Role of Climate Feedback in El Niño–Like SST Response to Global Warming

XIAOLIANG SONG AND GUANG J. ZHANG

Scripps Institution of Oceanography, University of California, San Diego, La Jolla, California

(Manuscript received 21 January 2014, in final form 18 May 2014)

ABSTRACT

Under global warming from the doubling of CO2, the equatorial Pacific experiences an El Niño–likewarming, as simulated by most global climate models. A new climate feedback and response analysis method

(CFRAM) is applied to 10 years of hourly output of the slab ocean model (SOM) version of the NCAR

Community Climate System Model, version 3.0, (CCSM3-SOM) to determine the processes responsible for

this warming. Unlike the traditional surface heat budget analysis, the CFRAM can explicitly quantify the

contributions of each radiative climate feedback and of each physical and dynamical process of a GCM to

temperature changes. The mean bias in the sum of partial SST changes due to each feedback derived with

CFRAM in the tropical Pacific is negligible (0.5%) compared to themean SST change from the CCSM3-SOM

simulations, with a spatial pattern correlation of 0.97 between the two. The analysis shows that the factors

contributing to the El Niño–like SST warming in the central Pacific are different from those in the eastern

Pacific. In the central Pacific, the largest contributor to El Niño–like SST warming is dynamical advection,

followed by PBL diffusion, water vapor feedback, and surface evaporation. In contrast, in the eastern Pacific

the dominant contributor to El Niño–like SST warming is cloud feedback, with water vapor feedback further

amplifying the warming.

1. Introduction

The equatorial Pacific sea surface temperature (SST)

changes associated with natural climate variability [e.g.,

El Niño–Southern Oscillation (ENSO)] affect not only

tropical cyclone activity, freshwater supplies, agricul-

ture, and ecosystems, but also the global patterns of

flood and drought and climate extremes worldwide

through atmospheric teleconnections (see review by

McPhaden et al. 2006; Alexander et al. 2002). This in-

dicates that the changes in equatorial Pacific SST are

highly influential for global climate. Therefore, an ac-

curate projection of the equatorial Pacific SST in re-

sponse to increased greenhouses gases is fundamentally

important to improving the projection of future global

and regional climate change. This is not only because the

changes in the background state of the equatorial Pacific

may influence the regional climate in many parts of the

world through atmospheric teleconnections, as illus-

trated by observed climate variability, but also because

they could influence the amplitude, frequency, spatial

pattern, and impacts of ENSO (Timmermann et al. 1999;

Fedorov and Philander 2000; Koutavas et al. 2002; Vecchi

andWittenberg 2010; Collins et al. 2010; McGregor et al.

2013).

How will the equatorial Pacific respond to increased

greenhouse gases?Will the pattern of SST changesmore

closely resemble an El Niño or a La Niña? This has beendebated for over a decade (Vecchi et al. 2008). Although

the majority of global climate models (GCMs) exhibit

an El Niño–like warming in the equatorial Pacific,

namely stronger warming in the central and eastern

equatorial Pacific than in the western equatorial Pacific

(Knutson and Manabe 1995; Meehl and Washington

1996; Cubasch et al. 2001; Yu and Boer 2002; Meehl

et al. 2007; Vecchi and Soden 2007; An et al. 2012; Yeh

et al. 2012; Lee and Wang 2014), there are also models

that project a La Niña–like warming, with the western

equatorial Pacific SST warming more than the central

and eastern Pacific (Liu et al. 2005; An et al. 2012), or no

significant trend toward either El Niño– or La Niña–likeconditions but an enhanced equatorial warming relative

to the subtropics (Collins 2005; Liu et al. 2005; DiNezio

et al. 2009). Note that some recent studies (DiNezio

et al. 2010; Collins et al. 2010; Xie et al. 2010) suggested

Corresponding author address: Dr. Xiaoliang Song, Scripps In-

stitution of Oceanography, 9500 Gilman Dr., La Jolla, CA 92093-

0221.

E-mail: xisong@ucsd.edu

1 OCTOBER 2014 SONG AND ZHANG 7301

DOI: 10.1175/JCLI-D-14-00072.1

� 2014 American Meteorological Society

that ‘‘El Niño (La Niña)–like climate change’’ is not an

appropriate expression because of a number of differ-

ences between ENSO events and the tropical Pacific

response to global warming, such as thermocline change,

seasonality, and atmospheric teleconnections, among

others. In this study, we use the term ‘‘El Niño–likewarming’’ only to denote the pattern ofmean SST change

in the equatorial Pacific, which resembles a present-day

ElNiño event. Currently it is still a challenge to accuratelydetermine the twentieth-century trend in the zonal SSTgradient across the equatorial Pacific from observations/reconstructions because of uncertainties related to chang-ing measurement techniques and analysis procedures(Deser et al. 2010). Recently Tokinaga et al. (2012a)

reconstructed the SST trend pattern for the period 1950–

2009 using only bucket measurements from the In-

ternational Comprehensive Ocean–Atmosphere Data

Set (ICOADS) and nighttime marine surface air tem-

perature. They found that the zonal SST gradient be-

tween western and eastern Pacific is reduced, physically

consistent with subsurface temperature observations.

Several mechanisms have been proposed to explain

the formation of El Niño–like warming in response to

increased greenhouse gases. First, the weakening of the

Walker circulation and associated Bjerknes (1969)

feedback was suggested (Meehl and Washington 1996;

Yu and Boer 2002). Observational (Zhang and Song

2006; Vecchi et al. 2006; Tokinaga et al. 2012b), theo-

retical (Held and Soden 2006; Tokinaga et al. 2012a;

Knutson and Manabe 1995), and modeling work

(Vecchi and Soden 2007; Meehl et al. 2007) all indicate

a weakening of the Walker circulation in response to

global warming. The associated weakening of the

equatorial trade winds may lead to a reduction in the

upwelling of cold water in the eastern Pacific, resulting

in a decrease of zonal SST gradient across the equatorial

Pacific, which in turn may further weaken the Walker

circulation and trade winds (Vecchi and Soden 2007;

Meehl et al. 2007). The second mechanism is evapora-

tion cooling (Knutson and Manabe 1995). Because of

the nonlinear dependence of saturation mixing ratio on

temperature through the Clausius–Clapeyron relation,

the sea surface with higher temperature would produce

more evaporative cooling in response to the same tem-

perature perturbation. Thus, there will be stronger

evaporative cooling in the western Pacific warm pool

than the eastern Pacific cold tongue, which can damp the

zonal SST gradient and result in El Niño–like warming

(Knutson and Manabe 1995). This evaporative cooling

argument was first proposed by Newell (1979) to explain

the regulation of the warm pool SST. However, it is not

supported by observations, which show low surface

evaporation in the western Pacific warm pool region

(Zhang and McPhaden 1995; Zhang et al. 1995; Zhang

and Grossman 1996). Meehl and Washington (1996)

showed stronger evaporative cooling in the eastern Pa-

cific than in the western Pacific in a GCM due to larger

SST increase in the east. In other words, evaporative

cooling may act to enhance the zonal SST gradient in-

stead of damping it. The third mechanism is cloud

feedback. Ramanathan and Collins (1991) suggested

that the increased highly reflective cirrus cloud associ-

ated with enhanced deep convection over the western

Pacific warm pool as SST increases can block out more

incoming solar radiation. This negative shortwave cloud

feedback over the warm pool can lead to a reduction in

east–west SST gradient across the equatorial Pacific. In

addition, the decrease in low-level stratus clouds in the

eastern Pacific as SSTs increase allows more incoming

solar radiation to reach the surface. This positive cloud

feedback in the eastern Pacific can also contribute to the

reduction in zonal SST gradient across the equatorial

Pacific (Meehl andWashington 1996). However, Yu and

Boer (2002) showed that bothElNiño–like warming and

negative shortwave cloud feedback in central and east-

ern Pacific occur in the Canadian Centre for Climate

Modelling andAnalysis (CCCma) coupled GCM, which

led them to conclude that cloud feedback is not a sig-

nificant factor in the El Niño–like warming response.

The role of ocean dynamical processes has also been

investigated. The SST is primarily regulated by ocean

dynamical thermostat (cold water upwelling) in the

eastern Pacific cold tongue (Clement et al. 1996; Cane

et al. 1997). Since the increased downward radiative flux

associated with a doubling of CO2 in the cold tongue

region may enhance the vertical ocean temperature

gradient and hence cold water upwelling, the weaker

warming in the eastern Pacific than in thewestern Pacific

will strengthen the east–west SST gradient across the

equatorial Pacific, which drives stronger Walker circu-

lation and surface easterly winds, and then results in

stronger upwelling cooling in the eastern Pacific. This in

turn further reinforces the zonal SST gradient and

Walker circulation, indicating that the ocean dynamical

process acts to enhance the zonal SST gradient. This

ocean-dynamics-triggered Bjerknes feedback has been

used to explain how a La Niña–like response can occur

(Cane et al. 1997; Vecchi et al. 2008). The comparison of

the tropical Pacific response to increased CO2 in 13

GCMs that participated in phase 3 of the Coupled

Model Intercomparison Project (CMIP3) of the World

Climate Research Programme shows that when the

model uses a nondynamic slab ocean component, the

equatorial Pacific exhibits an El Niño–like warming re-

sponse, whereas when a full dynamical ocean compo-

nent is used in the same model, the ‘‘El Niño-ness’’ of

7302 JOURNAL OF CL IMATE VOLUME 27

the SST response is diminished (Vecchi et al. 2008). This

demonstrates that the ocean dynamics indeed acts to

strengthen the east–west SST gradient and thus damps

the El Niño–like response, suggesting that atmospheric

processes are the drivers of the formation of El Niño–like warming. Therefore, although equatorial Pacific

SST variability is largely influenced by ocean dynamics,

the slab ocean model version of climate model is still

a good tool to help us understand the responsible

mechanisms of the El Niño–like SST warming.

Because the Bjerknes feedback is excluded in a non-

dynamic slab ocean model configuration, the appear-

ance of an El Niño–like warming response suggests that

the Bjerknes feedback from a weakened Walker circu-

lation is not a dominant mechanism for such a warming

response. The role of cloud feedback inmostmechanism

analyses is determined by cloud radiative forcing.

However, the changes in noncloud climate components

may affect (or contaminate) the change in cloud radia-

tive forcing, which may even change the sign of global

mean cloud feedback (Shell et al. 2008; Soden et al.

2008). Therefore, the contribution of cloud feedback to

El Niño–like warming is still not very clear. Previous

studies suggest that the ocean surface evaporation can

modulate the east–west SST gradient in the equatorial

Pacific. The changes in surface evaporation and circula-

tion may affect the moisture content in the atmosphere.

Can the associated water vapor feedback influence the

SST response in the equatorial Pacific? In addition, when

the Walker circulation is weakened, the intensity and

location of convection will change correspondingly. Will

the changes in temperature lapse rate affect the zonal

gradient of surface warming in the equatorial Pacific? All

these questions indicate that our understanding on the

mechanisms that govern the equatorial SST response to

increased greenhouse gases is still incomplete.

The SST changes are usually analyzed using the heat

budget equation of ocean mixed layer. Although the

contributions of surface turbulent and radiative heat

fluxes and ocean heat transport can be explicitly de-

termined, this method cannot isolate the contributions

of each radiative climate feedback (e.g., water vapor

feedback and lapse rate feedback) or the contributions

of each physical and dynamical process of GCMs. The

traditional climate feedback analysis methods (e.g.,

partial radiative perturbation method; Wetherald and

Manabe 1988) focus on the feedback parameter, the

ratio of the radiative perturbation at the top of the at-

mosphere due to a specific feedback to the total change

of the surface temperature, and do not provide a direct

estimate of surface temperature change caused by any

individual feedback. Therefore, it is difficult to directly

use these methods to quantify the contribution of

climate feedback to SST changes. Recently, Lu and Cai

(2009) formulated a coupled atmosphere–surface cli-

mate feedback and response analysis method

(CFRAM). This method can explicitly quantify the

contributions of both radiative and nonradiative climate

feedbacks to temperature changes. The decomposition

of nonradiative feedback is based on the GCM’s physi-

cal and dynamical processes so that the roles of GCM

physical and dynamical processes in climate change can

be easily understood. Song et al. (2014) applied the

CFRAM to 10 years of hourly output of the slab ocean

model (SOM) version of the National Center of Atmo-

spheric Research (NCAR) Community Climate System

Model, version 3.0, (CCSM3-SOM) to quantify the

contributions of climate feedbacks to global pattern of

surface warming with particular interests in the warming

contrast between land and ocean and between high and

low latitudes in response to a doubling of CO2. They

found that the CFRAM analysis with hourly model out-

put is highly accurate for quantifying the role of climate

feedbacks in the spatial distribution of global and re-

gional warming projected by GCMs. In this study, we use

the same method and data to understand the equatorial

Pacific SST response to a doubling of CO2 concentration.

The paper is organized as follows. Section 2 briefly

describes the model and experiment design, and the

procedure to apply the CFRAM to the NCAR CCSM3-

SOM simulations. Section 3 characterizes the El Niño–like response of the model in the equatorial Pacific. The

contributions of each feedback process to the equatorial

Pacific SST changes are examined in the section 4. A

summary and conclusions are then given in section 5.

2. Model setup and analysis approach

The slab ocean model version of the NCAR CCSM3.0

(Collins et al. 2006a) is used in this study. This configu-

ration allows for a fully interactive surface exchange be-

tween atmosphere model and ocean model. Therefore,

ocean temperature is a prognostic variable. However, the

horizontal ocean heat transport and deep-water heat

exchange are specified by an internal ocean mixed layer

heat flux (Q flux) to simplify the ocean model. The

thermodynamic sea ice component of the Community

Sea Ice Model (CSIM) is employed in this configuration

to calculate the sea ice concentration and thickness, sur-

face albedo, and surface exchange flux. The atmosphere

model component, the Community Atmosphere Model,

version 3 (CAM3; Collins et al. 2006b), is a global at-

mospheric general circulation model with T42 spectral

truncation (approximately 2.88 3 2.88 latitude/longitude)in the horizontal and 26 levels in the vertical from the

surface to 2.917hPa. The land surface model component

1 OCTOBER 2014 SONG AND ZHANG 7303

is the Community Land Model (CLM), which uses the

same horizontal grids as the atmospheric model.

The data from the same two 50-yr equilibrium simula-

tions as those in Song et al. (2014) are used in this study.

The present-day climate simulation (referred to as the

13CO2 run) uses the 1990 CO2 value of 355ppmv,

whereas the perturbation climate simulation (referred to

as the 23CO2 run) uses the doubled CO2 concentration of

710ppmv. Both simulations start from the same initial

conditions obtained from the NCAR repository. The

spatially varying oceanmixed layer depth is specified from

Levitus et al. (1998). The monthly mean distribution of

internal ocean mixed layer heat flux (Q flux) is derived

from the net ocean surface energy budget of 10 years

(1981–90) ofAtmosphericModel Intercomparison Project

(AMIP)-type simulation of the CAM3. The model attains

equilibrium with the doubled CO2 concentration after

25 years of time integration. Thus, the climate statistics and

climate feedback analysis presented in this paper are based

on hourly output from years 30 to 39 of the simulations.

The CFRAM analysis is applied to the CCSM3-SOM

model output as described in Song et al. (2014) to de-

termine the contributions of CO2 forcing and individual

feedback to the temperature changes. For application

details readers are referred to Song et al. (2014). The

CFRAM is an offline postprocessing diagnostic tool

based on the energy balance in an atmosphere–surface

column. Any dynamical, thermodynamic, or radiative

process that influences the local energy budget is con-

sidered a feedback process. The energy perturbation

due to any dynamical or physical process between the

13CO2 and 23CO2 runs is determined by the difference

in the corresponding energy flux convergence, which is

calculated by the CCSM3-SOM at each time step during

the model integration. The radiative energy perturba-

tion can be further decomposed into perturbations due

to the doubling of CO2, temperature change, water va-

por, cloud, and albedo feedbacks, which is accomplished

by a series of offline radiation calculations using radia-

tion code and hourly output from CCSM3-SOM with

a two-sided partial radiative perturbation method (Song

et al. 2014). Since the temperature change only influences

the longwave radiation, the partial temperature change

due to any individual feedback process is derived by re-

quiring the infrared radiation induced by the temperature

change alone to exactly balance the energy perturbation

due to the feedback process that is considered:

DTx5

�›R

›T

�21

DQx , (1)

where D stands for the difference between the two equi-

librium states; T5 (T1, T2, . . . , Tsurface )Tis the vector

of temperature from the surface to the top level of

the atmospheric model (TOM); and the superscript T

denotes transpose. Also, R5 (R1, R2, . . . , Rsurface )Tis

the vector of vertical divergence of net longwave radi-

ation fluxes. The Planck feedback matrix ›R/›T is de-

termined by the two-sided PRP calculation with 1-K

temperature perturbation from each layer. In addition,

Qx 5 (Qx1, Q

x2, . . . , Q

xsurface )

Tis the vector of energy

flux convergence associated with feedback process x.

Here the superscript x is substituted for different feed-

back processes, including conv for atmospheric convec-

tion, cond for large-scale condensation, ocn for ocean

process, shf for surface sensible heat flux, lhf for surface

latent heat flux, pbl for boundary layer process, gwd for

gravity wave drag, adv for dynamical advection process,

CO2 for the external forcing (doubling of CO2), wv

for water vapor feedback, cld for cloud feedback, and

albedo for albedo feedback.

3. El Niño–like warming in the CCSM3-SOM

Figure 1a displays the geographical distributions of

annual mean SST in the tropical Pacific from the

present-day climate simulation. The SST in the tropical

western Pacific is higher than in the tropical eastern

Pacific. In the equatorial Pacific, a narrow latitudinal

band of cold SSTs (cold tongue) appears in the central

and eastern Pacific, while the SST warm pool (SST .288C) is located west of 1708W. The SST changes due to

a doubling of CO2 from the CCSM3-SOM simulation

output (shadings) and the sum of the partial tempera-

ture changes due to external forcing and each feedback

process (contours) derived using Eq. (1) are shown in

Fig. 1b. The CFRAM analysis successfully reproduces

the SST change of the CCSM3-SOM in the tropical

Pacific. The mean bias in the sum of the partial SST

change in the tropical Pacific is 0.009K, which is negli-

gible (0.5%) compared to the mean SST change from

the CCSM3-SOM simulations (1.69K). The spatial

pattern correlation between the two is 0.97. The high

accuracy of the CFRAM analysis enables us to quantify

contributions of climate feedbacks to SST warming in

the tropical Pacific. The SST response in the equatorial

Pacific exhibits significant spatial variation with strong

warming up to 2.75K in the central and eastern Pacific

and weak warming (typically ,1.75K) in the western

Pacific. This warming pattern reduces the mean east–

west SST gradient in the equatorial Pacific. Figure 1c

further shows the longitudinal distribution of SST

change from CCSM3-SOM simulation and CFRAM

analysis over the equatorial Pacific (58S–58N). It is clear

that the CFRAM analysis successfully reproduces the

SST warming pattern of CCSM3-SOM in the equatorial

7304 JOURNAL OF CL IMATE VOLUME 27

Pacific. The SST warming in the equatorial Pacific has

a minimum in the western Pacific warm pool between

1608E and 1708W, and gradually increases eastward to

its maximum in eastern Pacific between 808 and 1008W.

The SST warming difference between the Niño-3 region(58S–58N, 908–1508W; 2.08K) and equatorial western

Pacific (58S–58N, 1408E–1708W; 1.57K) is about 0.51K,

with the largest warming gradient in the central Pacific.

Since one of the widely used El Niño definitions is thatthe SST anomalies of the Niño-3 region are 10.5K or

more for at least six consecutive months (Trenberth

1997), this suggests that the annual mean SST response

to a doubling of CO2 in the slab ocean model version of

CCSM3.0 resembles the reduction of SST gradient

across the equatorial Pacific during the El Niño event,indicating that the Pacific SST has a tendency towarda more El Niño–like condition as the climate warms.

Associated with the SST changes, the atmospheric

convection (precipitation) and circulation also change

correspondingly. In the control climate, the warm SST

drives deep convection and strong rising motion in the

western Pacific (Figs. 2a and 3a). The associated di-

vergent flow in the upper troposphere transports air

eastward to central and eastern Pacific in the equatorial

plane, which descends to produce a dry and high pres-

sure region over the eastern Pacific cold tongue. The sea

level pressure gradient between the high pressure over

eastern Pacific and the low pressure over western Pacific

FIG. 1. (a) Annual mean sea surface temperature (SST) distribution in present-day climate

simulation (13CO2), (b) annual mean SST change from 13CO2 to 23CO2 runs simulated by

CCSM3.0 (shadings) and the sum of partial SST change due to each feedback derived from

CFRAM analysis (contours), and (c) the longitudinal distribution of SST changes averaged

over the equatorial Pacific (58S–58N).

1 OCTOBER 2014 SONG AND ZHANG 7305

drives the low-level easterlies, closing the Walker cir-

culation loop (Figs. 2a,c). When the climate warms, the

warm SST extends eastward and convection shifts to the

central and eastern Pacific, resulting in significant pre-

cipitation increase in the central Pacific and the coastal

region of eastern Pacific (Fig. 3b). Correspondingly the

anomalous ascending motion occurs over the central

and eastern Pacific and anomalous descending motion

appears over the western Pacific Maritime Continent

(Fig. 2b). Meanwhile, the sea level pressure is increased

over the central and western Pacific and decreased over

the eastern Pacific (Fig. 2d). The weakened sea level

pressure gradient between the eastern and western Pacific

results in a reduction of low-level easterlies (Fig. 2b). The

equatorial westerly in the upper troposphere over central

and eastern Pacific is also decreased. All these changes in

the equatorial atmospheric circulation tend to oppose the

backgroundWalker circulation, slowing down theWalker

circulation in the warmer climate. Since the weakening of

easterly trade winds and the Walker circulation, the

eastward shift of convection over the central equatorial

Pacific, and the reduction in SST and SLP gradients across

the equatorial Pacific are all major characteristics of El

Niño, it suggests that the tropical circulation response re-sembles an El Niño in the CCSM3-SOM.

4. Contribution of climate feedbacks

The partial SST change due to cloud feedback aver-

aged over the 58S–58N latitude belt is shown in Fig. 4a.

The net cloud feedback produces SST cooling up to

3.5K west of 1058W, and SST warming up to 5K east of

1058W, indicating a negative net cloud feedback over the

western and central Pacific and positive cloud feedback

over the eastern Pacific. The negative net cloud feed-

back produces a stronger SST cooling in the central

equatorial Pacific than the western Pacific; however, the

SST change in CCSM3-SOM shows a greater SST

warming in the central Pacific than the western Pacific.

This indicates that the cloud feedback tends to damp the

El Niño–like warming in the central Pacific and only

contributes to the El Niño–like warming in the eastern

Pacific east of 1058W. The cloud feedback can be further

decomposed into shortwave cloud (cloud albedo) feed-

back and longwave cloud (cloud greenhouse) feedback.

The shortwave cloud feedback shows the same pattern

as that of net cloud feedback but with a slightly stronger

magnitude, indicating that the cloud feedback is domi-

nated by shortwave feedback and that the longwave

feedback tends to damp the impact of shortwave cloud

feedback. The shortwave cloud feedback produces SST

FIG. 2. (a) Longitude–pressure cross section of annualmean zonalwind (shadings;ms21) andwind vectors with vertical

velocity (0.5mbday21; 1mb5 1hPa) averaged between 58S and 58N in the present-day climate simulation. (b) As in (a),

but for the difference between 23CO2 and 13CO2. (c) Annual mean sea level pressure (hPa) averaged between 58S and

58N in the present-day climate simulation. (d) As in (c), but for the difference between 23CO2 and 13CO2.

7306 JOURNAL OF CL IMATE VOLUME 27

warming up to 6K in the eastern Pacific east of 1058W. It

can be largely attributed to the decrease in low-level

stratus cloud fraction (Fig. 4b) as a result of increased

deep convection in a warmer climate, which allows more

solar radiation into the surface. Although the increase of

high-level cloud fraction (Fig. 4d) associated with en-

hanced deep convection may block more incoming solar

radiation, the total cloud liquid water path is decreased

markedly (Fig. 4c) because of the decrease in low-level

cloud, leading to more solar radiation reaching the sur-

face. Both regional-mean cloud liquid water path and

high-level cloud fraction are increased in the western

and central equatorial Pacific because of more active

convection in a warmer climate. However, the increases

are larger over the central Pacific than western Pacific

resulting from the eastward shift of convection, which

blocks more incoming solar radiation and hence results

in stronger SST cooling in the central Pacific than

western Pacific. The SST change due to longwave cloud

feedback is much smaller than that induced by short-

wave cloud feedback. It is generally positive in the

western and central Pacific but becomes negative in the

eastern Pacific east of 1058W, indicating a reduced

contribution from longwave cloud feedback to the El

Niño–like warming in the eastern Pacific. Clouds absorb

upward longwave radiation from the surface and re-

emit it at their local temperature. Although clouds at any

height emit downward longwave radiation and warm the

surface, low-level clouds are more efficient because of

their closer proximity to the surface and higher temper-

ature than high-level clouds. In the western and central

Pacific, the low-level cloud fraction is increased (Fig. 4b)

so that it can absorb and re-emitmore longwave radiation

to the surface, resulting in a positive longwave cloud

feedback and surface warming. In the eastern Pacific east

of 1058W, the decrease of low-level cloud leads to less

downward longwave radiation onto the surface, resulting

in a negative longwave cloud feedback and surface

cooling.

Water vapor is the dominant greenhouse gas. The

positive water vapor feedback is the largest contributor

to global mean surface temperature increase in response

to a doubling of CO2 in the CCSM3-SOM (Song et al.

2014). Over the equatorial Pacific, the water vapor

feedback still plays a pivotal role in determining the

regional mean SST response. The mean SST change

over the equatorial Pacific due to water vapor feedback

is as high as 4.3K, owing to the abundant water vapor

supply in the tropics. The water vapor feedback (Fig. 5a)

also shows distinct longitudinal variation. The stronger

positive water vapor feedback produces larger SST

warming in the central and eastern equatorial Pacific

than the western Pacific, indicating a positive contribu-

tion to the El Niño–like warming pattern. The strength

of the water vapor feedback depends on changes of

moisture content in the atmosphere. Figure 5b shows the

FIG. 3. (a) Annual mean precipitation (mmday21) distribution in present-day climate simu-

lation and (b) annual mean precipitation (mmday21) change from the 13CO2 to 23CO2 run.

1 OCTOBER 2014 SONG AND ZHANG 7307

FIG. 4. (a) Partial SST change (K) due to cloud feedback, (b) change in low-level cloud

fraction (%), (c) change in cloud liquid water path (gm22), and (d) change in high-level cloud

fraction (%) from 1 3 CO2 to 2 3 CO2 conditions averaged between 58S and 58N.

7308 JOURNAL OF CL IMATE VOLUME 27

longitude–pressure cross section of specific humidity

changes averaged between 58S and 58N. It is clear that

the model simulates more increase in moisture over the

central and eastern Pacific than the western Pacific. The

stronger moistening of middle troposphere over the cen-

tral and eastern Pacific ismainly as a result of the eastward

shift of convection (Fig. 3b) and associated weakening of

the Walker circulation (Fig. 2b). The enhanced con-

vection and anomalous rising motion of the Walker

circulation over the central and eastern Pacific transport

more moisture upward. In the lower troposphere, the

weakened easterly (Fig. 2b, shadings) transports less

moisture from the central Pacific to the western Pacific,

resulting in more increase of moisture over the central

Pacific than the western Pacific. Over the eastern Pacific

east of 1008W, the change of wind speed in the lower

troposphere is relatively small compared to the central

Pacific. However, the much stronger ocean evaporation

(Fig. 6b) associated with more intense SST warming

(Fig. 1b) still supplies more moisture in the near-surface

atmosphere. In short, the changes in convection,Walker

circulation, and surface evaporation in response to a dou-

bling of CO2 produce stronger moistening of the tropo-

sphere over the central and eastern equatorial Pacific than

over the western Pacific. The associated water vapor

feedback results in greater SST warming there, leading to

an El Niño–like SST response. Similar changes in

moisture and water vapor feedback are also observed in

the 1987 ENSO warming from La Niña to El Niñoconditions (Soden 1997). Note that in Fig. 4b the low-

level cloud is decreased in the eastern Pacific east of

1058W even though specific humidity is increased in the

lower troposphere. In the CCSM3-SOM, the convective

and layered cloud fractions depend on convective mass

flux and relative humidity, respectively. The marine

stratocumulus clouds are diagnosed using an empirical

relationship between marine stratus cloud fraction and

lower tropospheric static stability defined by the differ-

ence in potential temperature between 700 hPa and the

surface. We found that the convective cloud fraction is

increased below 700 hPa over the eastern Pacific, cor-

responding to the enhanced convection (Fig. 3b). The

lower tropospheric static stability is increased slightly

but the relatively humidity is decreased by up to 3% in

the lower troposphere over eastern Pacific, indicating

that the decrease in low-level cloud fraction over the

eastern Pacific is caused by the reduction of relative

humidity.

The ocean surface latent heat flux affects SST in two

ways. First, it cools the ocean surface as a result of energy

loss from the ocean through evaporation. Second, it heats

the ocean surface through enhanced downward infrared

radiation because the atmosphere becomes warmer by

taking up energy from the ocean surface. The net effect of

latent heat flux in the CFRAM analysis is thus smaller

than that in conventional surface energy budget analyses

that only consider evaporative cooling. The partial SST

change due to latent heat flux averaged over the 58S–58Nlatitude belt is shown in Fig. 6a. It is generally positive in

the central Pacific and negative in thewestern and eastern

Pacific, meaning that the change in latent heat flux am-

plifies the SSTwarming in the central Pacific but damps it

in the western Pacific. Therefore, the latent heat flux can

contribute to theElNiño–like SSTwarming in the central

and western Pacific.

Figure 6b shows the surface latent heat flux changes.

There is a decrease of latent heat flux (evaporation)

from the sea surface in the central equatorial Pacific, and

an increase in the western and eastern Pacific. Because

evaporation depends on surface wind speed, SST, and

near-surface humidity, we can qualitatively understand

the pattern of latent heat flux change by analyzing the

changes of these factors. The SST warming along the

equator exhibits a maximum in the eastern Pacific and

a minimum in the western Pacific. Consequently the

model shows a larger enhancement of evaporation cool-

ing in the eastern Pacific than the western Pacific. In the

central Pacific where the surface easterly is reduced the

most, the evaporation is weakened mainly because of

FIG. 5. (a) Partial SST change (K) due to water vapor feedback

and (b) the longitude–pressure cross section of specific humidity

changes (g kg21) from 1 3 CO2 to 2 3 CO2 conditions averaged

between 58S and 58N.

1 OCTOBER 2014 SONG AND ZHANG 7309

lower wind speed, although SST is increased moderately.

Note that the central Pacific is also the region where

convection is increased the most. This negative relation-

ship between convection and surface evaporation is also

reported in the observational study of Zhang et al. (1995),

which shows that increased convection weakens the sur-

facewind through interactionwith large-scale circulation,

leading to a decrease in evaporation.

Although the nonradiative atmospheric processes do

not directly affect the surface energy budget and hence

SST, they can influence the atmospheric temperature

change. The associated changes in downward infrared

radiation to the surfacemay affect SST change. The total

effects of nonradiative atmospheric processes represent

the lapse rate feedback in the traditional radiative

feedback analysis. Therefore both surface and atmo-

spheric partial temperature changes due to nonradiative

atmospheric processes are analyzed to understand their

roles in the El Niño–like SSTwarming. Figure 7 presents

the partial temperature change resulting from convec-

tion averaged over 58S–58N in the atmosphere and at the

surface. There is large warming in the upper tropo-

sphere and cooling in the lower troposphere. Convec-

tion transports moist static energy upward, producing

a net convergence in the upper troposphere and a net

divergence in the lower troposphere. Thus, the enhanced

convection in the warmer climate results in a warming

perturbation in the upper troposphere and cooling per-

turbation in the lower troposphere, respectively. The

longitudinal distribution of partial temperature change is

consistent with that of precipitation (convection) change,

showing large decreases in temperature lapse rate in the

central Pacific between 1208 and 1708W and eastern Pa-

cific east of 1058W. The decrease of temperature lapse

rate means that more infrared radiation is emitted into

the space and less downward infrared radiation at the

surface; therefore the enhanced convection tends to cool

the surface. The cooling effect of convection on SST is

proportional to the magnitude of convection strength-

ening under 2 3 CO2 conditions. The more enhanced

convection leads to stronger SST cooling in the central

and eastern Pacific than in the western Pacific. Thus, it

tends to damp the El Niño–like warming.

The atmospheric dynamical process tends to damp the

energy perturbation induced by convection through

changing dynamical advection. Figure 8a shows the sum

of changes in adiabatic cooling (av), and vertical ad-

vection of sensible energy (2cpv›T/›p) and latent en-

ergy (2Lv›q/›p). Herev denotes the gridmean vertical

velocity in pressure coordinate p, a is specific volume

FIG. 6. (a) Partial SST change (K) resulting from surface latent heat flux change, and (b) the

change of surface latent heat flux (Wm22) from 1 3 CO2 to 2 3 CO2 conditions averaged

between 58S and 58N.

7310 JOURNAL OF CL IMATE VOLUME 27

(the inverse of the density), cp is specific heat at constant

pressure, L is latent heat of vaporization, and T and q

represent temperature and specific humidity, respec-

tively. The enhanced upward motion (Fig. 2b) and ver-

tical gradient of latent energy (as indicated by moisture

change in Fig. 5b) result in a positive energy advection in

the equatorial troposphere (not shown). Meanwhile the

adiabatic cooling term, which increases with height, is

also strengthened as a result of enhanced upwardmotion.

The net effect of changes in adiabatic cooling and vertical

advection of sensible and latent energy is positive energy

perturbation in the lower troposphere and negative per-

turbation in the upper troposphere. It generally explains

the pattern of temperature change associated to dynam-

ical advection in equatorial troposphere, namely warm-

ing in the lower troposphere and cooling in the upper

troposphere. The consequent increase of temperature

lapse rate leads to less emission of infrared radiation to

the space and more downward infrared radiation at the

surface, and thereby temperature warming at the surface.

The longitudinal distribution of the partial temperature

change associatedwith dynamical advection shows strong

SST warming in the central Pacific and relatively weak

warming in the western and eastern Pacific (Fig. 8c). This

suggests that the dynamical advection contributes to the

El Niño–like warming in the central Pacific.

The vertical mixing in the planetary boundary layer

(PBL) acts to transport energy from the surface upward

to the upper PBL. This would heat and moisten the

upper PBL and cool and dry the near surface layer. The

PBL is influenced by both upward energy flux from

the surface below and atmospheric convection. The sur-

face flux provides energy to the PBL and tends to

deepen the PBLwith stronger surface heating. The deep

convective updrafts act to vent mass/energy out of the

PBL, and the convective downdrafts act to dry and cool

the PBL. In addition, compensating subsidence around

convection in the free troposphere acts to dry the air

above the PBL height. Therefore, the stronger convec-

tion tends to reduce the height of the PBL. The PBL

change in a warmer climate can be approximately mea-

sured by the PBLheight. Themean change of PBL height

FIG. 7. Partial (a) atmospheric temperature (K) and (b) SST (K) changes due to convection

from 1 3 CO2 to 2 3 CO2 conditions averaged between 5S8 and 58N.

1 OCTOBER 2014 SONG AND ZHANG 7311

between 58S and 58N (Fig. 9a, solid) shows a decrease in

the western to central Pacific, withmaximum reduction in

the central Pacific, and an increase in the eastern Pacific

east of 958W. It generally correlates with the change of

convective precipitation (Fig. 9a, dotted) negatively in

the western and central Pacific, with the maximum de-

crease of the PBL height corresponding to the maximum

increase of convective precipitation in the central Pacific.

FIG. 8. (a) Sumof changes in adiabatic cooling andvertical advection of sensible and latent energy

(Wkg21), and partial (b) atmospheric temperature (K) and (c) SST (K) changes resulting from

dynamical advection, from 1 3 CO2 to 2 3 CO2 conditions averaged between 5S8 and 58N.

7312 JOURNAL OF CL IMATE VOLUME 27

In the eastern Pacific east of 958W, the convective pre-

cipitation is increased. However, the PBL height is not

decreased correspondingly. The increase of PBL height

can be largely attributed to the strong increase of surface

latent heat flux (Fig. 6b). The reduced vertical mixing in

the PBL transports less energy upward, resulting in

cooling perturbation in the upper PBL and heating per-

turbation in the layer near the surface in awarmer climate

(Fig. 9b). Consequently, the increase in temperature lapse

rate leads to less emission of infrared radiation upward and

more downward infrared radiation at the surface, and

thereby SST warming. In contrast, the enhanced PBL ver-

tical mixing over the eastern Pacific transports more energy

upward, leading to anomalous heating in the upper PBL

and anomalous cooling in the layer near the surface. The

consequent decrease in temperature lapse rate results in

SST cooling. Since the largest weakening of PBL process

occurs in the central Pacific, there is a stronger SST

FIG. 9. (a) PBL height (40m) and convective precipitation (mmday21) changes, and partial

(b) atmospheric and (c) sea surface temperature (K) changes resulting from PBL processes,

from 1 3 CO2 to 2 3 CO2 conditions averaged between 5S8 and 58N.

1 OCTOBER 2014 SONG AND ZHANG 7313

warming in the central Pacific than the western Pacific

(Fig. 9c). This suggests that the change in PBL processes

contributes to the El Niño–like warming in the cen-

tral Pacific. Because the cloud base of most of tropical

oceanic convection is located in the layer near the sur-

face, the anomalous heating of the layer near surface in

a warmer climate over the central Pacific explains the

enhanced convection there.

The contributions to El Niño–like SST warming from

other processes, such as albedo feedback, large-scale

condensation, sensible heat flux, gravity wave drag, CO2

forcing, and ocean processes, are relatively small and

will not be discussed in detail here. To quantify the

contribution of each feedback process to the El Niño–like SST warming in the equatorial Pacific, the re-

gionally averaged partial SST changes due to each

feedback process in the western Pacific (1408E–1708W),

central Pacific (1708–1108W), and eastern Pacific (1108–808W) and their differences are shown in Fig. 10. The

corresponding values of warming/cooling contribution

from each feedback are provided in Table 1. The major

players in the equatorial Pacific warming are CO2 forc-

ing, cloud radiative feedback, water vapor feedback,

convection, dynamical advection, PBL vertical diffu-

sion, and surface latent heat flux. However, their relative

importance in the western, central, and eastern Pacific

varies from region to region. Relative to the western

Pacific, changes in dynamical advection produce the

largest positive SST warming (1.46K) in the central Pa-

cific. The warming from PBL process and water vapor

feedback also has significant contributions (0.86 and

0.78K respectively). Latent heat flux further amplifies the

SST warming difference by 0.32K. These four processes

together accounts for a warming difference of 3.4K be-

tween the central and western Pacific. Convection and

cloud feedback cool the central Pacific by 21.48

and21.41K respectively, relative to the western Pacific.

For the eastern Pacific, the single most significant con-

tributor to the warming difference from the western Pa-

cific is cloud feedback at 3.76K followed by water vapor

feedback at 0.75K. The negative contributors are con-

vection and PBL processes. Note that PBL turbulence

mixing contributes positively to warming in the central

Pacific, but negatively in the eastern Pacific. In contrast,

cloud feedback contributes negatively in the central

Pacific, but positively in the eastern Pacific. Dynamical

advection has negligible contribution to the warming

difference between the eastern and western Pacific

whereas it has a large positive contribution in the central

Pacific. These contrasts clearly show that the mechanisms

of the El Niño–like SST warming are different between

FIG. 10. Regionally averaged partial SST change due to each feedback process between 58Sand 58N in the western Pacific (1408E–1708W), central Pacific (1708–1108W), and eastern Pacific

(1108–808W) and their differences (K).

TABLE 1. Regionally averaged partial SST changes due to each

feedback process between 58S and 58N in the western Pacific (WP;

1408E–1708W), central Pacific (CP; 1708–1108W), and eastern Pa-

cific (EP; 1108–808W) and their differences (in kelvin). The con-

tributions from albedo feedback, large-scale condensation, and

gravity wave drag are very close to zero and therefore not included

in the table.

WP CP EP CP 2 WP EP 2 WP

CO2 0.81 0.88 0.80 0.07 20.01

Cloud 21.29 22.70 2.47 21.41 3.76

Water vapor 3.84 4.62 4.59 0.78 0.75

Convection 24.19 25.67 26.59 21.48 22.4

PBL 0.57 1.43 20.65 0.86 21.22

Dynamical

advection

1.97 3.43 1.90 1.46 20.07

Latent

heat flux

20.25 0.07 20.27 0.32 20.02

Sensible

heat flux

0.035 20.002 20.002 20.037 20.037

Ocean process 0.08 20.14 0.06 20.22 20.02

Net change 1.57 1.92 2.30 0.35 0.73

7314 JOURNAL OF CL IMATE VOLUME 27

the central and eastern Pacific. To summarize, the

changes in dynamical advection is the largest contributor

to the El Niño–like SST warming in the central Pacific.

The changes in PBL diffusion and water vapor feedback

further enhance the warming gradient, with some con-

tributions from surface evaporation change. In the east-

ern Pacific, cloud feedback is the dominant contributor,

with additional contribution from water vapor feedback.

5. Concluding remarks

The simulation of the slab ocean version of CCSM3.0

in response to a doubling of CO2 exhibits an El Niño–like change in the mean state of the equatorial Pacific.

The SST warming in the central and eastern Pacific is

higher than in the western Pacific, with the largest

warming gradient in the central Pacific. Associated with

the SST changes, convection shifts to the central and

eastern Pacific, resulting in significant precipitation in-

crease there. Correspondingly the Walker circulation is

weakened markedly. The anomalous ascending motion

occurs over the central and eastern Pacific and anoma-

lous descending motion appears over the western Pacific

Maritime Continent. The low-level easterlies are re-

duced as a result of the weakened sea level pressure

gradient between the western and eastern Pacific.

The CFRAM is applied to 10 years of hourly output of

NCAR CCSM3-SOM to identify factors contributing to

the El Niño–like warming in the equatorial Pacific in

response to a doubling of CO2. Unlike the traditional

surface heat budget analysis, which cannot isolate the

contributions of each radiative climate feedback and

physical and dynamical process of GCMs, the CFRAM

considers every physical and dynamical process that

responds to temperature change and affects Earth’s

energy budget as a climate feedback process, and is able

to explicitly quantify the contributions of each feedback

to temperature changes. In the tropical Pacific, themean

bias in the sum of partial SST changes due to each

feedback derived with CFRAM is negligible (0.5%)

compared to the mean SST change from the CCSM3-

SOM simulations. The spatial pattern correlation be-

tween the two is as high as 0.97.

The analysis shows that the mechanism of El Niño–like SST warming in the central Pacific is different from

that in the eastern Pacific. The dynamical advection

feedback is the largest single contributor to El Niño–likewarming in the central Pacific. The net effect of en-

hanced adiabatic cooling and vertical advection of sen-

sible and latent energy in the central Pacific results in

more increase in temperature lapse rate in the central

Pacific, leading to greater SST warming in the central

Pacific than the western Pacific. The maximum

weakening of PBL vertical mixing occurs in the central

Pacific, making it the second largest contributor to El

Niño–like SST warming in the central Pacific. The

stronger moistening related to more enhanced convec-

tion, stronger anomalous rising motion, and weakened

easterlies result in greater positive water vapor feedback

in the central Pacific than in the western Pacific. Thus,

water vapor feedback also makes a significant contri-

bution to the El Niño–like warming in the central Pa-

cific. The reduced evaporation in the central Pacific due

to weakened easterlies leads to warming in the central

Pacific and enhanced evaporation as a result of increased

SST leads to cooling in the western Pacific, thus also

contributing to the El Niño–like warming in the central

Pacific. In contrast, the dominant contributor to El Niño–like SSTwarming in the eastern Pacific is cloud feedback.

Both positive cloud feedback associated with low-level

cloud decrease in the eastern Pacific and negative cloud

feedback due to the increase of cloud in the western

Pacific contribute to the El Niño–like warming. In addi-

tion, the stronger enhancement of evaporation and con-

vection in the eastern Pacific results in more increase in

moisture than the western Pacific, making water vapor

feedback also important to the El Niño–like warming in

the eastern Pacific.

How does the equatorial Pacific warming influence

the precipitation pattern? It is apparent that the pattern

of precipitation changes is significantly different from

that of SST changes. The maximum equatorial SST

warming occurs in the eastern Pacific east of 1008Wbetween 58S and 58N (Fig. 1b), while the significant in-

crease of precipitation occurs in the central Pacific and

the eastern Pacific within 38–58N, 808–908W (Fig. 3b). It

should be pointed out that the change of convection

depends not only on SST change but also on the value of

SST in a warmer climate. Studies have shown that

a necessary condition for the development and mainte-

nance of strong deep convection in the tropics is that the

local SST be about 288Cor greater (Graham andBarnett

1987; Clarke 2008). This high SST ensures that the

cloud-base air mass will be charged with the required

moist static energy so that convection can reach the

upper troposphere (Sud et al. 1999). The SST in the

eastern Pacific cold tongue between 58S and 28N is

generally between 228 and 268C in the 13CO2 simula-

tion, whereas it is greater than 26.58C in the central

Pacific west of 1508W and eastern Pacific within 38–58N,

808–908W (Fig. 1a). Thus even with maximum warming,

the SST in the eastern Pacific cold tongue is still below

288C, which is much lower than the SST in the central

Pacific west of 1508W and eastern Pacific within 38–58N,

808–908W. This qualitatively explains why a significant

increase in precipitation only occurs in the central

1 OCTOBER 2014 SONG AND ZHANG 7315

Pacific and northern equatorial eastern Pacific rather

than in the region with maximum warming. To quanti-

tatively represent the impact of warming pattern on

precipitation change, Fig. 11a shows the distribution of

annual mean changes of convective available potential

energy (CAPE), which is used as the closure for the

convection scheme (Zhang and McFarlane 1995; Zhang

et al. 1998) to determine the intensity of convection.

CAPE is calculated as the vertical integral of parcel’s

buoyancy from the level of free convection to the equi-

librium level, assuming that the parcel is lifted from the

lowest model level (;992hPa). It is increased everywhere

with tropical warming. The pattern of CAPE change

correlates well with that of precipitationwith a correlation

coefficient of 0.68, indicating that the precipitation change

in the equatorial Pacific can be generally determined by

CAPE change. However, the precipitation decrease over

the western Pacific Maritime Continent (Fig. 2b) cannot

be attributed to CAPE change, indicating that other fac-

tors may also influence the precipitation pattern change.

One such factor is the low-level moisture convergence.

Figure 11b shows the annual mean change of low-level

moisture convergence. The anomalous moisture conver-

gence occurs in in the central Pacific and northern equa-

torial eastern Pacific, contributing to the precipitation

increase, while the anomalous moisture divergence over

northern western Pacific west of 1408W results in the de-

crease of precipitation.

In this study the ocean dynamics are excluded from

the slab ocean model version of CCSM3.0. How will the

ocean dynamics influence the warming pattern and cli-

mate feedbacks in the fully dynamical ocean model

version of CCSM3.0 (CCSM3)? It should be noted that

the atmospheric feedback mechanisms (e.g., convection–

SST interaction) identified by the CFRAM in the

CCSM3-SOM represent the response and feedback of

atmospheric processes to temperature change, which are

intrinsic characteristics of atmosphere model and in-

dependent of the ocean dynamics. So if atmospheric

processes are the main drivers of the El Niño–likewarming—in other words, if the warming pattern is not

substantially changed in the CCSM3 relative to the

CCSM3-SOM—the contribution of atmospheric feed-

backs in the CCSM3 should be similar to that derived

from the CCSM3-SOM. Nevertheless, the ocean dy-

namical advection may help shape the SST warming

pattern. It is shown that the annual mean surface air

temperature changes between 1980–99 and 2080–99 in

the A1B scenario from the CCSM3 also display an El

Niño–like warming pattern (Meehl et al. 2006), with

greater warming in the Niño-3.4 region than the equa-torial western Pacific (Fig. 5a of Meehl et al. 2006).

However, the warming in the far eastern Pacific (east of

1008W) is comparable to that of the western Pacific.

Since themaximumwarming occurs in the equatorial far

eastern Pacific in the CCSM3-SOM, it suggests that

FIG. 11. Annual mean changes of (a) CAPE (J kg21) and (b) low-level (1000–700 hPa)moisture

convergence (1025 kgm22 s21) from 1 3 CO2 to 2 3 CO2 conditions.

7316 JOURNAL OF CL IMATE VOLUME 27

ocean dynamics mainly influence the El Niño–likewarming in the far eastern Pacific and acts to damp it.

Therefore, we expect the conclusions derived from the

CCSM3-SOM to be generally applicable to the fully

coupled CCSM3 in the western and central Pacific. In

the far eastern Pacific, low-level stratus clouds are

prevalent over cold SST tongue and the middle tropo-

sphere is very dry. In contrast, deep convective clouds

are dominant over the western Pacific warm SST so that

themiddle troposphere is moist. Therefore, even though

the warming amplitude is comparable between the far

eastern Pacific and the western Pacific in the CCSM3,

the decrease of low-level stratus clouds and increase of

moisture in the middle troposphere in the far eastern

Pacific can still be more effective than the western Pa-

cific. This indicates that cloud feedback and water vapor

feedback, the major contributors to the El Niño–likewarming in the eastern Pacific in the CCSM3-SOM, can

still produce stronger warming in the far eastern Pacific

in the CCSM3 as well. On the other hand, a similar initial

SST warming perturbation associated with a doubling of

CO2 may lead to much greater increase in vertical ocean

temperature gradient in the far eastern Pacific, where the

thermocline is shallow, than in the western Pacific. This

may in turn result in much stronger cold water upwelling

and surface cooling in the far eastern Pacific in the fully

coupled CCSM3. In this way the ocean dynamical process

may damp the tendency of enhancing the east–west SST

gradient caused by the cloud and water vapor feedback

even if the Walker circulation is weakened, resulting in

a comparable warming between the far eastern Pacific

and the western Pacific in the fully coupled CCSM3.

A recent study (Yeh et al. 2012) shows that the pat-

terns of tropical Pacific SST warming trend over the

second half of the twentieth century have changed from

a La Niña–like structure in the CMIP3 multimodel en-

semble dataset to an El Niño–like structure in the phase

5 of CMIP (CMIP5) dataset. Applying the CFRAM

climate feedback analysis to both CMIP3 and CMIP5

models should be able to help us identify which changes

in the CMIP5 model result in such change in the SST

warming pattern in the tropical Pacific, and improve our

understanding of the model uncertainties.

Acknowledgments. This research was supported by

Office of Science (BER), U.S. Department of Energy un-

der Grant DE-SC0008880, the U.S. National Oceanic and

Atmospheric Administration Grant NA11OAR4321098,

and theNational Science FoundationGrantsAGS1015964.

The computational support for this work was provided by

the NCAR Computational and Information Systems Lab-

oratory. The authors thank the anonymous reviewers for

their constructive comments.

REFERENCES

Alexander,M.A., I. Bladé,M.Newman, J.R.Lanzante,N.-C.Lau, and

J. D. Scott, 2002: The atmospheric bridge: The influence of ENSO

teleconnections on air–sea interaction over the global oceans.

J. Climate, 15, 2205–2231, doi:10.1175/1520-0442(2002)015,2205:

TABTIO.2.0.CO;2.

An, S.-I., J.-W. Kim, S.-H. Im, B.-M. Kim, and J.-H. Park, 2012:

Recent and future sea surface temperature trends in tropical

Pacific warm pool and cold tongue regions. Climate Dyn., 39,

1373–1383, doi:10.1007/s00382-011-1129-7.

Bjerknes, J., 1969: Atmospheric teleconnections from the equa-

torial Pacific. Mon. Wea. Rev., 97, 163–172, doi:10.1175/

1520-0493(1969)097,0163:ATFTEP.2.3.CO;2.

Cane,M. A., A. C. Clement, A. Kaplan, Y. Kushnir, D. Pozdnyakov,

R. Seager, S. E. Zebiak, and R. Murtugudde, 1997: Twentieth-

century sea surface temperature trends. Science, 275, 957–

960, doi:10.1126/science.275.5302.957.

Clarke, A. J., 2008:An Introduction to theDynamics of El Niño andthe Southern Oscillation. Academic Press, 324 pp.

Clement, A. C., R. Seager, M. A. Cane, and S. E. Zebiak, 1996: An

ocean dynamical thermostat. J. Climate, 9, 2190–2196, doi:10.1175/

1520-0442(1996)009,2190:AODT.2.0.CO;2.

Collins, M., 2005: El Niño– or La Niño–like climate change? Cli-

mate Dyn., 24, 89–104, doi:10.1007/s00382-004-0478-x.

——, and Coauthors, 2010: The impact of global warming on the

tropical Pacific Ocean and El Niño. Nat. Geosci., 3, 391–397,

doi:10.1038/ngeo868.

Collins, W. D., and Coauthors, 2006a: The Community Climate

System Model version 3 (CCSM3). J. Climate, 19, 2122–2143,

doi:10.1175/JCLI3761.1.

——, and Coauthors, 2006b: The formulation and atmospheric

simulation of the Community Atmosphere Model version 3

(CAM3). J. Climate, 19, 2144–2161, doi:10.1175/JCLI3760.1.Cubasch, U., and Coauthors, 2001: Projections of future climate

change. Climate Change 2001: The Scientific Basis, J. T.

Houghton et al., Eds., Cambridge University Press,

527–582.

Deser, C., A. S. Phillips, and M. A. Alexander, 2010: Twentieth

century tropical sea surface temperature trends revisited.

Geophys. Res. Lett., 37, L10701, doi:10.1029/2010GL043321.

DiNezio, P. N., A. C. Clement, G. A. Vecchi, B. J. Soden, B. P.

Kirtman, and S.-K. Lee, 2009: Climate response of the equa-

torial Pacific to global warming. J. Climate, 22, 4873–4892,

doi:10.1175/2009JCLI2982.1.

——, ——, and ——, 2010: Reconciling differing views of tropical

Pacific climate change.Eos, Trans. Amer. Geophys. Union, 91,

141–142, doi:10.1029/2010EO160001.

Fedorov, A. V., and S. G. H. Philander, 2000: Is El Niño changing?Science, 288, 1997–2002, doi:10.1126/science.288.5473.1997.

Graham, N. E., and T. P. Barnett, 1987: Sea surface temperature,

surface wind divergence, and convection over tropical oceans.

Science, 238, 657–659, doi:10.1126/science.238.4827.657.

Held, I. M., and B. J. Soden, 2006: Robust responses of the hy-

drological cycle to global warming. J. Climate, 19, 5686–5699,

doi:10.1175/JCLI3990.1.

Knutson, T. R., and S. Manabe, 1995: Time-mean response over

the tropical Pacific to increased CO2 in a coupled ocean–

atmosphere model. J. Climate, 8, 2181–2199, doi:10.1175/

1520-0442(1995)008,2181:TMROTT.2.0.CO;2.

Koutavas, A., J. Lynch-Stieglitz, T. M. Marchitto Jr., and J. P. Sachs,

2002: El Niño-like pattern in ice age tropical Pacific sea tem-perature. Science, 297, 226–230, doi:10.1126/science.1072376.

1 OCTOBER 2014 SONG AND ZHANG 7317

Lee, J. Y., and B. Wang, 2014: Future change of global monsoon in the

CMIP5.Climate Dyn., 42, 101–119, doi:10.1007/s00382-012-1564-0.

Levitus, S., T. Boyer, M. Conkright, D. Johnson, T. O’Brien,

J. Antonov, C. Stephens, and R. Gelfeld, 1998: Introduction.

Vol. 1, World Ocean Database 1998, NOAA Atlas NESDIS

18, 346 pp.

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, doi:10.1175/

JCLI3579.1.

Lu, J., and M. Cai, 2009: A new framework for isolating individual

feedback processes in coupled general circulation climate

models. Part I: Formulation. Climate Dyn., 32, 873–885,

doi:10.1007/s00382-008-0425-3.

McGregor, H. V., M. J. Fischer, M. K. Gagan, D. Fink, S. J. Phipps,

H.Wong, andC.D.Woodroffe, 2013:AweakElNiño/SouthernOscillation with delayed seasonal growth around 4,300 yearsago. Nat. Geosci., 6, 949–953, doi:10.1038/ngeo1936.

McPhaden, M. J., S. E. Zebiak, and M. H. Glantz, 2006: ENSO as

an integrating concept in Earth science. Science, 314, 1740–

1745, doi:10.1126/science.1132588.

Meehl, G. A., and W. M. Washington, 1996: El Niño-like climatechange in a model with increased atmospheric CO2 concen-

trations. Nature, 382, 56–60, doi:10.1038/382056a0.

——, and Coauthors, 2006: Climate change projections for the

twenty-first century and climate change commitment in the

CCSM3. J. Climate, 19, 2597–2616, doi:10.1175/JCLI3746.1.

——, and Coauthors, 2007: Global climate projections. Climate

Change 2007: The Physical Science Basis, S. Solomon and

Coauthors, Eds., Cambridge University Press, 747–845.

Newell, R. E., 1979: Climate of the ocean. Amer. Sci., 67, 405–416.

Ramanathan, V., andW. Collins, 1991: Thermodynamic regulation

of ocean warming by cirrus clouds deduced from observations

of the 1987 El Niño.Nature, 351, 27–32, doi:10.1038/351027a0.Shell, K.M., J. T. Kiehl, andC.A. Shields, 2008:Using the radiative

kernel technique to calculate climate feedback in NCAR’s

Community Atmospheric Model. J. Climate, 21, 2269–2282,doi:10.1175/2007JCLI2044.1.

Soden, B. J., 1997: Variations in the tropical greenhouse effect

during El Niño. J. Climate, 10, 1050–1055, doi:10.1175/

1520-0442(1997)010,1050:VITTGE.2.0.CO;2.

——, I.M.Held, R. Colman, K. Shell, J. T. Kiehl, andC. A. Shields,

2008: Quantifying climate feedbacks using radiative kernels.

J. Climate, 21, 3504–3520, doi:10.1175/2007JCLI2110.1.

Song, X., G. J. Zhang, and M. Cai, 2014: Characterizing the climate

feedback pattern in theNCARCCSM3-SOMusing hourly data.

J. Climate, 27, 2912–2930, doi:10.1175/JCLI-D-13-00567.1.

Sud, Y. C., G. K. Walker, and K.-M. Lau, 1999: Mechanisms reg-

ulating sea-surface temperatures and deep convection in the

tropics. Geophys. Res. Lett., 26, 1019–1022, doi:10.1029/

1999GL900197.

Timmermann, A., J. Oberhuber, A. Bacher, M. Esch, M. Latif, and

E. Roeckner, 1999: Increased El Niño frequency in a climatemodel forced by future greenhousewarming.Nature, 398, 694–

697, doi:10.1038/19505.

Tokinaga,H., S.-P.Xie,C.Deser,Y.Kosaka, andY.Okumura, 2012a:

Slowdown of the Walker circulation by tropical Indo–Pacific

warming. Nature, 491, 439–443, doi:10.1038/nature11576.

——,——, A. Timmermann, S. McGregor, T. Ogata, H. Kubota,

and Y. Okumura, 2012b: Regional patterns of tropical

Indo-Pacific climate change: Evidence of the Walker circula-

tion weakening. J. Climate, 25, 1689–1710, doi:10.1175/

JCLI-D-11-00263.1.

Trenberth,K. E., 1997: The definition ofElNiño.Bull. Amer.Meteor.

Soc., 78, 2771–2778, doi:10.1175/1520-0477(1997)078,2771:

TDOENO.2.0.CO;2.

Vecchi, G. A., and B. J. Soden, 2007: Global warming and the

weakening of the tropical circulation. J. Climate, 20, 4316–

4340, doi:10.1175/JCLI4258.1.

——, and A. T. Wittenberg, 2010: El Niño and our future climate:Where do we stand? Wiley Interdiscip. Rev.: Climate Change,

1, 260–270, doi:10.1002/wcc.33.

——, B. J. Soden, A. T. Wittenberg, I. M. Held, A. Leetmaa, and

M. J. Harrison, 2006: Weakening of tropical Pacific atmo-

spheric circulation due to anthropogenic forcing. Nature, 441,

73–76, doi:10.1038/nature04744.

——, A. Clement, and B. J. Soden, 2008: Examining the tropical

Pacific’s response to global warming. Eos, Trans. Amer. Ge-

ophys. Union, 89, 81–83, doi:10.1029/2008EO090002.

Wetherald, R., and S. Manabe, 1988: Cloud feedback processes in

a general circulation model. J. Atmos. Sci., 45, 1397–1415,doi:10.1175/1520-0469(1988)045,1397:CFPIAG.2.0.CO;2.

Xie, S.-P., C. Deser, G. A. Vecchi, J. Ma, H. Teng, and A. T.

Wittenberg, 2010: Global warming pattern formation: Sea

surface temperature and rainfall. J. Climate, 23, 966–986,

doi:10.1175/2009JCLI3329.1.

Yeh, S. W., Y. G. Ham, and J. Y. Lee, 2012: Changes in the tropical

Pacific SST trend from CMIP3 to CMIP5 and its implication of

ENSO. J. Climate, 25, 7764–7771, doi:10.1175/JCLI-D-12-00304.1.

Yu, B., and G. J. Boer, 2002: The roles of radiation and dynamical

processes in the El Niño-like response to global warming.Climate Dyn., 19, 539–554, doi:10.1007/s00382-002-0244-x.

Zhang, G. J., and N. A. McFarlane, 1995: Sensitivity of climate

simulations to the parameterization of cumulus convection in

the Canadian Climate Centre general circulation model. At-

mos.–Ocean, 33, 407–446, doi:10.1080/07055900.1995.9649539.

——, and M. J. McPhaden, 1995: The relationship between sea

surface temperature and latent heat flux in the equatorial Pacific.

J. Climate, 8, 589–605, doi:10.1175/1520-0442(1995)008,0589:

TRBSST.2.0.CO;2.

——,andR.L.Grossman, 1996: Surface evaporationduring theCentral

Equatorial Pacific Experiment: A climate scale perspective.

J. Climate, 9, 2522–2537, doi:10.1175/1520-0442(1996)009,2522:

SEDTCE.2.0.CO;2.

——, V. Ramanathan, and M. J. McPhaden, 1995: Convection–

evaporation feedback in the equatorial Pacific. J. Climate, 8, 3040–

3051, doi:10.1175/1520-0442(1995)008,3040:CEFITE.2.0.CO;2.

——, J. T. Kiehl, and P. J. Rasch, 1998: Response of climate

simulation to a new convective parameterization in the Na-

tional Center for Atmospheric Research Community Cli-

mate Model (CCM 3). J. Climate, 11, 2097–2115, doi:10.1175/

1520-0442-11.8.2097.

Zhang, M., and H. Song, 2006: Evidence of deceleration of atmo-

spheric vertical overturning circulation over the tropical

Pacific. Geophys. Res. Lett., 33, L12701, doi:10.1029/

2006GL025942.

7318 JOURNAL OF CL IMATE VOLUME 27