LETTERS
PUBLISHED ONLINE: 20 OCTOBER 2013 | DOI: 10.1038/NGEO1987
Contribution of ocean overturning circulation to
tropical rainfall peak in the Northern Hemisphere
Dargan M. W. Frierson
1
*
, Yen-Ting Hwang
1
, Neven S. Fu
ˇ
ckar
2,3
, Richard Seager
4
, Sarah M. Kang
5
,
Aaron Donohoe
6
, Elizabeth A. Maroon
1
, Xiaojuan Liu
1
and David S. Battisti
1
Rainfall in the tropics is largely focused in a narrow zonal
band near the Equator, known as the intertropical convergence
zone. On average, substantially more rain falls just north of
the Equator
1
. This hemispheric asymmetry in tropical rainfall
has been attributed to hemispheric asymmetries in ocean tem-
perature induced by tropical landmasses. However, the ocean
meridional overturning circulation also redistributes energy,
by carrying heat northwards across the Equator. Here, we use
satellite observations of the Earth’s energy budget
2
, atmo-
spheric reanalyses
3
and global climate model simulations to
study tropical rainfall using a global energetic framework. We
show that the meridional overturning circulation contributes
significantly to the hemispheric asymmetry in tropical rainfall
by transporting heat from the Southern Hemisphere to the
Northern Hemisphere, and thereby pushing the tropical rain
band north. This northward shift in tropical precipitation is
seen in global climate model simulations when ocean heat
transport is included, regardless of whether continents are
present or not. If the strength of the meridional overturning cir-
culation is reduced in the future as a result of global warming,
as has been suggested
4
, precipitation patterns in the tropics
could change, with potential societal consequences.
One of the defining features of the Earth’s climate is the fact
that tropical rainfall primarily maximizes within the Northern
Hemisphere (Fig. 1a,b). This is especially evident in the Atlantic
and eastern Pacific oceans, where precipitation near 5
◦
S is much
weaker than at the equivalent locations north of the Equator,
which are some of the rainiest regions on Earth in the intertropical
convergence zone. Throughout the tropics, societies as diverse as
subsistence farmers in the Sahel and the technology-based cities of
India are reliant, directly or indirectly, on monsoon rains. How
tropical rainfall evolves in the future will have important and
widespread social consequences.
Precipitation in the tropics exists largely within narrow zonal
bands owing to the large-scale atmospheric overturning known as
the Hadley circulation
5
. Within the lower branch of this circulation,
the trade winds converge near the Equator, drying the subtropics
and bringing moisture into the rain bands. This circulation is
thermally direct, and thus transports energy in the directions of its
upper branch. The Hadley circulation also plays a primary role in
determining the hemispheric asymmetry of tropical precipitation
because it transports copious amounts of moisture from the
Southern Hemisphere to the Northern Hemisphere. A northward
cross-equatorial mass transport in the moist lower branch of the
1
Department of Atmospheric Sciences, University of Washington, Seattle 98195, USA,
2
International Pacific Research Center, University of Hawaii,
Honolulu 96822, USA,
3
Institut Català de Ciències del Clima, Barcelona 08005, Spain,
4
Lamont-Doherty Earth Observatory of Columbia University,
Palisades, New York 10964, USA,
5
School of Urban and Environmental Engineering, Ulsan National Institute of Science and Technology, 689-798, Korea,
6
Department of Earth, Atmospheric, and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA.
*e-mail: dargan@atmos.washington.edu
circulation performs the moisture transport, and is compensated
aloft by a dry, southward cross-equatorial flow
6
.
Several recent studies
7–9
have argued that a useful attribution
of the hemispheric asymmetry of tropical precipitation can be
performed with an examination of the hemispheric asymmetry of
the atmospheric energy budget. Tropical precipitation responds to
atmospheric heating even from well outside the tropics by shifting
towards the heat input
7–14
. The influence of an extratropical heat
source is spread by baroclinic eddies into the tropics, where the
Hadley circulation responds with a cross-equatorial component.
The anomalous Hadley circulation causes an energy transport
away from the hemisphere with stronger atmospheric heating and
a moisture transport towards the heating. On the basis of this
energetic framework, we perform an attribution of the observed
energy input into the atmosphere at the top-of-atmosphere (TOA)
and surface, and claim that whatever causes the heating of the
Northern Hemisphere atmosphere to be larger is also what causes
precipitation to maximize in the Northern Hemisphere tropics.
Despite large differences in land coverage and cloudiness
between the hemispheres, much of the globe receives nearly
identical net TOA radiative input at a given latitude in both
the Northern Hemisphere and Southern Hemisphere (Fig. 2a).
However, when averaged over the hemispheres, the Southern
Hemisphere receives 1.5 W m
−2
more radiation than the Northern
Hemisphere, primarily owing to the area between 17
◦
and 35
◦
latitude, with an additional contribution from the polar cap. A map
of TOA net radiation (Fig. 2b) shows that the Northern Hemisphere
receives less radiation in the critical region of 17
◦
–35
◦
because the
Sahara and Arabian deserts are a strong net sink of TOA radiation.
Owing to their high albedo, these deserts are the only substantial
regions between 30
◦
N and 30
◦
S that lose energy through radiation,
a fact that was inferred in 1969 by the Nimbus III team. The polar
cap in the Southern Hemisphere receives more net radiation than
the Northern Hemisphere cap owing to the low temperature and
low outgoing longwave radiation over Antarctica. The hemispheric
difference in radiation can be alternatively partitioned into a similar
net shortwave irradiance in each hemisphere, and a larger longwave
flux out of the warmer Northern Hemisphere
15
.
As the Southern Hemisphere receives slightly more net radiation
than the Northern Hemisphere, TOA radiation cannot be the cause
of the present-day maximum of tropical rainfall north of the Equa-
tor. If there is a larger energy flux into the Northern Hemisphere
atmosphere than the Southern Hemisphere atmosphere, it must
instead be caused by a northward cross-equatorial ocean heat
940 NATURE GEOSCIENCE | VOL 6 | NOVEMBER 2013 | www.nature.com/naturegeoscience
© 2013 Macmillan Publishers Limited. All rights reserved
NATURE GEOSCIENCE DOI: 10.1038/NGEO1987
LETTERS
10
8
6
4
2
mm d
¬1
90° N
50° N
30° N
15° N
0°
15° S
30° S
50° S
90° S
02468
mm d
¬1
02468
0 6 12 18 24
mm d
¬1
a
bc
Latitude
90° N
50° N
30° N
15° N
0°
15° S
30° S
50° S
90° S
Latitude
Figure 1 | Observed and modelled annual mean precipitation. a, Annual mean precipitation, 2001–2010 from the Global Precipitation Climatology Project
Version 2.1 (ref. 1). b, Zonal average of a. c, Zonal mean precipitation from an aquaplanet GCM simulation using the GFDL AM2.1 model with prescribed
surface heat fluxes from observations (red), and the realistic geography ECHAM GCM using observed (solid grey) and hemispherically symmetrized
surface heat fluxes (dashed grey). The simulated tropical rainfall in the aquaplanet simulation maximizes at approximately 11
◦
N, whereas in observations
the peak is at approximately 5
◦
N.
transport (OHT), and fluxed into the atmosphere at the surface.
This is because, in the time mean, whereas the net surface flux over
land is zero, over ocean it is balanced by the divergence of OHT.
OHT can be estimated from satellite observations and at-
mospheric reanalyses
3
. The deduced zonally averaged Northern
Hemisphere ocean to atmosphere surface heat flux substantially
exceeds the Southern Hemisphere equivalent at nearly every lati-
tude poleward of 20
◦
(Fig. 2c,d). On average, the Northern Hemi-
sphere atmosphere receives 3.1 W m
−2
more upward energy flux
from the surface than the Southern Hemisphere corresponding
to a northward cross-equatorial OHT of 0.4 PW (Table 1). The
latitudes in which the Northern Hemisphere surface flux exceeds
the Southern Hemisphere by the most are regions of the Gulf
Stream and Kuroshio currents (25–40
◦
N), and also the latitudes
between 45
◦
and 70
◦
that encompass regions of deep water pro-
duction in the North Atlantic and wind-driven upwelling in the
Southern Ocean. In the deep tropics a region exists where the
Northern Hemisphere flux is larger, owing to upwelling differences
between the hemispheres.
The northward cross-equatorial oceanic heat transport occurs
in the Atlantic (Supplementary Fig. S1), where the surface heat
flux has a striking hemispheric asymmetry in the extratropics. The
meridional overturning circulation (MOC) facilitates the north-
ward cross-equatorial heat transport in the Atlantic, connecting
deep water production in the Nordic Seas and subpolar North
Atlantic and upwelling in the Southern Ocean
16,17
.
Simulations of the present climate from Coupled Model
Intercomparison Project Phase 5 (CMIP5; 20 models) and Phase 3
(CMIP3; 15 models) all have a larger climatological surface
heat flux into the Northern Hemisphere atmosphere than into
the Southern Hemisphere atmosphere as well (Supplementary
Table S1), providing further evidence that a northward cross-
equatorial OHT is a fundamental aspect of our present climate. The
hemispheric asymmetry of TOA radiation is less robust in CMIP3/5;
biases in extratropical clouds often lead to excessive heating of the
Southern Hemisphere, and thus to the southward intensification
of tropical precipitation that characterizes the double inter-tropical
convergence zone problem
8
. In some models, the TOA radiation
NATURE GEOSCIENCE | VOL 6 | NOVEMBER 2013 | www.nature.com/naturegeoscience 941
© 2013 Macmillan Publishers Limited. All rights reserved
LETTERS
NATURE GEOSCIENCE DOI: 10.1038/NGEO1987
90°
50°
30°
15°
0°
20° N
0°
20° S
20° N
0°
20° S
Latitude
90
50
30
15
EQ
Latitude
¬120 ¬100 ¬80 ¬60 ¬40 ¬20 0 20 40 60 80
W m
¬2
W m
¬2
W m
¬2
¬80 ¬60 ¬40 ¬20 20 40 60 800
¬40¬50 ¬30 ¬20 ¬10 0 10 20
a
bd
c
Southern Hemisphere
Northern Hemisphere
Southern Hemisphere
Northern Hemisphere
Figure 2 | Observed net TOA radiation and surface energy fluxes. a, Zonal average TOA net downward radiation in the Northern Hemisphere and
Southern Hemisphere from CERES EBAF, 2001–2010 (ref. 2). b, Latitude–longitude structure of a. c, Zonal average of the net surface flux (ocean to
atmosphere) in the Northern Hemisphere and Southern Hemisphere from CERES EBAF TOA net radiation minus ERA-Interim atmospheric transports,
2001–2010 (ref. 3). d, Latitude–longitude structure of c.
Table 1 | The hemispheric asymmetry (Northern Hemisphere minus Southern Hemisphere) of energy flux into the atmosphere and
its implied magnitude of cross-equatorial energy transport.
10-year mean Range Standard deviation
Hemispheric
asymmetry
(W m
−2
)
Implied cross-EQ
transport (PW)
Hemispheric
asymmetry
(W m
−2
)
Implied cross-EQ
transport (PW)
Hemispheric
asymmetry
(W m
−2
)
Implied cross-EQ
transport (PW)
Downward TOA flux −1.5 −0.2 −1.9 to −1.0 −0.2 to −0.1 0.4 0.05
Upward surface flux 3.1 0.4 2.4 to 3.8 0.3 to 0.5 0.7 0.09
The upward surface plus downward TOA implied transport value corresponds to a southward atmospheric transport. The range and standard deviation are calculated using annual means. Cross-EQ,
cross-equatorial.
biases can even overwhelm the northward cross-equatorial ocean
transport, leading to a Southern Hemisphere maximum of zonally
averaged tropical precipitation.
We argue that the structure of the oceanic MOC causes
tropical precipitation to maximize in the Northern Hemisphere,
by the following mechanism, illustrated in Fig. 3. The Northern
Hemisphere ocean releases heat to the extratropical atmosphere,
which is compensated for by a lesser atmospheric heat transport
from the subtropics to the extratropics in the Northern Hemisphere
than in the Southern Hemisphere. The hemispheric difference
between the atmospheric transport is especially striking between
20
◦
and 50
◦
latitude, where the Southern Hemisphere transports
around 0.5 PW more energy poleward than the Northern
Hemisphere (Supplementary Fig. S2). The atmosphere therefore
moves less heat poleward from the Northern Hemisphere lower
latitudes than from the Southern Hemisphere lower latitudes.
The tropical mean circulation responds by transporting energy
from the Northern Hemisphere to the Southern Hemisphere. The
southward energy transport is accomplished by a northerly cross-
equatorial flow in the upper branch and an accompanying southerly
cross-equatorial flow in the lower branch, which also transports
moisture northward, resulting in the Northern Hemisphere peak
of tropical precipitation.
An aquaplanet global climate model (GCM) simulation forced
with the observed surface heat flux has tropical precipitation
peaking clearly within the Northern Hemisphere, at a higher
latitude than the observed rain band (Fig. 1c). This simulation
shows that the OHT is more than sufficient to explain the northern
peak in zonal mean precipitation. To show that OHT is also
necessary to produce a Northern Hemisphere tropical maximum,
we use a GCM with full continental geometry and the observed
surface flux distribution (Fig. 1c). When the surface flux in this
GCM is symmetrized between hemispheres, the precipitation
becomes much larger in the Southern Hemisphere tropics than
in the Northern Hemisphere (Fig. 1c). These simulations support
our conclusions from observations, that TOA radiation effects
942 NATURE GEOSCIENCE | VOL 6 | NOVEMBER 2013 | www.nature.com/naturegeoscience
© 2013 Macmillan Publishers Limited. All rights reserved
NATURE GEOSCIENCE DOI: 10.1038/NGEO1987
LETTERS
More eddy
transport
F
F
Less energy
entering the
atmosphere
Southward cross-EQ
energy transport
Tropical
rain
belt
q
q
Less eddy
transport
More energy
entering the
atmosphere
Northward cross-EQ
energy transport
PS
Figure 3 | Schematic of the role of the oceanic MOC in forcing the
Northern Hemisphere maximum of tropical precipitation. Heat is released
from the ocean to the atmosphere in the Northern Hemisphere owing to
cross-equatorial OHT. The atmosphere responds through eddy energy
transports in the extratropics, and a cross-equatorial Hadley circulation,
which fluxes energy from the Northern Hemisphere to the Southern
Hemisphere. The moisture transport by the Hadley circulation is in the
opposite direction as the energy transport, so tropical precipitation moves
northwards. SP, South Pole; NP, North Pole; cross-EQ, cross-equatorial;
q, moisture transport; F, energy transport.
(along with the possible influence of land) would cause a Southern
Hemisphere rainfall maximum, and the observed OHT is a primary
cause of the Northern Hemisphere maximum.
Our simulations are from atmosphere-only GCMs, and there
is the potential for a rich interplay among ocean dynamics, land
properties and TOA radiation, which should be studied in detail.
For instance, Earth’s continental configuration plays an important
role in setting the direction and strength of the oceanic MOC.
A recent idealized coupled GCM study
14
showed that opening a
Southern Hemisphere Drake Passage-like channel anchors deep
water production and the tropical rain band in the Northern
Hemisphere, explicitly supporting a MOC-determined tropical rain
band location within a coupled context. This result has since been
shown in an aquaplanet coupled GCM setting
18
. These and other
coupled interactions should be further examined in comprehensive
GCMs and observations.
Our argument by no means precludes the existence of strong
local influences on tropical precipitation. Many well-documented
local mechanisms shape the east–west asymmetries of rainfall, the
seasonal cycle and interannual variability, and seek to explain
the Northern Hemisphere peak in tropical rain through tropical
processes and continental/topographic features
19–21
. Indeed, in
our comprehensive model simulations with symmetrized OHT,
whereas the Pacific and Atlantic intertropical convergence zones
clearly shift into the Southern Hemisphere, the Indian Ocean
rainfall location is affected much less (Supplementary Fig. S5),
indicating the primary role of local processes within that basin. The
links between coupled tropical and tropical-extratropical dynamics
and global energetics should be investigated further as well.
This study on what controls the present tropical precipitation
position has strong implications for our understanding of past
and future climates. Much palaeoclimate literature focuses on
the possibility of changes in the MOC structure and strength.
This analysis suggests that such changes would be associated with
latitudinal shifts in tropical precipitation. Previous studies with
GCMs have shown that MOC slowdowns are indeed accompanied
by southward shifts in rainfall over much of the tropics
11
.
The role of TOA radiation was also probably different
in past climates. For instance, the Sahara Desert was green
in the mid-Holocene
22
, which could on its own cause the
Northern Hemisphere to receive more radiation than the Southern
Hemisphere, and shift tropical rainfall further into the Northern
Hemisphere, unless other terms such as clouds could compensate.
In contrast, in glacial climates when ice covered much of the
Northern Hemisphere, the albedo effect would probably reduce
Northern Hemisphere net radiation and push tropical precipitation
southward
11
, although reduced outgoing longwave radiation over
ice sheets would have a compensating effect.
In the future, owing to global warming, we expect reductions
in the strength of the oceanic MOC (ref. 4), accompanied by
changes in surface albedo, clouds, aerosols and ocean heat uptake,
presenting a complex challenge for GCM forecasts of tropical
precipitation
7,23
. The complexity of the task is matched by the
importance to tropical societies of determining what will, in
fact, happen. The analysis of the coupling between energy fluxes,
circulation and precipitation presented here provides a useful
paradigm for addressing this critical global problem.
Methods
A 10-year mean of the Clouds and the Earth’s Radiant Energy System (CERES)
Energy Balanced and Filled (EBAF) satellite product
2
from 2001 to 2010 is
used to study TOA energetics, which includes contributions from downward
solar radiation, upward solar radiation (determined by the distribution of the
planetary albedo) and outgoing longwave radiation (determined primarily by the
temperature, greenhouse gas concentration, and cloudiness of each column).
Surface heat fluxes are not routinely measured at the ocean surface. The most
accurate method to derive the net surface heat flux is, therefore, an indirect one,
subtracting atmospheric energy transport estimates based on reanalysis data from
TOA satellite observations of net radiative flux. We estimate the implied OHT using
the method of ref. 24 applied to the recently released ERA-Interim data set
3
over
the same time period as the CERES data (2001–2010). Although this new reanalysis
data set is thought to be a significant improvement over previous estimates, there
of course remains some uncertainty in any such energy budget estimate. A basic
physical expectation can be used to infer the sign of the cross-equatorial OHT
without reference to reanalyses; an assumption of thermal directness of the deep
tropical overturning, along with the larger TOA flux into the Southern Hemisphere,
requires a northward cross-equatorial OHT.
An aquaplanet (ocean-covered) version of the Geophysical Fluid Dynamics
Laboratory (GFDL) AM2.1 atmospheric GCM (ref. 25) forced with the observed
surface heat flux (Fig. 2d) into the oceanic mixed layer is integrated to demonstrate
that the observed surface heat flux is more than sufficient to shift the maximum
of zonal mean tropical precipitation to its observed location in the Northern
Hemisphere. The ECHAM GCM version 4.6 with realistic geography is additionally
used, forced with the observed surface flux distribution in the control case. The
symmetrized case with the ECHAM GCM uses a surface flux distribution that is
zonally symmetric, with values chosen at each latitude such that the zonal integral
of the surface flux is equal to the average between the Northern Hemisphere and
Southern Hemisphere zonal integrals of the observed surface flux.
Received 3 May 2013; accepted 13 September 2013;
published online 20 October 2013
References
1. Huffman, G. J., Adler, R. F., Bolvin, D. T. & Gu, G. Improving the global
precipitation record: GPCP Version 2.1. Geophys. Res. Lett. 36, L17808 (2009).
2. Wielicki, B. A. et al. Clouds and the Earth’s radiant energy system (CERES):
An Earth observing system experiment. Bull. Am. Meteorol. Soc. 77,
853–868 (1996).
3. Dee, D. P. et al. The ERA-Interim reanalysis: Configuration and performance
of the data assimilation system. Q. J. R. Meteorol. Soc. 137, 553–597 (2011).
4. Weaver, A. J. et al. Stability of the Atlantic meridional overturning circulation:
A model intercomparison. Geophys. Res. Lett. 39, L20709 (2012).
5. Hadley, G. Concerning the cause of the general trade winds. Phil. Trans. R. Soc
29, 58–62 (1735).
6. Dima, I. M. & Wallace, J. M. On the seasonality of the Hadley Cell. J. Atmos. Sci.
60, 1522–1527 (2003).
7. Frierson, D. M. W. & Hwang, Y-T. Extratropical influence on ITCZ shifts in
slab ocean simulation of global warming. J. Clim. 25, 720–733 (2012).
8. Hwang, Y-T. & Frierson, D. M. W. A new look at the double ITCZ problem:
Connections to cloud bias over the Southern Ocean. Proc. Natl Acad. Sci. USA
110, 4935–4940 (2013).
9. Hwang, Y-T., Frierson, D. M. W. & Kang, S. M. Anthropogenic sulfate
aerosol and the southward shift of tropical precipitation in the 20th century.
Geophys. Res. Lett. 40, 1–6 (2013).
NATURE GEOSCIENCE | VOL 6 | NOVEMBER 2013 | www.nature.com/naturegeoscience 943
© 2013 Macmillan Publishers Limited. All rights reserved
LETTERS
NATURE GEOSCIENCE DOI: 10.1038/NGEO1987
10. Chiang, J. C. H. & Bitz, C. M. Influence of high latitude ice cover on the marine
Intertropical Convergence Zone. Clim. Dynam. 25, 477–496 (2005).
11. Zhang, R. & Delworth, T. L. Simulated tropical response to a substantial
weakening of the Atlantic thermohaline circulation. J. Clim. 18,
1853–1860 (2005).
12. Kang, S. M., Held, I. M., Frierson, D. M. W & Zhao, M. The response of the
ITCZ to extratropical thermal forcing: Idealized slab-ocean experiments with a
GCM. J. Clim. 21, 3521–3532 (2008).
13. Yoshimori, M. & Broccoli, A. J. On the link between Hadley circulation changes
and radiative feedback processes. Geophys. Res. Lett. 36, L20703 (2009).
14. Fučkar, N. S., Xie, S-P., Farneti, R., Maroon, E. A. & Frierson, D. M. W.
Influence of the extratropical ocean circulation on the intertropical
convergence zone in an idealized coupled general circulation model. J. Clim.
26, 4612–4629 (2013).
15. Voigt, A., Stevens, B., Bader, J. & Mauritsen, T. The observed hemispheric
symmetry in reflected shortwave irradiance. J. Clim. 26, 468–477 (2013).
16. Marshall, J. & Speer, K. Closure of the meridional overturning circulation
through Southern Ocean upwelling. Nature Geosci. 5, 171–180 (2012).
17. Lumpkin, R. & Speer, K. Global ocean meridional overturning. J. Phys.
Oceanogr. 37, 2550–2562 (2007).
18. Marshall, J., Donohoe, A., Ferreira, D. & McGee, D. The role of the
ocean circulation in setting the mean position of the ITCZ. Clim. Dyn.
http://dx.doi.org/10.1007/s00382-013-1767-z (2013).
19. Song, X. & Zhang, G. J. Convection parameterization, tropical Pacific double
ITCZ, and upper ocean biases in the NCAR CCSM3. Part II: Coupled feedback
and the role of ocean heat transport. J. Clim. 23, 800–812 (2010).
20. Xie, S-P. & Philander, S. G. H. A coupled ocean-atmosphere model of relevance
to the ITCZ in the eastern Pacific. Tellus 46A, 340–350 (1994).
21. Takahashi, K. & Battisti, D. S. Processes controlling the mean tropical pacific
precipitation pattern. part I: The Andes and the eastern Pacific ITCZ. J. Clim.
20, 3434–3451 (2007).
22. DeMenocal, P. B et al. Abrupt onset and termination of the African Humid
Period: Rapid climate response to gradual insolation forcing. Quat. Sci. Rev.
19, 347–361 (2000).
23. IPCC Climate Change 2007: The Physical Science Basis (eds Solomon, S. et al.)
(Cambridge Univ. Press, 2007).
24. Trenberth, K. E. & Fasullo, J. T. An observational estimate of ocean energy
divergence. J. Phys. Oceanogr. 38, 984–999 (2008).
25. Anderson, J. L. et al. The new GFDL global atmosphere and land model
AM2–LM2: Evaluation with prescribed SST simulations. J. Clim. 17,
4641–4673 (2004).
Acknowledgements
We acknowledge helpful conversations with L. Thompson, C. Bitz, K. Armour, B. Rose,
I. Held, R. Pierrehumbert, D. Hartmann, J. Scheff and M. Wallace. J. Fasullo provided
the ERA-Interim energy transports. We acknowledge the Program for Climate Model
Diagnosis and Intercomparison and the WCRP’s Working Group on Coupled Modelling
for their roles in making available the CMIP3 and CMIP5 data sets. D.M.W.F. and
Y-T.H. are supported by NSF Grants AGS-0846641 and AGS-0936069, and a University
of Washington Royalty Research Fund grant. N.S.F. is supported by the Japan Agency
for Marine-Earth Science and Technology (JAMSTEC), by NASA through grant
No. NNX07AG53G, and by NOAA through grant No. NA11NMF4320128, which
sponsor research at the International Pacific Research Center. R.S. is supported by
NSF award AGS-0804107. S.M.K. is supported by the Basic Science Research Program
through the National Research Foundation of Korea (NRF) funded by the Ministry of
Science, ICT and Future Planning (2013R1A1A3004589). A.D. is supported by the NOAA
Climate and Global Change Fellowship, administered by the University Corporation
for Atmospheric Research. E.A.M. is supported by the National Defense Science and
Engineering Graduate Fellowship Program.
Author contributions
D.M.W.F., N.S.F. and Y-T.H. designed the original diagnostics and experiments, with
frequent subsequent input on diagnostic techniques and experimental design from
all co-authors. Y-T.H. and D.M.W.F. analysed the observations. Y-T.H. performed
experiments with the GFDL model. X.L. and D.S.B. designed experiments with the
ECHAM model, and X.L. ran the ECHAM model experiments. D.M.W.F. led the writing
of the paper, with substantial input from all co-authors.
Additional information
Supplementary information is available in the online version of the paper. Reprints and
permissions information is available online at www.nature.com/reprints. Correspondence
and requests for materials should be addressed to D.M.W.F.
Competing financial interests
The authors declare no competing financial interests.
944 NATURE GEOSCIENCE | VOL 6 | NOVEMBER 2013 | www.nature.com/naturegeoscience
© 2013 Macmillan Publishers Limited. All rights reserved
