WELL THIS IS AWKWARD
http://www.nature.com/ncl...n2/full/nclimate2100.html
El Niño events are a prominent feature of climate variability
with global climatic impacts. The 1997/98 episode, often
referred to as ‘the climate event of the twentieth century’1,2
,
and the 1982/83 extreme El Niño3
, featured a pronounced
eastward extension of the west Pacific warm pool and
development of atmospheric convection, and hence a huge
rainfall increase, in the usually cold and dry equatorial
eastern Pacific. Such a massive reorganization of atmospheric
convection, which we define as an extreme El Niño, severely
disrupted global weather patterns, affecting ecosystems4,5
,
agriculture6
, tropical cyclones, drought, bushfires, floods and
other extreme weather events worldwide3,7–9
. Potential future
changes in such extreme El Niño occurrences could have
profound socio-economic consequences. Here we present
climate modelling evidence for a doubling in the occurrences
in the future in response to greenhouse warming. We estimate
the change by aggregating results from climate models in
the Coupled Model Intercomparison Project phases 3 (CMIP3;
ref. 10) and 5 (CMIP5; ref. 11) multi-model databases, and
a perturbed physics ensemble12. The increased frequency
arises from a projected surface warming over the eastern
equatorial Pacific that occurs faster than in the surrounding
ocean waters13,14, facilitating more occurrences of atmospheric
convection in the eastern equatorial region.
The 1982/83 and 1997/98 extreme El Niño events were characterized
by an exceptional warming, with sea surface temperatures
(SSTs) exceeding 28 ◦C extending into the eastern equatorial
Pacific2,3
. This led to an equatorward shift of the intertropical convergence
zone (ITCZ), and hence intense rainfall in the equatorial
eastern Pacific where cold and dry conditions normally prevail. This
major reorganization of atmospheric convection severely disrupted
global weather patterns and spurred major natural disasters. Catastrophic
floods occurred in the eastern equatorial region of Ecuador
and northern Peru3,7
, and neighbouring regions to the south and
north experienced severe droughts (Supplementary Fig. 1). The
anomalous conditions caused widespread environmental disruptions,
including the disappearance of marine life and decimation
of the native bird population in the Galapagos Islands15,16, and
severe bleaching of corals in the Pacific and beyond4,5
. The impacts
1CSIRO Marine and Atmospheric Research, Aspendale, Victoria 3195, Australia, 2Physical Oceanography Laboratory, Qingdao Collaborative Innovation
Center of Marine Science and Technology, Ocean University of China, Qingdao 266003, China, 3
Laboratoire d’Océanographie et du Climat:
Expérimentation et Approches Numériques (LOCEAN), IRD/UPMC/CNRS/MNHN, 75252 Paris Cedex 05, France, 4College of Engineering Mathematics
and Physical Sciences, Harrison Building, Streatham Campus, University of Exeter, Exeter EX1 3PB, UK, 5Geophysical Fluid Dynamics Laboratory/NOAA,
Princeton, New Jersey 08540-6649, USA, 6
IPRC, Department of Oceanography, SOEST, University of Hawaii, Honolulu, Hawaii 96822, USA, 7Australian
Research Council (ARC) Centre of Excellence for Climate System Science, Level 4 Mathews Building, The University of New South Wales, Sydney 2052,
Australia, 8NOAA/Pacific Marine Environmental Laboratory, Seattle, Washington 98115, USA, 9NCAS-Climate, University of Reading, Reading RG6 6BB,
UK, 10Department of Meteorology, SOEST, University of Hawaii, Honolulu, Hawaii 96822, USA. *e-mail:
[email protected]
extended to every continent, and the 1997/98 event alone caused
US$35–45 billion in damage and claimed an estimated 23,000
human lives worldwide17
.
The devastating impacts demand an examination of whether
greenhouse warming will alter the frequency of such extreme El
Niño events. Although many studies have examined the effects of
a projected warming on the Pacific mean state, El Niño diversity
and El Niño teleconnections18–21, the issue of how extreme El
Niños will change has not been investigated. Here we show
that greenhouse warming leads to a significant increase in the
frequency of such events.
We contrast the characteristics between the extreme and
moderate El Niño events using available data sets22,23, focusing
on December–January–February (DJF), the season in which
El Niño events peak. During moderate events, which include
canonical and Modoki El Niño24, the eastern boundary of the
warm pool (indicated by the 28 ◦C isotherm, purple, Fig. 1a)
and the atmospheric convective zone move eastwards to just
east of the Date Line. The ITCZ lies north of the Equator25
,
and the rainfall anomaly over the eastern equatorial Pacific
is small (Fig. 1a).
During extreme El Niño, the warm pool expands eastward and
eventually covers the entire equatorial Pacific (Fig. 1b), markedly
weakening the equatorial east–west and meridional SST gradients
(Fig. 1c,d); the latter being defined as the difference between
the northern off-equatorial (8◦ N, the ITCZ position) and the
equatorial Pacific. Consequently convection, which follows the
highest SSTs, extends eastward and the ITCZ shifts equatorward25
,
leading to atmospheric convection and extraordinary rainfall
(>5 mm per day, green, Fig. 1b) in the normally dry eastern
equatorial Pacific. There, easterly winds are replaced by westerlies,
which suppress the eastern equatorial upwelling2,3
, reinforcing the
exceptionally high SSTs in this region. In association, Niño3 areaaveraged
rainfall increases nonlinearly with Niño3 SST (Fig. 1c) and
the meridional SST gradient (Fig. 1d).
This striking rainfall nonlinearity is the distinct feature of an
extreme El Niño, reflecting the pronounced shift in convective
zone, with concurrent weakening of SST gradients in the eastern
equatorial Pacific (Supplementary Fig. 3). Niño3 rainfall is thus a
NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION |
www.nature.com/natureclimatechange 1
LETTERS NATURE CLIMATE CHANGE DOI:10.1038/NCLIMATE2100
Nov
Sep
Jul
May
Mar
Jan
Nov
Sep
Jul
May
Mar
Jan
Nov
Sep
Jul
May
Mar
Jan
Nov
Sep
Jul
May
Mar
Jan
Moderate EI Niño, 1979¬2010
Observed relationship, 1979¬2010 Observed relationship, 1979¬2010
Extreme EI Niño, 1979¬2010
180° 135° W 180° 135° W
¬2.5 ¬2.0 ¬1.5 ¬1.0 ¬0.5 0.0
(°C)
0.5 1.0 1.5 2.0 2.5
12
1997/98
1982/83
1997/98
1982/83
9
6
3
0
12
9
6
3
0
Niño3 rainfall (mm d¬1)
Niño3 SST (°C) Meridional SST gradient (°C) Niño3 rainfall (mm d¬1)
23 24 25 26 27 28 29 ¬2 ¬1 0 1 2 3
3
3
3
a b
c d
Figure 1 | Evolution and nonlinear characteristics of observed extreme El Niño events. a,b, Time–longitude diagram for composite moderate and extreme
El Niño events, respectively, of equatorial SST anomalies (colour scale) and rainfall anomalies (contour, at intervals of 3 mm per day), 28 ◦C isotherm
(purple curve) and total rainfall 5 mm per day isopleth (green curve). c,d, Relationship of eastern equatorial Pacific (Niño3 area: 5◦ S–5◦ N, 150◦ W–90◦ W)
DJF total rainfall with DJF Niño3 SST and meridional SST gradients in the Niño3 longitude range. The meridional SST gradient is defined as the average SST
over the off-equatorial region (5◦ N–10◦ N, 150◦ W–90◦ W) minus the average over the equatorial region (2.5
◦ S–2.5
◦ N, 150◦ W–90◦ W). Extreme El Niño
(defined as events for which austral summer rainfall is greater than 5 mm per day), moderate El Niño (defined as events with SST anomalies greater than
0.5 s.d. of that over the period since 1979 that are not extreme El Niño events), and La Niña and neutral events, are indicated by red, green and blue dots
respectively. During extreme El Niño, the meridional SST gradient diminishes, or reverses, shifting the ITCZ to the eastern equatorial Pacific.
good indicator of extreme El Niño25,26, particularly because it is
the anomalous convection and rainfall that in turn influence global
weather. Rainfall is also an excellent measure for condensational
heating of the atmosphere, thus providing an effective metric
for large-scale atmospheric circulation anomalies. The observed
rainfall nonlinearity can be measured by its skewness, which is
greater than one over the period since 1979. We define an extreme
El Niño as an event during which such massive reorganization
of atmospheric convection takes place, leading to Niño3 rainfall
that exceeds 5 mm per day; similar to a previous definition that
used a rainfall anomaly threshold in the eastern Pacific25. Our
definition distinctly identifies the 1982/83 and 1997/98 events
as extreme El Niños.
The extraordinary large-scale changes over the tropical Pacific
during an extreme El Niño mean that such events can induce
extreme swings of the South Pacific convergence zone (SPCZ)
(referred to as zonal SPCZ events), as occurred in 1982/83 and
1997/98 (refs 8,9). However, zonal SPCZ events can also occur
without extreme El Niños8
, as in 1991/92. During the 1991/92 event,
observations and reanalyses23,24 show that the 28 ◦C isotherm, the
5 mm per day rainfall isopleths, or ascending motion, did not cover
the entire Niño3 region, and the ITCZ was still situated north of the
Equator (Supplementary Fig. 4), in stark contrast to the 1982/83
and 1997/98 extreme El Niño events (Supplementary Fig. 2c). As
zonal SPCZ events can occur without extreme El Niños, as is also the
case in the climate models considered here (Supplementary Fig. 5),
2 NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION |
www.nature.com/natureclimatechange
NATURE CLIMATE CHANGE DOI:10.1038/NCLIMATE2100 LETTERS
Moderate EI Niño, 1891¬1990 Extreme EI Niño, 1891¬1990
180° 135° W 180° 135° W
¬2.5 ¬2.0 ¬1.5 ¬1.0 ¬0.5 0.0
(°C)
0.5 1.0 1.5 2.0 2.5
Modelled relationship, 1891¬1990 Modelled relationship, 1991¬2090
12
15
Moderate = 475
Extreme = 101
Moderate = 341
Extreme = 212
9
6
3
0
¬2 0 2 4 6
Niño3 rainfall (mm d¬1)
Meridional SST gradient (°C)
¬2 0 2 4 6
Meridional SST gradient (°C)
Nov
Sep
Jul
May
Mar
Jan
Nov
Sep
Jul
May
Mar
Jan
Nov
Sep
Jul
May
Mar
Jan
Nov
Sep
Jul
May
Mar
Jan
12
15
9
6
3
0
Niño3 rainfall (mm d¬1)
a
c
b
d
Figure 2 | Evolution and nonlinear characteristics of model extreme El Niño events, and changes in occurrences under greenhouse warming.
a,b, Time–longitude diagram for composite moderate and extreme El Niño events, respectively, of equatorial SST anomalies (colour scale) and rainfall
anomalies (contour, at intervals of 3 mm per day), 28 ◦C isotherm (purple curve) and total rainfall 5 mm per day isopleth (green curve) for the control
period, illustrating simulation of the observed evolution. c,d, Relationship between eastern equatorial Pacific (Niño3 area: 5◦ S–5◦ N, 150◦ W–90◦ W)
austral summer total rainfall and austral summer meridional SST gradient for the control and climate change periods, respectively. Red, green and blue dots
indicate extreme El Niño (defined as events for which austral summer rainfall is greater than 5 mm per day), moderate El Niño (defined as events with SST
anomalies greater than 0.5 s.d. of the control period that are not extreme El Niño events), and La Niña and neutral events, that is, all non-El Niño years,
respectively. The number of moderate El Niño and extreme El Niño events in each period is shown.
increased occurrences of zonal SPCZ events under global warming9
should not be used to infer a change in frequency of extreme El Niño
events, an issue that is specifically examined here.
Not all coupled general circulation models (CGCMs) simulate
the observed level of rainfall skewness. The CGCMs are forced with
historical anthropogenic and natural forcings, and future greenhouse
gas emission scenarios (Methods), each covering a 200-year
period. We determine Niño3 rainfall skewness over the 200-year
period in each model. Using skewness greater than 1 and Niño3
rainfall exceeding 5 mm per day as criteria for model selection, we
identify 9 CMIP3 and 11 CMIP5 CGCMs that can simulate an extreme
El Niño (Supplementary Figs 6–13 and Tables 1–2). For each
of these 20 CGCMs, we compare the frequency of extreme El Niño
in the first (1891–1990) and second (1991–2090) 100-year periods,
referred to as the control and climate change periods, respectively.
The models reproduce the contrasts between moderate and
extreme El Niño events (Fig. 2a,b) as seen in observations (Fig. 1a,b
and Supplementary Fig. 2), associated with large reductions in
meridional and zonal SST gradients (Supplementary Fig. 14). In
aggregation, the total number of El Niño events decreases slightly
but the total number of extreme El Niño events increases (Fig. 2c,d).
The frequency of extreme El Niños doubles from about one event
every 20 years (101 events in 2,000 years) in the control, to one
every 10 years (212 events in 2,000 years) in the climate change
period (Fig. 2c,d). This is statistically significant according to a
bootstrap test27, underscored by a strong inter-model consensus,
with 17 out of 20 models simulating an increase (Supplementary
Tables 1–2). These robust statistics are particularly compelling given
the large inter-model differences in convective parameterizations28
.
Sensitivity tests to varying definitions of extreme El Niño (for
NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION |
www.nature.com/natureclimatechange 3
LETTERS NATURE CLIMATE CHANGE DOI:10.1038/NCLIMATE2100
All samples
Anomaly (°C) Anomaly > 2 mm d¬1
Extreme EI Niño
500 70
60
50
40
30
20
10
0
400
300
Number of austral summers
Number of austral summers
70
60
50
40
30
20
10
0
Number of austral summers
70
60
50
40
30
20
10
0
Detrended Niño3 SST anomaly (s.d.)
Number of austral summers
200
100
0
¬2 ¬1 0 1
Meridional SST gradient (°C) Meridional SST gradient (°C)
23456 ¬2 ¬1 0 1 2 3 4 5 6
1891¬1990
1991¬2090
Detrended Niño3 rainfall anomaly (mm d¬1)
a b
c d
0 1 2 3 4 5 6 2 4 6 8 10 12
Figure 3 | Multi-model statistics associated with the increase in the frequency of extreme El Niño events. a,b, Multi-model histograms of the meridional
SST gradient in the eastern equatorial Pacific for all samples, and for extreme El Niño alone. The meridional SST gradient is defined as the average SST over
the east off-equatorial region (5◦ N–10◦ N, 150◦ E–90◦ W) minus the average over the eastern equatorial region (2.5
◦ S–2.5
◦ N, 155◦ E–120◦ W). All 2,000
samples in each period are distributed into 0.5
◦C bins, for the control (blue) and climate change (red) periods. The multi-model climatological values for
the control (blue dashed line) and the climate change (red dashed line) periods are indicated. c, Multi-model histogram of quadratically detrended SST
anomalies for extreme El Niño events alone. d, Multi-model histogram of quadratically detrended rainfall anomalies for El Niño defined as Niño3 rainfall
anomalies greater than 2 mm per day, showing changes in occurrences from control to the climate change period for given positive anomalies greater than
2 mm per day.
example, in conjunction with a diminishing meridional SST
gradient, or using Niño3 rainfall relative to the western Pacific
rainfall), or inclusion of all CGCMs, further support the robustness
of this result (Supplementary Tables 3–6).
We assess the potential impact of the well-known cold SST bias
using the HadCM3 CGCM, in which biases are corrected through a
flux adjustment12 in perturbed physics ensemble (PPE) experiments
that produce extreme El Niño events. There is a fourfold increase in
the frequency from one event in 60 years in the control to one event
in 15 years in the climate change period (Supplementary Fig. 15 and
Table 7). Thus, our conclusion remains valid in the absence of the
SST biases. Although flux adjustments are not sufficient to correct
all errors in models, this result does provide further evidence for
future increase in extreme El Niño frequency.
The more frequent establishment of atmospheric convection in
the eastern equatorial Pacific is induced by diminishing, or reversing,
meridional and zonal SST gradients (Supplementary Fig. 14),
rather than by a localized warming that exceeds the convective
threshold range of the control period. For the latter to be true,
the tropical convective area in recent decades and future climate
simulations must expand, but there is no systematic evidence
for this. Further, the CGCMs that are not selected, mostly with
SSTs below the convective threshold in the control period, are
unable to produce extreme El Niño events after the threshold is
reached in the climate change period (Supplementary Figs 12–13),
supporting the idea that convective threshold increases with
mean SSTs (ref. 29).
The weakening of the SST gradients is induced by faster
warming in the background state along the equatorial than in the
off-equatorial Pacific, and in the eastern equatorial Pacific than
in the west (Supplementary Fig. 16a)13,14, a feature produced even
without a dynamical ocean30. These slight changes in climatological
SST gradient translate into a large increase in the occurrences of a
diminished or reversed meridional SST gradient (Fig. 3a). This is associated
with more occurrences of maximum SSTs, and hence convection,
in the eastern equatorial Pacific for a given SST anomaly,
4 NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION |
www.nature.com/natureclimatechange
NATURE CLIMATE CHANGE DOI:10.1038/NCLIMATE2100 LETTERS
Normal condition
Extreme El Niño
Equator
Normal condition
Extreme El Niño
Equator
Present
Future
a
b
Figure 4 | Schematic depicting the mechanism for increased occurrences
of extreme El Niño under greenhouse warming. a,b, In both present-day
climate (a) and future climate (b), convection zones in the western Pacific
and the ITCZ latitudes shift from their normal positions (indicated by blue
clouds) to the eastern equatorial Pacific during an extreme El Niño event
(indicated by red clouds). Colour shading indicates mean SSTs and black
contours indicate SST anomalies. Under greenhouse-gas-induced warming
conditions, warming occurs everywhere but at a faster rate in the eastern
equatorial Pacific, diminishing the zonal and meridional SST gradients.
Strong SST gradients are a barrier to a shift in convection zones. Therefore,
in the future climate, shifts in convection zones can be facilitated by weaker
changes in SST and thus SST gradients (indicated by one black contour and
by green arrows), as compared with the present-day climate in which
stronger changes are required (indicated by two black contours and red
arrows).
leading to increased extreme El Niño occurrences (Fig. 3b) even
though neither the average amplitude of El Niño-related SST
anomalies nor the frequency of El Niño is substantially changed19
.
The increased extreme El Niño events do not simply result from
an increasing climatological rainfall, but from enhanced probability
of the establishment of atmospheric deep convection in the eastern
equatorial Pacific through the change in background conditions: as
the Equator warms more rapidly than the SSTs at the climatological
position of the ITCZ, it takes a relatively weaker SST anomaly
as compared with the control period to establish the warmest
water over the equatorial eastern Pacific (Fig. 4). Detrended SST
anomalies averaged over extreme El Niño events are indeed slightly
smaller in the climate change period than those in the control
period. There is virtually no change in occurrences concurrent with
high SST anomalies (for example, >2
◦C), and most of the increased
occurrences of extreme El Niño are associated with smaller SST
anomalies (Fig. 3c). The increased frequency of convection in the
eastern equatorial Pacific is further highlighted by a 66% increase in
the occurrences of detrended Niño3 rainfall anomalies greater than
2 mm per day, with a strong inter-model consensus (Fig. 3d and
Supplementary Table 8 and Fig. 17). In contrast, there is no robust
change following the commonly used definition based on similarly
detrended, standardized Niño3 SST anomalies (Supplementary
Table 9). The increased occurrences of atmospheric convection in
the eastern equatorial Pacific depict stronger air–sea interactions,
with a 25% increase in the sensitivity of detrended rainfall to
positive SST anomalies (Supplementary Fig. 18). Despite these
fundamental changes, the spatial pattern of the associated rainfall
teleconnection remains overall similar to that in the control period
(Supplementary Fig. 19), suggesting that, at a given location, past
extreme El Niño impacts will repeat more frequently in the future
as the planet warms.
In summary, our result of greenhouse-induced increased
occurrence of extreme El Niño events is in stark contrast with
previous findings of no consensus in El Niño change; our robust
results arise from the use of process-based metrics, such as
SST gradients and the impacts of reorganization of atmospheric
convection, that isolate the mechanism of extreme El Niño
events. With a projected large increase in extreme El Niño
occurrences, we should expect more occurrences of devastating
weather events, which will have pronounced implications for
twenty-first century climate.
Full methods and any associated references are available in the
Supplementary Information.
Methods
Diagnosis of extreme El Niño events. We use rainfall data in the satellite era
(1979–present)23, and SSTs from a global reanalysis22. DJF rainfall averaged over
the Niño3 region (150◦ W–90◦ W, 5◦
S–5◦ N) and meridional SST gradient in the
eastern Pacific (150◦ W–90◦ W), calculated as the difference between the average
over the off-equatorial (5◦ N–10◦ N) and equatorial box (2.5
◦
S–2.5
◦ N) regions,
are used as atmospheric and oceanic indices to characterize extreme El Niño events.
Rainfall increases nonlinearly with Niño3 SST, or with the meridional gradient.
The nonlinearity is measured by the skewness of Niño3 precipitation, which is
2.75 in observations. DJF Niño3 rainfall greater than 5 mm per day defines an
extreme El Niño event.
Selection of models. DJF Niño3 rainfall greater than 5 mm per day and rainfall
skewness greater than 1 are used as criteria for model selection from a total
of 19 CMIP3 (ref. 10) and 21 CMIP5 (ref. 11) CGCMs. One experiment (the
first simulation) from each model is used, covering the period 1891–2090 using
historical anthropogenic and natural forcings to 2000 for CMIP3 and 2005 for
CMIP5, and then a future emission scenario SRESA2 for CMIP3 and the RCP8.5
for CMIP5. In addition, 33 SST-bias-corrected PPE experiments, conducted
with the HadCM3 CGCM forced with historical radiative perturbations and a
1% per year CO2 increase12 for the future climate change runs, each covering a
200-year period, are used. Only 9 CMIP3 and 11 CMIP5 CGCMs meet the criteria
(Supplementary Tables 1 and 2), yielding a mean skewness close to the observed
(Supplementary Tables 1 and 2). The skewness criterion filters out models with an
overly wet or cold and dry model east equatorial Pacific (Supplementary Figs 10
and 11). These biases generally reduce the skewness, and are associated with SSTs
well below or above the convective threshold range of 26–28 ◦C (ref. 29), leading
to overly subdued or active Niño3 rainfall variability. Out of 33 PPE experiments,
25 meet the skewness criterion. We derive changes in the frequency of extreme
El Niño events by comparing the first 100 years (control period) to the later
(climate change period) years. We also test the sensitivity of our results to varying
definitions (Supplementary Tables 3–6).
Contrasts between extreme El Niño and zonal SPCZ events. Neither zonal SPCZ
nor extreme El Niño is a subset of the other (Supplementary Fig. 5a). This is
because zonal SPCZ events are more closely associated with the south off-equatorial
minus the equatorial meridional SST gradients over the central Pacific longitudes
(Supplementary Fig. 5b), instead of the north off-equatorial minus the equatorial
SST gradients over the eastern Pacific longitudes, which characterize extreme El
Niño (Fig. 2c,d and Supplementary Fig. 5c). An aggregation over the 20 selected
CGCMs (Supplementary Figs 6–13 and Tables 1–2) and over 200 years shows
that about 40% of all zonal SPCZ events are independent from extreme El Niño
events (green dots, Supplementary Fig. 5a and Table 10), analogous to the 1991/92
event, with generally lower Niño3 rainfall and larger north off-equatorial minus
equatorial SST gradients in the eastern Pacific, in contrast to those during extreme
El Niño events that can occur without zonal SPCZ events (about 20%, purple dots
in Supplementary Fig. 5a). Supplementary Fig. 5b–e further contrasts the SST and
rainfall anomaly patterns associated with independent zonal SPCZ events from
those during extreme El Niño events in which the anomalies extend farther east
into the Niño3 region.
Total rainfall change. The total rainfall change in the eastern equatorial Pacific
under greenhouse warming (1Raintotal) contains contributions from a change in
the annual cycle (1Rainannual-cycle), a long-term trend (1Rainlong-term), and a change
in the response of rainfall to changing El Niño/Southern Oscillation (ENSO;
1RainENSO). For a given season, the 1Rainannual-cycle and 1Rainlong-term terms can be
combined to a total long-term trend, 1Raintotal-long-term, such that
1Raintotal = 1Raintotal-long-term +1RainENSO
As ENSO is seasonally phase-locked, peaking in austral summer, if there is a trend
due to the response to ENSO, the total rainfall trend will include the contribution
from 1RainENSO, which would be at least partially removed by the detrending
process. To understand how the distribution of rainfall anomalies will change,
rainfall is quadratically detrended. The detrending process might partially remove
the rainfall increase due to the increased frequency of extreme El Niño events.
NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION |
www.nature.com/natureclimatechange 5
LETTERS NATURE CLIMATE CHANGE DOI:10.1038/NCLIMATE2100
Statistical significance test. We use a bootstrap method27 to examine whether
the change in frequency of the extreme El Niño events is statistically significant.
The 2,000 DJF samples from the 20 selected CGCMs in the control period are
re-sampled randomly with replacement to construct 10,000 realizations. The
standard deviation of the extreme El Niño frequency in the inter-realization is 9.8
events per 2,000 years, far smaller than the difference between the control and
the climate change periods at 111 events per 2,000 years (Fig. 2c,d), indicating
statistical significance of the difference.
Received 15 October 2013; accepted 11 December 2013;
published online 19 January 2014
References
1. Changnon, S. A. El Niño, 1997–1998: The Climate Event of the Century (Oxford
Univ. Press, 2000).
2. McPhaden, M. J. El Niño: The child prodigy of 1997–98. Nature 398,
559–562 (1999).
3. Philander, S. G. H. Anomalous El Niño of 1982–83. Nature 305, 16 (1983).
4. Glynn, P. W. & de Weerdt, W. H. Elimination of two reef-building hydrocorals
following the 1982–83 El Niño. Science 253, 69–71 (1991).
5. Aronson, R. B. et al. Coral bleach-out in Belize. Nature 405, 36 (2000).
6. Wilhite, D. A., Wood, D. A. & Meyer, S. J. in Climate Crisis (eds Glantz, M.,
Katz, R. & Krenz, M.) 75–78 (UNEP, 1987).
7. Vos, R., Velasco, M. & Edgar de Labastida, R. Economic and social effects of El
Niño in Ecuador, 1997–1998 (Inter-American Development Bank, Sustainable
Development Dept. Technical papers series POV-107, 1999).
8. Vincent, E. M. et al. Interannual variability of the South Pacific Convergence
Zone and implications for tropical cyclone genesis. Clim. Dynam. 36,
1881–1896 (2011).
9. Cai, W. et al. More extreme swings of the South Pacific convergence zone due
to greenhouse warming. Nature 488, 365–369 (2012).
10. Meehl, G. et al. The WCRP CMIP3 multimodel dataset: A new era in climate
change research. Bull. Am. Meteorol. Soc. 88, 1383–1394 (2007).
11. Taylor, K. E., Stouffer, R. J. & Meehl, G. A. An overview of CMIP5 and the
experimental design. Bull. Am. Meteorol. Soc. 93, 485–498 (2012).
12. Collins, M. et al. A comparison of perturbed physics and multi-model
ensembles: Model errors, feedbacks and forcings. Clim. Dynam. 36,
1737–1766 (2011).
13. Xie, S. P. et al. Global warming pattern formation: Sea surface temperature and
rainfall. J. Clim. 23, 966–986 (2010).
14. Tokinaga, H., Xie, S-P., Deser, C., Kosaka, Y. & Okumura, Y. M. Slowdown of
the Walker circulation driven by tropical Indo-Pacific warming. Nature 491,
439–443 (2012).
15. Valle, C. A. et al. The Impact of the 1982–1983 E1 Niño-Southern Oscillation
on Seabirds in the Galapagos Islands, Ecuador. J. Geophys. Res. 92,
14437–14444 (1987).
16. Merlen, G. The 1982–1983 El Niño: Some of its consequences for Galapagos
wildlife. Oryx 18, 210–214 (1984).
17. Sponberg, K. Compendium of Climatological Impacts, University
Corporation for Atmospheric Research Vol. 1 (National Oceanic and
Atmospheric Administration, Office of Global Programs, 1999);
http://www.cip.noaa.gov/docs/navimpact.pdf.
18. Timmermann, A. et al. Increased El Niño frequency in a climate model forced
by future greenhouse warming. Nature 398, 694–697 (1999).
19. Collins, M. et al. The impact of global warming on the tropical Pacific Ocean
and El Niño. Nature Geosci. 3, 391–397 (2010).
20. Yeh, S-W. et al. El Niño in a changing climate. Nature 461, 511–514 (2009).
21. Kug, J-S., An, S. I., Ham, Y. G. & Kang, I-S. Changes in El Niño and La Niña
teleconnection over North Pacific–America in the global warming simulations.
Theor. Appl. Climatol. 100, 275–282 (2010).
22. Rayner, N. A. et al. Global analyses of sea surface temperature, sea ice, and
night marine air temperature since the late nineteenth century. J. Geophys. Res.
108, 4407 (2003).
23. Adler, R. F. et al. The Version-2 Global Precipitation Climatology Project
(GPCP) Monthly Precipitation Analysis (1979–present). J. Hydrometeorol. 4,
1147–1167 (2003).
24. Ashok, K., Behera, S. K., Rao, S. A., Weng, H. & Yamagata, T. El Niño Modoki
and its possible teleconnection. J. Geophys. Res. 112, C11007 (2007).
25. Lengaigne, M. & Vecchi, G. A. Contrasting the termination of moderate and
extreme El Niño events in coupled general circulation models. Clim. Dynam.
35, 299–313 (2010).
26. Chiodi, A. M. & Harrison, D. E. Characterizing warm-ENSO variability in the
equatorial Pacific: An OLR perspective. J. Clim. 23, 2428–2439 (2010).
27. Austin, P. C. Bootstrap methods for developing predictive models. Am. Stat.
58, 131–137 (2004).
28. Kim, D. et al. El Niño–Southern Oscillation sensitivity to cumulus entrainment
in a coupled general circulation model. J. Geophys. Res. 116, D22112 (2011).
29. Johnson, N. C. & Xie, S-P. Changes in the sea surface temperature threshold
for tropical convection. Nature Geosci. 3, 842–845 (2010).
30. Vecchi, G. A. & Soden, B. J. Global warming and the weakening of the tropical
circulation. J. Clim. 20, 4316–4340 (2007).
Acknowledgements
W.C., S.B. and P.v.R. are supported by the Australian Climate Change Science
Program. W.C. is also supported by Goyder Research Institute, the CSIRO Office of
Chief Executive Science Leader award, and Pacific Australia Climate Change Science
Adaptation Programme. M.J.M. is supported by NOAA; PMEL contribution 4049.
M.C. is supported by the NERC SAPRISE project (NE/I022841/1); A.T. is supported
by NSF grant number 1049219; M.H.E. and A.S., by a grant under the ARC Laureate
Fellowship scheme (FL100100214); L.W. by China National Natural Science Foundation
Key Project(41130859); and E.G. by Agence Nationale pour la Recherche projects
ANR-10-Blanc-616 METRO.
Author contributions
W.C. conceived the study in discussion with M.L. and G.V., and wrote the initial
draft of the paper. S.B., P.v.R. and G.W. performed the analysis. M.C. conducted the
perturbed physics ensemble climate change experiments with the HadCM3 model. All
authors contributed to interpreting results, discussion of the associated dynamics, and
improvement of this paper.
Additional information
Supplementary information is available in the online version of the paper. Reprints and
permissions information is available online atwww.nature.com/reprints. Correspondence
and requests for materials should be addressed to W.C.
Competing financial interests
The authors declare no competing financial interests.
6 NATURE CLIMATE CHANGE |
