A preview of this full-text is provided by Wiley.
Content available from Journal of Geophysical Research: Atmospheres
This content is subject to copyright. Terms and conditions apply.
Estimating Contributions of Sea Ice and Land Snow
to Climate Feedback
Lei Duan
1
, Long Cao
1
, and Ken Caldeira
2
1
Department of Atmospheric Sciences, School of Earth Sciences, Zhejiang University, Hangzhou, China,
2
Department of
Global Ecology, Carnegie Institution, Stanford, CA, USA
Abstract In this study, we use the National Center for Atmospheric Research Community Earth System
Model to investigate the contribution of sea ice and land snow to the climate sensitivity in response to
increased atmospheric carbon dioxide content. We focus on the overall effect arising from the presence or
absence of sea ice and/or land snow. We analyze our results in terms of the radiative forcing and climate
feedback parameter. We find that the presence of sea ice and land snow decreases the climate feedback
parameter (and thus increases climate sensitivity). Adjusted radiative forcing from added carbon dioxide is
comparatively less sensitive to the presence of sea ice or land snow. The effect of sea ice on the climate
feedback parameter decreases as sea ice cover diminishes at higher CO
2
concentration. However, the
influence of both sea ice and land snow on the climate feedback parameter remains substantial under the
CO
2
concentration range considered here (to eight times preindustrial CO
2
content). Approximately, one
quarter of the effect of sea ice and land snow on the climate feedback parameter is a consequence of the effect
of these components on longwave feedback that is mainly associated with cloud change. Polar warming
in response to added CO
2
is approximately doubled by the presence of sea ice and land snow. Relative to the
case in which sea ice and land snow are absent in the model, in response to increased CO
2
concentrations,
the presence of sea ice and land snow results in an increase in global mean warming by over 40%.
Plain Language Summary Sea ice and land snow are two crucial components that affect the
climate response to external forcings. Feedbacks between ice/snow and climate change cause amplified
surface warming in high latitudes. In this study, we use a climate model to estimate the contribution of sea
ice and land snow to climate change in response to increased CO
2
concentrations. We compare the climate
response to increased CO
2
between the simulations with sea ice and/or land snow and the simulations
without them. We show that the existence of sea ice and land snow substantially amplifies the global
temperature response to increased CO
2
with sea ice having a stronger effect than land snow. Under higher
CO
2
levels, the effect of sea ice diminishes more rapidly than does the effect of land snow. About one
quarter of the total climate feedback from sea ice and land snow is associated with the change in longwave
radiation. Also we show that the effect of sea ice and land snow on the sensitivity of top‐of‐atmosphere
net energy flux to the global mean temperature change is approximately additive.
1. Introduction
Observations and modeling studies have recognized that the global warming from increased atmospheric
CO
2
concentrations is amplified in polar regions (e.g., Bekryaev et al., 2010; Bromwich et al., 2013;
Chapman & Walsh, 2007; Manabe et al., 1991; Manabe & Stouffer, 1980; Meehl et al., 2007; Pithan &
Mauritsen, 2014; Serreze & Francis, 2006; Solomon, 2006). Sea ice and land snow are two important
components contributing to substantial warming in middle and high latitudes, primarily from their surface
albedo feedback and the insulation feedback (Caldeira & Cvijanovic, 2014; Cess et al., 1991; Hall, 2004;
Holland et al., 2001; Holland & Bitz, 2003; Screen & Simmonds, 2010; Serreze et al., 2009; Stroeve et al.,
2012; Vavrus, 2007; Winton, 2006; Zhang, 2005). For example, sea ice and land snow retreats expose less
reflective land and ocean surface, increasing the absorption of solar insolation. The presence of sea ice
and land snow inhibits the energy exchange between atmosphere and surface. Other factors also contribute
to high‐latitude amplification of the temperature change, including the Planck feedback, changes in water
vapor content, atmospheric and oceanic heat transport, and cloud fraction in response to the imposed
forcing and the sea‐ice and land snow decline (e.g., Kay et al., 2012; Mahlstein & Knutti, 2011; Overland
& Wang, 2010; Södergren et al., 2018; Solomon, 2006; Taylor et al., 2013).
DUAN ET AL. 199
RESEARCH ARTICLE
10.1029/2018JD029093
Key Points:
•We use NCAR CESM to investigate
the overall contribution of sea ice
and land snow to climate feedback
to increased carbon dioxide content
•Sea ice and land snow lead to
substantial increases in polar
warming to increased CO
2
content
with sea ice playing a more
important role
•The presence of sea ice and land
snow decreases the climate feedback
parameter with about one quarter of
the effect from longwave feedback
Supporting Information:
•Supporting Information S1
Correspondence to:
L. Cao,
longcao@zju.edu.cn
Citation:
Duan, L., Cao, L., & Caldeira, K. (2019).
Estimating contributions of sea ice and
land snow to climate feedback. Journal
of Geophysical Research: Atmospheres,
124, 199–208. https://doi.org/10.1029/
2018JD029093
Received 31 MAY 2018
Accepted 14 DEC 2018
Accepted article online 7 JAN 2019
Published online 15 JAN 2019
©2019. American Geophysical Union.
All Rights Reserved.
The presence of sea ice and land snow exerts great impact on local climate, but their effects are not confined
to high‐latitude regions. The absence of sea ice and land snow has been shown to produce warming in tro-
pical regions (Cvijanovic & Caldeira, 2015; Vavrus, 2007). At the same time, the cooling effect induced by the
presence of additional high latitude ice could alter the location of the Intertropical Convergence Zone and
influence the tropical precipitation (Broccoli et al., 2006; Chiang & Bitz, 2005). Previous studies have con-
cluded that changes in sea ice and land snow are likely to affect extreme weather events in the midlatitudes
through their influence on large‐scale atmospheric circulation including storm tracks, jet streams, and pla-
netary waves (e.g., Francis et al., 2009; Francis & Vavrus, 2012; Liu et al., 2012; Peings & Magnusdottir,
2014). However, other studies (Barnes et al., 2014; Cohen et al., 2014; Screen & Simmonds, 2013; Wallace
et al., 2014) argued that the robustness of such correspondence remains uncertain.
The relative importance of sea ice and land snow on the climate feedback parameter (Gregory et al., 2004),
which is closely related to the top‐of‐atmosphere (TOA) net radiative flux (in the absence of any tempera-
ture change) divided by the equilibrium global mean temperature change, has been quantified in previous
modeling works. For instance, Ingram et al. (1989) found that under a doubling of atmospheric CO
2
, the
absence of sea ice feedback increases the climate feedback parameter from 0.77 to 0.95 W·m
−2
·K
−1
.
Similarly, Rind et al. (1998) found that under a doubling of atmospheric CO
2
, the climate feedback para-
meter increases from a default value of 0.95 to 1.51 W·m
−2
·K
−1
if sea ice is not allowed to change with
warming. Using the slab‐ocean configuration of CESM, Caldeira and Cvijanovic (2014) found that under
various CO
2
levels, on average the climate feedback parameter increased by 0.26 W·m
−2
·K
−1
(ranging from
0.22 to 0.3 W·m
−2
·K
−1
) in the absence of sea ice feedback. In their simulations, about one third of the
change in the climate feedback parameter caused by the presence of sea‐ice feedbacks is a consequence
of the change in longwave radiation.
The effect of snow on climate feedback was investigated in a number of studies. Vavrus (2007) found that
removing the entire snow cover on land resulted in 0.57 W m
−2
of radiative forcing, resulting in 0.84 K of
warming. An earlier model intercomparison project (Cess et al., 1991) of 17 general circulation models
examined the effect of snow retreat under a prescribed 4 K sea surface temperature warming and found that
changes in the cloud distribution and longwave feedback due to snow change contribute to the net snow
feedback in addition to surface albedo change. Qu and Hall (2014) analyzed the shortwave snow albedo feed-
back using results from 25 climate models that participated in the Coupled Model Intercomparison Project
Phase 5. Their results suggested that the removal of the snow‐albedo feedback would increase the climate
feedback parameter by about 0.08 W·m
−2
·K
−1
(ranging from 0.03 to 0.16 W·m
−2
·K
−1
).
The combined impact of sea ice and land snow on the climate response is also examined in a number of pre-
vious works. For example, Winton (2006) analyzed simulation results from 12 Earth system models with 1%/
year increase in CO
2
and found that the surface albedo feedback associated with sea ice and snow decreases
the climate feedback parameter by about 0.3 W·m
−2
·K
−1
. Holland et al. (2001) found about 17% and
Graversen and Wang (2009) found that about 15% of the global mean warming induced by a doubling of
atmospheric CO
2
is attributable to sea ice and land snow albedo feedbacks. Taylor et al. (2013) demonstrated
that the surface albedo feedback is the largest contributor to the extratropical warming in both hemispheres,
while Flanner et al. (2011) found that the Northern Hemisphere surface albedo feedback decreases the cli-
mate feedback parameter by 0.33 W·m
−2
·K
−1
. Crook and Forster (2014) analyzed satellite data and sug-
gested that current climate models underestimate the surface albedo feedback in the Northern
Hemisphere extratropical regions.
Table 1
Model‐Simulated Global Mean Changes in Surface Air Temperature for “None,”“Ice,”“Snow,”and “Both”Simulations Relative to the Corresponding 1 × CO
2
Case
With the Same Sea Ice and Snow Treatment
1×CO
2
2×CO
2
–1×CO
2
4×CO
2
–1×CO
2
6×CO
2
–1×CO
2
8×CO
2
–1×CO
2
Temperature (K) None 290.10 ± 0.01 1.94 ± 0.02 4.22 ± 0.02 5.86 ± 0.03 7.14 ± 0.02
Ice 288.89 ± 0.01 2.51 ± 0.02 5.14 ± 0.02 6.93 ± 0.02 8.35 ± 0.02
Snow 289.13 ± 0.01 2.23 ± 0.02 4.80 ± 0.02 6.63 ± 0.02 8.10 ± 0.02
Both 287.02 ± 0.01 3.17 ± 0.02 6.46 ± 0.03 8.53 ± 0.03 10.15 ± 0.02
Note. All results are calculated from the last 60‐year simulations of 100‐year slab‐ocean simulations. Uncertainty bars represent one standard error of the mean.
10.1029/2018JD029093
Journal of Geophysical Research: Atmospheres
DUAN ET AL. 200
In this study, we use the Community Earth System Model (CESM) to per-
form a series of slab‐ocean simulations with different sea ice and land
snow treatments to investigate the contribution of sea ice and land snow
to the climate feedback parameter and its shortwave and longwave com-
ponents. We compare the results from simulations with the presence of
sea ice and land snow separately and jointly to that with no existence of
sea ice and/or land snow. In contrast to previous studies, we analyze the
change in the climate feedback parameter associated with a complete loss
of sea ice and land snow, including changes in the energy flux, cloud, and
water vapor due to the change in sea ice and land snow. We also investi-
gate the contribution of sea ice and land snow at various CO
2
levels.
2. Methods
The model used in this study is the National Center for Atmospheric
Research (NCAR) Community Earth System Model version 1.2
(CESM1.2). In our simulations, the atmosphere is represented by the
Community Atmosphere Model version 4 (Neale et al., 2010). The
Community Atmosphere Model 4 uses a finite volume dynamic core with a hybrid sigma‐pressure vertical
coordinate. The horizontal resolution for the atmosphere is 1.9° latitude by 2.5° longitude with 26 vertical
levels. The land is represented by the Community Land Model version 4 (Oleson et al., 2010), which adopts
the same horizontal resolution as the Community Atmosphere Model 4 but with 15 vertical layers and up
to 5 snow/ice layers. Land snow in the Community Land Model 4 is represented by a series of state variables,
including mass of water (w
liq
), mass of ice (w
ice
), layer thickness (Δz), and layer temperature (T). Snow can
also exist on land surface without being represented by explicit snow layers when its thickness is less than
aspecified minimumdepth of0.01 m. The ocean component isrepresented by a mixed ocean layerwith depth
varying between 10 and 200 m and a prescribed ocean heat transport (Q‐flux). The slab ocean calculates sea
surface temperature changes by considering the energy imbalance at air‐sea interface and the prescribed heat
flux at the base of the mixed layer. The current version of the slab‐ocean configuration simulates sea ice
change by including an active sea ice component with dynamic and thermodynamic processes (Hunke
et al., 2010). The sea ice component calculates the ice growth rates, ice package transport, and interactions
between sea ice and external couplers. Using an earlier version of
CESM, the configuration with a slab‐ocean component was found to pro-
duce an equilibrium atmospheric response that is similar to that produced
by the configuration with a full ocean (Bitz et al., 2012).
To assess the overall effect of sea ice and land snow on the climate feed-
back parameter and climate sensitivity, we first spin up four types of simu-
lations under preindustrial CO
2
level (1 × CO
2
, 284.7 ppm). The CESM
slab‐ocean configuration reaches equilibrium within about 30 years, and
we perform each simulation for 100 years. In these simulations, sea ice
and land snow components are either included or disabled. These simula-
tions are denoted as follows: (1) “Both”: both sea ice and land snow com-
ponents are included as the default in CESM; (2) “None”: neither sea ice
nor land snow is included; (3) “Ice”: the sea ice and snow components
are included, but the land snow component is disabled; (4) “Snow”: the
land snow component is included, but the sea ice component is disabled.
In this study, we remove the sea ice component by setting the freezing
point of water in the ocean model to absolute zero. Caldeira and
Cvijanovic (2014) showed that results for radiative forcing and climate
feedback parameters were very similar regardless of whether sea ice feed-
backs were disabled by fixed sea ice cover and thickness or by setting the
freezing point of water to absolute zero. In the normal configuration
(“Both”), snow may appear on land or sea ice. We remove the snow
(and other forms of frozen water) on land by setting the freezing point
Figure 1. Model‐simulated global mean surface temperature change for
“None”,“Ice”,“Snow”, and “Both”simulations under various CO
2
levels.
All results are calculated using the last 60‐year results of 100‐year slab‐ocean
simulations with uncertainty represented by one standard error.
Figure 2. Model‐simulated changes in zonal distribution of surface air
temperature for “None,”“Ice,”“Snow,”and “Both”simulations under
4×CO
2
relative to the corresponding 1 × CO
2
case with the same sea ice and
snow treatment. All results are calculated from the last 60‐year simulations
of 100‐year slab‐ocean simulations. Ice and snow responses amplify the
temperature response to increased atmospheric CO
2
, especially in polar
regions.
10.1029/2018JD029093
Journal of Geophysical Research: Atmospheres
DUAN ET AL. 201
of water to absolute zero in the land model including lakes that are
allowed to supercool to melt snow and ice. In the Community Land
Model 4, the glacier land types are represented by a prescribed surface
fraction and surface albedo in each grid cell. Therefore, to change the
freezing points of water in the land model will not affect the state of
glacier. Our modifications of the model do not directly change cloud
processes in the atmosphere, which may involve phase changes of water.
Thus, snow and ice may be present in the atmosphere and fall on the
surface. All snow is immediately converted to liquid‐water equivalent
when reaching the surface, and the energy consumed to melt snow is
considered to decrease the surface temperature. Thus, energy is conserved
in our simulations.
To examine the contribution of sea ice and land snow to the climate feed-
back parameter under various CO
2
levels, we conduct simulations under
1×CO
2
,2×CO
2
,4×CO
2
,6×CO
2
, and 8 × CO
2
conditions for each
sea ice and land snow treatment described above. All simulations are
run for 100 years. We use the average results for the last 60‐year simula-
tions to represent climatological means. The TOA radiative forcing and
climate feedback parameter are diagnosed using the Gregory’s regression
method with all 100‐year simulation results (Gregory et al., 2004). To
improve the statistical power of the linear regression, for each combina-
tion of sea ice, land snow, and atmospheric CO
2
treatment, two additional
10‐year simulations are conducted with slightly altered initial condition as
done before in Caldeira and Cvijanovic (2014).
3. Results and Discussion
3.1. Sea Ice and Land Snow Effects on Temperature Response
The equilibrium global mean surface temperature at 1 × CO
2
concentra-
tion for “None,”“Ice,”“Snow,”and “Both”simulations are shown in
Table 1. In our simulations, adding sea ice to the “None”simulation cools
the planet by 1.21 ± 0.02 K. Adding land snow to the “None”simulation
cools the planet by 0.97 ± 0.02 K. Adding both sea ice and land snow to
the “None”simulation cools the planet by 3.08 ± 0.02 K. Global mean
temperature responses to 2 × CO
2
,4×CO
2
,6×CO
2
, and 8 × CO
2
forcing
are shown in Table 1 and Figure 1.
The presence of sea ice and land snow contributes to substantially larger
global warming in response to CO
2
increase. In our simulations, the tem-
perature rise with both sea ice and land snow is about 63%, 53%, 46%, and
42% greater than without sea ice or land snow for 2 × CO
2
,4×CO
2
,
6×CO
2
, and 8 × CO
2
forcing, respectively. The temperature rise with
sea ice only is about 30%, 22%, 18%, and 17% greater than without sea
ice or land snow for these forcing levels. With land snow only, the tem-
perature rise is about 15%, 14%, 12%, and 13% greater than without sea
ice or land snow for these forcing levels. The impact of sea ice and land
snow on global temperature is associated with the change in total sea
ice and land snow amount and the seasonal duration under different
CO
2
levels (Figure S1 in the supporting information). Our results indicate that both the sea ice (“Ice”) and
land snow (“Snow”) account for the larger surface temperature response with sea ice playing a more
important role.
Feedbacks from the sea ice and land snow components are crucial in shaping the latitudinal warming pat-
tern as well. Figure 2 shows the zonal distribution of change in surface temperature under 4 × CO
2
for dif-
ferent sea ice and land snow treatments. Results at other CO
2
levels and the corresponding normalized zonal
Figure 3. Gregory regressions of top‐of‐atmosphere (TOA) (a) net radiative
flux (positive downward), (b) net shortwave, and (c) net longwave radiation
versus global mean surface temperature change for “None,”“Ice,”“Snow,”
and “Both”simulations under 4 × CO
2
simulations relative to the corre-
sponding 1 × CO
2
case with the same sea ice and snow treatment. The
100‐year slab‐ocean simulations with two additional 10‐year simulations
with slightly altered initial condition are used for regression. Results for the
slope (climate feedback parameter) and vertical axis intercepts (adjusted
radiative forcing) of this regression are shown in Table 2. Panels for other
CO
2
levels are shown in supporting information Figure S4. The primary
influence of ice and snow on the climate feedback parameter is through the
shortwave component.
10.1029/2018JD029093
Journal of Geophysical Research: Atmospheres
DUAN ET AL. 202
distribution of change in surface temperature can be found in Figures S2 and S3. For 4 × CO
2
simulation,
there is an obvious equator‐to‐pole gradient with the surface temperature response where polar regions
warm substantially larger than the tropics. Compared to “None”simulations, the presence of sea ice and
land snow (“Both”simulations) leads to substantial increases in high latitudes warming. For example,
surface temperature increases by 13.92 and 11.84 K in the Northern and Southern Hemisphere polar
regions, which is about 98% and 102% larger than that in “None”simulations. In the model, warming in
tropical regions (between 30°S and 30°N) in response to a quadrupling of CO
2
is slightly larger (4.47 K) in
the “Both”simulation than that in the “None”simulation (3.46 K). Thus, the tropical response is affected
by ice and snow feedbacks, but the tropics are not affected by these feedbacks as strongly as the high
latitude regions (Figure 2). Ocean heat fluxes are the same in all of our simulations, and thus the
difference in tropical temperature is due only to the atmospheric communication of remote sea ice and
land snow effect. Use of a dynamic ocean would allow oceanic communication of high latitude ice and
snow effects and therefore would be expected to lead to somewhat different quantitative results.
By comparing only sea ice (“Ice”) or land snow (“Snow”) presence simulations with “None”simulations, it is
found that the sea ice component alone results in almost half of the warming observed in the “Both”
simulation in Arctic and the Southern Ocean, whereas land snow alone contributes most strongly to warm-
ing in the Northern Hemisphere midlatitudes (Figure 2). Such features are presented at all CO
2
levels
(Figures S2 and S3).
3.2. Sea Ice and Land Snow Contributions to Climate Feedback Parameters
We adopt a linear regression approach based on Gregory et al. (2004) to calculate the TOA radiative forcing
and the climate feedback parameter using the following formula:
N¼ERFCO2 −λΔT(1)
where Nis the TOA net radiative flux calculated as the net shortwave minus longwave radiation (all radia-
tive fluxes are set downward positive), ERF
CO2
is the effective radiative forcing due to increased CO
2
concen-
tration, λis the climate feedback parameter, and ΔTrepresents the change in global mean surface
temperature. For each of our simulation cases with increased CO
2
, we subtract results from the correspond-
ing 1 × CO
2
simulation with the same sea ice and land snow treatment. Then we apply the Gregory regres-
sion on the TOA net radiative flux versus the global mean surface temperature change. Regression results for
all cases are summarized in Figures 3 and S4, and Tables S1 to S3.
Applying Gregory regression to the “None”simulation 4 × CO
2
case, we estimate the climate feedback para-
meter λ
None,4 × CO2
to equal 1.48 ± 0.04 W·m
−2
·K
−1
(Table 2 and Figure 3). The estimated climate feedback
parameter for the “Both”simulation 4 × CO
2
case λ
Both,4 × CO2
is 0.90 ± 0.03 W·m
−2
·K
−1
. Comparison of
λ
Both,4 × CO2
and λ
None,4 × CO2
indicates that the presence of both sea ice and land snow decreases the climate
feedback parameter in response to CO
2
forcing by 39 ± 9%. The overall contribution of the sea ice and land
snow to the climate feedback parameter (λ
IceSnow
) can be estimated as the difference between climate feed-
back parameters diagnosed from “Both”and “None”simulations:
λIceSnow ¼λBoth−λNone (2)
As a consequence, we estimate λ
IceSnow
to be −0.58 ± 0.05 W·m
−2
·K
−1
under 4 × CO
2
case.
Table 2
Estimated Top‐of‐Atmosphere Radiative Forcing (Downward Positive) and the Climate Feedback Parameter (λ) for “None”,“Ice”,“Snow”, and “Both”Simulations at
4×CO
2
Levels Relative to 1 × CO
2
by Applying Gregory Regression on 100‐Year Simulation Results and Two Additional 10‐Year Simulations
4×CO
2
–1×CO
2
None Ice Snow Both
Radiative forcing (W m
−2
) 6.23 ± 0.17 5.79 ± 0.16 6.47 ± 0.16 5.79 ± 0.17
Climate feedback parameter (W·m
−2
·K
−1
) 1.48 ± 0.04 1.13 ± 0.03 1.35 ± 0.04 0.90 ± 0.03
10.1029/2018JD029093
Journal of Geophysical Research: Atmospheres
DUAN ET AL. 203
The climate feedback parameters derived from the corresponding “Ice”
and “Snow”simulations under 4 × CO
2
are 1.13 ± 0.03 and
1.35 ± 0.04 W·m
−2
·K
−1
, respectively. Using the same approach as for
“Both”and “None”simulations, we estimate the contribution of sea ice
(λ
Ice
) and land snow (λ
Snow
)tobe−0.35 ± 0.04
and −0.13 ± 0.04 W·m
−2
·K
−1
, respectively. Our results thus indicate a
stronger sea ice effect than land snow. Figure 4 shows the estimated con-
tribution of sea ice and land snow to the climate feedback parameter
under various CO
2
levels. It is clear that while the contribution of sea
ice to the climate feedback parameter is larger than that from land snow,
it reduces as sea ice area diminishes at higher CO
2
levels (Figures 4 and 5).
At the same time, land area covered by snow reduces from
40.43 ± 0.11 × 10
12
m
2
under 1 × CO
2
to 21.22 ± 0.14 × 10
12
m
2
under
8×CO
2
, maintaining a relatively unchanged land snow contribution to
the climate feedback parameter due largely to the relatively small climate
influence of land snow over land ice (see Figure S5). Under 8 × CO
2
case,
the sea ice and land snow show roughly equivalent contributions to the
climate feedback parameter of −0.23 ± 0.04 and
−0.13 ± 0.04 W·m
−2
·K
−1
, respectively. Contributions from sea ice and
land snow to the climate feedback parameter can be understood in terms of radiative forcing changes from
the change in sea ice and land snow area and the sensitivity of these areas to the global mean temperature
change (see Caldeira and Cvijanovic, 2014, and supporting information).
By comparing red and gray bars in Figure 4, we notice that the contribution of individual sea ice and land
snow component to climate feedback parameter is approximately additive. That is, the combined effect of
sea ice and land snow on climate feedback parameter is approximately equal to the linear sum of the indi-
vidual effect from the existence of sea ice or land snow. Our results are consistent with previous studies in
which the total climate feedback can be decomposed into linear combinations of independent feedback pro-
cesses (Colman, 2003; Soden & Held, 2006; Winton, 2006).
Gregory regressions can be performed independently on TOA net shortwave and longwave radiation (posi-
tive downward) to evaluate their individual contributions to the total climate feedback parameter. For
“None”simulation 4 × CO
2
case, we estimate the shortwave component feedback (λ
None,SW
)tobe
−0.38 ± 0.03 W·m
−2
·K
−1
and longwave component feedback (λ
None,LW
) to be 1.87 ± 0.02 W·m
−2
·K
−1
(Figure 3 and Tables S2 and S3). Opposite signs suggest that the shortwave
feedback tends to amplify the climate change while longwave tends to
damp the climate change. The corresponding shortwave and longwave
feedback for the “Both”simulation 4 × CO
2
case is −0.84 ± 0.02 and
1.74 ± 0.01 W·m
−2
·K
−1
, respectively.
Applying equation (2) on the shortwave and longwave feedbacks between
the “None”and “Both”simulation 4 × CO
2
cases, we estimate the contri-
bution of TOA net longwave feedback parameter due to the presence of
sea ice and land snow to be −0.13 ± 0.02 W·m
−2
·K
−1
, ~28% of the corre-
sponding shortwave contribution (−0.46 ± 0.04 W·m
−2
·K
−1
). We further
decompose the TOA net shortwave and longwave feedbacks into clear‐
sky and cloudy‐sky components. As shown in Table S4, the contribution
of the shortwave component between “None”and “Both”cases is mainly
associated with the surface albedo change and thus the clear‐sky short-
wave feedback, while the contribution of the longwave component is
mainly due to the cloud cover change that affects the cloudy‐sky
longwave feedback.
When considering all CO
2
levels, a decrease trend in the shortwave contri-
bution due to the presence of sea ice and land snow is simulated, whereas
the longwave contribution is less sensitive to the CO
2
change (Figure S6).
Figure 5. Model‐simulated sea ice area and land snow area as a function of
global mean surface temperature for “None,”“Ice,”“Snow,”and “Both”
simulations at various CO
2
levels. All results are calculated from the last 60‐
year simulations of 100‐year slab‐ocean simulations. The sea‐ice and land‐
snow areas are calculated by multiplying the sea‐ice and land‐snow fraction
in each grid cell with areas of that grid and then integrating over the globe.
Figure 4. Contributions of the sea ice and/or land snow components to the
climate feedback parameter for “Both,”“Ice,”and “Snow”simulations
relative to the “None”simulations. The “Ice”+“Snow”bar represents the
height of “Ice”bar stacked on top of the “Snow”bar. The comparison
between “‘Both”and “Ice + Snow”shows that the effects of ice and snow on
the climate feedback parameter are approximately additive.
10.1029/2018JD029093
Journal of Geophysical Research: Atmospheres
DUAN ET AL. 204
The contribution of longwave feedback to the climate feedback parameter
increases to ~45% of the shortwave contribution under 8 × CO
2
case.
Thus, our results confirm the importance of longwave feedback arising
from the sea ice and land snow change as reported in Caldeira and
Cvijanovic (2014) that examined sea ice effect only and emphasize the
necessity to consider the longwave feedback in the study of the climate
response to sea ice and land snow change.
3.3. Change of the Climate Feedback Parameter at Different
CO
2
Levels
Taking into account the results under various CO
2
levels, we find that the
climate feedback parameter decreases at higher CO
2
levels for the “None”
simulation (λ
None
, Figure 6), implying a more sensitive global mean tem-
perature response to CO
2
‐induced forcing. Decrease in λ
None
is mainly
associated with enhanced water vapor feedback. More water vapor is
added into the atmosphere for per degree global warming under higher
CO
2
levels (Figure S7b). Since more water vapor in the atmosphere traps
additional outgoing longwave radiation, it contributes to a smaller TOA
longwave feedback parameter and a smaller climate feedback parameter for “None”simulations
(Tables S1 and S3). The importance of water vapor feedback in determining the longwave feedback and
the climate feedback parameter has been recognized in many previous modeling studies (e.g., Raval et al.,
1994; Huang et al., 2007; Jonko et al., 2013; Meraner et al., 2013). Also, the cloud fraction change per degree
of the global mean temperature change differs at different CO
2
levels (Figure S7c and Table S4), contributing
to changes in the climate feedback parameter.
In contrast to “None”simulations, λ
Both
increases under higher CO
2
levels, suggesting a less sensitive cli-
mate response (Figure 6). As shown in Figure 6, the decrease in λ
Both
is primarily a result of diminishing
sea ice feedback at higher CO
2
levels. Values of the λ
Ice
and λ
Snow
from different sea ice and land snow treat-
ments approach each other at higher CO
2
concentrations mainly because the strength of the sea ice feedback
decreases due to the loss of ice cover under higher temperature.
4. Discussion and Conclusion
Using the NCAR CESM, we investigate the overall contribution arising from the presence or absence of sea
ice and/or land snow to the climate sensitivity in response to a series of carbon dioxide increase scenarios.
Feedbacks from the sea ice and land snow components are found to increase the global warming by over
40% relative to simulations where both of their effects are disabled. Sea ice and land snow are also crucial
in determining the substantially larger warming in high latitudes. Feedbacks from the sea ice component
only are found to increase the global warming by over 17%. We find that the sea ice component accounts
for nearly half of the temperature increase in the Arctic and Southern Ocean. When considering only land
snow feedbacks, global warming increases by over 12%. The primary effects of land snow are found mainly
in the Northern Hemisphere midlatitudes. Here we have focused on the climate effect of sea ice and land
snow, but it should be borne in mind that the loss of ice or snow will also have substantial impacts on other
components of the natural system.
Using the NCAR CESM, we estimated an averaged contribution of sea ice and land snow components to the
climate feedback parameter as −0.6 W·m
−2
·K
−1
considering the full set of feedbacks at various CO
2
levels,
larger than the estimates made in previous studies that considered only the surface albedo feedback (e.g.,
Graversen & Wang, 2009; Holland et al., 2001; Winton, 2006). We also find that approximately one quarter
of the warming effect from sea ice and land snow is associated with the longwave feedback.
Furthermore, we evaluate the contribution of sea ice and land snow to the climate feedback parameter using
the Gregory regression and equilibrium state sea ice and land snow results from “Ice,”“Snow,”and “None”
simulations, respectively. Our derived sea ice effect is stronger than that from the land snow but diminishes
at higher CO
2
levels due to decreases in sea ice cover. Under 8 × CO
2
case, the effect of sea ice and land snow
components becomes similar. Contributions from sea ice and land snow components to the climate feedback
Figure 6. Gregory regression results of the climate feedback parameter (λ)
versus radiative forcing for “None,”“Ice,”“Snow,”and “Both”simulations
at various CO
2
levels.
10.1029/2018JD029093
Journal of Geophysical Research: Atmospheres
DUAN ET AL. 205
parameter are roughly additive, consistent with previous studies that decompose the total climate feedback
as the linear sum of independent processes (e.g., Andrews & Webb, 2018; Colman, 2003; Hansen et al., 1984;
Jonko et al., 2013; Soden & held, 2006; Wetherald & Manabe, 1988).
We used a single climate model to examine the sea ice and land snow effects on the climate sensitivity.
Results from the Coupled Model Intercomparison Project have shown that climate feedback parameters
are highly model dependent and vary significantly among models (e.g., Andrews et al., 2012; Winton,
2006). Using the Coupled Model Intercomparison Project Phase 5 ensemble results, Massonnet et al.
(2012) showed that there is a large spread of the sea ice responses among different models. In our simula-
tions, a slab‐ocean configuration with specified ocean heat transport is used to represent the ocean response,
thus excluding the interaction of climate with deep water and oceanic heat transport. These interactions are
found to affect the global warming pattern as well as the atmospheric feedback (Boer & Yu, 2003; Mahlstein
& Knutti, 2011; Polyakov et al., 2005). In addition, the sea ice and land snow feedbacks can be quite sensitive
to ocean processes that are not represented in the slab‐ocean mode (Johns et al., 2006). Therefore, caution
should be exercised when interpreting quantitative conclusions presented in this paper.
Key qualitative conclusions reached in this study are as follows: (1) Sea ice has a stronger influence on the
climate feedback parameter than does land snow, but both of them substantially decrease the climate feed-
back parameter, particularly at high latitudes. (2) Sea ice and snow feedbacks involve a substantial long-
wave component. (3) The sum of the effects of sea ice and land snow taken individually on the climate
feedback parameter is approximately equal to the effect of sea ice and land snow taken jointly on the cli-
mate feedback parameter, as would be expected of a quasilinear system. (4) The influence of sea ice on
the climate feedback parameter diminishes with increasing atmospheric CO
2
more rapidly than does the
influence of land snow on the climate feedback parameter. It would be useful to test if effects of sea ice
and land snow on climate sensitivity are similar in other models.
References
Andrews, T., Gregory, J. M., Webb, M. J., & Taylor, K. E. (2012). Forcing, feedbacks and climate sensitivity in CMIP5 coupled atmosphere‐
ocean climate models. Geophysical Research Letters,39, L09712. https://doi.org/10.1029/2012GL051607
Andrews, T., & Webb, M. J. (2018). The dependence of global cloud and lapse rate feedbacks on the spatial structure of tropical Pacific
warming. Journal of Climate,31(2), 641–654. https://doi.org/10.1175/JCLI‐D‐17‐0087.1
Barnes, E. A., Dunn‐Sigouin, E., Masato, G., & Woollings, T. (2014). Exploring recent trends in Northern Hemisphere block ing. Geophysical
Research Letters,41,638–644. https://doi.org/10.1002/2013GL058745
Bekryaev, R. V., Polyakov, I. V., & Alexeev, V. A. (2010). Role of polar amplification in long‐term surface air temperature variations and
modern Arctic warming. Journal of Climate,23(14), 3888–3906. https://doi.org/10.1175/2010JCLI3297.1
Bitz, C. M., Shell, K. M., Gent, P. R., Bailey, D. A., Danabasoglu, G., Armour, K. C., et al. (2012). Climate sensitivity of the community
climate system model, version 4. Journal of Climate,25(9), 3053–3070. https://doi.org/10.1175/JCLI‐D‐11‐00290.1
Boer, G., & Yu, B. (2003). Climate sensitivity and response. Climate Dynamics,20(4), 415–429. https://doi.org/10.1007/s00382‐002‐0283‐3
Broccoli, A. J., Dahl, K. A., & Stouffer, R. J. (2006). Response of the ITCZ to Northern Hemisphere cooling. Geophysical Research Letters,33,
L01702. https://doi.org/10.1029/2005GL024546
Bromwich, D. H., Nicolas, J. P., Monaghan, A. J., Lazzara, M. A., Keller, L. M., Weidner, G. A., & Wilson, A. B. (2013). Central West
Antarctica among the most rapidly warming regions on Earth. Nature Geoscience,6(2), 139–145. https://doi.org/10.1038/ngeo1671
Caldeira, K., & Cvijanovic, I. (2014). Estimating the contribution of sea ice response to climate sensitivity in a climate model. Journal of
Climate,27(22), 8597–8607. https://doi.org/10.1175/JCLI‐D‐14‐00042.1
Cess, R. O., Potter, G. L., Zhang, M. H., Blanchet, J. P., Chalita, S., Colman, R., et al. (1991). Interpretation of snow‐climate feedback as
produced by 17 general circulation models. Science,253(5022), 888–892. https://doi.org/10.1126/science.253.5022.888
Chapman, W. L., & Walsh, J. E. (2007). Simulations of Arctic temperature and pressure by global coupled models. Journal of Climate,20(4),
609–632. https://doi.org/10.1175/JCLI4026.1
Chiang, J. C., & Bitz, C. M. (2005). Influence of high latitude ice cover on the marine Intertropical Convergence Zone. Climate Dynamics,
25(5), 477–496. https://doi.org/10.1007/s00382‐005‐0040‐5
Cohen, J., Screen, J. A., Furtado, J. C., Barlow, M., Whittleston, D., Coumou, D., et al. (2014). Recent Arctic amplification and extreme mid‐
latitude weather. Nature Geoscience,7(9), 627–637. https://doi.org/10.1038/ngeo2234
Colman, R. (2003). A comparison of climate feedbacks in general circulation models. Climate Dynamics,20(7–8), 865–873. https://doi.org/
10.1007/s00382‐003‐0310‐z
Crook, J. A., & Forster, P. M. (2014). Comparison of surface albedo feedback in climate models and observations. Geophysical Research
Letters,41, 1717–1723. https://doi.org/10.1002/2014GL059280
Cvijanovic, I., & Caldeira, K. (2015). Atmospheric impacts of sea ice decline in CO
2
induced global warming. Climate Dynamics,44(5–6),
1173–1186.
Flanner, M. G., Shell, K. M., Barlage, M., Perovich, D. K., & Tschudi, M. A. (2011). Radiative forcing and albedo feedback from the Northern
Hemisphere cryosphere between 1979 and 2008. Nature Geoscience,4(3), 151–155. https://doi.org/10.1038/ngeo1062
Francis, J. A., Chan, W., Leathers, D. J., Miller, J. R., & Veron, D. E. (2009). Winter Northern Hemisphere weather patterns remember
summer Arctic Sea‐ice extent. Geophysical Research Letters,36, L07503. https://doi.org/10.1029/2009GL037274
Francis, J. A., & Vavrus, S. J. (2012). Evidence linking Arctic amplification to extreme weather in mid‐latitudes. Geophysical Research
Letters,39, L06801. https://doi.org/10.1029/2012GL051000
10.1029/2018JD029093
Journal of Geophysical Research: Atmospheres
DUAN ET AL. 206
Acknowledgments
This work is supported by the National
Key Basic Research Program of China
(2015CB953601), the National Natural
Science Foundation of China (41422503
and 41675063), and the Fundamental
Research Funds for the Central
Universities. We also received support
from the Department of Global
Ecology, Carnegie Institution for
Science. The model simulations were
performed out at the High‐Performance
Computing Facility “MAZAMA”
funded by the Department of Global
Ecology, Carnegie Institution for
Science. CESM and files needed to run
it are available from NCAR. The data
set used to produce figures and tables in
this paper is archived on Zenodo and is
available at http://doi.org/10.5281/
zenodo.1482309.
Graversen, R. G., & Wang, M. (2009). Polar amplification in a coupled climate model with locked albedo. Climate Dynam ics,33(5), 629–643.
https://doi.org/10.1007/s00382‐009‐0535‐6
Gregory, J. M., Ingram, W. J., Palmer, M. A., Jones, G. S., Stott, P. A., Thorpe, R. B., et al. (2004). A new method for diagnosing radiative
forcing and climate sensitivity. Geophysical Research Letters,31, L03205. https://doi.org/10.1029/2003GL018747
Hall, A. (2004). The role of surface albedo feedback in climate. Journal of Climate,17(7), 1550–1568. https://doi.org/10.1175/1520‐
0442(2004)017<1550:TROSAF>2.0.CO;2
Hansen, J., Lacis, A., Rind, D., Russell, G., Stone, P., Fung, I., et al. (1984). Climate sensitivity: Analysis of feedback mechanisms. Climate
processes and climate sensitivity, 130–163. https://doi.org/10.1029/GM029p0130
Holland, M. M., & Bitz, C. M. (2003). Polar amplification of climate change in coupled models. Climate Dynamics,21(3–4), 221 –232. https://
doi.org/10.1007/s00382‐003‐0332‐6
Holland, M. M., Bitz, C. M., & Weaver, A. J. (2001). The influence of sea ice physics on simulations of climate change. Journal of Geophysical
Research,106(C9), 19639–19655. https://doi.org/10.1029/2000JC000651
Huang, Y., Ramaswamy, V., & Soden, B. (2007). An investigation of the sensitivity of the clear‐sky outgoing longwave radiation to atmo-
spheric temperature and water vapor. Journal of Geophysical Research,112, D05104. https://doi.org/10.1029/2005JD006906
Hunke, E. C., Lipscomb, W. H., Turner, A. K., Jeffery, N., & Elliott, S. (2010). CICE: the Los Alamos sea ice model documentation and
software user’s manual version 4.1 LA‐CC‐06‐012. T‐3 Fluid Dynamics Group, Los Alamos National Laboratory, 675.
Ingram, W. J., Wilson, C. A., & Mitchell, J. F. B. (1989). Modeling climate change: An assessment of sea ice and surface albedo feedbacks.
Journal of Geophysical Research,94(D6), 8609–8622. https://doi.org/10.1029/JD094iD06p08609
Johns, T. C., Durman, C. F., Banks, H. T., Roberts, M. J., McLaren, A. J., Ridley, J. K., et al. (2006). The new Hadley Centre climate model
(HadGEM1): Evaluation of coupled simulations. Journal of Climate,19(7), 1327–1353. https://doi.org/10.1175/JCLI3712.1
Jonko, A. K., Shell, K. M., Sanderson, B. M., & Danabasoglu, G. (2013). Climate feedbacks in CCSM3 under changing CO
2
forcing. Part II:
Variation of climate feedbacks and sensitivity with forcing. Journal of Climate,26(9), 2784–2795. https://doi.org/10.1175/JCLI‐D‐12‐
00479.1
Kay, J. E., Holland, M. M., Bitz, C. M., Blanchard‐Wrigglesworth, E., Gettelman, A., Conley, A., & Bailey, D. (2012). The influence of local
feedbacks and northward heat transport on the equilibrium Arctic climate response to increased greenhouse gas forcing. Journal of
Climate,25(16), 5433–5450. https://doi.org/10.1175/JCLI‐D‐11‐00622.1
Liu, J., Curry, J. A., Wang, H., Song, M., & Horton, R. M. (2012). Impact of declining Arctic Sea ice on winter snowfall. Proceedings of the
National Academy of Sciences,109(11), 4074–4079. https://doi.org/10.1073/pnas.1114910109
Mahlstein, I., & Knutti, R. (2011). Ocean heat transport as a cause for model uncertainty in projected Arctic warming. Journal of Climate,
24(5), 1451–1460. https://doi.org/10.1175/2010JCLI3713.1
Manabe, S., & Stouffer, R. J. (1980). Sensitivity of a global climate model to an increase of CO
2
concentration in the atmosphere. Journal of
Geophysical Research,85(C10), 5529–5554. https://doi.org/10.1029/JC085iC10p05529
Manabe, S., Stouffer, R. J., Spelman, M. J., & Bryan, K. (1991). Transient responses of a coupled ocean–atmosphere model to gradual
changes of atmospheric CO
2
. Part I. Annual mean response. Journal of Climate,4(8), 785–818. https://doi.org/10.1175/1520‐
0442(1991)004<0785:TROACO>2.0.CO;2
Massonnet, F., Fichefet, T., Goosse, H., Bitz, C. M., Philippon‐Berthier, G., Holland, M. M., & Barriat, P. Y. (2012). Constraining projecti ons
of summer Arctic Sea ice. The Cryosphere,6(6), 1383–1394. https://doi.org/10.5194/tc‐6‐1383‐2012
Meehl, G. A., Stocker, T. F., Collins, W. D., Friedlingstein, P., Gaye, T., Gregory, J. M., et al. (2007). Global climate projections.
Meraner, K., Mauritsen, T., & Voigt, A. (2013). Robust increase in equilibrium climate sensitivity under global warming. Geophysical
Research Letters,40,5944–5948. https://doi.org/10.1002/2013GL058118
Neale, R. B., Richter, J. H., Conley, A. J., Park, S., Lauritzen, P. H., Gettelman, A., et al. (2010) Description of the NCAR Community
Atmosphere Model (CAM 4.0) (Natl Cent Atmos Res, Boulder, CO), NCAR/TN‐485 + STR.
Oleson, K. W., Lawrence, D. M., Gordon, B., Flanner, M. G., Kluzek, E., Lawrence, P. J., et al. (2010). Technical description of version 4.0 of
the Community Land Model (CLM).
Overland, J. E., & Wang, M. (2010). Large‐scale atmospheric circulation changes are associated with the recent loss of Arctic Sea ice. Tellus
A,62(1), 1–9. https://doi.org/10.1111/j.1600‐0870.2009.00421.x
Peings, Y., & Magnusdottir, G. (2014). Response of the wintertime Northern Hemisphere atmospheric circulation to current and pro-
jected Arctic Sea ice decline: A numerical study with CAM5. Journal of Climate,27(1), 244–264. https://doi.org/10.1175/JCLI‐D‐13‐
00272.1
Pithan, F., & Mauritsen, T. (2014). Arctic amplification dominated by temperature feedbacks in contemporary climate models. Nature
Geoscience,7(3), 181–184. https://doi.org/10.1038/ngeo2071
Polyakov, I. V., Beszczynska, A., Carmack, E. C., Dmitrenko, I. A., Fahrbach, E., Frolov, I. E., et al. (2005). One more step toward a warmer
Arctic. Geophysical Research Letters,32, L17605. https://doi.org/10.1029/2005GL023740
Qu, X., & Hall, A. (2014). On the persistent spread in snow‐albedo feedback. Climate Dynamics,42(1–2), 69–81.
Raval, A., Oort, A. H., & Ramaswamy, V. (1994). Observed dependence of outgoing longwave radiation on sea surface temperature and
moisture. Journal of Climate,7(5), 807–821. https://doi.org/10.1175/1520‐0442(1994)007<0807:ODOOLR>2.0.CO;2
Rind, D., Healy, R., Parkinson, C., & Martinson, D. (1998). The role of sea ice in 2 × CO
2
climate model sensitivity: Part II: Hemispheric
dependencies. Oceanographic Literature Review,1(45), 37.
Screen, J. A., & Simmonds, I. (2010). The central role of diminishing sea ice in recent Arctic temperature amplification. Nature,464(7293),
1334–1337. https://doi.org/10.1038/nature09051
Screen, J. A., & Simmonds, I. (2013). Exploring links between Arctic amplification and mid‐latitude weather. Geophysical Research Letters,
40, 959–964. https://doi.org/10.1002/grl.50174
Serreze, M. C., Barrett, A. P., Stroeve, J. C., Kindig, D. N., & Holland, M. M. (2009). The emergence of surface‐based Arctic amplification.
The Cryosphere,3(1), 11–19. https://doi.org/10.5194/tc‐3‐11‐2009
Serreze, M. C., & Francis, J. A. (2006). The Arctic amplification debate . Climatic Change,76(3–4), 241–264. https://doi.org/10.1007/s10584‐
005‐9017‐y
Soden, B. J., & Held, I. M. (2006). An assessment of climate feedbacks in coupled ocean–atmosphere models. Journal of Climate,19(14),
3354–3360. https://doi.org/10.1175/JCLI3799.1
Södergren, A. H., McDonald, A. J., & Bodeker, G. E. (2018). An energy balance model exploration of the impacts of interactions between
surface albedo, cloud cover and water vapor on polar amplification. Climate Dynamics,51(5–6), 1639–1658.
Solomon, A. (2006). Impact of latent heat release on polar climate. Geophysical Research Letters,33, L07716. https://doi.org/10.1029/
2005GL025607
10.1029/2018JD029093
Journal of Geophysical Research: Atmospheres
DUAN ET AL. 207
Stroeve, J. C., Serreze, M. C., Holland, M. M., Kay, J. E., Malanik, J., & Barrett, A. P. (2012). The Arctic’s rapidly shrinking sea ice cover: A
research synthesis. Climatic Change,110(3–4), 1005–1027. https://doi.org/10.1007/s10584‐011‐0101‐1
Taylor, P. C., Cai, M., Hu, A., Meehl, J., Washington, W., & Zhang, G. J. (2013). A decomposition of feedback contributions to polar
warming amplification. Journal of Climate,26(18), 7023–7043. https://doi.org/10.1175/JCLI‐D‐12‐00696.1
Vavrus, S. (2007). The role of terrestrial snow cover in the climate system. Climate Dynamics,29(1), 73–88. https://doi.org/10.1007/s00382‐
007‐0226‐0
Wallace, J. M., Held, I. M., Thompson, D. W., Trenberth, K. E., & Walsh, J. E. (2014). Global warming and winter weather. Science,
343(6172), 729–730. https://doi.org/10.1126/science.343.6172.729
Wetherald, R. T., & Manabe, S. (1988). Cloud feedback processes in a general circulation model. Journal of the Atmospheric Sciences,45(8),
1397–1416. https://doi.org/10.1175/1520‐0469(1988)045<1397:CFPIAG>2.0.CO;2
Winton, M. (2006). Surface albedo feedback estimates for the AR4 climate models. Journal of Climate,19(3), 359–365. https://doi.org/
10.1175/JCLI3624.1
Zhang, T. (2005). Influence of the seasonal snow cover on the ground thermal regime: An overview. Reviews of Geophysics,43, RG4002.
https://doi.org/10.1029/2004RG000157
10.1029/2018JD029093
Journal of Geophysical Research: Atmospheres
DUAN ET AL. 208