Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

1 FEBRUARY 2023 KAUR ET AL.

775

Why Is Climate Sensitivity for Solar Forcing Smaller than for


an Equivalent CO2 Forcing?

HARPREET KAUR,a GOVINDASAMY BALA,a AND ASHWIN K. SESHADRIa,b


a
Centre for Atmospheric and Oceanic Sciences, Indian Institute of Science, Bangalore, India
b
Divecha Centre for Climate Change, Indian Institute of Science, Bangalore, India

(Manuscript received 29 December 2021, in final form 5 August 2022)

ABSTRACT: Previous studies have shown that climate sensitivity, defined as the global mean surface temperature change
per unit radiative forcing, is smaller for solar radiative forcing compared to an equivalent CO2 radiative forcing. We investi-
gate the causes for this difference using the NCAR CAM4 model. The contributions to the climate feedback parameter,
which is inversely related to climate sensitivity, are estimated for water vapor, lapse rate, Planck, albedo, and cloud feed-
backs using the radiative kernel technique. The total feedback estimated for CO2 and solar radiative forcing from our
model simulations is 21.23 and 21.45 W m22 K21, respectively. We find that the difference in feedback between the
two cases is primarily due to differences in lapse rate, water vapor, and cloud feedbacks, which together explain 65% of the
difference in total feedback. The rest comes from Planck and albedo feedbacks. The differences in feedbacks arise mainly
from differences in the horizontal (meridional) structure of forcing and the consequent warming. Our study provides
important insights into the effects of the meridional structure of forcing on climate feedback, which is important for evalu-
ating global climate change from different forcing agents.

SIGNIFICANCE STATEMENT: An increase in atmospheric CO2 concentration or an increase in incoming solar ra-
diation leads to a rise in the radiative budget and consequent climate warming, which is amplified by the presence of
multiple climate feedbacks. These feedbacks, from changes in surface albedo, combined effect of water vapor and the
vertical lapse rate of temperature, and changes in clouds, differ between solar and CO2 forcing. Using radiative kernels,
this study quantifies these individual feedbacks for an equivalent radiative change caused by an increase in CO2 or in-
coming solar radiation, showing how the differences arise from differences in the meridional patterns of warming. In
agreement with prior studies, these differences can explain the smaller efficacy of solar forcing compared to CO2
forcing.

KEYWORDS: Climate change; Climate sensitivity; Radiative fluxes; Radiative forcing; Climate models;
General circulation models

1. Introduction Shindell et al. 2015; Modak et al. 2016, 2018; Modak and Bala
2019).
An understanding of the climate response to different radi-
Several previous studies have shown that climate is more
ative forcing agents is fundamental for climate change studies.
sensitive to CO2 radiative forcing than solar radiative forcing
Accurate projections of the effects of mitigation of green-
(Forster et al. 2000; Hansen et al. 2005; Lambert and Faull
house gases as well as solar radiation modification to offset
2007; Bala et al. 2010; Schmidt et al. 2012; Modak et al. 2016;
global warming depend on improved estimation of climate
Ceppi and Gregory 2019). This implies that the same increase
feedbacks in response to CO2 increase as well as solar irradi-
of radiative flux at the top of the atmosphere (TOA) due to
ance reductions. To quantify the climate response to pertur-
CO2 forcing would result in larger global mean surface warm-
bations in various radiative forcing agents, the concept of
ing compared to solar forcing. Additionally, the magnitude of
climate sensitivity has been introduced (Cess et al. 1990; Bony
the difference between the climate sensitivities for two forcing
et al. 2006; Hansen et al. 1997). The climate sensitivity is de-
agents also depends on the definition of radiative forcing
fined as the ratio between global mean surface temperature
(Modak et al. 2018; Richardson et al. 2019). Different defini-
change and the radiative forcing (Shine 2000), and its esti-
mates have been found to depend on the radiative forcing tions of radiative forcing have emerged over the years. These
agent (Hansen et al. 1997; Joshi et al. 2003; Shine et al. 2003; include instantaneous radiative forcing, stratosphere-adjusted
Hansen et al. 2005; Ban-Weiss et al. 2012; Shindell 2014; radiative forcing, effective radiative forcing, and regressed
radiative forcing (Gregory et al. 2004; Hansen et al. 2005;
Ban-Weiss et al. 2012; Forster et al. 2016; Modak et al. 2018).
The instantaneous radiative forcing refers to the instanta-
Supplemental information related to this paper is available neous change in radiative flux at the tropopause immediately
at the Journals Online website: https://doi.org/10.1175/JCLI-D-21- after the forcing agent is introduced into the climate system,
0980.s1.
with the state of the entire climate system held fixed (Hansen
et al. 2005). Once radiative forcing is imposed, the strato-
Corresponding author: Harpreet Kaur, harpreetkaur@iisc.ac.in sphere adjusts rapidly and attains equilibrium to the altered

DOI: 10.1175/JCLI-D-21-0980.1
Ó 2023 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright
Policy (www.ametsoc.org/PUBSReuseLicenses).
Brought to you by University of Maryland, McKeldin Library | Unauthenticated | Downloaded 01/19/23 06:55 PM UTC
776 JOURNAL OF CLIMATE VOLUME 36

flux within a few months. Following this, the radiative flux and the corresponding radiative flux change at the TOA
change at TOA and tropopause become equal. The change in (Shell et al. 2008; Soden et al. 2008). In the case of cloud feed-
radiative flux at the tropopause once the stratosphere has backs, there are significant nonlinearities primarily arising
adjusted, while the state of the troposphere is unchanged, is from the overlaps between clouds at different levels (Song
called the stratospheric-adjusted forcing (Hansen et al. 2005). et al. 2014), due to which the radiative kernel technique does
The instantaneous radiative forcing, as well as strato- not give accurate results (Shell et al. 2008; Soden et al. 2008).
spheric-adjusted radiative forcing, exclude the effects of tro- Therefore, we estimate cloud feedbacks by calculating the ad-
pospheric and land surface changes, considering them to be justed change in cloud radiative forcing (Soden et al. 2008).
included in the climate response. However, subsequent stud- This involves estimating the change in cloud radiative forcing
ies (Andrews and Forster 2008; Gregory and Webb 2008; between perturbed and control simulations, which are subse-
Dong et al. 2009; Doutriaux-Boucher et al. 2009) have shown quently adjusted using radiative kernels to correct for the
that there are relatively rapid land surface and tropospheric masking effect of clouds on the other feedbacks.
adjustments, including cloud changes, that have effects on Previous studies point to fast tropospheric adjustments as
time scales considerably shorter than a month. Thus, another well as convective and low cloud feedbacks as the primary
definition of radiative forcing was introduced to estimate the causes of the difference between the climate sensitivity for so-
radiative forcing as the change in radiative fluxes at the top of lar and CO2 radiative forcing (Forster et al. 2000; Bony and
the atmosphere after the stratosphere, land surface, and tro- Dufresne 2005; Lambert and Faull 2007; Andrews et al. 2009;
posphere have adjusted (Myhre et al. 2013; Boucher et al. Bala et al. 2010; Cao et al. 2012; Modak et al. 2016; Zelinka
2013; Sherwood et al. 2015; Forster et al. 2016; Ramaswamy et al. 2020). In this paper, we estimate ERF using Hansen’s
et al. 2019). Radiative forcing estimated in this manner is prescribed-SST method (Hansen et al. 2005) and exclude the
called the “effective radiative forcing” (ERF) and is found to fast adjustments in our estimation of feedback parameters so
be a better predictor of global mean surface temperature that the differences in climate feedback parameters reported
change (Shine et al. 2003; Hansen et al. 2005; Ban-Weiss et al. in this paper reflect differences in the slow response.
2012; Myhre et al. 2013; Forster et al. 2016; Richardson et al. Specifically, we estimate the feedback parameter lX associ-
2019). ERF is formally defined as “the change in the net TOA ated with change DX in variable X, by dividing the corre-
downward radiative flux after allowing for atmospheric tem- sponding radiative flux change by the slow response in global
peratures, water vapor and clouds to adjust, but with surface mean surface temperature. The slow response in global mean
temperature or a portion of surface conditions unchanged” surface temperature is calculated as the difference between
(Myhre et al. 2013). The ERF definition of radiative forcing global mean surface warming from the slab ocean simulations
incorporates the effects of rapid adjustments that do not scale and the land surface warming from the prescribed-SST ex-
with the global mean surface temperature change, allowing periments. Correspondingly, climate sensitivity is evaluated as
one to exclude fast adjustments from the total climate re- the slow response in global mean surface warming, in re-
sponse. Since it is difficult to prescribe the land surface tem- sponse to a unit ERF. The calculations performed thus allow
perature in complex climate model simulations, some studies us to render estimates of individual feedback contributions af-
also allow adjustments to the land temperature change while ter excluding the effects of fast tropospheric and land surface
calculating ERF which removes the radiative effects of land adjustments.
warming on climate feedback (Richardson et al. 2019; Andrews
et al. 2021). The other part of the climate response that scales
2. Model and experiments
with the global mean surface temperature change is called
slow response. In the ERF framework, the slow response per To estimate radiative forcing and study the climate re-
unit global mean surface temperature change is called the sponse to solar and CO2 radiative forcing, we use the NCAR
feedback, and the climate feedback parameter is estimated as CESM1.0.4 model in two configurations: 1) the prescribed sea
the sensitivity of outgoing radiative flux at TOA to unit sur- surface temperature (SST) configuration, where the atmo-
face temperature change. As TOA radiative response should spheric component of the CESM, the Community Atmo-
be equal to ERF at equilibrium, the climate sensitivity for any sphere Model version 4 (CAM4) (Neale et al. 2010) is driven
forcing agent is the (negative) inverse of the climate feedback by prescribed SST for estimation of the effective radiative
parameter. forcing, and 2) the slab ocean model (SOM) configuration,
The climate feedback parameter can be decomposed as the where CAM4 is coupled with the slab ocean model for analy-
sum of feedback parameters describing individual climate sis of the climate response to the two forcings. The simula-
feedbacks. The value of individual climate feedbacks and, tions are performed at a horizontal resolution of 1.98 3 2.58
thus, the climate sensitivity differs between climate forcing (latitude 3 longitude), with 18 levels in the troposphere and
agents. In this paper, we quantify and compare the individual 8 levels in the stratosphere. In both configurations, CAM4
climate feedbacks (albedo, Planck, lapse rate, water vapor, is coupled to the land model, the Community Land Model
and cloud) for solar and CO2 radiative forcing. We estimate version 4 (CLM4) (Oleson et al. 2010).
individual feedback parameters using the radiative kernels ap- We perform a set of three experiments: 1) the control simula-
proach (Shell et al. 2008; Soden et al. 2008; Jonko et al. 2012; tion (Ctrl) with preindustrial CO2 concentration of 284.7 ppm
Sanderson and Shell 2012; Yue et al. 2016; Liu et al. 2018), and solar constant of 1361 W m22, 2) the “2XCO2” simulation
which assumes linearity between the changes in each variable where atmospheric CO2 concentration is doubled to 569.4 ppm,

Brought to you by University of Maryland, McKeldin Library | Unauthenticated | Downloaded 01/19/23 06:55 PM UTC
1 FEBRUARY 2023 KAUR ET AL. 777

TABLE 1. Effective radiative forcing (ERF), slow global mean surface temperature change (DTS), albedo (la), Planck (lPlanck),
lapse rate (lLapse), water vapor (lWV), the shortwave (lCldsw) and longwave (lCldlw) cloud feedback parameters, and the sum of all
feedback parameters, as well as the total feedback parameter (l*) estimated from the model. The ERF is estimated from the
prescribed-SST simulations. The slow global mean surface temperature change is estimated as the difference between the total global
mean surface temperature change (estimated from slab ocean simulations) and the fast global mean surface temperature change
(estimated from the prescribed-SST simulation). Global-mean values of individual feedback parameters (W m22 K21) are calculated
using the radiative kernel technique (Shell et al. 2008). Cloud feedback is calculated using the adjusted change in cloud radiative
forcing (Soden et al. 2008). The uncertainties in the estimates are represented by the standard deviation of the annual mean,
estimated from a 20-yr time series for ERF, and 50 years for temperature change and the feedback parameters.

2XCO2 Solar Solar 2 2XCO2


22
ERF (W m ) 3.59 6 0.05 4.08 6 0.07 0.49 6 0.04
DTS (K) 2.93 6 0.013 2.82 6 0.007 20.11 6 0.016
la (W m22 K21) 0.32 6 0.019 0.28 6 0.019 20.04 6 0.02
lPlanck (W m22 K21) 22.95 6 0.01 22.99 6 0.01 20.04 6 0.01
lLapse (W m22 K21) 20.14 6 0.004 20.29 6 0.003 20.15 6 0.005
lWV (W m22 K21) 1.33 6 0.004 1.42 6 0.003 0.09 6 0.008
lCldsw (W m22 K21) 0.07 6 0.009 0 6 0.006 20.07 6 0.008
lCldlw (W m22 K21) 0.14 6 0.003 0.12 6 0.002 20.02 6 0.003
Sum (W m22 K21) 21.23 6 0.01 21.46 6 0.01 20.23 6 0.018
l* (W m22 K21) 21.23 6 0.006 21.45 6 0.008 20.22 6 0.005

and solar constant maintained at 1361 W m22 and 3) the “Solar” and cloud feedbacks (Shell et al. 2008; Soden et al. 2008;
simulation where CO2 concentration is prescribed at the prein- Pendergrass et al. 2018a):
dustrial level of 284.7 ppm and solar constant increased by 2%
l 5 la 1 lP 1 lLR 1 lWV 1 lC 1 Residual:
(27.22 W m22). Increasing the solar constant by 2% yields a radi-
ative forcing value close to the radiative forcing (Table 1) in the Here, la is albedo feedback, lP is Planck feedback, lLR is
2XCO2 simulation. Both the forcing experiments, 2XCO2 and lapse rate feedback, lWV is water vapor feedback, and lC is
Solar, are performed in both prescribed-SST and SOM configu- cloud feedback.
rations. The prescribed-SST simulations last for 40 years, and the To identify the sources of differences in climate sensitivity
last 20 years are used for estimating the radiative forcing. The for solar and CO2 radiative forcing, we compute the individ-
SOM simulations last for 100 years, and the last 50 years of simu- ual feedback parameters for each of these forcings. The radia-
lation are used to estimate the equilibrium climate response. tive kernel approach (Shell et al. 2008; Soden et al. 2008;
Pendergrass et al. 2018a) is used for the estimation of all non-
cloud feedbacks. The kernel approach makes a linear approxi-
3. Methodology mation for the climate feedbacks and decomposes the
The global mean temperature response to radiative forcing feedback parameter corresponding to variable X
is measured by the climate feedback parameter l. The feed- DRX
back parameter l is defined as the change in radiative flux at lX 5
DTS
the top of the atmosphere caused by an increase in global
mean surface temperature of 1 K (Collins at al. 2013). Once into two factors, the radiative kernel and the climate re-
this feedback parameter is known, it can be used to estimate sponse, as
the climate sensitivity, which is defined as the global warming  
per unit radiative forcing. The radiative forcing is related to DX
lX 5 KX ,
the climate feedback parameter l as DTS

F 5 2lDTS , where radiative kernel KX 5 DRX/DX represents the change


in net incoming TOA radiative flux for a unit standard
where F is the effective radiative forcing and DTS is the slow change in a variable X (variable X can be albedo, tempera-
change in global mean surface temperature at equilibrium. ture, lapse rate, water vapor, or clouds). Since the estima-
The changes in TOA radiative flux and the global mean sur- tion of the kernels is computationally expensive and these
face temperature between the corresponding forcing experi- kernels are similar for all climate models (Soden et al. 2008),
ment and preindustrial control climate are calculated from we use precalculated radiative kernels that have been made
prescribed-SST and slab ocean model simulations, respec- publicly available (Shell et al. 2008; Pendergrass et al. 2018b).
tively, to estimate climate sensitivity and thus, the climate The kernels used in our study are based on the NCAR Com-
feedback parameters. Assuming that the different feedbacks munity Atmospheric Model (CAM3) with a horizontal reso-
are additive, the climate feedback parameter l is approxi- lution of 2.88 latitude and 2.88 longitude with 17 vertical
mated as the sum of albedo, Planck, lapse rate, water vapor, layers (Shell et al. 2008).

Brought to you by University of Maryland, McKeldin Library | Unauthenticated | Downloaded 01/19/23 06:55 PM UTC
778 JOURNAL OF CLIMATE VOLUME 36

For calculating the feedback parameters, we use the are characterized by their standard deviations, estimated from
monthly average of the slow response in all variables involved corresponding time series of 50 annual means for each of these
in climate feedback processes (albedo, Planck, lapse rate, and variables. The time series of annual means for slow change for
water vapor). The slow response for a climate variable is then each variable is evaluated as the difference between the time se-
normalized by the slow response in global mean surface tem- ries of the total response and the average of the fast response
perature change (the difference between total and fast surface calculated from the last 20 years of the prescribed-SST
temperature response), for each grid point and vertical level simulations.
for all 12 months of the year. Multiplying this climate re-
sponse by the radiative kernel for the corresponding grid 4. Results
point and vertical level, summing across levels and averaging
Table 1 shows the effective radiative forcing, slow global
in the horizontal, we obtain the month-wise contribution to
mean surface temperature change, individual feedback pa-
the feedback, which is then averaged across months to obtain
rameters, and their sum, as well as the net climate feedback
the annually averaged feedback.
parameter for solar and CO2 forcings. The total or net climate
As the kernel method of estimating feedbacks is based on
feedback factor l* is estimated from model simulations as the
assuming linearity between the perturbation and the resulting
ratio of radiative flux change from the prescribed-SST experi-
change in TOA radiative flux, which does not hold for clouds,
ments and the slow global mean surface temperature change
we cannot use the kernel technique directly for cloud feed-
(change in slab ocean experiments minus the change in
backs. Instead, the cloud feedback is estimated as the differ-
prescribed-SST experiments). The individual feedbacks are cal-
ence of cloud radiative forcing (CRF) (Ramanathan et al.
culated using the radiative kernel approach as discussed in the
1989) between the experiment and the Ctrl run, which we de-
previous section, with modifications for cloud feedbacks. Net
note as dCRF. CRF is estimated by subtracting the TOA radi-
feedback factors are negative in both 2XCO2 and solar forcing
ative flux in clear-sky conditions (without clouds) from the
cases, and the feedback factor for solar forcing is larger in
corresponding TOA radiative flux under all-sky conditions
magnitude (21.45 W m22 K21) compared to CO2 forcing
(including clouds). The cloud feedback dCRF estimated by (21.23 W m22 K21). The net feedback factors are closely
this method differs from the generally more accurate calcula- approximated by the sum of individual feedback factors:
tions of the cloud feedbacks that result from the partial radia- 21.46 W m22 K21 for solar forcing and 21.23 W m22 K21 for
tive perturbation method (Wetherald and Manabe 1988; CO2 radiative forcing.
Zhang et al. 1994; Colman 2003; Soden et al. 2004). This dis- Individual feedbacks have the same signs for the two forc-
crepancy in results comes from the masking of the clear-sky ings (Table 1). Albedo, water vapor, and longwave cloud
response of noncloud feedbacks when one subtracts the clear- feedbacks are positive in both cases, while Planck and lapse
sky forcing to obtain the dCRF (Zhang et al. 1994; Colman rate feedbacks are negative. Much of the difference in the
2003; Soden et al. 2004). By masking the surface (Soden et al. sum of feedbacks (20.23 W m22 K21) is explained by a few
2008; Shell et al. 2008), for instance, clouds diminish the effects individual feedbacks: with 65% of this explained by the lapse
of the other kernels, including temperature, albedo, and water rate, water vapor, and cloud feedbacks. The water vapor and
vapor on the clear-sky fluxes. The partial radiative perturba- lapse rate feedback are closely related and tend to compen-
tion method seeks to isolate the effects of cloud changes on ra- sate each other (Zhang et al. 1994; Cess 1975; Soden and
diative forcing, while keeping the other variables fixed at their Held 2006; Held and Shell 2012). While solar forcing causes a
values in the unperturbed climate. In contrast, the dCRF is cal- stronger negative lapse rate feedback, it also leads to a slightly
culated with all variables changing between the experiment more positive water vapor feedback. As a result, the sum of
and the control run. This disparity is adjusted by adding a cor- lapse rate and water vapor feedbacks for the two forcing
rection term to the value of dCRF, giving rise to an adjusted agents have only a small difference, with a value that is more
change in cloud radiative forcing (Soden et al. 2008). negative by 0.06 W m22 K21 for solar radiative forcing. Both
albedo feedback and Planck feedback, which have opposite
lC 5 dCRF 1 (KT0 2 KT )dT 1 (KW
0
2 KW )dW signs, are respectively less positive and more negative for so-
1 (Ka0 2 Ka )da 1 (F 0 2 F), lar forcing, by 0.04 W m22 K21. The standard deviation for
differences in many individual feedbacks between the Solar
dCRF 5 dCRFExp 2 dCRFCtrl , and 2XCO2 cases is smaller than the standard deviation for
the individual feedbacks (Table 1). This is because the indi-
where CRFExp is the cloud radiative forcing for the perturba- vidual feedbacks are correlated with each other which results
tion experiment and CRFCtrl is cloud radiative forcing for the in the standard deviation value for the difference to be less
control simulation (Ctrl), (F 0 2 F) is the difference between than the sum of individual standard deviation.
clear-sky and cloudy sky radiative forcing, and (KX0 2 KX ) is The solar radiative forcing decreases rapidly toward the
the difference between clear-sky and all-sky radiative kernels poles (Fig. 1). In comparison, the spatial pattern of CO2 radia-
of variable X (T is temperature, W is water vapor, and a is tive forcing is more uniform across latitudes (Hansen et al.
surface albedo). 1997). As a result, high-latitude warming is more pronounced
The uncertainties in estimates of radiative forcing and climate in the CO2 forcing case as compared to solar forcing (Fig. 2).
feedbacks, as well as the slow change in surface temperature, In contrast, tropical warming is larger for solar forcing for

Brought to you by University of Maryland, McKeldin Library | Unauthenticated | Downloaded 01/19/23 06:55 PM UTC
1 FEBRUARY 2023 KAUR ET AL. 779

FIG. 1. Zonal mean distributions of effective radiative forcing


(normalized by slow change in global mean surface temperature)
in the (a) 2XCO2 (black) and (b) Solar (red) simulations relative
to the Ctrl simulation and the difference (dotted black) estimated FIG. 2. Model-simulated spatial patterns of changes in surface
from the prescribed-SST simulations. The shaded green region rep- temperature per unit global mean surface temperature change in
resents the positive change, and the shaded blue region represents (a) 2XCO2 and (b) Solar simulations, and (c) the difference be-
the negative change in the difference value of the zonally averaged tween the two cases (Solar minus 2XCO2). The results shown are
solar and CO2 radiative forcing. All the results are annual mean the slow surface temperature response estimated as the total re-
values averaged over years 21–40 out of the 40-yr simulations. sponse minus the fast adjustment. The total response is calculated
as the annual mean values over the last 50 years of the 100-yr
CAM4 slab ocean simulations and the fast adjustment is esti-
similar values of global mean radiative forcing (Fig. 2). These
mated as annual mean values over the last 20 years of the 40-yr
differences in the warming patterns further give rise to the dif- CAM4 prescribed-SST simulations. The stippling in the difference
ferences in individual climate feedbacks. plot in (c) represents statistically significant differences, at the
95% confidence level, estimated using Student’s t test based on
a. Surface albedo feedback
the corresponding 50-yr annual mean samples.
The surface albedo feedback contributes 17% to the global
mean difference in the sum of feedbacks (Table 1). The differ- patterns of surface warming. As discussed in section 4a,
ence in surface albedo feedback between the Solar and 2XCO2 low-latitude warming is larger for solar forcing compared to
cases is related to the difference in the latitudinal distribution CO2 forcing, whereas CO2 forcing leads to larger warming
of radiative forcing and surface warming (Figs. 1 and 2). As in the high latitudes (Fig. 2). The larger warming in the
discussed above, high-latitude warming is larger in the CO2 tropics for solar forcing causes a larger increase in outgoing
forcing compared to solar forcing (Fig. 2). In contrast, tropical longwave radiation compared to CO2 radiative forcing. In
warming is larger for solar forcing compared to CO2 forcing
contrast, the increase in outgoing longwave radiation and,
for similar values of global mean radiative forcings. These dif-
thereby, the magnitude of Planck feedback in the high lati-
ferences translate to differences in the albedo response to
tudes is larger for CO2 forcing (Fig. 4). Since the magnitude
radiative forcing, as shown in Fig. 3. Much of the albedo feed-
of the Planck feedback (;4sT3) increases with the tempera-
back results from albedo changes caused by melting of sea ice
ture of the region, and given the much larger area of the low
and snow in high latitudes (Hall 2004). Since CO2 forcing
latitudes where the Planck feedback is stronger for solar
leads to larger warming in these latitudes, the albedo feed-
forcing (Fig. 4), the overall contribution from the tropics is
back is larger in the 2XCO2 case (Fig. 3). Figure 3 shows the
much larger than from polar regions. This results in the
spatial patterns of differences in albedo change per unit slow
magnitude of Planck response being larger and the resulting
response in global mean surface temperature change and cor-
feedback being more negative in the case of solar forcing, as
responding differences in the albedo feedback parameter be-
shown in Table 1.
tween the two forcings (Solar minus 2XCO2). Clearly, these
differences are concentrated in higher latitudes, where the
c. Lapse rate feedback
surface albedo is sensitive to differences in warming.
The lapse rate feedback is a measure of TOA radiative flux
b. Planck feedback change in response to the departure of tropospheric warming
The atmospheric temperature feedback is decomposed into from a vertically uniform profile and is related to the change
two parts: 1) Planck feedback and 2) lapse rate feedback. The in tropospheric lapse rate (Soden et al. 2008). The lapse rate
part of temperature feedback that is attributed to the verti- changes in the perturbed climate relative to the Ctrl climate
cally uniform component of tropospheric warming is the are different between the two forcings and this causes differ-
“Planck” feedback. The Planck feedback depends on spatial ing lapse rate feedbacks (Table 1). In prior studies, the

Brought to you by University of Maryland, McKeldin Library | Unauthenticated | Downloaded 01/19/23 06:55 PM UTC
780 JOURNAL OF CLIMATE VOLUME 36

FIG. 3. Model-simulated differences in the spatial patterns of (a) albedo response and (b) albedo feedbacks for solar
and CO2 radiative forcing. The albedo feedback is calculated as a product of two factors: the slow response in albedo
per unit global mean slow surface temperature change and the CAM3 (Shell et al. 2008) radiative kernels for the sur-
face albedo feedback. Slow response for each variable is evaluated here as the difference between total response and
the fast adjustment. The total response is calculated as the variable annual mean values over the last 50 years of the
100-yr CAM4 slab ocean simulations and the fast adjustment is estimated as variable annual mean values over the last
20 years of the 40-yr CAM4 prescribed-SST simulations. The stippling in difference plots represents statistically signif-
icant differences, at the 95% confidence level, estimated using Student’s t test based on corresponding 50-yr annual
mean samples.

difference between the lapse rate feedback for the two forcings Schlesinger and Mitchell 1987; Colman et al. 2001; Crook et al.
has been attributed to different background climate states es- 2011; Atwood et al. 2016; Feldl et al. 2017). This effect is larger
tablished after fast adjustments (Modak et al. 2016), as well as for CO2 radiative forcing as it causes more warming in these
the different SST spatial patterns (Cronin and Jansen 2016; latitudes. These differences can be seen in Fig. 5. The overall
Rugenstein et al. 2016; Andrews and Webb 2018; Ceppi and result of these contrasting lapse rate responses is that the lapse
Gregory 2019). As we calculate the feedback parameters using rate feedback is more negative in low latitudes and less positive
only the slow response in this paper, the differences due to fast in high latitudes for solar forcing, as compared to CO2 forcing.
adjustments do not contribute to the differences in individual This contributes to the more negative lapse rate feedback for
feedback parameters. solar forcing (Table 1).
Several previous studies have shown that the difference Table S1 and Figs. S2–S4 show an estimate of lapse rate
in the vertical profiles of slow temperature change (Fig. S1b feedback from the tropical (208S–208N), Arctic (608–908N)
in the online supplemental material) is due to the SST pattern and Antarctic regions (908–608S). The net contribution to the
effect (Armour et al. 2013; Andrews et al. 2015; Rugenstein differences between lapse rate feedbacks for solar and
et al. 2016; Ceppi and Gregory 2017; Zhou et al. 2017; Andrews 2XCO2 experiments from the Arctic, tropics, and Antarctic
and Webb 2018; Ceppi and Gregory 2019). The SST pattern regions are 20.20, 20.12, and 20.14 W m22 K21 respectively.
effect refers to dependence of the global radiative response on It can be seen that both the tropical and polar regions contrib-
time-varying spatial patterns of warming (Stevens et al. 2016; ute to the feedback differences in total lapse rate feedback be-
Armour 2017; Proistosescu and Huybers 2017; Andrews and tween 2xCO2 and solar forcing. Moreover, smaller surface
Webb 2018; Marvel et al. 2018; Dong et al. 2019; Gregory et al. area in the Arctic reduces the contribution of polar region to
2020; Rugenstein et al. 2020). In the tropics, the troposphere is total lapse rate feedback, which further enhances the contri-
close to radiative–convective equilibrium (Sobel et al. 2001). bution from tropical regions.
The vertical temperature structure is approximately determined
d. Water vapor feedback
by the moist adiabatic lapse rate (Manabe and Stouffer 1980;
Santer et al. 2005), which decreases with increasing surface tem- Atmospheric water vapor is the most important greenhouse
perature and relative humidity. Since solar forcing leads to gas (Colman 2003; Schmidt et al. 2010; Reay et al. 2007). An
larger surface warming in the tropics, the lapse rate becomes increase in atmospheric water vapor increases absorption of
correspondingly smaller, resulting in more negative lapse rate both longwave and near-infrared solar radiation and thereby
feedback compared to the CO2 case. In contrast, in the polar re- yields positive feedbacks for both CO2 and solar forcing. As
gions, the air is cold and stable to convection, and warming is shown in Table 1, the global water vapor feedback resulting
mostly confined to the lower levels of the atmosphere (Manabe from solar radiative forcing is larger by 0.09 W m22 K21 com-
and Wetherald 1975; Screen et al. 2012). As a result, the lapse pared to CO2 radiative forcing. With constant relative humi-
rate feedback in these latitudes is positive (Ramanathan 1977; dity, any change in the logarithm of specific humidity is

Brought to you by University of Maryland, McKeldin Library | Unauthenticated | Downloaded 01/19/23 06:55 PM UTC
1 FEBRUARY 2023 KAUR ET AL. 781

FIG. 4. Model-simulated spatial patterns of vertically integrated


Planck feedback in (a) 2XCO2 and (b) Solar simulations and (c)
the difference between the two cases (Solar minus 2XCO2). The
feedback is calculated as a product of two factors: 1) the slow
change in uniform atmospheric temperature per unit global
mean slow surface temperature change and 2) the CAM3 (Shell
et al. 2008) radiative kernels for the Planck feedback. The slow
response of atmospheric temperature is evaluated here as the
difference between the total response and the fast adjustment.
The total response is calculated as the annual mean values over FIG. 5. The zonal mean lapse rate feedback [W m 22 K 21
the last 50 years of the 100-yr CAM4 slab ocean simulations (100 hPa) 21 ] in the (a) 2XCO2 and (b) Solar forcings, and (c) the
and the fast adjustment is estimated as annual mean values difference between the two cases (Solar minus 2XCO2). The lapse
over the last 20 years of the 40-yr CAM4 prescribed-SST simu- rate feedback is calculated as a product of two factors: 1) the de-
lations. The stippling in the difference plot in (c) represents sta- parture of slow temperature response from a vertically uniform re-
tistically significant differences, at the 95% confidence level, sponse per unit global mean slow surface temperature change and
estimated using Student’s t test based on the corresponding 2) the CAM3 (Shell et al. 2008) radiative kernels for the lapse rate
50-yr annual mean samples. feedback. The slow response of atmospheric temperature here is
evaluated as the difference of the total response and the fast ad-
justment. The total response is calculated as the annual mean
approximately proportional to the tropospheric temperature values over the last 50 years of the 100-yr CAM4 slab ocean
change, following the Clausius–Clapeyron relation. As can be simulations and the fast adjustment is estimated as annual mean
seen in Fig. S5, the specific humidity response closely follows values over the last 20 years of the 40-yr CAM4 prescribed-SST
the differences in the temperature response (Figs. S1a–c). So- simulations. The stippling in the difference plot in (c) represents
statistically significant differences, at the 95% confidence level, es-
lar forcing, with its warmer middle-to-upper troposphere
timated using Student’s t test based on the corresponding 50-yr an-
(Figs. S1a–c), leads to higher specific humidity in the upper nual mean samples.
tropospheric levels. This leads to a larger water vapor feed-
back in the case of solar forcing, as shown in Fig. 6c. The wa-
e. Cloud feedback
ter vapor feedback, and the difference in the feedback
between the two forcings (Fig. 6c), mainly arises in tropical re- Cloud feedback generally involves the net result of various
gions, where the water vapor content is largest, and the atmo- effects of changing clouds on the shortwave and longwave ra-
sphere is closer to saturation (Held and Soden 2000; Soden diative fluxes. The net cloud feedback is positive in both the
et al. 2008; Shell et al. 2008; Jonko et al. 2013). There is a Solar and 2XCO2 cases (Table 1). We can decompose the to-
strong interhemispheric asymmetry for water vapor feedback tal cloud feedback into longwave and shortwave components.
for solar and CO2 radiative forcing (Fig. 6). The difference in Table 1 shows that the global shortwave cloud feedback is
water vapor feedback in the NH and in the SH originates positive for CO2 forcing and almost zero for the Solar case.
from the interhemispheric asymmetry in the kernel and water Longwave cloud feedback is positive for the two forcings and
vapor change for the two forcings (Fig. S6). However, the has a comparable magnitude in both cases. Thus, the differ-
asymmetry is more prominent in water vapor change with ence in net cloud feedback between the two cases is primarily
larger increase in the SH than in the NH, which is likely due due to differing shortwave cloud feedbacks.
to the asymmetric distribution of land and ocean areas be- The sign and magnitude of cloud feedback would depend
tween the NH and SH. on the change in altitude of clouds and change in fraction and

Brought to you by University of Maryland, McKeldin Library | Unauthenticated | Downloaded 01/19/23 06:55 PM UTC
782 JOURNAL OF CLIMATE VOLUME 36

FIG. 7. The zonal mean slow response in cloud fraction (%) per
FIG. 6. The zonal mean water vapor feedback [W m22 K21 unit global mean slow surface temperature change in (a) 2XCO2
(100 hPa)21] in the (a) 2XCO2 and (b) Solar simulations, and and (b) Solar simulations, and (c) the difference between the two
(c) the difference between the two (Solar minus 2XCO2). The (Solar minus 2XCO2). The slow cloud response is estimated as the
water vapor feedback is calculated as the product of two factors: difference of the total response and the fast adjustment. The total
slow water vapor response per unit global mean slow surface tem- response is calculated as the annual mean values over the last
perature change and the CAM3 (Shell et al. 2008) radiative kernels 50 years of the 100-yr CAM4 slab ocean simulations and the fast ad-
for the lapse rate feedback. The slow response of atmospheric tem- justment is estimated from annual mean values over the last 20 years
perature here is evaluated as the difference between total response of the 40-yr CAM4 prescribed-SST simulations. The stippling in the
and the fast adjustment. The total response is calculated as the an- difference plot in (c) represents statistically significant differences, at
nual mean values over the last 50 years of the 100-yr CAM4 slab the 95% confidence level, estimated using Student’s t test based on
ocean simulations and the fast adjustment is estimated from annual the corresponding 50-yr annual mean samples.
mean values over the last 20 years of the 40-yr CAM4 prescribed-
SST simulations. The stippling in difference plot in (c) represents
statistically significant differences, at the 95% confidence level, esti-
Figures 7a and 8a show that CO2 forcing leads to decrease
mated using Student’s t test based on the corresponding 50-yr in low clouds in the subtropical and midlatitude regions
annual mean samples. (308–608) in both hemispheres, decreases in high clouds around
300 mb in the tropical regions, and an increase in the overall
height of high clouds in the upper troposphere (Figs. 7 and 9).
optical depth of clouds. An increase in altitude of high-level This is consistent with prior work (Zelinka et al. 2012), which
clouds (pressure , 400 mb; 1 mb 5 1 hPa) results in positive shows that warming leads to deepening of the troposphere,
longwave cloud feedback due to the increase in the green- thereby leading to an increase in the altitude of cloud tops. As
house effect of clouds (Ceppi and Gregory 2017). In contrast, a result, the overall feedback of clouds from CO2 forcing is
increases in cloud fraction and optical depth particularly of positive (Table 1).
low-level clouds increases the reflection of incoming short- In the Solar case, the slow cloud response shows a decrease of
wave radiation by clouds and hence lead to negative short- low clouds in the subtropical and midlatitude regions (Figs. 7b
wave cloud feedback (Zelinka et al. 2016; McCoy et al. 2017). and 8b; see also Fig. S7b) and an increase in the height of high
The overall cloud feedback depends on how much of the clouds in the upper troposphere (Figs. 7b and 9). The cloud re-
change occurs for low versus high clouds. Since low clouds sponse to radiative forcing strongly depends on SST pattern
have a larger shortwave effect of reflecting radiation to space, change through various dynamic, thermodynamic, and radia-
the contribution of an increase in low-level clouds is negative tive processes. Many studies have shown a strong correlation
feedback. In contrast, high clouds have a larger greenhouse between the low-cloud cover and lower tropospheric stability
effect, and thus an increase in high clouds contributes posi- (LTS) (Klein and Hartmann 1993; Miller 1997; Larson et al.
tively to cloud feedbacks. 1999; Williams et al. 2008; Qu et al. 2015a; Zhou et al. 2016;

Brought to you by University of Maryland, McKeldin Library | Unauthenticated | Downloaded 01/19/23 06:55 PM UTC
1 FEBRUARY 2023 KAUR ET AL. 783

FIG. 8. Model-simulated spatial patterns of vertically integrated slow change in cloud fraction per unit global mean
slow surface temperature change, showing low cloud changes in the (a) 2XCO2 and (b) Solar simulations and high
cloud changes in the (c) 2XCO2 and (d) Solar simulations, as well as the difference between them (Solar minus
2XCO2) for (e) low clouds and (f) high clouds. The slow cloud response is estimated as the difference of the total re-
sponse and the fast adjustment. The total response is calculated as the annual mean values over the last
50 years of the 100-yr CAM4 slab ocean simulations and the fast adjustment is estimated from annual mean values
over the last 20 years of the 40-yr CAM4 prescribed-SST simulations. The stippling in the difference plots in (e) and
(f) represents statistically significant differences, at the 95% confidence level, estimated using Student’s t test based on
the corresponding 50-yr annual mean samples.

Ceppi and Gregory 2017; Zhou et al. 2017; Andrews and The decrease in low clouds is smaller in the Solar case than
Webb 2018). The LTS is estimated as the difference in poten- in the 2XCO2 case (Figs. 7c and 8e), but the decrease in
tial temperature at 700 hPa and near the surface and is a high clouds around 300 mb is larger (Figs. 7c, 9c), implying
measure of the inversion strength of the planetary boundary that both longwave and shortwave cloud feedbacks would
layer. The LTS increases in a warm climate leading to an in- be smaller (i.e., less positive) for solar radiative forcing
crease in low cloud cover, which results in negative short- than CO2 radiative forcing. These changes in cloud fraction
wave cloud feedback (Figs. S7 and S9) (Miller 1997; Qu et al. are related to differential warming of the troposphere (Fig. S1):
2015b). We do find some evidence for this in the subtropical warming is larger in the upper troposphere in the Solar case,
region in our simulations (Fig. S8). However, there are sev- leading to a larger decrease in high clouds around 300 mb
eral other factors, such as changes in relative humidity, verti- (Fig. 7c). In the 2XCO2 case, in contrast, warming is larger
cal velocity, and near-surface wind speed, which could also in the lower troposphere (Fig. S1), causing a larger decline
have competing effects on cloud fraction change and hence, in low clouds (Figs. 7c and 9c). The spatial and meridional
cloud feedbacks (Fig. S7) (Myers et al. 2021). A detailed in- patterns of longwave, shortwave, and total cloud feedbacks
vestigation of the effects of these processes is beyond the (Figs. S9 and S10 and Fig. 10) in the 2XCO2 experiment are
scope of our study. similar to those of previous studies (Gettelman et al. 2012).

Brought to you by University of Maryland, McKeldin Library | Unauthenticated | Downloaded 01/19/23 06:55 PM UTC
784 JOURNAL OF CLIMATE VOLUME 36

FIG. 9. Vertical profile of globally averaged change in cloud fraction in the 2XCO2 and Solar simulations: (a) the fast adjustments as es-
timated from the prescribed-SST simulations, (b) the slow response calculated as the total change minus fast adjustments, and (c) the total
cloud change from slab ocean simulations. Results are computed from annual mean values over the last 20 (50) years of the 40-yr (100-yr)
prescribed-SST (slab ocean) simulations for the estimation of fast adjustments (total change).

The difference in net cloud feedback (Table 1) between the regions in both hemispheres, and a stronger zonal mean signa-
2XCO2 and Solar cases arises mainly from the differing short- ture in the Southern Hemisphere owing to reduction in low
wave cloud feedbacks (Fig. 10). The shortwave cloud feed- clouds. The feedback is slightly less positive in the Solar case
back (Fig. 10b; see also Figs. S9a,b) in the 2XCO2 and Solar (Fig. 10b; see also Fig. S9e) in these regions, which is consistent
cases is consistent with the low cloud and LTS changes dis- with a slightly smaller reduction of low clouds (Figs. 7c and 8e;
cussed above (Figs. 7 and 8a,b; see also Figs. S7b and S8), see also Fig. S7b). The longwave cloud feedback (Figs. S9c,d)
with low cloud changes correlating to changes in the LTS giv- has large spatial variability, especially in tropical regions. How-
ing rise to positive feedback in the subtropical and midlatitude ever, averaged around the globe, the magnitudes of long-
wave cloud feedback are similar in the two cases (Table 1).
Thus, we conclude that a slightly smaller reduction of low
clouds in the subtropical and midlatitude regions in the case
of Solar forcing, compared to the 2XCO2 case, leads to a
slightly less positive overall (shortwave 1 longwave) cloud
feedback in the Solar case compared to the 2XCO2 case
(0.12 vs 0.21 W m22 K21).

5. Discussion and conclusions


In this paper, we have examined the climate system re-
sponse to solar and CO2 radiative forcings using a climate
modeling framework. We have identified how the climate sen-
sitivity to solar radiative forcing is smaller than to CO2 radia-
tive forcing because of the differences in climate feedbacks,
which is related to the difference in the latitudinal pattern of
warming between the two cases (Fig. 2). We have used the ra-
diative kernel technique (Shell et al. 2008; Soden et al. 2008;
Pendergrass et al. 2018a), finding that differences between the
total climate feedback parameter for solar and CO2 radiative
forcings mainly arise from lapse rate, water vapor, and cloud
FIG. 10. Distributions of zonal mean (a) longwave cloud feed-
feedbacks. We further characterize how the differing meridio-
back, (b) shortwave cloud feedback, and (c) total cloud feedback
(longwave 1 shortwave) in the 2xCO2 and Solar simulations. The
nal patterns of surface warming for solar and CO2 radiative
difference between the two cases in shown by the dotted lines. The forcing lead to differences between albedo, Planck, lapse rate,
cloud feedback is calculated by adjusting the change in cloud radia- water vapor, and cloud feedbacks and provide process-based
tive forcing, following the method described in Soden et al. (2008), accounts of these differences for each feedback. Our analy-
as discussed in the main text. sis shows that the smaller climate sensitivity of solar

Brought to you by University of Maryland, McKeldin Library | Unauthenticated | Downloaded 01/19/23 06:55 PM UTC
1 FEBRUARY 2023 KAUR ET AL. 785

radiative forcing results from the differences in the meridi- detailed mechanistic understanding of the spatial pattern of
onal profiles of warming. For solar forcing, larger warming cloud response to the surface temperature changes merits
in tropical latitudes compared to CO2 forcing leads to less future investigations.
positive albedo feedback, more negative Planck feedback, Newer versions of CESM are available with different equi-
more negative lapse rate feedback, more positive water va- librium climate sensitivity (3.2 K for CAM4, 4 K for CAM5,
por feedback, and less positive cloud feedback. The net ef- and 5.3 K for CAM6). Differences in equilibrium climate
fect is that, together, these feedbacks cause a smaller sensitivities for CAM4, CAM5, and CAM6 arise mainly from
global mean surface warming in the case of solar forcing cloud feedbacks and the direct radiative forcing of CO2
for the same change in effective radiative forcing. (Gettelman et al. 2012, 2019). The water vapor, temperature,
Our study has examined the dependence of the slow re- and albedo feedback values have remained fairly constant for
sponse on the forcing agent, which has been the subject of re- different versions of CAM (Gettelman et al. 2012, 2019). We
cent research. We exclude the fast adjustments in our believe that the qualitative conclusions of our studies are in-
calculation of climate feedbacks and applied the effective ra- sensitive to the model version used. Nevertheless, it would be
diative forcing (ERF) concept to estimate the radiative flux useful to evaluate the robustness of our results using different
imbalance and calculate radiative forcing: it is calculated once models in future studies. Furthermore, this paper uses the sur-
the stratosphere, land surface, and troposphere have adjusted. face temperature over land and ocean, rather than surface
Our analysis shows that the differences in slow response con- air temperature for our calculations. The latest IPCC report
tribute significantly to the difference in climate sensitivities shows that there is no systematic difference between changes
between solar and CO2 radiative forcing (Table 1). The fast in global mean surface air temperature and surface tempera-
adjustment, which does not scale with temperature change, ture (Gulev et al. 2021). The differences between surface tem-
has been excluded from the climate feedback and climate sen- perature and surface air temperature anomalies are small and
sitivity calculations. Instead of the total global mean surface less than 10%.
temperature change, which includes the fast effect of land Our results are based on a single climate model. A multi-
warming, we have used the slow global mean surface temper-
model intercomparison would be required to confirm the
ature change as a normalizing factor in climate feedback
robustness of these results. While quantitative differences be-
calculations.
tween the climate responses to solar and CO2 forcing are
For calculations of feedback, we have used precalculated
likely to be model dependent, the qualitative conclusion that
radiative kernels of Shell et al. (2008). The adoption of any
solar forcing has lower efficacy than CO2 forcing arises from
one set of kernels, derived from a single model, for different
basic physical principles, which we have sought to describe in
forcing experiments might lead to uncertainties of around
this study. There is no experiment in the latest phase of Cou-
10% in the results owing to intermodel uncertainties (Soden
pled Model Intercomparison Project (CMIP6; Eyring et al.
et al. 2008). However, a computationally and practically more
2016) that exactly corresponds to our Solar simulation where
feasible approach than the kernel technique is yet to be devel-
the solar constant is increased by 2%. However, such an
oped. Furthermore, our climate response calculations are esti-
experiment is available under the Precipitation Driver and
mated from equilibrium simulations using an atmospheric
Response Model Intercomparison Project (PDRMIP; Myhre
model coupled to a slab ocean model, which lacks ocean dy-
namics. The warming patterns in full Earth system models et al. 2017). Solar dimming experiments are also available
(ESMs) that include a dynamic ocean component would under the Geoengineering Model Intercomparison Project
evolve with time due to the transient heat uptake by the (GeoMIP6; Kravitz et al. 2015). Although we have not used
ocean and internal variability. Furthermore, the slab ocean these simulations in this study, we recognize the importance
model lacks effects of changes in ocean circulation and heat of multimodel intercomparison for testing the robustness of
transport, which can influence the melting of high-latitude our results. Future studies should address this important re-
sea ice and thereby affect the albedo feedback (Shell 2013; search gap. Further, a systematic and detailed investigation of
Yoshimori et al. 2018). Investigation of the individual feed- the dependence of the individual feedbacks on warming pat-
back differences in ESMs that simulate transient climate terns caused by different radiative forcing agents using a mul-
change caused by different forcing agents merits future timodel framework is needed.
studies. The present study highlights the role of differences in
Previous studies have indicated that the TOA radiative slow response for different forcing agents, and it is unlikely
flux is quite sensitive to rapid cloud adjustments (Colman that the differences would be affected by different magni-
and McAvaney 2011; Zelinka et al. 2013), and uncertainty tudes of radiative forcing (Table 1). Our study has practical
in cloud feedbacks is primarily the cause for the large un- relevance for assessing the impact of perturbations caused
certainty in net feedback (Soden and Held 2006; Dufresne by different radiative forcing agents on the climate system.
and Bony 2008; Vial et al. 2013). Nevertheless, it is reassur- The study also has important implications for the estimation
ing that the spatial patterns of slow cloud response and of the reduction of total solar irradiance in solar radiation mod-
cloud feedbacks in our 2XCO2 simulations are broadly sim- ification or solar geoengineering options that are proposed to
ilar to previous studies (Bitz et al. 2012; Gettelman et al. partially or fully offset the global mean warming caused by
2012). In this study, we have not investigated the complex CO2 (Bala et al. 2010; Schmidt et al. 2012; Niemeier et al. 2013;
spatial pattern of cloud changes and feedbacks, and a Kravitz et al. 2014; Modak and Bala 2014; Irvine et al. 2016;

Brought to you by University of Maryland, McKeldin Library | Unauthenticated | Downloaded 01/19/23 06:55 PM UTC
786 JOURNAL OF CLIMATE VOLUME 36

Kravitz et al. 2017; Tilmes et al. 2017; Duan et al. 2018; Nalam carbon aerosols. Climate Dyn., 38, 897–911, https://doi.org/10.
et al. 2018). 1007/s00382-011-1052-y.
Bitz, C. M., K. M. Shell, P. R. Gent, D. A. Bailey, G. Danabasoglu,
K. C. Armour, M. M. Holland, and J. T. Kiehl, 2012: Climate
Acknowledgments. This work was supported by the National
sensitivity of the Community Climate System Model, version
Supercomputer Mission (reference number DST/NSM/R&D_ 4. J. Climate, 25, 3053–3070, https://doi.org/10.1175/JCLI-D-11-
HPC Applications/2021/03.09) under the department of sci- 00290.1.
ence and technology (DST), Government of India. We Bony, S., and J. L. Dufresne, 2005: Marine boundary layer clouds
thank Dr. Karen Shell for making the radiative kernels at the heart of tropical cloud feedback uncertainties in cli-
freely available. We acknowledge the Supercomputer Educa- mate models. Geophys. Res. Lett., 32, L20806, https://doi.org/
tion and Research Centre (SERC), Indian Institute of Sci- 10.1029/2005GL023851.
ence, Bengaluru, for providing the computational resources }}, and Coauthors, 2006: How well do we understand and eval-
to perform CAM4 simulations. The first author gratefully ac- uate climate change feedback processes? J. Climate, 19,
knowledges the MHRD Fellowship provided by the Indian 3445–3482, https://doi.org/10.1175/JCLI3819.1.
Boucher, O., and Coauthors, 2013: Clouds and aerosols. Climate
Institute of Science. Three reviewers of the manuscript are
Change 2013: The Physical Science Basis, T. F. Stocker et al.,
acknowledged for their helpful suggestions.
Eds., Cambridge University Press, 571–657, https://doi.org/10.
1017/CBO9781107415324.016.
Data availability statement. The radiative kernels used in Cao, L., G. Bala, and K. Caldeira, 2012: Climate response to
our research are publicly available and downloaded from changes in atmospheric carbon dioxide and solar irradiance
https://climatedataguide.ucar.edu/climate-data/radiative-kernels- on the time scale of days to weeks. Environ. Res. Lett., 7,
climate-models. The NCL codes and model outputs used for 034015, https://doi.org/10.1088/1748-9326/7/3/034015.
our calculations are available on request from the corresponding Ceppi, P., and J. M. Gregory, 2017: Relationship of tropospheric
author. stability to climate sensitivity and Earth’s observed radiation
budget. Proc. Natl. Acad. Sci. USA, 114, 13 126–13 131,
https://doi.org/10.1073/pnas.1714308114.
REFERENCES }}, and }}, 2019: A refined model for the Earth’s global en-
ergy balance. Climate Dyn., 53, 4781–4797, https://doi.org/10.
Andrews, T., and P. M. Forster, 2008: CO2 forcing induces semi- 1007/s00382-019-04825-x.
direct effects with consequences for climate feedback inter- Cess, R. D., 1975: Global climate change: An investigation of at-
pretations. Geophys. Res. Lett., 35, L04802, https://doi.org/10. mospheric feedback mechanisms. Tellus, 27, 193–198, https://
1029/2007GL032273. doi.org/10.3402/tellusa.v27i3.9901.
}}, and M. J. Webb, 2018: The dependence of global cloud and }}, and Coauthors, 1990: Intercomparison and interpretation of
lapse rate feedbacks on the spatial structure of tropical climate feedback processes in 19 atmospheric general circula-
Pacific warming. J. Climate, 31, 641–654, https://doi.org/10. tion models. J. Geophys. Res., 95, 16601, https://doi.org/10.
1175/JCLI-D-17-0087.1. 1029/JD095iD10p16601.
}}, P. M. Forster, and J. M. Gregory, 2009: A surface energy Collins, M., and Coauthors, 2013: Long-term climate change: Pro-
perspective on climate change. J. Climate, 22, 2557–2570, jections, commitments and irreversibility. Climate Change
https://doi.org/10.1175/2008JCLI2759.1. 2013: The Physical Science Basis. T. F. Stocker et al., Eds.,
}}, J. M. Gregory, and M. J. Webb, 2015: The dependence of Cambridge University Press, 1029–1136.
radiative forcing and feedback on evolving patterns of surface Colman, R., 2003: A comparison of climate feedbacks in general
temperature change in climate models. J. Climate, 28, 1630– circulation models. Climate Dyn., 20, 865–873, https://doi.org/
1648, https://doi.org/10.1175/JCLI-D-14-00545.1. 10.1007/s00382-003-0310-z.
}}, C. J. Smith, G. Myhre, P. M. Forster, R. Chadwick, and D. }}, and B. J. McAvaney, 2011: On tropospheric adjustment to
Ackerley, 2021: Effective radiative forcing in a GCM with forcing and climate feedbacks. Climate Dyn., 36, 1649–1658,
fixed surface temperatures. J. Geophys. Res. Atmos., 126, https://doi.org/10.1007/s00382-011-1067-4.
e2020JD033880, https://doi.org/10.1029/2020JD033880. }}, J. Fraser, and L. Rotstayn, 2001: Climate feedbacks in a gen-
Armour, K. C., 2017: Energy budget constraints on climate sensi- eral circulation model incorporating prognostic clouds. Climate
tivity in light of inconstant climate feedbacks. Nat. Climate Dyn., 18, 103–122, https://doi.org/10.1007/s003820100162.
Change, 7, 331–335, https://doi.org/10.1038/nclimate3278. Cronin, T. W., and M. F. Jansen, 2016: Analytic radiative-advective
}}, C. M. Bitz, and G. H. Roe, 2013: Time-varying climate sen- equilibrium as a model for high-latitude climate. Geophys.
sitivity from regional feedbacks. J. Climate, 26, 4518–4534, Res. Lett., 43, 449–457, https://doi.org/10.1002/2015GL067172.
https://doi.org/10.1175/JCLI-D-12-00544.1. Crook, J. A., P. M. Forster, and N. Stuber, 2011: Spatial patterns
Atwood, A. R., E. Wu, D. M. W. Frierson, D. S. Battisti, and J. P. of modeled climate feedback and contributions to tempera-
Sachs, 2016: Quantifying climate forcings and feedbacks over ture response and polar amplification. J. Climate, 24, 3575–
the last millennium in the CMIP5-PMIP3 models. J. Climate, 3592, https://doi.org/10.1175/2011JCLI3863.1.
29, 1161–1178, https://doi.org/10.1175/JCLI-D-15-0063.1. Dong, B., J. M. Gregory, and R. T. Sutton, 2009: Understanding
Bala, G., K. Caldeira, and R. Nemani, 2010: Fast versus slow re- land–sea warming contrast in response to increasing green-
sponse in climate change: Implications for the global hydro- house gases. Part I: Transient adjustment. J. Climate, 22,
logical cycle. Climate Dyn., 35, 423–434, https://doi.org/10. 3079–3097, https://doi.org/10.1175/2009JCLI2652.1.
1007/s00382-009-0583-y. Dong, Y., C. Proistosescu, K. C. Armour, and D. S. Battisti, 2019:
Ban-Weiss, G. A., L. Cao, G. Bala, and K. Caldeira, 2012: Depen- Attributing historical and future evolution of radiative feed-
dence of climate forcing and response on the altitude of black backs to regional warming patterns using a Green’s function

Brought to you by University of Maryland, McKeldin Library | Unauthenticated | Downloaded 01/19/23 06:55 PM UTC
1 FEBRUARY 2023 KAUR ET AL. 787

approach: The preeminence of the western Pacific. J. Climate, Held, I. M., and B. J. Soden, 2000: Water vapor feedback and
32, 5471–5491, https://doi.org/10.1175/JCLI-D-18-0843.1. global warming. Annu. Rev. Energy Environ., 25, 441–475,
Doutriaux-Boucher, M., M. J. Webb, J. M. Gregory, and O. https://doi.org/10.1146/annurev.energy.25.1.441.
Boucher, 2009: Carbon dioxide induced stomatal closure }}, and K. M. Shell, 2012: Using relative humidity as a state
increases radiative forcing via a rapid reduction in low variable in climate feedback analysis. J. Climate, 25, 2578–
cloud. Geophys. Res. Lett., 36, L02703, https://doi.org/10. 2582, https://doi.org/10.1175/JCLI-D-11-00721.1.
1029/2008GL036273. Irvine, P. J., B. Kravitz, M. G. Lawrence, and H. Muri, 2016: An
Duan, L., L. Cao, G. Bala, and K. Caldeira, 2018: Comparison of overview of the Earth system science of solar geoengineering.
the fast and slow climate response to three radiation manage- Wiley Interdiscip. Rev.: Climate Change, 7, 815–833, https://
ment geoengineering schemes. J. Geophys. Res. Atmos., 123, doi.org/10.1002/wcc.423.
11 980–12 001, https://doi.org/10.1029/2018JD029034. Jonko, A. K., K. M. Shell, B. M. Sanderson, and G. Danabasoglu,
Dufresne, J. L., and S. Bony, 2008: An assessment of the primary 2012: Climate feedbacks in CCSM3 under changing CO2 forc-
sources of spread of global warming estimates from coupled ing. Part I: Adapting the linear radiative kernel technique to
atmosphere–ocean models. J. Climate, 21, 5135–5144, https:// feedback calculations for a broad range of forcings. J. Climate,
doi.org/10.1175/2008JCLI2239.1. 25, 5260–5272, https://doi.org/10.1175/JCLI-D-11-00524.1.
Eyring, V., S. Bony, G. A. Meehl, C. A. Senior, B. Stevens, R. J. }}, }}, }}, and }}, 2013: Climate feedbacks in CCSM3
Stouffer, and K. E. Taylor, 2016: Overview of the Coupled under changing CO2 forcing. Part II: Variation of climate
Model Intercomparison Project Phase 6 (CMIP6) experimen- feedbacks and sensitivity with forcing. J. Climate, 26, 2784–
tal design and organization. Geosci. Model Dev., 9, 1937– 2795, https://doi.org/10.1175/JCLI-D-12-00479.1.
1958, https://doi.org/10.5194/gmd-9-1937-2016. Joshi, M., K. Shine, M. Ponater, N. Stuber, R. Sausen, and L. Li,
Feldl, N., S. Bordoni, and T. M. Merlis, 2017: Coupled high-latitude 2003: A comparison of climate response to different radiative
climate feedbacks and their impact on atmospheric heat trans- forcings in three general circulation models: Towards an im-
port. J. Climate, 30, 189–201, https://doi.org/10.1175/JCLI-D-16- proved metric of climate change. Climate Dyn., 20, 843–854,
0324.1. https://doi.org/10.1007/s00382-003-0305-9.
Forster, P. M., M. Blackburn, R. Glover, and K. P. Shine, 2000: Klein, S. A., and D. L. Hartmann, 1993: The seasonal cycle of low
An examination of climate sensitivity for idealised climate stratiform clouds. J. Climate, 6, 1587–1606, https://doi.org/10.
change experiments in an intermediate general circulation
1175/1520-0442(1993)006,1587:TSCOLS.2.0.CO;2.
model. Climate Dyn., 16, 833–849, https://doi.org/10.1007/
Kravitz, B., and Coauthors, 2014: A multi-model assessment of
s003820000083.
regional climate disparities caused by solar geoengineering.
}}, and Coauthors, 2016: Recommendations for diagnosing ef-
Environ. Res. Lett., 9, 74 013–74 020, https://doi.org/10.1088/
fective radiative forcing from climate models for CMIP6. J.
1748-9326/9/7/074013.
Geophys. Res. Atmos., 121, 12 460–12 475, https://doi.org/10.
}}, and Coauthors, 2015: The Geoengineering Model Intercom-
1002/2016JD025320.
parison Project Phase 6 (GeoMIP6): Simulation design and
Gettelman, A., J. E. Kay, and K. M. Shell, 2012: The evolution of
preliminary results. Geosci. Model Dev., 8, 3379–3392, https://
climate sensitivity and climate feedbacks in the Community
doi.org/10.5194/gmd-8-3379-2015.
Atmosphere Model. J. Climate, 25, 1453–1469, https://doi.org/
}}, D. G. MacMartin, M. J. Mills, J. H. Richter, S. Tilmes, J. F.
10.1175/JCLI-D-11-00197.1.
Lamarque, J. J. Tribbia, and F. Vitt, 2017: First simulations
}}, and Coauthors, 2019: High climate sensitivity in the Com-
of designing stratospheric sulfate aerosol geoengineering to
munity Earth System Model version 2 (CESM2). Geophys.
meet multiple simultaneous climate objectives. J. Geophys.
Res. Lett., 46, 8329–8337, https://doi.org/10.1029/2019GL083978.
Gregory, J. M., and M. Webb, 2008: Tropospheric adjustment in- Res. Atmos., 122, 12 616–12 634, https://doi.org/10.1002/
duces a cloud component in CO2 forcing. J. Climate, 21, 58– 2017JD026874.
71, https://doi.org/10.1175/2007JCLI1834.1. Lambert, F. H., and N. E. Faull, 2007: Tropospheric adjustment:
}}, and Coauthors, 2004: A new method for diagnosing radia- The response of two general circulation models to a change
tive forcing and climate sensitivity. Geophys. Res. Lett., 31, in insolation. Geophys. Res. Lett., 34, 2–6, https://doi.org/10.
L03205, https://doi.org/10.1029/2003GL018747. 1029/2006GL028124.
}}, T. Andrews, P. Ceppi, T. Mauritsen, and M. J. Webb, 2020: Larson, K., D. L. Hartmann, and S. A. Klein, 1999: The role of
How accurately can the climate sensitivity to CO2 be esti- clouds, water vapor, circulation, and boundary layer structure
mated from historical climate change? Climate Dyn., 54, 129– in the sensitivity of the tropical climate. J. Climate, 12, 2359–
157, https://doi.org/10.1007/s00382-019-04991-y. 2374, https://doi.org/10.1175/1520-0442(1999)012,2359:
Gulev, S. K., and Coauthors, 2021: Changing state of the cli- TROCWV.2.0.CO;2.
mate system. Climate Change 2021: The Physical Science Liu, R., H. Su, K. N. Liou, J. H. Jiang, Y. Gu, S. C. Liu, and C. J.
Basis. V. Masson-Delmotte et al., Eds., Cambridge Uni- Shiu, 2018: An assessment of tropospheric water vapor feed-
versity Press, 287–422, https://www.ipcc.ch/report/ar6/wg1/ back using radiative kernels. J. Geophys. Res. Atmos., 123,
downloads/report/IPCC_AR6_WGI_Chapter02.pdf. 1499–1509, https://doi.org/10.1002/2017JD027512.
Hall, A., 2004: The role of surface albedo feedback in climate. J. Manabe, S., and R. T. Wetherald, 1975: The effects of doubling
Climate, 17, 1550–1568, https://doi.org/10.1175/1520-0442(2004) the CO2 concentration on the climate of a general circulation
017,1550:TROSAF.2.0.CO;2. model. J. Atmos. Sci., 32, 3–15, https://doi.org/10.1175/1520-
Hansen, J., M. Sato, and R. Ruedy, 1997: Radiative forcing and 0469(1975)032,0003:TEODTC.2.0.CO;2.
climate response. J. Geophys. Res., 102, 6831–6864, https:// }}, and R. J. Stouffer, 1980: Sensitivity of a global climate
doi.org/10.1029/96JD03436. model to an increase of CO2 concentration in the atmo-
}}, and Coauthors, 2005: Efficacy of climate forcings. J. Geophys. sphere. J. Geophys. Res., 85, 5529–5554, https://doi.org/10.
Res., 110, D18104, https://doi.org/10.1029/2005JD005776. 1029/JC085iC10p05529.

Brought to you by University of Maryland, McKeldin Library | Unauthenticated | Downloaded 01/19/23 06:55 PM UTC
788 JOURNAL OF CLIMATE VOLUME 36

Marvel, K., R. Pincus, G. A. Schmidt, and R. L. Miller, 2018: In- climatedataguide.ucar.edu/climate-data/radiative-kernels-


ternal variability and disequilibrium confound estimates of climate-models.
climate sensitivity from observations. Geophys. Res. Lett., 45, Proistosescu, C., and P. J. Huybers, 2017: Slow climate mode rec-
1595–1601, https://doi.org/10.1002/2017GL076468. onciles historical and model-based estimates of climate sensi-
McCoy, D. T., R. Eastman, D. L. Hartmann, and R. Wood, 2017: tivity. Sci. Adv., 3 (7), 1–7, https://doi.org/10.1126/sciadv.
The change in low cloud cover in a warmed climate inferred 1602821.
from AIRS, MODIS, and ERA-Interim. J. Climate, 30, 3609– Qu, X., A. Hall, S. A. Klein, and A. M. DeAngelis, 2015a: Posi-
3620, https://doi.org/10.1175/JCLI-D-15-0734.1. tive tropical marine low-cloud cover feedback inferred from
Miller, R. L., 1997: Tropical thermostats and low cloud cover. cloud-controlling factors. Geophys. Res. Lett., 42, 7767–7775,
J. Climate, 10, 409–440, https://doi.org/10.1175/1520-0442 https://doi.org/10.1002/2015GL065627.
(1997)010,0409:TTALCC.2.0.CO;2. }}, }}, }}, and P. M. Caldwell, 2015b: The strength of the
Modak, A., and G. Bala, 2014: Sensitivity of simulated climate to tropical inversion and its response to climate change in 18
latitudinal distribution of solar insolation reduction in solar CMIP5 models. Climate Dyn., 45, 375–396, https://doi.org/10.
radiation management. Atmos. Chem. Phys., 14, 7769–7779, 1007/s00382-014-2441-9.
https://doi.org/10.5194/acp-14-7769-2014. Ramanathan, V., 1977: Interactions between ice-albedo, lapse-rate
}}, and }}, 2019: Efficacy of black carbon aerosols: The role and cloud-top feedbacks: An analysis of the nonlinear re-
of shortwave cloud feedback. Environ. Res. Lett., 14, 084029, sponse of a GCM climate model. J. Atmos. Sci., 34, 1885–1897,
https://doi.org/10.1088/1748-9326/ab21e7. https://doi.org/10.1175/1520-0469(1977)034,1885:IBIALR.
}}, }}, L. Cao, and K. Caldeira, 2016: Why must a solar forc- 2.0.CO;2.
ing be larger than a CO2 forcing to cause the same global }}, and Coauthors, 1989: Cloud-radiative forcing and climate:
mean surface temperature change? Environ. Res. Lett., 11, Results from the Earth’s radiation budget. Phys. Today, 42,
044013, https://doi.org/10.1088/1748-9326/11/4/044013. 22–32, https://doi.org/10.1063/1.881167.
}}, }}, K. Caldeira, and L. Cao, 2018: Does shortwave Ramaswamy, V., and Coauthors, 2019: Radiative forcing of climate:
absorption by methane influence its effectiveness? Climate The historical evolution of the radiative forcing concept, the
Dyn., 51, 3653–3672, https://doi.org/10.1007/s00382-018-
forcing agents and their quantification, and applications. A
4102-x.
Century of Progress in Atmospheric and Related Sciences:
Myers, T. A., R. C. Scott, M. D. Zelinka, S. A. Klein, J. R. Norris,
Celebrating the American Meteorological Society Centennial,
and P. M. Caldwell, 2021: Observational constraints on low
Meteor. Monogr., No. 59, Amer. Meteor. Soc., https://doi.org/
cloud feedback reduce uncertainty of climate sensitivity. Nat.
10.1175/AMSMONOGRAPHS-D-19-0001.1.
Climate Change, 11, 501–507, https://doi.org/10.1038/s41558-
Reay, D., C. Sabine, P. Smith, and G. Hymus, 2007: Climate
021-01039-0.
change 2007: Spring-time for sinks. Nature, 446, 727–728,
Myhre, G., and Coauthors, 2013: Anthropogenic and natural radi-
https://doi.org/10.1038/446727a.
ative forcing. Climate Change 2013: The Physical Science
Richardson, T. B., and Coauthors, 2019: Efficacy of climate forc-
Basis, T. F. Stocker et al., Eds., Cambridge University Press,
ings in PDRMIP models. J. Geophys. Res. Atmos., 124,
659–740, https://doi.org/10.1017/CBO9781107415324.018.
}}, and Coauthors, 2017: PDRMIP: A Precipitation Driver and 12 824–12 844, https://doi.org/10.1029/2019JD030581.
Rugenstein, M. A. A., K. Caldeira, and R. Knutti, 2016: Depen-
Response Model Intercomparison Project}Protocol and pre-
dence of global radiative feedbacks on evolving patterns of
liminary results. Bull. Amer. Meteor. Soc., 98, 1185–1198, https://
doi.org/10.1175/BAMS-D-16-0019.1. surface heat fluxes. Geophys. Res. Lett., 43, 9877–9885,
Nalam, A., G. Bala, and A. Modak, 2018: Effects of Arctic geoen- https://doi.org/10.1002/2016GL070907.
gineering on precipitation in the tropical monsoon regions. }}, and Coauthors, 2020: Equilibrium climate sensitivity esti-
Climate Dyn., 50, 3375–3395, https://doi.org/10.1007/s00382- mated by equilibrating climate models. Geophys. Res. Lett.,
017-3810-y. 47, e2019GL083898, https://doi.org/10.1029/2019GL083898.
Neale, R. B., and Coauthors, 2010: Description of the NCAR Sanderson, B. M., and K. M. Shell, 2012: Model-specific radiative
Community Atmosphere Model (CAM5.0). NCAR Tech. kernels for calculating cloud and noncloud climate feedbacks.
Note NCAR/TN-486-STR, 268 pp., www.cesm.ucar.edu/ J. Climate, 25, 7607–7624, https://doi.org/10.1175/JCLI-D-11-
models/cesm1.1/cam/docs/description/cam5_desc.pdf. 00726.1.
Niemeier, U., H. Schmidt, K. Alterskjær, and J. E. Kristjánsson, Santer, B. D., and Coauthors, 2005: Atmospheric science: Amplifi-
2013: Solar irradiance reduction via climate engineering: Im- cation of surface temperature trends and variability in the
pact of different techniques on the energy balance and the tropical atmosphere. Science, 309, 1551–1556, https://doi.org/
hydrological cycle. J. Geophys. Res. Atmos., 118, 11 905– 10.1126/science.1114867.
11 917, https://doi.org/10.1002/2013JD020445. Schlesinger, M. E., and J. F. B. Mitchell, 1987: Climate model sim-
Oleson, K. W., and Coauthors, 2010: Technical description of ver- ulations of the equilibrium climatic response to increased car-
sion 4.0 of the Community Land Model (CLM) NCAR Tech. bon dioxide. Rev. Geophys., 25, 760–798, https://doi.org/10.
Note NCAR/TN-4781STR, 257 pp., https://doi.org/10.5065/ 1029/RG025i004p00760.
D6FB50WZ. Schmidt, H., R. A. Ruedy, R. L. Miller, and A. A. Lacis, 2010: At-
Pendergrass, A. G., A. Conley, and F. M. Vitt, 2018a: Surface and tribution of the present-day total greenhouse effect. J. Geo-
top-of-atmosphere radiative feedback kernels for CESM- phys. Res., 115, D20106, https://doi.org/10.1029/2010JD014287.
CAM5. Earth Syst. Sci. Data, 10, 317–324, https://doi.org/10. }}, and Coauthors, 2012: Solar irradiance reduction to counter-
5194/essd-10-317-2018. act radiative forcing from a quadrupling of CO2: Climate re-
}}, and Coauthors, Eds., 2018b: The Climate Data Guide: sponses simulated by four earth system models. Earth Syst.
Radiative kernels from climate models. UCAR, https:// Dyn., 3, 63–78, https://doi.org/10.5194/esd-3-63-2012.

Brought to you by University of Maryland, McKeldin Library | Unauthenticated | Downloaded 01/19/23 06:55 PM UTC
1 FEBRUARY 2023 KAUR ET AL. 789

Screen, J. A., C. Deser, and I. Simmonds, 2012: Local and remote Tilmes, S., J. H. Richter, M. J. Mills, B. Kravitz, D. G. Macmartin,
controls on observed Arctic warming. Geophys. Res. Lett., 39, F. Vitt, J. J. Tribbia, and J. F. Lamarque, 2017: Sensitivity of
L10709, https://doi.org/10.1029/2012GL051598. aerosol distribution and climate response to stratospheric
Shell, K. M., 2013: Consistent differences in climate feedbacks be- SO2 injection locations. J. Geophys. Res. Atmos., 122, 12 591–
tween atmosphere–ocean GCMs and atmospheric GCMs 12 615, https://doi.org/10.1002/2017JD026888.
with slab-ocean models. J. Climate, 26, 4264–4281, https://doi. Vial, J., J. L. Dufresne, and S. Bony, 2013: On the interpretation
org/10.1175/JCLI-D-12-00519.1. of inter-model spread in CMIP5 climate sensitivity estimates.
}}, J. T. Kiehl, and C. A. Shields, 2008: Using the radiative ker- Climate Dyn., 41, 3339–3362, https://doi.org/10.1007/s00382-
nel technique to calculate climate feedbacks in NCAR’s 013-1725-9.
Community Atmospheric Model. J. Climate, 21, 2269–2282, Wetherald, R., and S. Manabe, 1988: Cloud feedback processes in
https://doi.org/10.1175/2007JCLI2044.1. a general circulation model. J. Atmos. Sci., 45, 1397–1416,
Sherwood, S. C., S. Bony, O. Boucher, C. Bretherton, P. M. https://doi.org/10.1175/1520-0469(1988)045,1397:CFPIAG.2.0.
Forster, J. M. Gregory, and B. Stevens, 2015: Adjustments in CO;2.
the forcing-feedback framework for understanding climate Williams, K. D., W. J. Ingram, and J. M. Gregory, 2008: Time var-
change. Bull. Amer. Meteor. Soc., 96, 217–228, https://doi.org/ iation of effective climate sensitivity in GCMs. J. Climate, 21,
10.1175/BAMS-D-13-00167.1. 5076–5090, https://doi.org/10.1175/2008JCLI2371.1.
Shindell, D. T., 2014: Inhomogeneous forcing and transient cli- Yoshimori, M., A. Abe-Ouchi, H. Tatebe, T. Nozawa, and A.
mate sensitivity. Nat. Climate Change, 4, 274–277, https://doi. Oka, 2018: The importance of ocean dynamical feedback
org/10.1038/nclimate2136. for understanding the impact of mid-high-latitude warming
}}, G. Faluvegi, L. Rotstayn, and G. Milly, 2015: Spatial pat- on tropical precipitation change. J. Climate, 31, 2417–2434,
terns of radiative forcing and surface temperature response.
https://doi.org/10.1175/JCLI-D-17-0402.1.
J. Geophys. Res. Atmos., 120, 5385–5403, https://doi.org/10.
Yue, Q., B. H. Kahn, E. J. Fetzer, M. Schreier, S. Wong, X. Chen,
1002/2014JD022752.
and X. Huang, 2016: Observation-based longwave cloud radi-
Shine, K. P., 2000: Radiative forcing of climate change. Space Sci.
ative kernels derived from the A-Train. J. Climate, 29, 2023–
Rev., 94, 363–373, https://doi.org/10.1023/A:1026752230256.
2040, https://doi.org/10.1175/JCLI-D-15-0257.1.
}}, J. Cook, E. J. Highwood, and M. M. Joshi, 2003: An alterna-
Zelinka, M. D., S. A. Klein, and D. L. Hartmann, 2012: Comput-
tive to radiative forcing for estimating the relative importance
ing and partitioning cloud feedbacks using cloud property his-
of climate change mechanisms. Geophys. Res. Lett., 30, 2047,
tograms. Part I: Cloud radiative kernels. J. Climate, 25, 3715–
https://doi.org/10.1029/2003GL018141.
3735, https://doi.org/10.1175/JCLI-D-11-00248.1.
Sobel, A. H., J. Nilsson, and L. M. Polvani, 2001: The weak tem-
}}, }}, K. E. Taylor, T. Andrews, M. J. Webb, J. M. Gregory,
perature gradient approximation and balanced tropical mois-
ture waves. J. Atmos. Sci., 58, 3650–3665, https://doi.org/10. and P. M. Forster, 2013: Contributions of different cloud types
1175/1520-0469(2001)058,3650:TWTGAA.2.0.CO;2. to feedbacks and rapid adjustments in CMIP5. J. Climate, 26,
Soden, B. J., and I. M. Held, 2006: An assessment of climate feed- 5007–5027, https://doi.org/10.1175/JCLI-D-12-00555.1.
backs in coupled ocean-atmosphere models. J. Climate, 19, }}, C. Zhou, and S. A. Klein, 2016: Insights from a refined de-
3354–3360, https://doi.org/10.1175/JCLI3799.1. composition of cloud feedbacks. Geophys. Res. Lett., 43,
}}, A. J. Broccoli, and R. S. Hemler, 2004: On the use of cloud 9529–9269, https://doi.org/10.1002/2016GL069917.
forcing to estimate cloud feedback. J. Climate, 17, 3661–3665, }}, T. A. Myers, D. T. McCoy, S. Po-Chedley, P. M. Caldwell,
https://doi.org/10.1175/1520-0442(2004)017,3661:OTUOCF. P. Ceppi, S. A. Klein, and K. E. Taylor, 2020: Causes of
2.0.CO;2. higher climate sensitivity in CMIP6 models. Geophys. Res. Lett.,
}}, }}, R. C. Colman, K. M. Shell, J. T. Kiehl, and C. A. 47, e2019GL085782, https://doi.org/10.1029/2019GL085782.
Shields, 2008: Quantifying climate feedbacks using radiative Zhang, M. H., J. J. Hack, J. T. Kiehl, and R. D. Cess, 1994: Diag-
kernels. J. Climate, 21, 3504–3520, https://doi.org/10.1175/ nostic study of climate feedback processes in atmospheric
2007JCLI2110.1. general circulation models. J. Geophys. Res., 99, 5525–5537,
Song, X., G. J. Zhang, and M. Cai, 2014: Quantifying contribu- https://doi.org/10.1029/93JD03523.
tions of climate feedbacks to tropospheric warming in the Zhou, C., M. D. Zelinka, and S. A. Klein, 2016: Impact of decadal
NCAR CCSM3.0. Climate Dyn., 42, 901–917, https://doi.org/ cloud variations on the Earth’s energy budget. Nat. Geosci.,
10.1007/s00382-013-1805-x. 9, 871–874, https://doi.org/10.1038/ngeo2828.
Stevens, B., S. C. Sherwood, S. Bony, and M. J. Webb, 2016: Pros- }}, }}, and }}, 2017: Analyzing the dependence of global
pects for narrowing bounds on Earth’s equilibrium climate cloud feedback on the spatial pattern of sea surface temperature
sensitivity. Earth’s Future, 4, 512–522, https://doi.org/10.1002/ change with a Green’s function approach. J. Adv. Model. Earth
2016EF000376. Syst., 9, 2174–2189, https://doi.org/10.1002/2017MS001096.

Brought to you by University of Maryland, McKeldin Library | Unauthenticated | Downloaded 01/19/23 06:55 PM UTC

You might also like