Global and Planetary Change 228 (2023) 104205

Contents lists available at ScienceDirect

Global and Planetary Change

journal homepage: www.elsevier.com/locate/gloplacha

North Indian Ocean sea level rise in the past and future: The role of climate
change and variability
J. Jyoti a, b, c, *, P. Swapna a, R. Krishnan a
a
Centre for Climate Change Research, Indian Institute of Tropical Meteorology, Pune 411008, India
b
Center for Climate Physics, Institute for Basic Science, Busan, Republic of Korea
c
Pusan National University, Busan, Republic of Korea

A R T I C L E I N F O A B S T R A C T

Editor: Dr. Alan Haywood Sea-level rise is an inevitable consequence of climate change due to increased external forcings and potentially
devastating effects on the coastal population, low-lying islands, and marine ecosystem. To better understand and
Keywords: quantify the impact of climate change and variability on the global and Indian Ocean sea level, we analyze
Thermosteric sea level rise observed, reanalysis, and climate model experiments from phase five and recent phase six of the Coupled Model
Total sea level
Intercomparison Project (CMIP). Our analysis based on CMIP5 and CMIP6 models reveals that the global and
External forcing
North Indian Ocean (NIO) sea-level rise is primarily driven by anthropogenic forcing and internal variability
Internal variability
North Indian Ocean likely to have a secondary contribution. Contrary to the global sea-level rise, mainly driven by mass contribution,
Future projections the past NIO sea-level rise is dominated by the thermosteric contribution caused by the collaborative contri­
bution of anthropogenic forcing and internal variability. The observed long-term trend in thermosteric sea level
during 1955–2005 in the NIO is comparable to the global mean, while the projected sea-level rise is higher than
the global mean. Since the thermosteric sea level dominates the past Indian Ocean sea level, it is crucial to
investigate the future sea-level projection in the Indian Ocean under global warming. Notably, by the end of the
21st century, the most significant change in total sea level in the Indian Ocean is primarily in the Arabian Sea,
with a rise of about 0.76 m in the NIO and 0.75 m in the global ocean based on high emission scenario (RCP 8.5)
from CMIP5 simulation. Our study highlights the dominant contributors to the total sea-level rise and its pro­
jected rise over the 21st century in the densely populated Indian Ocean region, demanding better adaptation
strategies and policymaking.

1. Introduction variability. Recent studies have shown that about 93% of the excess heat
in the climate system has been absorbed by the oceans, leading to
Understanding twentieth-century global and regional sea-level rise is increased heat content, thermal expansion, and sea-level rise (IPCC AR5
paramount, as they affect the livelihood of millions of people in the and AR6). There is high confidence that GMSL rise has increased, and
world’s coastal regions. Earth’s climate system is influenced by internal human-caused climate change has substantially contributed to the in­
climate variability and external forcing. External forcing refers to the crease since 1900. Church et al. (2013) documented that it is very likely
anthropogenic greenhouse gases (GHGs), aerosols, ozone concentra­ that in the 21st century and beyond, the sea-level change will have a
tions, land-use land-cover changes, the natural forcing of solar irradi­ strong regional pattern, with some places experiencing significant de­
ance, volcanic aerosol concentrations, etc. The internal variability, i.e., viations of local and regional sea-level change from the global mean
dominant climate modes such as El Niño Southern Oscillation (ENSO), change. Additionally, previous studies (Tokarska et al., 2019) have
Pacific Decadal Oscillation (PDO), Atlantic Multidecadal Oscillation shown that changes in ocean warming are detectable and distinct from
(AMO), etc. also contributes to the changes in the climate system the internal variability of the climate system.
(Marcos and Amores, 2014; Slangen et al., 2014). Like other dominant Existing literature based on tide gauges and altimetry reveal an in­
climate modes, which contribute to internal variability, the monsoon crease in sea level globally at a rate of 1.7 mm yr− 1 during 1880–2009
circulation in the Indian Ocean significantly contributes to its internal and in the NIO at 1.06–1.75 mm yr− 1 during 1874–2004 (Church and

* Corresponding author at: Centre for Climate Change Research, Indian Institute of Tropical Meteorology, 411008, India.
E-mail addresses: jyotijadhav@tropmet.res.in, jyoti@pusan.ac.kr (J. Jyoti).

https://doi.org/10.1016/j.gloplacha.2023.104205
Received 13 October 2020; Received in revised form 14 July 2023; Accepted 21 July 2023
Available online 25 July 2023
0921-8181/© 2023 Elsevier B.V. All rights reserved.
J. Jyoti et al. Global and Planetary Change 228 (2023) 104205

White, 2011; Unnikrishnan et al., 2006; Unnikrishnan and Shankar, 2. Data and methods
2007). The dominant cause of the GMSL rise since 1970 is anthropogenic
forcing (Church et al., 2013), and the mass contribution (i.e., added 2.1. Observations
water from melting ice sheets and glaciers) dominates the change. Un­
like the GMSL, Swapna et al., 2017 notice that during the recent past The Satellite-derived sea-level anomaly from the Archiving, Valida­
thermosteric changes dominated the NIO sea-level rise, and the rate of tion, and Interpretation of Satellite Oceanographic (Aviso; Ducet et al.,
TSL rise is about 0.68 ± 0.03 mm yr− 1 during 1958–2015 and has 2000) of the global domain for 1993–2019 is utilized for the analysis.
accelerated to 2.3 ± 0.09 mm yr− 1 during 1993–2015. Additionally, we use available long-term tide-gauge records in the In­
There has been significant progress in understanding the GMSL dian Ocean obtained from the archives of Permanent Service for Mean
response to external forcing (anthropogenic and natural sources). Sea Level (PSMSL; Woodworth and Player, 2003). To get TSL (followed
However, external forcing’s effect on the regional sea-level rise, espe­ eq. (1)) resulting from temperature, we considered different reanalysis
cially the past and future Indian Ocean sea-level rise, remains unex­ and observational datasets such as ORAS4 (Balmaseda et al., 2013) from
plored. The regional sea-level trend and variability significantly deviate 1958 to 2017, Ishii et al. (2017) updated version 7.2 from 1955 to 2018,
from the global mean. The Indian Ocean is warming faster than the EN4 (Good et al., 2013) version 4.2.0 for the period 1900–2019, and
global mean (IPCC AR6; Chapter 9), leading to a significant sea-level rise NOAA (Levitus et al., 2012). The NOAA datasets are available as yearly
in the NIO. Studies have also shown convincing evidence that the NIO mean thermosteric anomalies from the surface to 700 m depth during
sea level has accelerated for the last 4–5 decades (Han et al., 2010; 1950–2019. The methodology used to estimate the steric sea level and
Swapna et al., 2017) and are mainly controlled by dynamic changes in components for the upper 700 m of the ocean following standard
circulation. Regional sea-level change can potentially be far more sub­ methods, such as those detailed in Swapna et al. (2017) and Jyoti et al.
stantial than the global mean rise and depend on many processes (2019), are mentioned as eq. (1) below. Since 2005, sufficient Argo
involving the ocean, atmosphere, geosphere, and cryosphere (Church products have been available; thus, in this study, in addition to rean­
et al., 2013). alysis, we use the temperature profiles of Argo to compute TSL. We
The projected sea-level rise is spatially non-uniform; therefore, dif­ utilize three different gridded monthly Argo products of temperature
ferences in the spatial distribution of sea-level rise make regional sea- down to 700 m, such as the Japan Agency for Marine-Earth Science and
level projections challenging. GMSL is projected to accelerate under Technology (JAMSTEC; Hosoda et al., 2008), CSIO (BOA), and the In­
all scenarios, and the rising rate strongly depends on the scenario fol­ ternational Pacific Research Centre (IPRC).
lowed. The GMSL will rise between 0.43 m (0.29–0.59 m; under low
emission scenario of RCP2.6) and 0.84 m (0.61–1.10 m; RCP8.5) by
2100 relative to 1986–2005, and it will continue to rise beyond 2100 for 2.2. Climate models data
centuries due to deep-ocean heat uptake and mass loss of the ice-sheet
(Chapter 4; IPCC, 2019: Summary for Policymakers. A study by Jevre­ We make use of climate model simulations from the CMIP5 and
jeva et al., 2020 estimated the contribution of thermal expansion to CMIP6 fields (Eyring et al., 2016; Taylor et al., 2012) archives that had
GMSL future projection and discussed the differences between global potential seawater temperature (“thetao,” variable name described in
mean thermosteric sea level (GMTSL) from CMIP5 and CMIP5 also CMIP) data availability. The monthly profiles of potential temperature
explain large uncertainties in the simulation of GMTSL. While very few from external forcing experiments are further used to obtain the esti­
studies, like Lyu et al., 2020, emphasized SSH’s importance to regional mates of the TSL. The standard equation used for the estimation of TSL is
sea-level projections. The SSH reflects the redistribution of heat, salt, detailed in Swapna et al. (2017) and Jyoti et al. (2019):
and mass in the ocean, contributing significantly to the regional sea
1∑
level. For the Indian Ocean, providing the projections of SSH is essential ηthermosteric (τ) = η(τr ) − dz⌊ρ(θ, Sr , pr ) − ρ(θr , Sr , pr ) ⌋ (1)
ρ0
because no independent studies have previously quantified the regional
sea-level projections. Where ηthermosteric (τ) is the TSL at the time (τ), variables with an “r”
This study addresses the possible impact of external forcing and the superscript refer to reference state values so that η(τr ) is the sea level at
internal variability on the past sea-level rise in the NIO resulting from the reference state, ρ0 is a globally constant Boussinesq reference den­
ocean temperature changes, termed thermosteric sea level (TSL), or sity, θ is the potential temperature of seawater, S is the salinity and p
ocean thermal expansion. The likely contribution of internal variability pressure. We choose the upper 700 m for TSL calculation, as this depth
to NIO is discussed based on the discrepancies caused between the all- range corresponds to that with the most available observational data.
forced and multi-model mean and the observations from the differ­ Specifically, for model output, we take into account the ensemble
ences (supported by a spread of ensemble members) between observed mean of the following available simulations: historical, i.e., fully forced
and simulated sea levels. Additionally, it provides a quantitative (all-forcing) and individual-forcing simulations, i.e., natural-only (Nat-
assessment of the projected sea level in the NIO and the global total sea only) and greenhouse-gas only forcing (GHG-only). The Nat-only forcing
level (SSH), defined as the geodetic height of the sea surface above the includes variations of aerosol concentrations due to volcanic eruptions
reference ellipsoid (a negative value if below; Gregory et al., 2019) by and changes in solar activity, while GHG has anthropogenic variations of
the 21st century based on CMIP5 estimates. greenhouse gas concentrations. Note that aerosol-only simulation is
To fill the gap, in the present work, we analyze available observa­ indirectly estimated as the difference between anthropogenic-only and
tions, reanalysis, and climate model simulations to identify the role of GHG-only simulations. At the same time, the anthropogenic-only
external forcing on the Indian Ocean and global ocean subsurface tem­ experiment is the difference between historical and natural-only ex­
perature. Also, we investigate the agreement (ensemble spread) between periments (assuming the additivity of the response of different forcing).
different climate models and examine the multimodel ensemble mean of This provides a consistent set of model experiments for all contributors.
various external forcing for the NIO and global Ocean Sea level. The climate models currently available and used for the study from the
Furthermore, the paper is structured as follows; the first section in­ CMIP5 and CMIP6 are listed in tables S1 and S2, respectively (refer to
vestigates contributions of the external forcing from CMIP5 and CMIP6 supplementary information). We use historical simulations from CMIP5
to the past sea-level rise in the global and NIO; in section 2, we give a from 1950 to 2005 and for CMIP6 from 1950 to 2014. At the same time,
brief description of the data used; section 3 deals with results, and the the future scenarios from CMIP5 are considered from 2006 to 2100. The
conclusion and implications of the study are discussed in section 5. modeled Indian Ocean CMIP5 and CMIP6 TSL of the historical simula­
tions are compared against the Ishii reconstruction.
For this study, we used only the ensemble mean. For multimodel

2
J. Jyoti et al. Global and Planetary Change 228 (2023) 104205

analyses and intermodal comparison, we interpolated all model data √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅

onto the common observational grid, a global grid of 1o latitude x 1o s=
1 ∑
dif t (2)
longitude. Only the first ensemble member ‘r1i1p1’ of CMIP5 and NT − 1 t
‘r1i1p1f1’ of CMIP6 models have been used for historical and future
scenarios. To compare observed mean fields to those from CMIP5 and The summed difference between ensemble variability and variability
CMIP6 simulations, we consider 1950–2005 as a common climatological among individual products (dift) is computed as:
period for both simulations. dif t
∑[
Pn,t − Enst
]2
To analyze the future scenarios from CMIP5, we use corrected data n
(Church et al., 2013) files that contain the ocean and ice components,
A value below 1 indicates that the spread between products is smaller
sums, and uncertainties used in preparing the IPCC AR5 report (2014)
than the amplitude of ensemble variability in those products.
with some slight modifications. One choice is made to combine the
ocean and inverse barometer effect into one field, both for the mean and
3. Results
the uncertainty. The corrected data include files of annual time series of
the median, 5th, and 95th percentile for each contribution to sea-level
3.1. Global and regional sea-level response to a warming climate
rise and sum. The sum is considered as the total sea level, which is the
addition of all the contributors (i.e., Thermal expansion, Glaciers,
3.1.1. Observed sea-level rise in recent decades
Greenland ice sheet, surface mass balance (SMB), Antarctica ice sheet
Since the observed sea-level rise in the NIO is dominated by TSL
SMB, Greenland ice sheet dynamic change, Antarctica ice sheet dynamic
(Srinivasu et al., 2017; Swapna et al., 2017) compared to mass contri­
change, Land water storage, Greenland ice sheet, Antarctica ice sheet).
bution in the global ocean, we analyze the TSL in the NIO and global
This study provides a projection for the total sea level. The supple­
ocean. Fig.1 shows the annual mean TSL anomalies from observation
mentary information presents one map for the thermosteric plus dy­
(Fig. 1c, d) and reanalysis (Fig. 1a, b) datasets in the global and NIO from
namic sea surface height plus the inverse barometer (IB) signal minus
1950 to 2019. Fig. 1 reveals that the TSL rises in the NIO and the global
the model spread of all three. Therefore, the global mean is not the
ocean, which is higher in the recent decade than in the previous period.
thermosteric global signal minus the thermosteric spread but also con­
We observe that the TSL time series closely follows the total sea level in
tains the dynamic sea surface height and IB ensemble spreads (which do
the NIO. Note that ORAS4 SSH (black curve) for the global ocean in
not have a zero global mean).
panel a is not shown because the global sea level is consistent with the
Likewise, CMIP5 corrected datasets; we use CMIP6 future scenarios
altimeter estimates (Balmaseda et al., 2013). However, partitioning into
(Hermans et al., 2021) corrected for model drift for GMSL and used in
volume and mass variations was more debatable before 1993, as inferred
preparing the IPCC AR6 report (2021). To remove drift, Hermans et al.
by discrepancies between ORAS4 and GRACE-derived bottom pressure.
(2021) apply the least-squares quadratic fit to the entire control exper­
Therefore, the present analysis confirms previous results confirming the
iment of each model and subtract the overlapping part of the fit from the
dominance of TSL in the NIO (Swapna et al., 2017; Jyoti et al., 2019).
historical and scenario runs. Although they chose a quadratic fit for
The observed Indian Ocean Sea level shows decadal to multi-decadal
consistency with IPCC AR5 (Church et al., 2013), the drift is nearly
sea-level variability forced by regional wind changes (Han et al.,
linear for most CMIP6 models. They use CMIP6 ensembles, including 20
2010; Srinivasu et al., 2017; Thompson et al., 2016; Swapna et al.,
models, to correct model drift and select each model’s first available
2017).
simulation variant (“r1i1p1f1”). Unfortunately, Hermans et al. (2021)
A recent study by Swapna et al., 2017 has shown that long-term
updated data does not provide spatially varying GMSL, so we could not
multi-decadal changes in the NIO sea-level rise are accompanied by a
check the regionally varying sea level for the Indian Ocean, unlike the
weakening of summer monsoon circulation, resulting in increased heat
CMIP5. Details about the Climate models used to prepare CMIP5 data for
retention in the NIO and a corresponding rise in TSL. However, the
future projections are shown in Table S1, and those from CMIP6 are
weakening of large-scale summer monsoon circulation has been attrib­
shown in Table S2.
uted to increasing anthropogenic aerosol forcing (Bollasina et al., 2011;
To assess the spread in variability among products in CMIP5 and
Krishnan et al., 2016; Ramanathan et al., 2005). Henceforth, we address
CMIP6 simulations, we refer to the method detailed in the supplemen­
the role of external forcing on the NIO and GMSL from CMIP5 and
tary information of Jyoti et al. (2019). At a given grid point, we note P(n,
recently available CMIP6 simulations.
t), the value of the product n at time t, where n varies from 1 to N (total
number of products) and t varies from 1 to T (total number of time
3.1.2. Enhanced warming in the deeper ocean and the role of external
steps), The ensemble average at a given time t is computed as:
forcing
1∑ Wide-range of evidence shows that Earth has been warming in recent
Enst = Pn,t
N n decades, and over 93% of the total heating of the Earth’s system has
gone into the ocean, which further leads to an increased ocean volume
The temporal standard deviation of this ensemble average is
(Church and White, 2011; Levitus et al., 2012; Rhein et al., 2014).
computed as follows:
Recent studies have shown that the upper global ocean’s heat content
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
1 ∑ has increased over the last 5–6 decades. Also, the warming rate varies
σ= [Enst − Enst ]2 considerably among the ocean basins. The warming and the corre­
T− 1 t
sponding sea-level rise is seen from the observed and reanalysis TSL
Where, (Fig.1) can be caused by external forcings and internal variability. The
critical question is how much of the observed TSL rise in the NIO can be
1 ∑ due to anthropogenic causes, natural fluctuations, and has any contri­
Enst = Enst
T− 1 t bution due to internal variability? To answer this, we analyze the
Finally, the spread (s) over the entire period is estimated as the observed ocean warming in the NIO and the global ocean using an in­
square root of the mean difference (i.e., averaged over products and dividual model (Fig. 2; only CMIP5 simulations are shown) and a multi-
time) between the product ensemble and individual variability: model ensemble mean response (Fig. S2) to natural-only and all-forcing
simulations from CMIP5 and CMIP6. The details of simulations and
models used are discussed in section 2.2 and tables S1 and S2.
The global and NIO warming signal is shown by the temperature

3
J. Jyoti et al. Global and Planetary Change 228 (2023) 104205

Fig. 1. Time series of annual mean thermosteric sea-level anomalies (mm) from reanalysis for the (a) Global mean, (b) North Indian Ocean (50◦ E-110◦ E, 5oS-25oN).
Observed thermosteric (Argo) and satellite-derived (Aviso) annual mean sea-level anomalies for the (c) Global and (d) North Indian Ocean. Colour labeling for panel
(b) is mentioned in panel (a), and that for panel (d) is in (c).

trend from Ishii reanalysis (black curves) from 1955 to 2014, CMIP5 all- simulations for global (Fig. 2, S2c) and NIO (Fig.2, S2d). The NIO tem­
forcing individual models (tan curves), natural-only (red curves; Fig. 2c, perature trend due to natural-only forcing does not show any trend, is
d), and all-forcing (red curves; Fig. 2a,b) multi-model ensemble mean inconsistent with observation, and has a minimal contribution. We as­
simulations. It is observed that the warming extends to about 700 m in sume that the discrepancies between observed and all-forcing simula­
the NIO and is also higher than the global mean from observation (black tions is due to internal variability. Also, the multi-model ensemble mean
curve). But the warming signal in the NIO is mainly confined to the smoothed out the contribution due to internal variability.
upper 300 m. Only a few models show a temperature trend as significant Thus, based on Figs. 2, S2, and S3, it is clear that neither natural-only
as or more than the observed trend (Fig. 2b). Below 75 m, the global forcing nor internal variability can explain all of the observed signals,
ocean presents a more pronounced warming than NIO for the all-forcing and the spread in all-forcings simulation is consistent. This support the
simulation ensemble mean from CMIP5 (red curve) and CMIP6 (blue conclusion that the warming in the NIO is primarily influenced by
curve), shown in Fig. S2a, b. This supports internal variability’s possible anthropogenic forcing and is likely to have a secondary contribution
or likely contribution to the NIO TSL rise. We assume this may be the from internal variability. In contrast, anthropogenic forcing is respon­
case, but they could also be caused by model error or differences due to sible for the global ocean’s observed warming (temperature trend).
different climate sensitivities.
Further, to support the possible role of internal variability and 3.1.3. Spatial trend pattern and temporal evolution
anthropogenic forcing and for their comparison in CMIP5 and CMIP6 This section assessed the observed ocean thermal expansion caused
simulations, we separately show a similar figure like Fig. 2 in the sup­ by excess heat absorbed by the ocean as the climate warms. Due to the
plementary information (Fig. S2). We observed that the results from anthropogenic forcing in the climate system, sea-level rise occurs on a
CMIP5 are consistent with the CMIP6 simulations. This is also supported broad range of temporal and spatial scales, with many contributing
by a subsurface temperature trend spatial map from observation and all- factors making it an integral measure of climate change.
forcing simulations (Fig. S1) from CMIP5 and CMIP6. Though the We analyze the spatial trend in TSL from the ensemble mean of
warming trend in all-forcing simulations from CMIP5 and CMIP6 is CMIP5 and CMIP6 for all-forcing and natural-only simulations (Fig.3)
much closer to the observed trend in the global and NIO, we need to and their corresponding ensemble spread (Fig. S4). The trend is
assess whether it is contributed solely by anthropogenic, natural-only computed for the period 1950–2005 (CMIP5) and 1950–2014 (CMIP6),
forcing, and or internal variability. First, we look at the possible role and its statistical significance was performed with Student’s t-test at a
of natural-only simulation and the internal variability to ensure that a 90% confidence level (denoted by Stipple). Ocean heat uptake is a
change is due to anthropogenic factors. spatially heterogeneous (Stammer et al., 2013), and this is evident from
Here, we estimate the natural-only simulation temperature trend for the spatial pattern of the observed TSL trend (Fig. 3a) of the upper water
the individual models (tan curves; Fig. 2c, d), the ensemble means of column (700 m) from 1950 to 2014. The spatial trends of the ensemble
CMIP5 (1955–2005; Fig. S2c, d), and CMIP6 (1950–2014, Fig. S2c, d) mean of all-forcing simulations from CMIP6 (Fig. 3b) from 1950 to 2014
with observation (red curves). The temperature trend from the natural- and CMIP5 (Fig. 3d) from 1950 to 2005 are consistent region-wise with
only multi-model ensemble means from CMIP5 and CMIP6, as well as the observed trend estimated from 1950 to 2014. However, the all-
the response of individual models, shows that the observed warming forcing simulation from CMIP5 and CMIP6 underestimates the
trend (black curve) is much higher than that of natural-only forcing observed warming magnitude, and thus we cannot rule out the role of

4
J. Jyoti et al. Global and Planetary Change 228 (2023) 104205

Fig. 2. The linear temperature trend (oC yr− 1)

for the upper 700 m averaged over the global
ocean in the (a) all-forcings, (c) natural-only
simulations, and averaged over the North In­
dian Ocean in the (b) all-forcings and (d) natural-
only simulations from CMIP5 climate models
from 1950 to 2005. Plotted are individual models
(tan curves), ensemble averages (black curves),
and observations from Ishii et al. (2017) data­
base are also included (red curves). (For inter­
pretation of the references to colour in this figure
legend, the reader is referred to the web version
of this article.)

internal variability. Except for the northern Pacific, the spatial pattern between CMIP5 and CMIP6 models over those regions is more promi­
observed in CMIP5 is consistent with CMIP6. Conversely, TSL trends for nent than in the Indian Ocean. Confirming the robustness of the NIO
the upper 700 m are significantly smaller than in all-forcing (Fig. 3c, e) signal in the all-forcing and natural-only forcing with good agreement
in the CMIP’s natural-only simulation. Indicating no significant contri­ among CMIP5 and CMIP6 models used for the analysis.
bution from naturally arising forcings to the Indian and most of the Time-series of the ensemble mean TSL anomalies computed for the
global ocean region. But some areas, like North Atlantic, show a rela­ upper 700 m for the NIO and global ocean from the all-forcings and
tively high contribution in the naturally forced run for the period natural-only climate model simulations from CMIP5 (red curve for
investigated from CMIP5 and CMIP6. 1950–2005), CMIP6 (blue curve for 1950–2014) are shown in Fig. 4. The
The robustness of the signal observed in the all-forcing and natural- all-forcing multi-model ensemble means from CMIP5 (Fig.4a, c) and
only simulations is further verified through the spread in variability CMIP6 show an increasing trend similar to observations and is also
among all available models. The spread between ensemble members, within the standard error shown based on observation. However, the
calculated based on standard deviation, and the mathematical formu­ multi-decadal variability seen in the observed NIO (Swapna et al., 2017)
lation for its estimation is described as eq.(2). The spread between sea level is forced by the wind-dynamic response (internal variability)
ensemble members of all-forcing and natural-only simulations for and is smoothed out in all-forcing simulation from CMIP5, CMIP6 due to
CMIP5 and CMIP6 are shown in Fig. S4. The spread of less than one the multi-model ensemble mean. Therefore, we also analyze the time
indicates that the spread between the ensemble members is smaller than series of the individual model and their response in all-forcing and
the amplitude of ensemble variability, thus establishing a better agree­ natural-only simulations and compare it with observation (figure not
ment between the ensembles. The Pacific and the Atlantic Oceans show shown). We observed that TSL in the NIO is more significant than most
a sea-level rise in the all-forcing simulation; however, the spread of the all-forced models. Though large variability exists in all-forced

5
J. Jyoti et al. Global and Planetary Change 228 (2023) 104205

Fig. 3. The spatial trend of thermosteric sea-level (mm yr− 1) for the top 700 m for 1950–2014 computed from (a) Ishii reanalysis, (b) all-forcings, and (c) natural-
only simulations from the ensemble mean of CMIP6 simulations, and that for 1950–2005 ensemble mean of (d) all-forcing and (e) natural-only CMIP5 simulations.
The regions at a 90% significance level are stipple.

models, they are within the error bars from observations. On the con­ of CMIP5 and CMIP6 natural-only (Fig. 4b, d) simulations. The number
trary, the natural-only simulations from individual models could be of available models that provide natural-only simulations is less than all-
more consistent outside the observed variability. Hence, we cannot forcing simulations. The natural-only CMIP5 (CMIP6) multi-model
discard the role of internal variability in the NIO TSL rise. ensemble average trend is − 0.09 ± 0.01 (− 0.06 ± 0.03) mm yr− 1 and
From observation, the sea-level trend for the global ocean is 0.43 ± -0.04 ± 0.02 (− 0.09 ± 0.02) mm yr− 1 for the global and NIO respec­
0.02 mm yr− 1, and that from the all-forcing CMIP5 (CMIP6) simulation tively during the period 1955–2005. Compared to all-forcing, the
is 0.42 ± 0.02 (0.21 ± 0.03) mm yr− 1. The NIO observation trend is 0.43 natural-only simulations from CMIP show no significant long-term
± 0.07 mm yr− 1, and the ensemble means trend from the CMIP5 trend, resulting in a slight net negative TSL rise from 1955 to 2005 for
(CMIP6) historical simulation is 0.43 ± 0.03 (0.28 ± 0.03) mm yr− 1. the global and NIO. Thus, the observed TSL in the global ocean is solely
The trend estimates are computed for the common period 1955–2005. associated with anthropogenic forcing. However, the NIO shows a pri­
The uncertainty in the ensemble averages corresponds to the 95% con­ mary contribution due to anthropogenic forcing, and internal variability
fidence interval, and observation corresponds to the standard error. The is likely to have a secondary because the contribution due to natural-
analysis reveals that the increase in TSL in the NIO is comparable to the only forcing is relatively small.
global mean during the last 4–5 decades (1955–2005). To summarise, the model simulations driven by natural-only forcing
We have also investigated the time evolution of the ensemble mean do not reproduce the observed TSL rise. This indicates that

6
J. Jyoti et al. Global and Planetary Change 228 (2023) 104205

Fig. 4. Time evolution of ensemble mean

of thermosteric sea level anomalies from
CMIP5 (red curves) and CMIP6 (blue
curves) climate models simulations for
the period 1950–2005 and 1950–2014,
respectively, in the global ocean esti­
mated for (a) all-forcing, and (b) natural-
only; and that in the North Indian Ocean
for (c) all-forcing and (d) natural-only.
Ishii observations (black curves) are
shown for 1950–2018. The rate of sea-
level rise is estimated for the common
period 1955–2005 for observation and
the ensemble mean of CMIP5 and CMIP6
models. The error bars based on one
standard deviation (light pink shading)
are shown for observations. (For inter­
pretation of the references to colour in
this figure legend, the reader is referred
to the web version of this article.)

anthropogenic forcings are majorly responsible for the globally observed et al., 2016). The rate of TSL rise for the GHG-only experiment has
TSL rise. In the case of NIO, anthropogenic forcing has a primary increased rapidly (Fig. 5a-d; blue curves) since 1950, and the TSL in­
contribution, and internal variability is likely to have a secondary crease in the global (Myhre et al., 2013), and NIO closely follows the
contribution in the twentieth century. The rate of TSL increase in the increase in atmospheric GHG radiative forcing both in CMIP5 and
NIO is comparable to the global. CMIP6. Compared to the observation and all-forcing simulations, the
GHG-only forced experiment shows enhance TSL rise in the global and
3.1.4. Contribution of external forcings to Indian Ocean sea-level rise NIO.
We have seen that external forcing contributes primarily to TSL rise Likewise, the TSL from aerosol-only simulations (Fig. 5a-d; tan
both for the global and the NIO. To get further insight, we investigate the curve) shows a long-term negative trend, resulting in a small net nega­
role of individual external forcings to sea-level rise in the NIO. We have tive change for the global and NIO. Note that the combined response of
utilized the CMIP5 and CMIP6 historical simulation (all-forcing), individual simulations is equivalent to an all-forcing simulation. Albeit
natural-only, and GHG-only simulations. The various simulations from the aerosol’s counteracting effect, the all-forcing simulation shows a
CMIP5 and CMIP6 are presented as multi-model ensemble means. substantial TSL rise in the global and NIO (Fig. 5a-b; red curves). The
Although the significant contributor to the twentieth-century GMSL model ensemble means from CMIP5 and CMIP6 are consistent among all
change is known, the response of individual forcing to the NIO sea level simulations. The variability observed in TSL in the all-forcing simulation
is unknown. Here, we determine the underlying drivers of TSL change is further used to investigate the NIO sea-level rise.
and quantify the impact of a range of external forcings. We focus on To quantify the contribution of the different forcings on the temporal
understanding the independent contribution of various external forcings variability of TSL, we examine the long-term trends from CMIP5
to TSL rise. Where the all-forcing experiment is the addition of natural- (Fig. 5e) and CMIP6 (Fig. 5f) and compare the global and NIO. The
only and anthropogenic-only (equal to GHG-only plus aerosol-only) quantification is perform for 1980 through 2005, as the TSL rise was
simulations. We consider aerosol-only (all-forcing– GHG-only – observed globally, and the NIO sea level was more prominent during this
natural-only) as the difference between all-forcing, GHG-only, and period (Fig. 5a-d). The time series of all-forcing and observations have
natural-only simulations. The response of individual multi-model means nearly similar variations in the GMSL with precisely the same median
from CMIP5 and CMIP6 for different forcing experiments for the global values. While the NIO TSL variations and median values in all-forcing
Ocean and NIO are shown in Fig. 5. simulations are slightly underestimated in the CMIP5 and CMIP6
The natural-only simulations, driven by the solar cycle and volcanic multi-model ensemble time series, this might be due to decadal, multi-
eruption changes, show a decrease in TSL and no significant long-term decadal time scales disagreement between observations and all-forcing
trend for the global (consistent with earlier studies; Slangen et al., simulation in the NIO. Interestingly, except for other simulations, the
2015, 2016) and NIO (Fig. 5; green curves). Volcanic eruptions have a all-forcing CMIP6 response for the global ocean shows large variability
temporary cooling effect, i.e., negative radiative forcing after volcanic compare to CMIP5. The variability in the observations and ensemble
eruptions causes long-term perturbations in the ocean, which causes mean of all-forcing simulations significantly differs from that of GHG-
decreased oceanic temperatures and a sudden fall in TSL (Church and only, natural-only, and aerosol-only simulations in the global and NIO
White, 2006; Gleckler et al., 2006; Gregory et al., 2005). (Fig. 5e, f). Also, the variations and magnitude in the natural-only and
In the GHG-only simulations, enhanced thermal expansion occurs aerosol-only simulations are more substantial in the NIO than in the
due to the increasing GHG concentrations, as seen from the 1960s on­ global ocean, and the values are negative. This supports that the
wards, with a growing response to increasing GHG forcings (Slangen anthropogenic GHG-only simulations dominate the NIO TSL rise,

7
J. Jyoti et al. Global and Planetary Change 228 (2023) 104205

Fig. 5. The annual ensemble means thermosteric sea-level anomalies (mm) from CMIP6 model simulations averaged over (a) the Global and (b) North Indian Ocean,
and that from CMIP5 simulations averaged over (c) the Global and (d) North Indian Ocean. Comparison of the standard deviation of the multi-model ensemble mean
time series (mm) from (e) CMIP5 and (f) CMIP6 simulations for the period of 1980–2005 and 1980–2014, respectively, box and whiskers indicate median (black line
over the box), 25–75% (box), and 5–95% (whiskers).

evidenced by CMIP5 and CMIP6. 3.2. Projected sea-level rise in the 21st century
Till this section, we investigate the contribution of anthropogenic,
natural-only, and greenhouse gases for ocean warming (“thermal 3.2.1. Sea level projections
expansion”) over the historical period and provide respective estimates The previous section shows that the thermal expansion causes the
of the Indian and Global Ocean. Though the TSL component is dominant present-day sea-level rise in the NIO due to increased GHG concentra­
for the past Indian Ocean sea level, that may or may not be the scenario tions in recent decades. In this section, we focus on projected changes in
in the future. Hence, projection analysis is performed hereafter for the SSH. To quantify the projected SSH, we use corrected sea-level data from
total sea level. We believe that projection of SSH for the regional Ocean CMIP5 models, which have been utilized to prepare the IPCC AR5 report
(Indian Ocean) is preferable by decision-makers to assess coastal risk, (details in the data and methodology section). Additionally, we have
develop resilient climate communities and plan vital infrastructure in used the corrected CMIP6 future simulation (Hermans et al., 2021) to
low-elevation coastal zones. prepare the IPCC AR6 report. Throughout the analysis, low-end (RCP
2.6/SSP 1–2.6), mid-range (RCP 4.5/SSP 2–4.5), and high-end (RCP

8
J. Jyoti et al. Global and Planetary Change 228 (2023) 104205

8.5/SSP 5–8.5) emission scenarios from CMIP5, CMIP6 are used to un­ 1986–2005 from CMIP5 for the low, mid, and high-end scenarios in the
derstand the spatial and temporal changes in projected sea level and near future (2041–2060), far future (2081–2100), and by 2100 are listed
quantify the sea-level projections for the global and NIO. in Table 1. We observed that the NIO’s SSH and dynamic sea-level mean
Global sea level is projected to rise 20 cm to as much as 60 cm above values are higher than the global ocean in low-end, mid-range, and high-
current levels by the end of the 21st century (Church et al., 2013; IPCC end scenarios. However, the difference between NIO and the global
AR5) and is further projected to rise even after the decline of CO2 ocean median value projection for SSH is about 0.1 m (higher for NIO),
concentrations (IPCC AR6; chapter 9). Therefore, reliable projections of and that of dynamic sea level is 0.1 m to 0.3 m. Surprisingly the future
sea level in the 21st century are essential for planning adaptation or dynamic sea level, i.e., the contribution from thermosteric and circula­
mitigation measures. However, significant differences in regional pro­ tion change, indicates nearly 35% (41%), 38% (45%), and 44% (48%)
jections from the global mean indicate that regional sea-level pro­ contribution to SSH in RCP 2.6, RCP 4.5, and 8.5 respectively for the
jections are necessary for better adaptation planning, but this is lacking global ocean (NIO). This clearly shows that the global ocean has a
for the Indian Ocean. So, it is worth estimating the NIO’s SSH projections dominant contribution due to mass added to the ocean over dynamic
based on CMIP5 models. plus thermosteric sea level, and vice-versa for the NIO.
The multi-model ensemble means for different scenarios project SSH Henceforth we compare the GMSL projections from CMIP6 with
rise for the global (Figs. 6, 7) and NIO (Fig. 6) in the 21st century. As the CMIP5 because Hermans et al. (2021) provided the corrected SSH and its
ocean integrates the surface heat flux, sea-level change following component’s global mean. Similar to Fig. 6a, Fig. 7a shows the annual
different scenarios remains relatively high for the first several decades mean time series of the corrected global SSH from Hermans et al. (2021)
(Fig. 6a,b) since 2006. Therefore, to understand the difference in SSH for three scenarios (SSP1–2.6, SSP2–4.5, and SSP5–8.5). The results
response to strong-mitigation and non-mitigation scenarios, we also from CMIP6 are consistent with the CMIP5, and the differences are
consider the RCP 2.6 and 4.5, 8.5 scenarios, particularly for this analysis. relatively modest. Interestingly, in contrast to the CMIP5, the GMSL is
An RCP 2.6 scenario response shows a reduced rate of SSH change higher in the CMIP6 due to higher climate sensitivity in CMIP6 models.
beyond about 2040 for the global and NIO. The SSH change in RCP 2.6 (Zelinka et al., 2020).
by 2100 is just below half of RCP 8.5 (Fig. 6a,b; red and green curves) for Following the IPCC AR5 methods (Church et al., 2013), CMIP6
the NIO and the global ocean. This indicates that the sea level change simulation results in median ensemble projections of GMSL rise are 0.76,
depends on the pathways of CO2 emissions. We observed that the SSH 0.56, and 0.47 m in 2100 (see Table 2) in SSP5–8.5, SSP2–4.5, and
rise for all the scenarios in the NIO is always slightly higher than the SSP1–2.6, respectively (Fig. 7a). This represents a modest increase of 2,
global ocean and supported through the corresponding ensemble mean 3, and 3 cm (+3%, +6%, and + 7%) relative to CMIP5. The 5%–95%
projections of the time-averaged SSH changes with reference period range shifted from 0.43 to 1.13 m, 0.29–0.81 m, and 0.23–0.70 m to
1986–2005 for two different scenarios shown in Fig. 6c-f. Interestingly, 0.52–1.05 m, 0.38–0.76 m, and 0.30–0.64 m. Thus, updating existing
the variability observed in the annual mean time series of SSH in the NIO GMSL projections with CMIP6 data affects projections at 2100 only
for all three scenarios is extensive compared to the global ocean. moderately but has a more pronounced impact beyond 2100.
Apart from time variability, the SSH change for mid and high-end In this section, we examine and compare CMIP5 and CMIP6 (only
scenarios estimated for the near future (2041–2060) and far future GMSL) temporal and spatial change (from CMIP5 only) of regional SSH
(2081–2100) are shown in Fig. 6c-f. The spatial distribution of regional and dynamic sea-level (Supplementary figure) projection by the end of
sea level projected changes in the world oceans, derived from 21 CMIP5 the 21st century. From the CMIP5 multi-model ensemble mean SSH
models for the RCP 4.5 and RCP 8.5. The projected sea level in both spatial pattern and dynamic sea-level projections, we observe non-
scenarios reveals a generally rising GMSL. Notably, in the Indian Ocean, uniform sea-level rise globally, with substantial dynamic sea-level rise,
the most significant changes are seen in the north and western tropical particularly in the North-western Indian Ocean, North Pacific, North
Indian Ocean (Fig. 6c-f), with an SSH change of about 0.46 m (RCP 2.6), Atlantic, Arctic, and middle latitude of the southern ocean.
0.55 m (RCP 4.5) – 0.76 m (RCP 8.5) by the end of the 21st century
(Table1). The north and western tropical Indian Ocean is one of the few 3.2.2. Sea level rise along coastal Indian Ocean: past to future projections
oceanic regions where CMIP climate models predict maximum changes Here, we attempt to understand the regional sea-level projections for
in phase five. In a similar study, Carson et al. (2015) showed that these the NIO with the corrected CMIP5 future projections, which have yet to
projected changes in mean sea level by the 21st century are considerably be addressed. We estimates the annual mean sea-level rise by 2100
more significant than the noise and suggested to have an anthropogenic under the RCP 2.6, RCP 4.5, and 8.5 scenarios relative to 1986–2005 at
origin. The global SSH contribution by 2100 for RCP 4.5 is 0.54 m and available long-term tide gauge locations in the NIOs coastal regions
0.55 m in the NIO. The projected sea level gradually increased from the (Fig. 8; the circle on the spatial map denotes tide gauge stations). The
low-end scenario and reached a maximum in the RCP 8.5 in the far observed mean sea level anomalies analyzed from the satellite altimetry
future (Fig. 6f). datasets are also shown for the exact locations (Fig. 8; purple curves).
The TSL, the wind-driven circulation (contribution due to both we Fig. 8 illustrates the observed and projected mean sea-level rise for in­
mentioned here as dynamic sea level), is an essential component of the dividual coastal stations utilizing tide-gauge data, satellite observations,
regional sea-level projections. The dynamic sea level determined jointly and CMIP5 projections (corrected mean sea level used for IPCC AR5)—
by ocean density and wind-driven circulation is defined as the height of selected locations according to their societal relevance and the time
the sea surface above the geoid with zero global means (Gregory et al., series of tide gauges available for a more extended period. Most of the
2019). Therefore, to understand the future contribution due to dynamic location’s observational records reveal the presence of strong decadal to
sea level, we provide a similar analysis like Fig. 6 for the dynamic sea multi-decadal variability. However, this variability is more or less
level and shown in supplementary information as Fig. S5. The spatial captured by many sites in all scenarios by the end of the 21st century.
and temporal change of the multi-model ensemble mean of dynamic sea Although the altimetry time series is very short relative to the tide gauge
level reveals increasing sea level in the future in both scenarios, i.e., RCP records and the projections, there is no apparent disagreement between
4.5 and 8.5 (Fig. S4). Interestingly, in the Indian Ocean, the dominant the time series slopes where they coincide. Except for the station,
changes are seen in the western Indian Ocean, especially in the Arabian Dimond Harbour, where the tide gauge mean sea level disagrees with
Sea. Note that the Arabian Sea is where we witness the dominance of the the satellite observations because these are not corrected for vertical
monsoon circulation and corresponding changes in the cross-equatorial land movement and are consistent with the previous studies (Unnik­
cell (responsible for the heat transport mechanism in the Indian Ocean). rishnan and Shankar, 2007). However, Diamond Harbour lies in the
The median values and ranges for projections for the SSH and dy­ delta of the River Ganga (a short distance upstream of the Hooghly es­
namic sea level of the global and NIO sea-level rise relative to tuary), whose subsidence rates are up to 4 mm yr− 1 (Goodbred and

9
J. Jyoti et al. Global and Planetary Change 228 (2023) 104205

Fig. 6. Time series of total sea level (m) for RCP 2.6 (green curve), RCP 4.5 (blue curve), and RCP 8.5 (red curve) scenarios for the (a) Global mean, and (b) North
Indian Ocean; shaded areas show the 5%–95% confidence interval with respective colors. Ensemble mean projections of the time-averaged total sea level changes
computed from 21 CMIP5 climate models (in meters), with reference period 1986–2005 for different scenarios as RCP 4.5 for the (c) near future (1941–2060), and (d)
far future (2081–2100); and from RCP 8.5 for the (e) near future (1941–2060), and (f) far future (2081–2100). The global and NIO rate of sea-level change estimates
for 2005–2100 are shown for three scenarios in panels a and b. Colour bars for panels (c) and (e) are common, shown on the left-hand side of the figure at the end,
while panels (d) and (f) are on the right-hand side of the figure. (For interpretation of the references to colour in this figure legend, the reader is referred to the web
version of this article.)

10
J. Jyoti et al. Global and Planetary Change 228 (2023) 104205

anthropogenic forcing (GHG forcing, offset by aerosol-induced change)

component - but note that internal variability is more important than
natural variability. Results confirm that the global mean and NIO TSL
pattern show robust responses to anthropogenic forcings with a higher
rise rate in the NIO than the global in the latter half of the twentieth
century in GHG forcing simulation. The rate of rising from the CMIP5
multi-model ensemble mean average for all-forcings (natural-only)
simulation for the NIO is 0.43 ± 0.03 (− 0.04 ± 0.02) mm yr− 1, and that
for the global ocean is 0.42 ± 0.03 (− 0.09 ± 0.02) mm yr− 1 during
1955–2005. This shows a major anthropogenic forcing component - but
note that internal variability is more important than natural variability
for the global and NIO TSL since 1950.
Subsequently, in a warming climate, the mean sea level is expected
to increase at a higher rate in the coming centuries. We examine the
Fig. 7. Time series of total sea level (m) for SSP1–2.6 (green curve), SSP2–4.5
regional (NIO) and global SSH projections under low, medium, and
(blue curve), and SSP5–8.5 (red curve) scenarios for the (a) Global mean; high-emission pathways using three RCP scenarios (RCP 2.6, RCP 4.5,
shaded areas show the 5%–95% confidence interval with respective colors. The RCP 8.5) with nominal forcing levels in each set of scenarios reaching
global mean sea-level estimates rate for 2005–2100 is shown for three scenarios approximately 2.6, 4.5, and 8.5 W m− 2 in 21st century, respectively
inside the figure. (For interpretation of the references to colour in this figure (O’Neill et al., 2016). The projected regional sea level shows a consid­
legend, the reader is referred to the web version of this article.) erable increase for all the scenarios and reveals generally higher SSH
projections for the NIO than the global. Notably, in the Indian Ocean, the
Kuehl, 2000). more considerable SSH changes are seen in the north and western
The linear trend estimates of future mean sea-level projection for the tropical Indian Ocean (Fig. 6c-f) with an SSH change of about 0.55 m
individual stations are shown in Fig. 8. The linear mean sea level trend (RCP 4.5) – 0.76 m (RCP 8.5) in CMIP5 by the end of the 21st century.
estimates range (minimum and maximum selected among all stations) The detailed analysis of the comprehensive set of climate simulations
for RCP 2.6 from 2.5 mm yr− 1 to 5.12 mm yr− 1, for RCP 4.5 from 3.55 demonstrates two main results. First, natural variations have little or no
mm yr− 1 to 6.08 mm yr− 1, and that for RCP 8.5 from 5.54 mm yr− 1 to contribution to the NIO TSL rise; the anthropogenic forcing is the
8.48 mm yr− 1. Among all the analyzed stations, Kandla shows a low dominant contributor and is likely to have a secondary contribution due
mean sea-level rise, while Seychelles shows very high estimates to internal variability. Second, because of the delayed response in the
compared to other stations. The mean sea-level rise for future pro­ deep ocean, global SSH has risen in the 21st century. The NIO SSH
jections shows higher values over the southern coastal stations than the projections are higher than the global by the end of the 21st century.
northern stations on India’s east and west coasts. Thus, the sea-level rise-related risk for the natural and human systems is
higher along NIO coastlines than globally.
4. Discussion and conclusions

The sea-level rise is a catastrophic consequence of anthropogenic

climate change and internal variability (more significant for regional
than global sea levels). In the present study, we assessed sea-level Table 2
changes in the global ocean and NIO, focusing mainly on ocean vol­ Median values and ranges are for projections of Global mean sea level and TSL in
ume changes (Past analysis) and total sea level (future research only). meters (m) relative to 1986–2005 in the near future (2041–2060), far future
(2081–2100), and in 21st-Centuary from CMIP6 three different scenarios as
We learned from Swapna et al. (2017) that the TSL changes had
SSP1–2.6, SSP2–4.5, and SSP5–8.5.
contributed substantially to regional sea-level rise, especially in the NIO.
Therefore, in this study, we first demonstrate the influence of external Year Global mean

forcings and internal variability on the past global and NIO TSL. We used SSP1–2.6 SSP2–4.5 SSP5–8.5
simulations from CMIP5 archives for the past since 1950–2005 and Estimates from CMIP6 relative to 1986–2005 for the GMSL (m)
2006–2100 for future projections. We estimated the rate of TSL rise in 2041–2060 0.22 (0.16–0.29) 0.23 (0.17–0.30) 0.25 (0.19–0.32)
the global and NIO based on the CMIP5 model all-forcings and natural- 2081–2100 0.42 (0.28–0.57) 0.49 (0.34–0.66) 0.64 (0.44–0.87)
only simulations. In 2100 0.47 (0.30–0.64) 0.56 (0.38–0.76) 0.76 (0.52–1.05)
Estimates from CMIP6 relative to 1986–2005 for the TSL
Based on Figs. 2, S2, S3, and 5e,f, it is clear that neither natural 2041–2060 0.09 (0.06–0.11) 0.09 (0.07–0.12) 0.10 (0.08–0.13)
forcing nor internal variability can explain all observed signals, as the 2081–2100 0.15 (0.10–0.20) 0.19 (0.14–0.25) 0.27 (0.19–0.35)
spread of the natural-only CMIP5 and CMIP6 ensemble can’t explain the In 2100 0.16 (0.11–0.21) 0.21 (0.15–0.28) 0.32 (0.23–0.42)
observations. Thus, it is defensible that NIO TSL rise has a major

Table 1
Median values and ranges for projections of Global and North Indian Ocean total sea level and TSL + Dynamic sea level in meter (m) relative to 1986–2005 in the near
future (2041–2060), far future (2081–2100), and in 21st-Centuary for the RCP 2.6, RCP 4.5, 8.5 scenarios.
Year RCP 2.6 RCP 4.5 RCP 8.5

Global mean North Indian Ocean Global mean North Indian Ocean Global mean North Indian Ocean

Estimates from CMIP5 relative to 1986–2005 for the SSH (m)

2041–2060 0.23 (0.12–0.34) 0.23 (0.13–0.33) 0.24 (0.12–036) 0.24 (0.14–0.34) 0.26 (0.14–0.39) 0.27 (0.16–0.38)
2081–2100 0.41 (0.21–0.62) 0.41 (0.22–0.61) 0.48 (0.26–0.72) 0.49 (0.28–0.71) 0.64 (0.36–0.95) 0.65 (0.40–0.94)
In 2100 0.46 (0.23–0.70) 0.46 (0.25–0.68) 0.54 (0.29–0.81) 0.55 (0.31–0.81) 0.75 (0.43–1.13) 0.76 (0.47–1.13)
Estimates from CMIP5 relative to 1986–2005 for the TSL+ Dynamic sea level + IB (m)
2041–2060 0.09 (0.02–0.17) 0.10 (0.04–0.17) 0.10 (0.02–0.18) 0.11 (0.04–0.18) 0.12 (0.04–0.20) 0.13 (0.06–0.20)
2081–2100 0.15 (0.05–0.25) 0.17 (0.07–0.27) 0.19 (0.08–0.30) 0.22 (0.11–0.33) 0.28 (0.14–0.43) 0.31 (0.18–0.44)
In 2100 0.16 (0.05–0.28) 0.19 (0.08–0.29) 0.21 (0.09–0.33) 0.24 (0.12–0.36) 0.33 (0.17–0.50) 0.37 (0.22–0.51)

11
J. Jyoti et al. Global and Planetary Change 228 (2023) 104205

Fig. 8. Time series of annual mean sea-level anomalies (m) from tide gauges (black curves), satellite-derived Aviso (magenta curves), and future projections from
RCP 2.6 (green curves), RCP 4.5 (blue curves), and RCP 8.5 (red curves) from CMIP5 for selected tide gauge stations over the east and west coast of India and island
region over the Indian Ocean. The middle colour pattern shows the spatial trend of mean sea level from the RCP 8.5 scenario for 2005–2100. Tide gauge stations are
marked from 1 to 14 on the spatial map in the middle. The mean sea-level trends for all three scenarios for 2005–2100 at the locations of the tide gauge stations are
mentioned in the corresponding map. The anomalies for AVISO and model simulations are calculated for the common period 1993–2019. (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of this article.)

Declaration of Competing Interest study are publicly available and are described in the data sets section
and tabulated as in Table S1 and S2 in supporting information. Data
The authors declare no competing financial interests. analysis and graphing were done with a licensed version of the Pyferret
and Python programs.
Data availability
Appendix A. Supplementary data
Data will be made available on request.
Supplementary data to this article can be found online at https://doi.
Acknowledgments org/10.1016/j.gloplacha.2023.104205.

The authors thank the Director, IITM, for providing support to carry
References
out this research. We thank editor and anonymous reviewers to critically
go through the manuscript and provide us valuable suggestions to Balmaseda, M.A., Mogensen, K., Weaver, A.T., 2013. Evaluation of the ECMWF Ocean
improve the content of the manuscript. All the datasets used for this reanalysis system ORAS4. Q. J. R. Meteorol. Soc. 139, 1132–1161. https://doi.org/
10.1002/qj.2063.

12
J. Jyoti et al. Global and Planetary Change 228 (2023) 104205

Bollasina, M.A., Ming, Yi, Ramaswamy, V., 2011. Science 334 (502–505). https://doi. Marcos, M., Amores, A., 2014. Quantifying anthropogenic and natural contributions to
org/10.1126/science.1204994. thermosteric sea level rise. Geophys. Res. Lett. 41, 2502–2507. https://doi.org/
Church, J.A., White, N.J., 2006. A 20th century acceleration in global sea-level rise. 10.1002/2014GL059766.
Geophys. Res. Lett. 33 https://doi.org/10.1029/2005GL024826. Myhre, G., Samset, B.H., Schulz, M., Balkanski, Y., Bauer, S., Berntsen, T.K., Bian, H.,
Church, J.A., White, N.J., 2011. Sea-Level rise from the late 19th to the early 21st Bellouin, N., Chin, M., Diehl, T., Easter, R.C., Feichter, J., Ghan, S.J.,
Century. Surv. Geophys. 32, 585–602. https://doi.org/10.1007/s10712-011-9119-1. Hauglustaine, D., Iversen, T., Kinne, S., Kirkevåg, A., Lamarque, J.-F., Lin, G., Liu, X.,
Church, J.A., White, N.J., Konikow, L.F., Domingues, C.M., Graham Cogley, J., Lund, M.T., Luo, G., Ma, X., Noije, T., Penner, J.E., Rasch, P.J., Ruiz, A., Seland, Ø.,
Rignot, E., Gregory, J.M., Van Den Broeke, M.R., Monaghan, A.J., Velicogna, I., Skeie, R.B., Stier, P., Takemura, T., Tsigaridis, K., Wang, P., Wang, Z., Xu, L., Yu, H.,
2013. Erratum: Revisiting the Earth’s sea-level and energy budgets from 1961 to Yu, F., Yoon, J.-H., Zhang, K., Zhang, H., Zhou, C., 2013. Radiative forcing of the
2008. Geophys. Res. Lett. 40 (4066) https://doi.org/10.1002/grl.50752. direct aerosol effect from AeroCom phase II simulations. Atmos. Chem. Phys. 13,
Ducet, N., Traon, P.Y.L., Reverdin, G., 2000. Global high-resolution mapping of ocean 1853–1877. https://doi.org/10.5194/acp-13-1853-2013.
circulation from TOPEX/Poseidon and ERS-1 and -2. J. Geophys. Res. Oceans 105, O’Neill, B.C., Coauthors, 2016. The Scenario Model Intercomparison Project
19477–19498. https://doi.org/10.1029/2000JC900063. (ScenarioMIP) for CMIP6. Geosci. Model Dev. 9, 3461–3482. https://doi.org/
Eyring, V., Bony, S., Meehl, G.A., Senior, C.A., Stevens, B., Stouffer, R.J., Taylor, K.E., 10.5194/gmd-9-3461-2016.
2016. Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) Ramanathan, V., Chung, C., Kim, D., Bettge, T., Buja, L., Kiehl, J.T., Washington, W.M.,
experimental design and organization. Geosci. Model Dev. 9, 1937–1958. https:// Fu, Q., Sikka, D.R., Wild, M., 2005. Atmospheric brown clouds: Impacts on south
doi.org/10.5194/gmd-9-1937-2016. Asian climate and hydrological cycle. PNAS 102 (15), 5326–5333. https://doi.org/
Gleckler, P.J., Wigley, T.M.L., Santer, B.D., Gregory, J.M., AchutaRao, K., Taylor, K.E., 10.1073/pnas.0500656102.
2006. Krakatoa’s signature persists in the ocean. Nature 439, 675. https://doi.org/ Rhein, et al., 2014. Observations: Ocean. Climate Change 2013 – The Physical Science
10.1038/439675a. Basis: Working Group I Contribution to the Fifth Assessment Report of the
Good, S.A., Martin, M.J., Rayner, N.A., 2013. EN4: Quality controlled ocean temperature Intergovernmental Panel on Climate Change. https://doi.org/10.1017/
and salinity profiles and monthly objective analyses with uncertainty estimates. CBO9781107415324.010.
J. Geophys. Res. Oceans 118, 6704–6716. https://doi.org/10.1002/2013JC009067. Slangen, A.B.A., Church, J.A., Zhang, X., Monselesan, D., 2014. Detection and attribution
Goodbred, S.L., Kuehl, S.A., 2000. The significance of large sediment supply, active of global mean thermosteric sea level change. Geophys. Res. Lett. 41, 5951–5959.
tectonism, and eustasy on margin sequence development: Late Quaternary https://doi.org/10.1002/2014GL061356.
stratigraphy and evolution of the Ganges–Brahmaputra delta. Sediment. Geol. 133, Slangen, A.B.A., Church, J.A., Zhang, X., Monselesan, D.P., 2015. The Sea Level Response
227–248. https://doi.org/10.1016/S0037-0738(00)00041-5. to External Forcings in Historical Simulations of CMIP5 climate Models. J. Clim. 28,
Gregory, J.M., Coauthors, 2019. Concepts and terminology for sea level: mean, 8521–8539. https://doi.org/10.1175/JCLI-D-15-0376.1.
variability and change, both local and global. Surv. Geophys. 40, 1251–1289. Slangen, A.B.A., Church, J.A., Agosta, C., Fettweis, X., Marzeion, B., Richter, K., 2016.
https://doi.org/10.1007/s10712-019-09525-z. Anthropogenic forcing dominates global mean sea-level rise since 1970. Nat. Clim.
Gregory, J.M., Dixon, K.W., Stouffer, R.J., Weaver, A.J., Driesschaert, E., Eby, M., Chang. 6, 701–705. https://doi.org/10.1038/nclimate2991.
Fichefet, T., Hasumi, H., Hu, A., Jungclaus, J.H., Kamenkovich, I.V., Levermann, A., Srinivasu, U., Ravichandran, M., Han, W., Sivareddy, S., Rahman, H., Li, Y., Nayak, S.,
Montoya, M., Murakami, S., Nawrath, S., Oka, A., Sokolov, A.P., Thorpe, R.B., 2005. 2017. Causes for the reversal of North Indian Ocean decadal sea level trend in recent
A model intercomparison of changes in the Atlantic thermohaline circulation in two decades. Clim. Dyn. 49, 3887–3904. https://doi.org/10.1007/s00382-017-
response to increasing atmospheric CO2 concentration. Geophys. Res. Lett. 32 3551-y.
https://doi.org/10.1029/2005GL023209. Stammer, D., Cazenave, A., Ponte, R.M., Tamisiea, M.E., 2013. Causes for contemporary
Han, W., Meehl, G., Rajagopalan, B., et al., 2010. Patterns of Indian Ocean sea-level regional sea level changes. Annu. Rev. Mar. Sci. 5, 21–46. https://doi.org/10.1146/
change in a warming climate. Nature Geosci. 3, 546–550. https://doi.org/10.1038/ annurev-marine-121211-172406.
ngeo901. Swapna, P., Jyoti, J., Krishnan, R., Sandeep, N., Griffies, S.M., 2017. Multi-decadal
Hermans, T.H.J., Gregory, J.M., Palmer, M.D., Ringer, M.A., Katsman, C.A., Slangen, A.B. weakening of Indian Summer Monsoon circulation induces an increasing Northern
A., 2021. Projecting global mean sea-level change using CMIP6 models. Geophys. Indian Ocean Sea Level. Geophys. Res. Lett. 44, 10,560–10,572. https://doi.org/
Res. Lett. 48, e2020GL092064 https://doi.org/10.1029/2020GL092064. 10.1002/2017GL074706.
Hosoda, S., Ohira, T., Nakamura, T., 2008. A monthly mean dataset of global oceanic Taylor, K.E., Stouffer, R.J., Meehl, G.A., 2012. An overview of CMIP5 and the experiment
temperature and salinity derived from Argo float observations. JAMSTEC-R 8, design. Bull. Am. Meteor. Soc. 93, 485–498. https://doi.org/10.1175/BAMS-D-11-
47–59. https://doi.org/10.5918/jamstecr.8.47. 00094.1.
Ishii, M., Fukuda, Y., Hirahara, S., Yasui, S., Suzuki, T., Sato, K., 2017. Accuracy of Thompson, P.R., Piecuch, C.G., Merrifield, M.A., McCreary, J.P., Firing, E., 2016. Forcing
Global Upper Ocean Heat Content Estimation Expected from present Observational of recent decadal variability in the Equatorial and North Indian Ocean. J. Geophys.
Data Sets. Sola 13, 163–167. https://doi.org/10.2151/sola.2017-030. Res. Oceans 121, 6762–6778. https://doi.org/10.1002/2016JC012132.
Jyoti, J., Swapna, P., Krishnan, R., Naidu, C.V., 2019. Pacific modulation of accelerated Tokarska, K., Hegerl, G., Schurer, A., Ribes, A., Fasullo, John, 2019. Quantifying human
South Indian Ocean Sea level rise during the early 21st Century. Clim. Dyn. https:// contributions to past and future ocean warming and thermosteric sea level rise.
doi.org/10.1007/s00382-019-04795-0. Environ. Res. Lett. 14, 074020 https://doi.org/10.1088/1748-9326/ab23c1.
Krishnan, R., Sabin, T.P., Vellore, R., Mujumdar, M., Sanjay, J., Goswami, B.N., Unnikrishnan, A.S., Shankar, D., 2007. Are sea-level-rise trends along the coasts of the
Hourdin, F., Dufresne, J.L., Terray, P., 2016. Deciphering the desiccation trend of the North Indian Ocean consistent with global estimates? Glob. Planet. Chang. 57,
south Asian monsoon hydroclimate in a warming world. Clim. Dyn. 47 (3–4), 301–307. https://doi.org/10.1016/j.gloplacha.2006.11.029.
1007–1027. https://doi.org/10.1007/s00382-015-2886-5. Unnikrishnan, A.S., Kumar, K.R., Fernandes, S.E., Michael, G.S., Patwardhan, S.K., 2006.
Levitus, S., Antonov, J.I., Boyer, T.P., Baranova, O.K., Garcia, H.E., Locarnini, R.A., Sea Level Changes along the Indian Coast: Observations and Projections.
Mishonov, A.V., Reagan, J.R., Seidov, D., Yarosh, E.S., Zweng, M.M., 2012. World Woodworth, P.L., Player, R., 2003. The permanent service for mean sea level: an update
Ocean heat content and thermosteric sea level change (0–2000 m), 1955–2010. to the 21st century. J. Coast. Res. 19, 287–295.
Geophys. Res. Lett. 39 https://doi.org/10.1029/2012GL051106. Zelinka, M.D., Myers, T.A., McCoy, D.T., Po-Chedley, S., Caldwell, P.M., Ceppi, P., et al.,
Lyu, K., Zhang, X., Church, J.A., 2020. Regional dynamic sea level simulated in the 2020. Causes of higher climate sensitivity in CMIP6 models. Geophys. Res. Lett. 47,
CMIP5 and CMIP6 models: mean biases, future projections, and their linkages. e2019GL085782 https://doi.org/10.1029/2019GL085782.
J. Clim. 33, 6377–6398. https://doi.org/10.1175/JCLI-D-19-1029.1.

13