its p value are indicated in each panel. Regressions of the NDJF (g) buoyancy due to freshwater flux (N m−2 day−1; counted positive upward) and (h) SSS (psu) biases onto the OND zonal surface wind stress index (normalized by its standard deviation, counted positive westward). The dots in (g) and (h) are at a significance level of p < 0.05. The green and black boxes in (g) and (h) denote the region of the BoB and central equatorial Indian Ocean, respectively.
Citation: Journal of Climate 37, 1; 10.1175/JCLI-D-23-0366.1
Previous studies have indeed underlined that the Indian Ocean equatorial easterly bias and zonal gradients of SST and rainfall are linked through the positive Bjerknes feedback in CMIP3 and CMIP5 (e.g., Cai and Cowan 2013; Annamalai et al. 2017). Figures 8g and 8h confirm this is also the case in CMIP6: a stronger equatorial easterly wind bias is associated with a stronger dry (Fig. 8g) and cold (not shown) bias over the EEIO and southern BoB. One would expect a dry and salty bias to deepen the MLD: Fig. 8f confirms that this is the case, by identifying a significant linear relation between the NDJF MLD bias and the simultaneous haline buoyancy flux Bw bias [slope of −1.9 m (N m−2 day−1)−1; p < 0.01]. This confirms that an increasing downward haline buoyancy flux will increase the deep MLD bias in the BoB.
Overall, the equatorial easterly wind bias controls the BLT bias diversity in the BoB through two mechanisms. First, it regulates the ILD bias through ILD signal that propagates from the equatorial band to the BoB through the coastal waveguide. But the intensity of the equatorial easterly wind bias also controls the dry bias over the southern BoB through the Bjerknes feedback loop. As a result, an enhanced easterly wind bias leads to a negative buoyancy flux over the BoB, and deeper MLD. These two physical processes add up to produce a clear increase of the BoB BLT shallow bias as the easterly bias increases.
Contrary to the intermodel spread, the BoB MMM BLT bias is only influenced by the MLD bias (the MMM ILD bias is close to zero; Figs. 5b,e). This can be related to the MMM dry bias (Fig. 2f) and too-salty SSS (Fig. 10c) leading to a too-deep MLD (Fig. 1i).
d. SEAS
The SEAS BLT intermodel spread is primarily due to the MLD bias (Figs. 5c,f). To determine the causes of this MLD bias, we have regressed the NDJF wind stress, surface currents, and SSS biases onto the simultaneous SEAS MLD index (defined in section 2c: normalized and counted positive downward; Figs. 9a–c). The SEAS MLD bias is associated with a southwesterly wind over the northern and western BoB (Fig. 9a). Previous studies have established that southwesterly anomalies in the BoB such as those of Fig. 9a drive an upwelling coastal Kelvin wave that propagates clockwise around India, inducing an anticlockwise circulation such as that in Fig. 9b (e.g., McCreary et al. 1996; Shetye et al. 2008; Suresh et al. 2016), hence weakening the background climatological NDJF circulation (e.g., Schott and McCreary 2001). This southwesterly wind therefore weakens the clockwise circulation around India (anomalous anticlockwise circulation in Fig. 9b), leading to the less low-salinity water into the SEAS and hence the saltier SSS (Fig. 9c). Additionally, the anticyclonic wind stress pattern in the Arabian Sea on Fig. 9a is also associated with Ekman transport from the central and northern Arabian Sea into the SEAS box (Figs. S4 and S5), bringing salty water and probably also contributing to the SSS bias, although this relation is less statistically significant than that with the salt transport by the West India Coastal Current (p ∼ 0.1 vs ∼0.01, Fig. 9 and Fig. S5). The mechanism is confirmed by the scatterplots between the NDJF MLD bias and the simultaneous horizontal salt advection bias over the SEAS [Fig. 9f, regression coefficient of 61.41 m (psu month−1)−1, p < 0.05]. Southwesterly wind biases in the BoB (Fig. 9a) lead to a reduced clockwise circulation around India (Fig. 9b), less BoB freshwater inputs, and a higher salinity (Fig. 9c), deeper MLD (Fig. 9f), and shallower BLT (Fig. 9d) in the SEAS. The intermodel diversity in SEAS BLT biases is thus controlled by biases in the winter monsoon intensity over the BoB, with an underestimated northeast monsoon leading to a too-shallow BLT in the SEAS.
Controls of the CMIP6 BLT bias in the SEAS. Regressions of the NDJF (a) surface wind stress (N m−2), (b) surface currents (m s−1), and (c) SSS biases onto the NDJF SEAS MLD index (normalized by its standard deviation, counted positive downward). The black boxes in (a)–(c) denote the region of the SEAS. The dots and red arrows in (a)–(c) are at a significance level of p < 0.05. Scatterplots of the NDJF horizontal salt advection bias (psu month−1) with the NDJF (d) BLT (m), (e) ILD (m), and (f) MLD (m) biases in the SEAS. The black solid lines in (d)–(f) indicate the least-squares fit, and the slope [m (psu month−1)−1] and its p value are indicated in each panel.
Citation: Journal of Climate 37, 1; 10.1175/JCLI-D-23-0366.1
The shallow MMM BLT bias in the SEAS is due to a deep MLD bias (Figs. 1c,i) in the SEAS. This deep MLD bias is due to too-salty SSS in the SEAS (Figs. 10c and 11c), which originates from a MMM underestimated inflow of BoB low-salinity water (positive MMM salinity advection bias of 0.15 psu month−1; Figs. 9d–f). This reduced BoB freshwater input has probable contributions from both the too-weak MMM winter clockwise circulation around India (Fig. 10f) and the too-salty BoB (Fig. 10c). One can note that the MMM wind stress does not display a southwesterly wind anomaly in the BoB as highlighted for the diversity (Fig. 9a), but that the MMM equatorial easterly wind will drive an upwelling Kelvin wave and anomalous anticlockwise circulation around India as noted by previous studies (e.g., McCreary et al. 1996). But overall, the same mechanisms associated with the input of BoB freshwater to the SEAS explain the MMM shallow BLT bias and its diversity in CMIP6.
Observed and average CMIP6 NDJF sea surface salinity (SSS) and surface currents climatologies. NDJF (a)–(c) SSS (ISAS-15 analysis based on Argo data; psu), (d)–(f) surface ocean current fields (GLORYS12 data; m s−1; vectors) and current speeds (computed by surface current velocity from GLORYS12 data; m s−1; shading) in the (left) observations, (center) CMIP6 MMM, and (right) bias.
Citation: Journal of Climate 37, 1; 10.1175/JCLI-D-23-0366.1
Observed and average CMIP6 vertical salinity bias in the SEAS. Depth–longitude diagram of NDJF salinity (ISAS-15 analysis based on Argo data; psu) along the section for (a) observations, (b) CMIP6 MMM, and (c) bias. The section (yellow line) is shown in the map inset of (a) (the red spot is the initial position: 6.5°N, 71°E).
Citation: Journal of Climate 37, 1; 10.1175/JCLI-D-23-0366.1
5. Impact on SST
In the previous section, we have discussed the mechanisms by which biases in atmospheric forcing and oceanic dynamics give rise to a too-shallow BL during NDJF in the NIO in most CMIP6 models (Fig. 3). Below, we test the hypothesis that those shallow BLT biases contribute to the cold SST bias in the BoB in CMIP6, in relation to an enhanced vertical entrainment and turbulent mixing of cold subsurface water into the ML, as hypothesized by Breugem et al. (2008) for the tropical Atlantic. It is not straightforward to compute the vertical entrainment and turbulent mixing in CMIP6, as these processes are parameterized and a reconstruction based on available monthly mean model output would be inaccurate (Breugem et al. 2008). We can however easily estimate the temperature difference between the core and the base of the ML,
Figures 12a–c display the seasonal cycle of those temperature differences in the EEIO, BoB, and SEAS. In observations, there is a strong reduction of the temperature gradient at the ML bottom in winter (blue shading), when the BLT is thick, in the BoB and SEAS, as expected. This effect is less pronounced in the EEIO. In the BoB, the strong postmonsoon temperature stratification can even sustain temperature inversions as documented in the literature (e.g., Thadathil et al. 2002; Thompson et al. 2006; Girishkumar et al. 2013; Thadathil et al. 2016). The CMIP6 MMM broadly simulates the ΔT seasonal variations but strongly overestimates it throughout the year for the EEIO and BoB and in winter and spring for the SEAS (this is also generally true for individual models; not shown). This is consistent with the underestimated BLT in most models, which results in colder temperatures below the ML, and arguably more entrainment cooling. Let us now investigate if a shallower BLT results in a larger cold SST bias.
Overestimated temperature contrast at the bottom of the mixed layer in CMIP6. Temperature difference (°C) between core and base of the mixed layer (i.e.,
Citation: Journal of Climate 37, 1; 10.1175/JCLI-D-23-0366.1
Figures 13a–c present scatterplots between the NDJF SST and BLT biases in the three regions with a thick observed BLT and strong shallow BLT bias. The linear relationship is insignificant in the EEIO (Fig. 13a). In contrast, there is a strong negative regression coefficient in the SEAS (−0.04°C m−1; p < 0.01; Fig. 13c) and a weaker one in the BoB (−0.02°C m−1; p < 0.1; Fig. 13b). This indicates that, contrary to the Breugem et al. (2008) hypothesis and common perception of the role of the BL, shallow BLT favors warmer, not colder SSTs; that is, they suppress cold SST biases in CMIP6 models and do not promote them. This suggests that the BLT effect on vertical entrainment, if any, is counterbalanced by other opposite effects.
Links between the NDJF SST and BLT/MLD biases in CMIP6. Scatterplots of the NDJF SST bias (°C) in the (left) EEIO, (center) BoB, and (right) SEAS with the simultaneous (a)–(c) BLT (m) and (d)–(f) MLD (m) biases. The black solid lines indicate the least-squares fit, and the slope (°C m−1) and its p value are indicated in each panel.
Citation: Journal of Climate 37, 1; 10.1175/JCLI-D-23-0366.1
In the previous section, we saw that deep MLD biases contribute to the shallow BLT bias diversity in the BoB and SEAS. Figures 13d–f indicate that those BoB and SEAS deep MLD biases are associated with weaker cold SST biases (p < 0.01 in both regions, no significant signal in the EEIO). This could be related to the results of Krishnamohan et al. (2019) and Akhil et al. (2023): those studies found that the salinity effect on SST in the SEAS and BoB was dominated by the ML cooling rate by surface fluxes, rather than by vertical mixing. During winter, net surface heat fluxes are negative in the BoB and SEAS in observations and CMIP6 models (not shown). The too-deep ML in CMIP6 has a higher heat capacity than in observations, resulting in a weaker winter cooling rate by negative surface heat fluxes. Therefore, in CMIP6 models, the shallow BLT bias influence on entrainment cooling does not dominate. Rather, it seems that the associated deep MLD bias in the BoB and SEAS alleviates the cold SST bias in those regions, through “diluting” the winter surface cooling over a thicker layer.
6. Summary and discussion
a. Summary
Using CMIP6 historical simulation datasets, we investigate how current state-of-the-art climate models from CMIP6 simulate the BLT climatology over the NIO. Most CMIP6 models reproduce the BLT observed seasonal cycle and spatial pattern over the NIO. Most models however display a shallow bias during NDJF season over the EEIO, BoB, and SEAS, where and when observed climatological BLT is thickest. We thus examine the causes for the MMM shallow BLT bias in those regions, as well as those of intermodel diversity of the bias.
The Fig. 14a schematic summarizes how equatorial easterly wind biases in climate models lead to the NIO shallow BLT bias. CMIP6 models simulate an equatorial easterly wind bias which is accompanied by a cold SST and dry bias in the southern BoB via the Bjerknes feedback, as previously documented in CMIP3 and CMIP5 (Cai and Cowan 2013; Annamalai et al. 2017). Such an equatorial easterly wind bias induces an upward thermocline tilt toward the EEIO, controlling the shallow MMM ILD and BLT biases, and their intermodel spread.
Controls of NIO NDJF BLT biases and their impact on other CMIP6 biases. (a) Schematic of the mechanisms of the BLT biases in CMIP6 models, linking the equatorial easterly wind bias to shallow BLT biases in the EEIO, BoB, and SEAS. The background map shows the MMM CMIP6 BLT bias. (b) Schematic of how salinity biases in CMIP6 may alleviate the other biases through the Bjerknes feedback. The background map shows the MMM CMIP6 MLD bias.
Citation: Journal of Climate 37, 1; 10.1175/JCLI-D-23-0366.1
The EEIO shallow ILD bias propagates into BoB through the coastal waveguide, hence leading to a larger BoB shallow ILD bias as the easterly wind bias strengthens. At the same time, the equatorial easterly wind bias is also linked to the BoB cold and dry bias through the Bjerknes feedback. Stronger equatorial easterlies hence lead to an underestimated BoB rainfall and too-deep MLD, which adds up to the shallow ILD bias to strengthen the shallow BLT bias. The MLD deep bias is the main contributor to the MMM BoB shallow BLT bias, while both the MLD and ILD contribute to the BoB BLT bias diversity, as detailed above.
The SEAS MMM shallow BLT bias and its intermodel spread are both dominated by the MLD deep bias. The SEAS deep MLD biases can be traced back to a salty bias that results from underestimated freshwater input from the BoB due to the underestimated clockwise winter monsoon currents around India. The intensity of the winter monsoon northeasterly winds controls the diversity of the SEAS shallow BLT bias, while the MMM shallow BLT bias is probably related to the remote effect of equatorial winds on the monsoonal currents around India.
Our results hence highlight that the Bjerknes feedback loop that links the equatorial easterly winds, and zonal SST and rainfall gradient biases in CMIP further lead to biases in the entire NIO haline structure in CMIP6.
b. Discussion
Breugem et al. (2008) proposed that the BLT systematic biases in climate models have a significant impact on their SST bias through the BL effect on the ML cooling through vertical processes, which in turn can affect atmospheric convection, and the entire tropical climate. We also find a bias in the temperature gradient at the bottom of the ML that suggests an overestimated entrainment cooling related to the shallow BLT biases in the EEIO, BoB, and SEAS. However, we do not find a consistent SST cooling in the NIO in CMIP6 simulations. On the contrary, our study suggests that the dry, saline bias in the BoB and the associated deep MLD bias lead to a diminished cooling rate in response to winter negative surface heat fluxes and thus alleviate the existing cold BoB SST bias (Fig. 14b). This is hence consistent with salinity effects alleviating, rather than strengthening, coupled model biases in the NIO. Namely, the salinity effects in the BoB are prone to reduce the cold BoB SST bias, which in turn weakens the local dry bias through the SST–rainfall relationship and attenuates the equatorial easterly wind bias through the Bjerknes feedback (Fig. 14b), as demonstrated here for CMIP6, and before for CMIP3 and CMIP5 (Cai and Cowan 2013; Annamalai et al. 2017). This study shows that the equatorial easterly wind bias plays an important role in the development of shallow NIO BLT bias in CMIP6 models. Furthermore, the potential negative feedback mechanism we propose implies that BoB SST and rainfall biases and equatorial zonal wind bias already present in climate models are diminished rather than reinforced by the BoB salinity effects. This underlines that the role of salinity in the tropical climate should not be underestimated.
Acknowledgments.
We acknowledge the Program for Climate Model Diagnosis and Intercomparison and the World Climate Research Programme (WCRP) Working Group on Coupled Modeling for their roles in making available the WCRP CMIP6 multimodel dataset. The authors thank all the reviewers, whose comments and suggestions improved the manuscript. This study was supported by the Fundamental Research Funds for the Central Universities (B230205012). S. Pang was also supported by the Postgraduate Research and Practice Innovation Program of Jiangsu Province (KYCX22_0587), the scholarship from China Scholarship Council (202206710108), and the Project on Excellent Postgraduate Dissertation of Hohai University (B230205012). S. Pang gratefully acknowledges Sorbonne University for providing a quality working environment.
Data availability statement.
The data supporting the findings of this study are openly available. The CMIP6 data used in this study are available online at https://esgf-node.llnl.gov/projects/cmip6/. The used gridded fields of temperature and salinity are from the ISAS-15 product (https://www.seanoe.org/data/00412/52367/). The observation-based climatologies of ILD/MLD/BLT are available at https://cerweb.ifremer.fr/deboyer/mld/gridded_data.php. The surface ocean currents are GLORYS12 (https://data.marine.copernicus.eu/product/GLOBAL_MULTIYEAR_PHY_001_030/description). The 10-m horizontal wind is collected from the ERA5 reanalysis product (https://www.ecmwf.int/en/forecasts/dataset/ecmwf-reanalysis-v5). Evaporation data are obtained from the OAFlux project (https://rda.ucar.edu/datasets/ds260.1/dataaccess/). The precipitation data are derived from GPCP (https://psl.noaa.gov/data/gridded/data.gpcp.html).
REFERENCES
Adler, R. F., and Coauthors, 2003: The version-2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present). J. Hydrometeor., 4, 1147–1167, https://doi.org/10.1175/1525-7541(2003)004<1147:TVGPCP>2.0.CO;2.
Agarwal, N., R. Sharma, A. Parekh, S. Basu, A. Sarkar, and V. K. Agarwal, 2012: Argo observations of barrier layer in the tropical Indian Ocean. Adv. Space Res., 50, 642–654, https://doi.org/10.1016/j.asr.2012.05.021.
Akhil, V. P., M. Lengaigne, K. S. Krishnamohan, M. G. Keerthi, and J. Vialard, 2023: Southeastern Arabian Sea salinity variability: Mechanisms and influence on surface temperature. Climate Dyn., 61, 3737–3754, https://doi.org/10.1007/s00382-023-06765-z.
Ando, K., and M. J. McPhaden, 1997: Variability of surface layer hydrography in the tropical Pacific Ocean. J. Geophys. Res., 102, 23 063–23 078, https://doi.org/10.1029/97JC01443.
Annamalai, H., B. Taguchi, J. P. McCreary, M. Nagura, and T. Miyama, 2017: Systematic errors in South Asian monsoon simulation: Importance of equatorial Indian Ocean processes. J. Climate, 30, 8159–8178, https://doi.org/10.1175/JCLI-D-16-0573.1.
Balaguru, K., P. Chang, R. Saravanan, L. R. Leung, Z. Xu, M. Li, and J.-S. Hsieh, 2012: Ocean barrier layers’ effect on tropical cyclone intensification. Proc. Natl. Acad. Sci. USA, 109, 14 343–14 347, https://doi.org/10.1073/pnas.1201364109.
Breugem, W.-P., P. Chang, C. J. Jang, J. Mignot, and W. Hazeleger, 2008: Barrier layers and tropical Atlantic SST biases in coupled GCMs. Tellus, 60A, 885–897, https://doi.org/10.1111/j.1600-0870.2008.00343.x.
Bryan, F., 1986: High-latitude salinity effects and interhemispheric thermohaline circulations. Nature, 323, 301–304, https://doi.org/10.1038/323301a0.
Cai, W., and T. Cowan, 2013: Why is the amplitude of the Indian Ocean dipole overly large in CMIP3 and CMIP5 climate models? Geophys. Res. Lett., 40, 1200–1205, https://doi.org/10.1002/grl.50208.
Cai, W., A. Pan, D. Roemmich, T. Cowan, and X. Guo, 2009: Argo profiles a rare occurrence of three consecutive positive Indian Ocean dipole events, 2006–2008. Geophys. Res. Lett., 36, L08701, https://doi.org/10.1029/2008GL037038.
Cane, M. A., and E. S. Sarachik, 1981: The response of a linear baroclinic equatorial ocean to periodic forcing. J. Mar. Res., 39, 651–693, https://elischolar.library.yale.edu/journal_of_marine_research/1564.
Chaitanya, A. V. S., J. Vialard, M. Lengaigne, F. d’Ovidio, J. Riotte, F. Papa, and R. A. James, 2021: Redistribution of riverine and rainfall freshwater by the Bay of Bengal circulation. Ocean Dyn., 71, 1113–1139, https://doi.org/10.1007/s10236-021-01486-5.
Cronin, M. F., and M. J. McPhaden, 2002: Barrier layer formation during westerly wind bursts. J. Geophys. Res., 107, 8020, https://doi.org/10.1029/2001JC001171.
de Boyer Montégut, C., J. Mignot, A. Lazar, and S. Cravatte, 2007: Control of salinity on the mixed layer depth in the world ocean: 1. General description. J. Geophys. Res., 112, C06011, https://doi.org/10.1029/2006JC003953.
Di Sante, F., E. Coppola, R. Farneti, and F. Giorgi, 2019: Indian summer monsoon as simulated by the regional Earth system model RegCM-ES: The role of local air–sea interaction. Climate Dyn., 53, 759–778, https://doi.org/10.1007/s00382-019-04612-8.
Drushka, K., J. Sprintall, and S. T. Gille, 2014: Subseasonal variations in salinity and barrier-layer thickness in the eastern equatorial Indian Ocean. J. Geophys. Res. Oceans, 119, 805–823, https://doi.org/10.1002/2013JC009422.
Durand, F., D. Shankar, F. Birol, and S. S. C. Shenoi, 2009: Spatiotemporal structure of the East India Coastal Current from satellite altimetry. J. Geophys. Res., 114, C02013, https://doi.org/10.1029/2008JC004807.
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.
Fathrio, I., S. Iizuka, A. Manda, Y.-M. Kodama, S. Ishida, Q. Moteki, H. Yamada, and Y. Tachibana, 2017: Assessment of western Indian Ocean SST bias of CMIP5 models. J. Geophys. Res. Oceans, 122, 3123–3140, https://doi.org/10.1002/2016JC012443.
Foltz, G. R., and M. J. McPhaden, 2009: Impact of barrier layer thickness on SST in the central tropical North Atlantic. J. Climate, 22, 285–299, https://doi.org/10.1175/2008JCLI2308.1.
Gaillard, F., T. Reynaud, V. Thierry, N. Kolodziejczyk, and K. von Schuckmann, 2016: In situ–based reanalysis of the global ocean temperature and salinity with ISAS: Variability of the heat content and steric height. J. Climate, 29, 1305–1323, https://doi.org/10.1175/JCLI-D-15-0028.1.
Gill, A. E., 1982: Atmosphere–Ocean Dynamics. Academic Press, 662 pp.
Girishkumar, M. S., M. Ravichandran, and M. J. McPhaden, 2013: Temperature inversions and their influence on the mixed layer heat budget during the winters of 2006–2007 and 2007–2008 in the Bay of Bengal. J. Geophys. Res. Oceans, 118, 2426–2437, https://doi.org/10.1002/jgrc.20192.
Han, W., and J. P. McCreary Jr., 2001: Modeling salinity distributions in the Indian Ocean. J. Geophys. Res., 106, 859–877, https://doi.org/10.1029/2000JC000316.
Han, W., J. P. McCreary Jr., and K. E. Kohler, 2001: Influence of precipitation minus evaporation and Bay of Bengal rivers on dynamics, thermodynamics, and mixed layer physics in the upper Indian Ocean. J. Geophys. Res., 106, 6895–6916, https://doi.org/10.1029/2000JC000403.
Haskins, R. K., K. I. C. Oliver, L. C. Jackson, R. A. Wood, and S. S. Drijfhout, 2020: Temperature domination of AMOC weakening due to freshwater hosing in two GCMs. Climate Dyn., 54, 273–286, https://doi.org/10.1007/s00382-019-04998-5.
Hersbach, H., and Coauthors, 2020: The ERA5 global reanalysis. Quart. J. Roy. Meteor. Soc., 146, 1999–2049, https://doi.org/10.1002/qj.3803.
Jackson, L. C., and R. A. Wood, 2018: Timescales of AMOC decline in response to fresh water forcing. Climate Dyn., 51, 1333–1350, https://doi.org/10.1007/s00382-017-3957-6.
Jensen, T. G., 2001: Arabian Sea and Bay of Bengal exchange of salt and tracers in an ocean model. Geophys. Res. Lett., 28, 3967–3970, https://doi.org/10.1029/2001GL013422.
Jungclaus, J. H., H. Haak, M. Esch, E. Roeckner, and J. Marotzke, 2006: Will Greenland melting halt the thermohaline circulation? Geophys. Res. Lett., 33, L17708, https://doi.org/10.1029/2006GL026815.
Keigwin, L. D., G. A. Jones, S. J. Lehman, and E. A. Boyle, 1991: Deglacial meltwater discharge, North Atlantic deep circulation, and abrupt climate change. J. Geophys. Res., 96, 16 811–16 826, https://doi.org/10.1029/91JC01624.
Kolodziejczyk, N., A. Prigent-Mazella, and F. Gaillard, 2021: ISAS temperature, salinity, dissolved oxygen gridded fields. SEANOE, accessed 2021, https://doi.org/10.17882/52367.
Kraus, E. B., and J. S. Turner, 1967: A one-dimensional model of the seasonal thermocline: II. The general theory and its consequences. Tellus, 19A, 98–106, https://doi.org/10.3402/tellusa.v19i1.9753.
Krishnamohan, K. S., and Coauthors, 2019: Is there an effect of Bay of Bengal salinity on the northern Indian Ocean climatological rainfall? Deep-Sea Res. II, 166, 19–33, https://doi.org/10.1016/j.dsr2.2019.04.003.
Kumari, A., S. P. Kumar, and A. Chakraborty, 2018: Seasonal and interannual variability in the barrier layer of the Bay of Bengal. J. Geophys. Res. Oceans, 123, 1001–1015, https://doi.org/10.1002/2017JC013213.
Levine, R. C., A. G. Turner, D. Marathayil, and G. M. Martin, 2013: The role of northern Arabian Sea surface temperature biases in CMIP5 model simulations and future projections of Indian summer monsoon rainfall. Climate Dyn., 41, 155–172, https://doi.org/10.1007/s00382-012-1656-x.
Li, G., S.-P. Xie, and Y. Du, 2015: Monsoon-induced biases of climate models over the tropical Indian Ocean. J. Climate, 28, 3058–3072, https://doi.org/10.1175/JCLI-D-14-00740.1.
Lukas, R., and E. Lindstrom, 1991: The mixed layer of the western equatorial Pacific Ocean. J. Geophys. Res., 96, 3343–3357, https://doi.org/10.1029/90JC01951.
Maes, C., J. Picaut, Y. Kuroda, and K. Ando, 2004: Characteristics of the convergence zone at the eastern edge of the Pacific warm pool. Geophys. Res. Lett., 31, L11304, https://doi.org/10.1029/2004GL019867.
Maes, C., J. Picaut, and S. Belamari, 2005: Importance of the salinity barrier layer for the buildup of El Niño. J. Climate, 18, 104–118, https://doi.org/10.1175/JCLI-3214.1.
Marathayil, D., A. G. Turner, L. C. Shaffrey, and R. C. Levine, 2013: Systematic winter sea-surface temperature biases in the northern Arabian Sea in HiGEM and the CMIP3 models. Environ. Res. Lett., 8, 014028, https://doi.org/10.1088/1748-9326/8/1/014028.
Masson, S., P. Delecluse, J.-P. Boulanger, and C. Menkes, 2002: A model study of the seasonal variability and formation mechanisms of the barrier layer in the eastern equatorial Indian Ocean. J. Geophys. Res., 107, 8017, https://doi.org/10.1029/2001JC000832.
Masson, S., and Coauthors, 2005: Impact of barrier layer on winter-spring variability of the southeastern Arabian Sea. Geophys. Res. Lett., 32, L07703, https://doi.org/10.1029/2004GL021980.
McCreary, J. P., P. K. Kundu, and R. L. Molinari, 1993: A numerical investigation of dynamics, thermodynamics and mixed-layer processes in the Indian Ocean. Prog. Oceanogr., 31, 181–244, https://doi.org/10.1016/0079-6611(93)90002-U.
McCreary, J. P., W. Han, D. Shankar, and S. R. Shetye, 1996: Dynamics of the East India Coastal Current: 2. Numerical solutions. J. Geophys. Res., 101, 13 993–14 010, https://doi.org/10.1029/96JC00560.
McPhaden, M. J., and B. A. Taft, 1988: Dynamics of seasonal and intraseasonal variability in the eastern equatorial Pacific. J. Phys. Oceanogr., 18, 1713–1732, https://doi.org/10.1175/1520-0485(1988)018<1713:DOSAIV>2.0.CO;2.
Meehl, G. A., G. J. Boer, C. Covey, M. Latif, and R. J. Stouffer, 1997: Intercomparison makes for a better climate model. Eos, Trans. Amer. Geophys. Union, 78, 445–451, https://doi.org/10.1029/97EO00276.
Meehl, G. A., G. J. Boer, C. Covey, M. Latif, and R. J. Stouffer, 2000: The Coupled Model Intercomparison Project (CMIP). Bull. Amer. Meteor. Soc., 81, 313–318, https://doi.org/10.1175/1520-0477(2000)081<0313:TCMIPC>2.3.CO;2.
Meehl, G. A., C. Covey, B. McAvaney, M. Latif, and R. J. Stouffer, 2004: Overview of the Coupled Model Intercomparison Project (CMIP). Bull. Amer. Meteor. Soc., 86, 89–53.
Meehl, G. A., C. Covey, T. Delworth, M. Latif, B. McAvaney, J. F. B. Mitchell, R. J. Stouffer, and K. E. Taylor, 2007: The WCRP CMIP3 multimodel dataset: A new era in climate change research. Bull. Amer. Meteor. Soc., 88, 1383–1394, https://doi.org/10.1175/BAMS-88-9-1383.
Neetu, S., M. Lengaigne, J. Vialard, G. Samson, S. Masson, K. S. Krishnamohan, and I. Suresh, 2019: Premonsoon/postmonsoon Bay of Bengal tropical cyclones intensity: Role of air-sea coupling and large-scale background state. Geophys. Res. Lett., 46, 2149–2157, https://doi.org/10.1029/2018GL081132.
Parekh, A., J. S. Chowdary, O. Sayantani, T. S. Fousiya, and C. Gnanaseelan, 2016: Tropical Indian Ocean surface salinity bias in climate forecasting system coupled models and the role of upper Ocean processes. Climate Dyn., 46, 2403–2422, https://doi.org/10.1007/s00382-015-2709-8.
Prasanna Kumar, S., P. M. Muraleedharan, T. G. Prasad, M. Gauns, N. Ramaiah, S. N. de Souza, S. Sardesai, and M. Madhupratap, 2002: Why is the Bay of Bengal less productive during summer monsoon compared to the Arabian Sea? Geophys. Res. Lett., 29, 2235, https://doi.org/10.1029/2002GL016013.
Qiu, Y., and Coauthors, 2019: Upper-ocean response to the super Tropical Cyclone Phailin (2013) over the freshwater region of the Bay of Bengal. J. Phys. Oceanogr., 49, 1201–1228, https://doi.org/10.1175/JPO-D-18-0228.1.
Qu, T., and G. Meyers, 2005: Seasonal variation of barrier layer in the southeastern tropical Indian Ocean. J. Geophys. Res., 110, C11003, https://doi.org/10.1029/2004JC002816.
Rahaman, H., and Coauthors, 2020: An assessment of the Indian Ocean mean state and seasonal cycle in a suite of interannual CORE-II simulations. Ocean Modell., 145, 101503, https://doi.org/10.1016/j.ocemod.2019.101503.
Reeves Eyre, J. E. J., L. Van Roekel, X. Zeng, M. A. Brunke, and J.-C. Golaz, 2019: Ocean barrier layers in the energy exascale Earth system model. Geophys. Res. Lett., 46, 8234–8243, https://doi.org/10.1029/2019GL083591.
Ren, L., and S. C. Riser, 2009: Seasonal salt budget in the northeast Pacific Ocean. J. Geophys. Res., 114, C12004, https://doi.org/10.1029/2009JC005307.
Saha, A., N. Serra, and D. Stammer, 2021: Growth and decay of northwestern tropical Atlantic barrier layers. J. Geophys. Res. Oceans, 126, e2020JC016956, https://doi.org/10.1029/2020JC016956.
Sandeep, S., and R. S. Ajayamohan, 2014: Origin of cold bias over the Arabian Sea in climate models. Sci. Rep., 4, 6403, https://doi.org/10.1038/srep06403.
Schneider, E. K., B. Huang, and J. Shukla, 1995: Ocean wave dynamics and El Niño. J. Climate, 8, 2415–2439, https://doi.org/10.1175/1520-0442(1995)008<2415:OWDAEN>2.0.CO;2.
Schott, F., and J. P. McCreary Jr., 2001: The monsoon circulation of the Indian Ocean. Prog. Oceanogr., 51 (1), 1–123, https://doi.org/10.1016/S0079-6611(01)00083-0.
Schott, F., J. Reppin, J. Fischer, and D. Quadfasel, 1994: Currents and transports of the monsoon current south of Sri Lanka. J. Geophys. Res., 99, 25 127–25 141, https://doi.org/10.1029/94JC02216.
Seo, H., S.-P. Xie, R. Murtugudde, M. Jochum, and A. J. Miller, 2009: Seasonal effects of Indian Ocean freshwater forcing in a regional coupled model. J. Climate, 22, 6577–6596, https://doi.org/10.1175/2009JCLI2990.1.
Shenoi, S. S. C., D. Shankar, and S. R. Shetye, 2002: Differences in heat budgets of the near-surface Arabian Sea and Bay of Bengal: Implications for the summer monsoon. J. Geophys. Res., 107, 3052, https://doi.org/10.1029/2000JC000679.
Shetye, S. R., 1998: West India Coastal Current and Lakshadweep high/low. Sadhana, 23, 637–651, https://doi.org/10.1007/BF02744586.
Shetye, S. R., A. D. Gouveia, D. Shankar, S. S. C. Shenoi, P. N. Vinayachandran, D. Sundar, G. S. Michael, and G. Nampoothiri, 1996: Hydrography and circulation in the western Bay of Bengal during the northeast monsoon. J. Geophys. Res., 101, 14 011–14 025, https://doi.org/10.1029/95JC03307.
Shetye, S. R., I. Suresh, D. Shankar, D. Sundar, S. Jayakumar, P. Mehra, R. G. Prabhudesai, and P. S. Pednekar, 2008: Observational evidence for remote forcing of the West India Coastal Current. J. Geophys. Res., 113, C11001, https://doi.org/10.1029/2008JC004874.
Smith, R. S., and J. M. Gregory, 2009: A study of the sensitivity of ocean overturning circulation and climate to freshwater input in different regions of the North Atlantic. Geophys. Res. Lett., 36, L15701, https://doi.org/10.1029/2009GL038607.
Stevenson, J. W., and P. P. Niiler, 1983: Upper ocean heat budget during the Hawaii-to-Tahiti shuttle experiment. J. Phys. Oceanogr., 13, 1894–1907, https://doi.org/10.1175/1520-0485(1983)013<1894:UOHBDT>2.0.CO;2.
Suresh, I., J. Vialard, T. Izumo, M. Lengaigne, W. Han, J. McCreary, and P. M. Muraleedharan, 2016: Dominant role of winds near Sri Lanka in driving seasonal sea level variations along the west coast of India. Geophys. Res. Lett., 43, 7028–7035, https://doi.org/10.1002/2016GL069976.
Taylor, K. E., 2001: Summarizing multiple aspects of model performance in a single diagram. J. Geophys. Res., 106, 7183–7192, https://doi.org/10.1029/2000JD900719.
Thadathil, P., V. V. Gopalakrishna, P. M. Muraleedharan, G. V. Reddy, N. Araligidad, and S. Shenoy, 2002: Surface layer temperature inversion in the Bay of Bengal. Deep-Sea Res. I, 49, 1801–1818, https://doi.org/10.1016/S0967-0637(02)00044-4.
Thadathil, P., P. M. Muraleedharan, R. R. Rao, Y. K. Somayajulu, G. V. Reddy, and C. Revichandran, 2007: Observed seasonal variability of barrier layer in the Bay of Bengal. J. Geophys. Res., 112, C02009, https://doi.org/10.1029/2006JC003651.
Thadathil, P., and Coauthors, 2008: Seasonal variability of the observed barrier layer in the Arabian Sea. J. Phys. Oceanogr., 38, 624–638, https://doi.org/10.1175/2007JPO3798.1.
Thadathil, P., I. Suresh, S. Gautham, S. Prasanna Kumar, M. Lengaigne, R. R. Rao, S. Neetu, and A. Hegde, 2016: Surface layer temperature inversion in the Bay of Bengal: Main characteristics and related mechanisms. J. Geophys. Res. Oceans, 121, 5682–5696, https://doi.org/10.1002/2016JC011674.
Thompson, B., C. Gnanaseelan, and P. S. Salvekar, 2006: Seasonal evolution of temperature inversions in the North Indian Ocean. Curr. Sci., 90, 697–704, https://www.jstor.org/stable/24089118.
Vialard, J., and P. Delecluse, 1998a: An OGCM study for the TOGA decade. Part I: Role of salinity in the physics of the western Pacific fresh pool. J. Phys. Oceanogr., 28, 1071–1088, https://doi.org/10.1175/1520-0485(1998)028<1071:AOSFTT>2.0.CO;2.
Vialard, J., and P. Delecluse, 1998b: An OGCM study for the TOGA decade. Part II: Barrier-layer formation and variability. J. Phys. Oceanogr., 28, 1089–1106, https://doi.org/10.1175/1520-0485(1998)028<1089:AOSFTT>2.0.CO;2.
Vialard, J., P. Delecluse, and C. Menkes, 2002: A modeling study of salinity variability and its effects in the tropical Pacific Ocean during the 1993–1999 period. J. Geophys. Res., 107, 8005, https://doi.org/10.1029/2000JC000758.
Wang, G., W. Cai, and A. Santoso, 2021: Simulated thermocline tilt over the tropical Indian Ocean and its influence on future sea surface temperature variability. Geophys. Res. Lett., 48, e2020GL091902, https://doi.org/10.1029/2020GL091902.
Wang, H., S. Legg, and R. Hallberg, 2018: The effect of Arctic freshwater pathways on North Atlantic convection and the Atlantic meridional overturning circulation. J. Climate, 31, 5165–5188, https://doi.org/10.1175/JCLI-D-17-0629.1.
Wang, X., G. Han, Y. Qi, and W. Li, 2011: Impact of barrier layer on typhoon-induced sea surface cooling. Dyn. Atmos. Oceans, 52, 367–385, https://doi.org/10.1016/j.dynatmoce.2011.05.002.
Wei, H.-H., A. C. Subramanian, K. B. Karnauskas, C. A. DeMott, M. R. Mazloff, and M. A. Balmaseda, 2021: Tropical Pacific air-sea interaction processes and biases in CESM2 and their relation to El Niño development. J. Geophys. Res. Oceans, 126, e2020JC016967, https://doi.org/10.1029/2020JC016967.
Yu, L., and R. A. Weller, 2007: Objectively analyzed air–sea heat fluxes for the global ice-free oceans (1981–2005). Bull. Amer. Meteor. Soc., 88, 527–540, https://doi.org/10.1175/BAMS-88-4-527.
