Abstract

Global sea level rise due to the thermal expansion of the warming oceans and freshwater input from melting glaciers and ice sheets is threatening to inundate low-lying islands and coastlines worldwide. At present the global mean sea level rises at 3.1 ± 0.7 mm yr−1 with an accelerating tendency. However, the magnitude of recent decadal sea level trends varies greatly spatially, attaining values of up to 10 mm yr−1 in some areas of the western tropical Pacific. Identifying the causes of recent regional sea level trends and understanding the patterns of future projected sea level change is of crucial importance. Using a wind-forced simplified dynamical ocean model, the study shows that the regional features of recent decadal and multidecadal sea level trends in the tropical Indo-Pacific can be attributed to changes in the prevailing wind regimes. Furthermore, it is demonstrated that within an ensemble of 10 state-of-the-art coupled general circulation models, forced by increasing atmospheric CO2 concentrations over the next century, wind-induced redistributions of upper-ocean water play a key role in establishing the spatial characteristics of projected regional sea level rise. Wind-related changes in near-surface mass and heat convergence near the Solomon Islands, Tuvalu, Kiribati, the Cook Islands, and French Polynesia oppose—but cannot cancel—the regional signal of global mean sea level rise.

1. Introduction

Sea level in the tropical oceans varies considerably as a result of interannual (Wyrtki 1975) to decadal (Qiu and Chen 2006) wind stress forcing. Wind stress curl anomalies cause changes in the convergence of near-surface Ekman transports. Depending on the sign of the related Ekman pumping velocity, warm water is either pumped down into the ocean, which leads to a deepening of the tropical thermocline and an increase in local sea level, or cold water is sucked up to the near-surface layer, which leads to a shoaling of the thermocline and a sea level decrease. Moreover, wind stress curl forcing triggers long westward-propagating oceanic Rossby waves that are accompanied by sea level anomalies along their propagation pathways. Thus, wind-induced changes in upper-ocean heat content in the tropical oceans strongly contribute to the regional characteristics of sea level anomalies on interannual to centennial time scales.

Recent assessments of future sea level rise have focused on global mean sea level rise and the contributions due to thermal expansion, glacier melting, and the disintegration of ice sheets (e.g., Church and White 2006; Rahmstorf 2007; Bindoff et al. 2007). However, there is still a major uncertainty with respect to the regional characteristics of projected future sea level rise, as stated both in the third and fourth assessment reports of Working Group I of the Intergovernmental Panel on Climate Change (IPCC; e.g., Christensen et al. 2007). Identifying the causes of this uncertainty and developing a better understanding of the drivers of future regional sea level change, in comparison to the global mean estimates, would be very beneficial for policy makers in many different countries and island nations.

Further to this, with many low-lying South Pacific islands already experiencing significant positive sea level trends and local inundations, it is of utmost importance to better understand the reasons for these trends and how they will project into the next decades. The goal of our paper is to elucidate the effects of long-term wind changes on the regional characteristics of past and future sea level trends in the tropical southern Indo-Pacific region and to compare these regional projections with recent estimates of global mean future sea level rise.

The paper is organized as follows. In section 2 we describe the simplified modeling approach applied to recent wind data and future wind projections from IPCC-type models to quantify their effects on thermocline depth and regional change. In section 3 we present the main results and address the question how recent and future wind stress changes have affected and will affect the regional sea level rise in the tropical southern Indo-Pacific. Section 4 summarizes our main results and concludes with a brief discussion.

2. Methods

To quantify the direct forcing effects of wind changes on regional sea level, a linear reduced-gravity shallow-water model (SWM) in Cartesian coordinates is used. With this model we study the thermocline and equivalent sea level response to historical daily wind forcing obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF) analysis dataset (for the period 1993–2008) and the 40-yr ECMWF Re-Analysis (ERA-40) dataset (for the period 1958–2001; Uppala et al. 2005) as well as to wind projections from an ensemble of future climate change simulations (scenarios A1B and B1) conducted with state-of-the art coupled general circulation models (CGCMs). Our shallow-water model features the ocean dynamical response to wind changes while keeping the total water mass conserved. A sea level increase at one location has to be balanced by a decrease at another location. Global mean sea level rise is not captured by such a model.

a. The simplified ocean model

The ocean model used here is a 1.5-layer reduced-gravity model of the stratified ocean, titled the SWM. The upper and lower layers of the model are separated by an interface that approximates the sharp tropical thermocline separating the warm surface waters from the cold waters of the deep ocean. Motion in the upper layer is driven by the applied anomalous wind stresses (per unit density), while the lower layer is assumed motionless and infinitely deep. The associated response of the ocean is characterized by the vertical displacement of the thermocline and the horizontal components of upper-layer flow velocity. The upper-layer ocean dynamics are described by the linear reduced-gravity form of the shallow-water equations (McGregor et al. 2007). We prescribe the reduced-gravity parameter gSW = 0.0265 m s−2 and the mean depth of the upper layer as 300 m, so the first baroclinic-mode Kelvin wave speed is 2.8 m s−1. The model has a 1° resolution and is configured for the low–mid latitude Indo-Pacific Ocean (51°S and 51°N, 40°E and 60°W). It also includes realistic continental boundaries that were calculated as the locations where the bathymetric dataset (Smith and Sandwell 1997) has a depth of less than the model mean thermocline depth of 300 m.

1) Translating thermocline depth of the shallow-water model into sea level

To translate thermocline anomalies h, predicted by the shallow-water model into sea level anomalies η, we use the linear relationship η(x, y, t) = [g′(x, y)/g]h(x, y, t), where the parameter g′(x, y) = α(x, y)gSW is obtained empirically from a set of linear regressions between different simulated thermocline anomalies hj and sea level data ηj(j = 1, … , 5). Note that gSW is the SWM reduced-gravity parameter that is constant in space and time.

Three different SWM hindcast runs were conducted with the wind stress anomalies of the National Centers for Environmental Prediction (NCEP; Kalnay et al. 1996), ERA-40 (Uppala et al. 2005), and ECMWF analyses (see data acknowledgment) for the periods 1948–2008, 1958–2001, and 1986–2008, respectively. The SWM thermocline anomalies of these three experiments hi, (i = 1, … , 3) are then regressed against the Archiving, Validation, and Interpretation of Satellite Oceanographic data (AVISO) observed monthly-mean sea level anomaly data ηi at each grid point in the model domain using ηi = αi(x, y)(gSW/g)hi(i = 1, … , 3) for the respective overlap periods. This results in three separate spatial maps of regression coefficients αi(x, y) = (g/gSW)〈ηihi〉/〈hi2〉(i = 1, … , 3).

Furthermore, the linear scaling relationship also uses thermocline and sea level information from the eddy-resolving global OGCM for the Earth Simulator (OFES) hindcast (Masumoto et al. 2004) (h4, η4) and the ECMWF operational ocean analysis/reanalysis system (ORA-S3; Balmaseda et al. 2008) (h5, η5). The OGCM component of our thermocline–sea level scaling parameter α(x, y) was derived by regressing the modeled 20°C isotherm depth anomalies in OFES (h4) and ORA-S3 (h5) against the respective modeled sea level anomalies (η4, η5) from these hindcast simulations for the entire simulation period (1950–2001), thus giving ηi = αi(x, y)(gSW/g)hi for OFES (i = 4) and for the ORA-S3 OGCM hindcast (i = 5). This calculation provides an extra two spatial maps of regression coefficients, α4 and α5. No SWM data were used for this component.

The mean of these five thermocline–sea level regression coefficient spatial maps is shown in the lower panels of Fig. 1 and will be used as scaling between simulated SWM thermocline anomalies and sea level. With very few zonal features in the regression, we have decided to take the zonally averaged domain regression coefficients and apply a Gaussian fit in latitude (Fig. 1, lower right). The domain is defined here as the SWM domain (see Fig. 1). This latitudinal Gaussian fit is then used to convert SWM dynamic height to realistic sea level. One clear feature in both the spatial map and zonally averaged regression coefficients is the decrease in their value with increasing latitude. This feature is apparent in all five of the regression coefficient spatial maps αi(i = 1, … , 5) that were averaged to make the map displayed in Fig. 1 (bottom left). This fall off in regression coefficient value as latitude increases is realistic and occurs because of the decreasing stratification with increasing latitude.

Fig. 1.

(top left) Correlation between observed (detrended) sea level anomalies for the period 1993–2008 (AVISO) and the model results that were obtained by forcing the SWM with the historical wind stress data for this period, obtained from different wind stress products. (bottom left) Average regression map between thermocline and sea level anomalies. The average was obtained from various regression fields that were obtained by regressing simulated thermocline and simulated sea level anomalies in the ORA-S3 OGCM reanalysis product, the eddy-resolving OFES model hindcast, and by regressing the observed sea level data (AVISO) with the SWM results forced by the ERA-40 and NCEP wind stress data. (top right) Zonally averaged correlation coefficient (zonal average only taken over domain shown) as a function of latitude. (bottom right) Zonally averaged regression coefficient (zonal average only taken over domain shown) and a simple Gaussian fit to this zonal mean with respect to latitude. The empirical Gaussian latitudinal fit is used to convert simulated shallow-water sea level to realistic sea level.

Fig. 1.

(top left) Correlation between observed (detrended) sea level anomalies for the period 1993–2008 (AVISO) and the model results that were obtained by forcing the SWM with the historical wind stress data for this period, obtained from different wind stress products. (bottom left) Average regression map between thermocline and sea level anomalies. The average was obtained from various regression fields that were obtained by regressing simulated thermocline and simulated sea level anomalies in the ORA-S3 OGCM reanalysis product, the eddy-resolving OFES model hindcast, and by regressing the observed sea level data (AVISO) with the SWM results forced by the ERA-40 and NCEP wind stress data. (top right) Zonally averaged correlation coefficient (zonal average only taken over domain shown) as a function of latitude. (bottom right) Zonally averaged regression coefficient (zonal average only taken over domain shown) and a simple Gaussian fit to this zonal mean with respect to latitude. The empirical Gaussian latitudinal fit is used to convert simulated shallow-water sea level to realistic sea level.

2) Experiments

(i) Sea level hindcasts

To understand the effects of recent wind stress changes on the regional patterns of sea level rise, we conducted two SWM hindcast experiments. In one case the SWM was forced by the ERA-40 reanalysis wind stress from 1958 to 2001 (Uppala et al. 2005); in the other case, ECMWF wind stress analysis data (see data acknowledgment) were used to force the SWM for the period 1993–2008. Both model simulations capture the interannual variations in sea level, as observed–reconstructed in different sea level datasets (including tide gauges), very well. Correlations in the tropical Indo-Pacific are typically higher than 0.6 (see also Fig. 1, upper-left panel). Moreover, these forced SWM experiments also capture the long-term decadal to multidecadal trends quite realistically (see Fig. 2), suggesting that recent regional sea level trends in the tropical Pacific and Indian ocean are to a large extent controlled by wind forcing.

Fig. 2.

Linear sea surface height trends (mm yr−1) for the period 1993–2008 in (a) satellite altimeter data and (c) as simulated by reduced-gravity SWM (forced by ECMWF analysis winds.). (b) Linear sea surface height trends (mm yr−1) for the period 1958–2001 derived from SODA POP 1.4.2 (Carton et al. 2005) that uses ERA-40 wind forcing. (d) As in (b), but derived from a reduced-gravity SWM that was forced by ERA-40 winds (Uppala et al. 2005).

Fig. 2.

Linear sea surface height trends (mm yr−1) for the period 1993–2008 in (a) satellite altimeter data and (c) as simulated by reduced-gravity SWM (forced by ECMWF analysis winds.). (b) Linear sea surface height trends (mm yr−1) for the period 1958–2001 derived from SODA POP 1.4.2 (Carton et al. 2005) that uses ERA-40 wind forcing. (d) As in (b), but derived from a reduced-gravity SWM that was forced by ERA-40 winds (Uppala et al. 2005).

(ii) Climate change projections

To estimate the effect of future wind changes on regional sea level trends, two ensembles of experiments were carried out with the shallow-water model. The first ensemble included 14 SWM simulations that were each forced with the linear wind stress trend, calculated between 2001 and 2100, projected from 14 CGCM simulations forced with IPCC-prescribed A1B scenario greenhouse gas concentrations. The 14 IPCC CGCMs analyzed for their projected wind stress trends were as follows: Bjerknes Center for Climate Research-Bergen Climate Model 2.0 (BCCR-BCM2.0), National Center for Atmospheric Research Community Climate System Model, version 3 (NCAR CCSM3), Centre National de Recherches Météorologiques Coupled Global Climate Model, version 3 (CNRM-CM3), Commonwealth Scientific and Industrial Research Organisation Mark version 3.0 (CSIRO Mk3.0), CSIRO Mk3.5, Geophysical Fluid Dynamics Laboratory Climate Model version 2.1 (GFDL CM2.1), Goddard Institute for Space Studies Atmosphere–Ocean Model (GISS-AOM), GISS Model E-R (GISS-ER), L’Institut Pierre-Simon Laplace Coupled Model, version 4 (IPSL CM4), Model for Interdisciplinary Research on Climate 3.2, medium-resolution version [MIROC3.2(medres)], Model for Interdisciplinary Research on Climate 3.2, high-resolution version [MIROC3.2(hires)], Meteorological Institute of the University of Bonn, ECHO-G Model (MIUBECHOG), Max Planck Institute (MPI) ECHAM5, and the Meteorological Research Institute Coupled General Circulation Model, version 2.3.2a (MRI CGCM2.3.2a) (only these 14 CGCMs provided wind stress data for the A1B and B1 emission scenarios). The second ensemble of SWM simulations again included 14 SWM simulations. However, in this case, they were each forced with the linear wind stress trend projected from 14 CGCM simulations forced with IPCC-prescribed B1 scenario greenhouse gas concentrations between 2001 and 2100.

b. The simplified atmospheric model

To study the effect of the multimodel ensemble mean climate change signal in sea surface temperature (SST) on the wind stress, we employ a low-order dynamical atmospheric model (Kleeman 1991). This model is a two-layer system (250 and 750 mb) on the equatorial beta plane that was linearized around a state of rest. It adopts Rayleigh friction and Newtonian cooling in the dynamical and thermodynamical equations, respectively. The diabatic forcing parameterization includes a moist static energy threshold, and the system of dynamical and thermodynamical equations is solved on an approximately 2.8° global horizontal grid. The tropical Indo-Pacific SST anomaly forcing (between 30°S and 30°N) for the model is obtained by computing the SST difference from an ensemble of 14 A1B simulations (mentioned earlier) using the averages over the periods 2081–2100 and 2001–20. The linearized model is not developed to simulate the response to globally averaged warming. Hence, for our atmospheric model response experiment, we subtracted the domain-averaged (30°S and 30°N) SST value from the multimodel ensemble mean surface-warming pattern for the A1B scenario.

3. Results

a. Recent sea level trends and their relation to wind forcing

The short-term and large-scale trend for the period 1993–2008 simulated by the shallow-water model forced by the wind stress from the ECWMF analysis matches the satellite-derived trend qualitatively well (Figs. 2a,c), in accordance with a recent study by Qiu and Chen (2006). This suggests that tropical wind variations are largely responsible for the positive sea level trends in the southwestern and northwestern tropical Pacific (Carton et al. 2005).

Considering that the zonal changes in equatorial thermocline depth during El Niño (La Niña) years act to lower (raise) sea levels in the western equatorial Pacific, a large fraction of this recent trend (1993–2008) can be explained by the sign, magnitude, and frequency of ENSO events in the trend period. For instance, the change from the El Niño–dominated conditions in the first 5 yr of the trend period, which also includes the largest El Niño event on record in 1997/98, to the more normal El Niño event magnitude and frequency of the later trend period results in an increasing (deepening) trend of western equatorial Pacific sea level (thermocline depth).

The observed–reconstructed longer-term sea level trend from 1958 to 2001 is also reproduced qualitatively well by the wind-forced shallow-water model, in comparison to estimates of sea level trends obtained directly from hydrographic data (Ishii et al. 2006) or from ocean reanalysis products such as the Simple Ocean Data Assimilation Parallel Ocean Program (SODA POP; Carton et al. 2005; Fig. 2c). The overall zonal redistribution of mass in the equatorial band with decreasing (increasing) sea level in the west (east) over the longer period is consistent with the ocean’s response to the observed slowdown of the Walker circulation that was reported in Vecchi et al. (2006).

Our analysis confirms previous results (Qiu and Chen 2006) that suggest that many large-scale features of both decadal and multidecadal sea level trends in the Pacific can be explained in terms of the ocean’s response to variations in Ekman pumping and the propagation of oceanic Rossby waves.

b. Projected future sea level trends and their relation to wind forcing

Estimating the regional aspects of future sea level rise in the tropical oceans requires a proper assessment and understanding of the dominant wind stress trends due to greenhouse warming (Lowe and Gregory 2006). Long-term wind stress anomaly patterns in the tropical oceans will be governed to a large extent by changes of SST in response to increasing greenhouse gas concentrations. These tropical SST changes in turn generate local and remote wind stress curl anomalies (Gill 1980) that will cause Ekman pumping and Rossby wave propagation in the ocean and subsequently thermocline and sea level anomalies.

To identify the robust tropical wind stress response to increasing greenhouse gas concentrations over the next 100 yr, we computed the ensemble mean wind stress trend using 14 individual CGCMs that were forced by greenhouse gas emission scenarios A1B and B1 (Meehl et al. 2007; see section 2). For both greenhouse gas emission scenarios, the reduced warming in the southeastern tropical Pacific is accompanied by an intensification of the southeasterly trades (Figs. 3e,f).

Fig. 3.

(a) Simulated future sea level trends (mm yr−1) derived from the ensemble mean of 14 SWM simulations that were forced for 100 yr with the wind stress trends obtained from a 14-member ensemble of CGCM A1B greenhouse warming experiments. (b) As in (a), but using wind stress trend forcing from a 14-member ensemble of CGCM experiments, subject to the B1 greenhouse gas emission scenario. (c) Simulated future sea level trends (Indo-Pacific spatial average between 30°S and 30°N subtracted) obtained from the multimodel ensemble mean of 10 CGCMs that were forced for the period 2001–2100 with the greenhouse gas emission scenarios A1B. (d) As in (c), but using the greenhouse gas emission scenario B1. Stippling denotes regions where greater than 66% of the models utilized agree on the sign of the sea level change. (e),(f) The IPCC A1B and B1 scenario 14-member multimodel ensemble mean wind stress linear trend for the period 2001–2100, respectively. Vector units are Pa × 10−3 yr−1, and vector color indicates the sign and magnitude (3 × 10−6 m s−1 100 yr−1) of the wind-stress-induced Ekman pumping.

Fig. 3.

(a) Simulated future sea level trends (mm yr−1) derived from the ensemble mean of 14 SWM simulations that were forced for 100 yr with the wind stress trends obtained from a 14-member ensemble of CGCM A1B greenhouse warming experiments. (b) As in (a), but using wind stress trend forcing from a 14-member ensemble of CGCM experiments, subject to the B1 greenhouse gas emission scenario. (c) Simulated future sea level trends (Indo-Pacific spatial average between 30°S and 30°N subtracted) obtained from the multimodel ensemble mean of 10 CGCMs that were forced for the period 2001–2100 with the greenhouse gas emission scenarios A1B. (d) As in (c), but using the greenhouse gas emission scenario B1. Stippling denotes regions where greater than 66% of the models utilized agree on the sign of the sea level change. (e),(f) The IPCC A1B and B1 scenario 14-member multimodel ensemble mean wind stress linear trend for the period 2001–2100, respectively. Vector units are Pa × 10−3 yr−1, and vector color indicates the sign and magnitude (3 × 10−6 m s−1 100 yr−1) of the wind-stress-induced Ekman pumping.

The projected southeasterly trade wind intensification generates Ekman suction in large areas of the tropical South Pacific and the southern Indian Oceans (Figs. 3e,f) and Ekman pumping northeast of Australia and the northern Indian Ocean. We forced the linear reduced-gravity shallow-water model for the period 2001–2100 with the wind stress trends of each of the 14 CGCMs and for both greenhouse gas emission scenarios. The resulting 14-member ensemble mean sea level trends exhibit robust features (Figs. 3a,b) for both emission scenarios that can be directly related to the respective Ekman pumping anomalies (Figs. 3e,f): Negative sea level trends of about −0.5 mm yr−1 occur in large areas of the southern tropical Pacific, including the tropical Islands of Tuvalu, parts of French Polynesia, Kiribati, the Cook Islands, and the Solomon Islands. Wind-induced sea level rise occurs south of the Solomon Islands, in Vanuatu and Fiji, as well as in the Indian Ocean between 5°S and 15°S. These features are reproduced by more than 66% of the simple shallow-water model simulations (Figs. 3a,b, stippling). While the amplitude of the wind-induced sea level trends in the B1 scenario is smaller than in the A1B scenario (Figs. 3a,b), their resulting wind-forced sea level trend patterns are very similar.

To identify the regional patterns of simulated future sea level change in the multimodel ensemble of CGCM greenhouse warming simulations (Meehl et al. 2007), we computed the multimodel ensemble mean trend of sea level for the A1B and B1 scenarios from the 10 phase 3 of the Coupled Model Intercomparison Project (CMIP3) model simulations that archived sea level data. To highlight the regional patterns of sea level change in the tropical Pacific and the Indian Ocean, the domain-averaged sea level trend was subtracted in Figs. 3c,d. The large-scale features of sea level rise in the South Pacific, as simulated by the CGCMs (Figs. 3c,d), correspond well to the wind-induced sea level trends generated by the shallow-water model (Figs. 3a,b). More than 66% of the 10 CGCMs utilized in this analysis show a drop of sea level (relative to the domain mean sea level rise) in the trade wind region of the South Pacific. Furthermore, east of Australia, a consistent projected acceleration of the global mean sea level trend is found (Figs. 3c,d) that can also be explained in terms of the Ekman pumping anomalies (Figs. 3e,f). However, one main region of discrepancy between the SWM model and CGCM ensemble means is the eastern tropical Pacific. In this region the SWM ensemble mean displays a nonrobust slight increase in the sea level trend (relative to the domain mean), which is consistent with the ensemble mean thermocline depth projection of the CGCMs presented in Vecchi and Soden (2007), while the CGCM ensemble displays a robust change toward lower-than-normal sea level rise. This indicates that this local reduction in sea level rise cannot be directly attributed to wind-induced changes in the upper-layer thickness and as such must be attributed to a change in density (a process the SWM cannot produce) because of changed freshwater fluxes, or advection.

As demonstrated earlier, one major cause for the projected regional patterns of sea level in the South Pacific relative to the global mean sea level rise is the southeasterly wind intensification that was noted previously (Liu et al. 2005; Vecchi and Soden 2007). This large-scale circulation anomaly is a robust feature in the 24 model simulations that participated in the CMIP3 (Meehl et al. 2007) A1B and B1 greenhouse warming experiments. To further elucidate its origin, we forced a simple atmosphere model that captures the first baroclinic Kelvin and Rossby modes (Kleeman 1991) by the multimodel ensemble mean SST pattern that emerges in response to increasing greenhouse gas concentrations following the A1B emission scenario (Fig. 4). Here the ensemble mean SST response was determined just from the 14 CGCM experiments that were used to create Fig. 3e.

Fig. 4.

SST response (°C) (shading, upper color bar) after 100 yr to increasing greenhouse gas concentrations as simulated by 14 CGCMs following the A1B scenario. Wind stress (vector) (Pa) and Ekman pumping response (vector colors, bottom color bar) (3 × 10−6 m s−1) to SST greenhouse warming signal (after subtracting the spatial-domain-averaged SST of 2.18°C) using a simple shallow water atmosphere model (Kleeman 1991) that captures Kelvin and Rossby wave dynamics.

Fig. 4.

SST response (°C) (shading, upper color bar) after 100 yr to increasing greenhouse gas concentrations as simulated by 14 CGCMs following the A1B scenario. Wind stress (vector) (Pa) and Ekman pumping response (vector colors, bottom color bar) (3 × 10−6 m s−1) to SST greenhouse warming signal (after subtracting the spatial-domain-averaged SST of 2.18°C) using a simple shallow water atmosphere model (Kleeman 1991) that captures Kelvin and Rossby wave dynamics.

The resulting tropical wind stress pattern (Fig. 4) simulated by the simplified atmospheric model resembles the one obtained from the 14 CGCM simulations (Figs. 3e,f). Thus, the southeasterly trade wind intensification in the South Pacific can be related to the fact that the SST increase is strongest on the equator, whereas it has a smaller magnitude in the South Pacific. Moreover, the simulated southeasterly wind anomalies would generate evaporative cooling and hence reduce the greenhouse warming signal in the South Pacific through the wind–evaporation–SST feedback (Xie 1996). As part of the simulated southeasterly wind intensification and its related wind convergence toward the equator, the SST-forced simplified atmospheric model also generates a cyclonic wind anomaly in the southwestern tropical Pacific between 12°S and the equator that would explain the Ekman suction and hence reduce (relative to the global average) the relative sea level rise east of Papua New Guinea and encompass the region near Tuvalu, the Cook Islands, and Kiribati. This reduction can be attributed to the first-order baroclinic atmospheric Rossby wave response to the projected future SST changes and the corresponding ocean Ekman pumping changes.

In a recent study (Liu et al. 2005), the minimum projected warming in off-equatorial regions, relative to the maximum warming on the equator (Timmermann et al. 2004, their Fig. 4), was explained in terms of increased latent cooling, due to higher climatological mean winds in off-equatorial regions and CO2-induced changes in the sea-air-specific humidity gradient (Seager and Murtugudde 1997), and increased low-level clouds, due to an increased atmospheric static stability. However, Xie et al. (2010) show that the equatorial Pacific Ocean maximum SST warming results instead from a reduction of the Newtonian cooling toward the equator. This reduced equatorial Newtonian cooling is set by the mean evaporation, which is determined by mean SST, wind speed, and relative humidity.

The multimodel ensemble mean warming (Meehl et al. 2007; Vecchi and Soden 2007) also exhibits a pronounced meridional asymmetry in the eastern equatorial Pacific with enhanced warming in the northeastern tropical Pacific compared to the southeastern tropical Pacific. This warming asymmetry can be partly related to the decreased heat capacity of the Northern Hemisphere, due to the asymmetric land–sea distribution (Timmermann et al. 2004; Liu et al. 2005), as well as to fact that the climatological southeasterly trade winds in the South Pacific are far more extensive than their Northern Hemispheric counterparts and thus can upwell cool subsurface water in regions of Ekman suction and provide a more efficient way of the spreading temperature anomalies equatorward via the wind–evaporation–SST feedback.

4. Summary and discussion

Summarizing our main results (see Fig. 6 for a schematic illustration), we found that the future greenhouse warming pattern of SST in the tropical and subtropical Pacific generates a robust wind response in the tropical South Pacific. The related change in wind stress curl induces local Ekman suction and hence a decrease of sea level relative to the global mean signal in many areas in the South Pacific and Southern Indian Oceans. While the wind-induced negative sea level changes projected for the next 100 yr are quite considerable for many low-lying Pacific Islands (Fig. 5), they are relatively small (10%–30%) compared to recent global mean sea level rise estimates (Bindoff et al. 2007; Rahmstorf 2007).

Fig. 5.

Future 100-yr regional projections of the contribution (%) of wind-forced sea level trends relative to two different estimates of global mean sea level rise (Bindoff et al. 2007; Rahmstorf 2007) for greenhouse gas emission scenario A1B. Bindoff et al. (2007; left abscissa) and Rahmstorf (2007; right abscissa) estimates of sea level rise are shown. The wind contributions were estimated by forcing the linear ocean SWM with the wind stress trends from 14 CGCMs subject to emission scenario A1B. The crosses represent the medians of the SWM projections, and the bars characterize the 5%–95% uncertainty ranges within the 14-member ensemble of SWM simulations. The regions are defined as follows: Seychelles: 4°–10°S, 46°–56°E; Maldives: 0°–8°N, 72°–74°E; Solomon Island: 5°–11°S, 154°–163°E; Tuvalu: 5°–10°S, 176°–180°E; North Cook Islands: 8°–12°S, 195°–205°E; South Cook Islands: 17°–20°S, 196°–203°E; South Line Islands: 9°–11°S, 208°–211°E; and French Polynesia: 14°–24°S, 205°–225°E.

Fig. 5.

Future 100-yr regional projections of the contribution (%) of wind-forced sea level trends relative to two different estimates of global mean sea level rise (Bindoff et al. 2007; Rahmstorf 2007) for greenhouse gas emission scenario A1B. Bindoff et al. (2007; left abscissa) and Rahmstorf (2007; right abscissa) estimates of sea level rise are shown. The wind contributions were estimated by forcing the linear ocean SWM with the wind stress trends from 14 CGCMs subject to emission scenario A1B. The crosses represent the medians of the SWM projections, and the bars characterize the 5%–95% uncertainty ranges within the 14-member ensemble of SWM simulations. The regions are defined as follows: Seychelles: 4°–10°S, 46°–56°E; Maldives: 0°–8°N, 72°–74°E; Solomon Island: 5°–11°S, 154°–163°E; Tuvalu: 5°–10°S, 176°–180°E; North Cook Islands: 8°–12°S, 195°–205°E; South Cook Islands: 17°–20°S, 196°–203°E; South Line Islands: 9°–11°S, 208°–211°E; and French Polynesia: 14°–24°S, 205°–225°E.

A potential caveat of the model projections discussed here could be the fact that the southeasterly trade wind intensification occurs in an area near the South Pacific convergence zone (SPCZ). In most CMIP-3 CGCMs, the meridional extent of the SPCZ is not simulated realistically (Kim et al. 2008). Estimating the effects of such model biases on the spatial characteristics of the anthropogenic climate signal is a general problem in climate modeling that requires further experimentation with improved or flux-corrected CGCMs. However, flux corrections have other disadvantages, such as affecting the stability properties of the coupled atmosphere–ocean system (Neelin and Dijsktra 1995).

In addition to the long-term anthropogenic sea level rise, low-lying islands in the South Pacific experience other climatic threats that may affect the nature of coastal inundations. Among them are changes in the probability of extreme weather events, such as westerly wind bursts, typhoons, and changes of interannual sea level variability associated with the El Niño–Southern Oscillation phenomenon (Yeh et al. 2009). More research needs to be conducted to provide more accurate estimates of projected changes of higher-order statistical moments of sea level in the South Pacific.

Fig. 6.

Schematic illustration of how the projected surface warming affects winds and eventually sea level.

Fig. 6.

Schematic illustration of how the projected surface warming affects winds and eventually sea level.

Acknowledgments

This work was sponsored by the Japan Agency for Marine-Earth Science and Technology (JAMSTEC) through its sponsorship of the International Pacific Research Center, NOAA through Grant NA08OAR4320910, and the Office of Science (BER), U.S. Department of Energy, through Grant DE-FG02-07ER64469.

The altimeter products used in this study were produced by SSALTO/DUACS and distributed by Aviso, with support from CNES (available online at http://www.aviso.oceanobs.com/duacs/). The ECMWF analysis data used in this study are described in detail at ECMWF (available online at http://www.ecmwf.int/products/data/operational_system/) and were sourced from the Asia-Pacific Data-Research Center (APDRC). The original data are available from the Research Data Archive (RDA) at the National Center for Atmospheric Research (available online at http://dss.ucar.edu) in dataset number ds111.1.

REFERENCES

REFERENCES
Balmaseda
,
M.
,
A.
Vidard
, and
D.
Anderson
,
2008
:
ECMWF ocean analysis system: ORA-S3.
Mon. Wea. Rev.
,
136
,
3018
3034
.
Bindoff
,
N. L.
, and
Coauthors
,
2007
:
Observations: Oceanic climate change and sea level.
Climate Change 2007: The Physical Science Basis, S. Solomon et al., eds., Cambridge University Press, 385–432
.
Carton
,
J. A.
,
B. S.
Giese
, and
S. A.
Grodsky
,
2005
:
Sea level rise and the warming of the oceans in the Simple Ocean Data Assimilation (SODA) ocean reanalysis.
J. Geophys. Res.
,
110
,
C09006
.
doi:10.1029/2004JC002817
.
Christensen
,
J. H.
, and
Coauthors
,
2007
:
Regional climate projections.
Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 847–940
.
Church
,
J. A.
, and
N. J.
White
,
2006
:
A 20th century acceleration in global sea-level rise.
Geophys. Res. Lett.
,
33
,
L01602
.
doi:10.1029/2005GL024826
.
Gill
,
A.
,
1980
:
Some simple solutions for heat-induced tropical circulation.
Quart. J. Roy. Meteor. Soc.
,
106
,
447
462
.
Ishii
,
M.
,
M.
Kimoto
,
K.
Sakamoto
, and
S-I.
Iwasaki
,
2006
:
Steric sea level changes estimated from historical ocean subsurface temperature and salinity analyses.
J. Oceanogr.
,
62
,
155
170
.
Kalnay
,
E.
, and
Coauthors
,
1996
:
The NCEP/NCAR 40-Year Reanalysis Project.
Bull. Amer. Meteor. Soc.
,
77
,
437
471
.
Kim
,
H.
,
B.
Wang
, and
Q.
Ding
,
2008
:
The global monsoon variability simulated by CMIP3 coupled climate models.
J. Climate
,
21
,
5271
5294
.
Kleeman
,
R.
,
1991
:
A simple model of the atmospheric response to ENSO sea surface temperature anomalies.
J. Atmos. Sci.
,
38
,
3
18
.
Liu
,
Z.
,
S.
Vavrus
,
F.
He
,
N.
Wen
, and
Y.
Zhong
,
2005
:
Rethinking tropical ocean response to global warming: The enhanced equatorial warming.
J. Climate
,
18
,
4684
4700
.
Lowe
,
J. A.
, and
J. M.
Gregory
,
2006
:
Understanding projections of sea level rise in a Hadley Centre coupled climate model.
J. Geophys. Res.
,
111
,
C11014
.
doi:10.1029/2005JC003421
.
Masumoto
,
Y.
, and
Coauthors
,
2004
:
A fifty-year eddy-resolving simulation of the world ocean: Preliminary outcomes of OFES (OGCM for the Earth Simulator).
J. Earth Simul.
,
1
,
35
56
.
McGregor
,
S.
,
N.
Holbrook
, and
S.
Power
,
2007
:
Interdecadal SST variability in the equatorial Pacific Ocean. Part I: The role of off-equatorial wind stresses and oceanic Rossby waves.
J. Climate
,
20
,
2643
2658
.
Meehl
,
G.
,
C.
Covey
,
K. E.
Taylor
,
T.
Delworth
,
R. J.
Stouffer
,
M.
Latif
,
B.
McAvaney
, and
J. F. B.
Mitchell
,
2007
:
The WCRP CMIP3 multimodel dataset: A new era in climate change research.
Bull. Amer. Meteor. Soc.
,
88
,
1383
1394
.
Neelin
,
J. D.
, and
H. A.
Dijsktra
,
1995
:
Ocean–atmosphere interaction and the tropical climatology. Part I: The dangers of flux correction.
J. Climate
,
8
,
1325
1342
.
Qiu
,
B.
, and
S.
Chen
,
2006
:
Decadal variability in the large-scale sea surface height field of the South Pacific Ocean: Observations and causes.
J. Phys. Oceanogr.
,
36
,
1751
1761
.
Rahmstorf
,
S.
,
2007
:
A Semi-Empirical Approach to Projecting Future Sea-Level Rise.
Science
,
315
,
368
370
.
Seager
,
R.
, and
R.
Murtugudde
,
1997
:
Ocean dynamics, thermocline adjustment, and regulation of tropical SST.
J. Climate
,
10
,
521
534
.
Smith
,
W.
, and
D.
Sandwell
,
1997
:
Global seafloor topography from satellite altimetry and ship depth soundings.
Science
,
277
,
1956
1962
.
Timmermann
,
A.
,
F-F.
Jin
, and
M.
Collins
,
2004
:
Intensification of the annual cycle in the tropical Pacific due to greenhouse warming.
Geophys. Res. Lett.
,
31
,
L12208
.
doi:10.1029/2004GL019442
.
Uppala
,
S. M.
, and
Coauthors
,
2005
:
The ERA-40 Re-Analysis.
Quart. J. Roy. Meteor. Soc.
,
131
,
2961
3012
.
Vecchi
,
G.
, and
B.
Soden
,
2007
:
Global warming and the weakening of the tropical circulation.
J. Climate
,
20
,
4316
4340
.
Vecchi
,
G.
,
B.
Soden
,
A.
Wittenberg
,
I.
Held
,
A.
Leetmaa
, and
M.
Harrison
,
2006
:
Weakening of tropical Pacific atmospheric circulation due to anthropogenic forcing.
Nature
,
441
,
73
76
.
Wyrtki
,
K.
,
1975
:
El Niño—The dynamic response of the equatorial Pacific Ocean to atmospheric forcing.
J. Phys. Oceanogr.
,
5
,
572
584
.
Xie
,
S-P.
,
1996
:
Westward propagation of latitudinal asymmetry in a coupled ocean–atmosphere model.
J. Atmos. Sci.
,
53
,
3236
3250
.
Xie
,
S-P.
,
C.
Deser
,
G.
Vecchi
,
J.
Ma
,
H.
Teng
, and
A.
Wittenberg
,
2010
:
Global warming pattern formation: Sea surface temperature and rainfall.
J. Climate
,
23
,
966
986
.
Yeh
,
S-W.
,
J-S.
Kug
,
B.
Dewitte
,
M-H.
Kwon
,
B.
Kirtman
, and
F-F.
Jin
,
2009
:
El Niño in a changing climate.
Nature
,
461
,
511
514
.

Footnotes

Corresponding author address: Axel Timmermann, IPRC, SOEST, University of Hawaii at Manoa, 2525 Correa Road, Honolulu, HI 96822. Email: axel@hawaii.edu

* International Pacific Research Center Publication Number 686 and School of Ocean and Earth Science and Technology Publication Number 7920.