PERSPECTIVE

PUBLISHED ONLINE: 22 JUNE 2015 | DOI: 10.1038/NCLIMATE2657

Attribution of climate extreme events

Kevin E. Trenberth1, John T. Fasullo1 and Theodore G. Shepherd2

There is a tremendous desire to attribute causes to weather and climate events that is often challenging from a physical
standpoint. Headlines attributing an event solely to either human-induced climate change or natural variability can be mis-
leading when both are invariably in play. The conventional attribution framework struggles with dynamically driven extremes
because of the small signal-to-noise ratios and often uncertain nature of the forced changes. Here, we suggest that a differ-
ent framing is desirable, which asks why such extremes unfold the way they do. Specifically, we suggest that it is more use-
ful to regard the extreme circulation regime or weather event as being largely unaffected by climate change, and question
whether known changes in the climate system’s thermodynamic state affected the impact of the particular event. Some exam-
ples briefly illustrated include ‘snowmaggedon’ in February 2010, superstorm Sandy in October 2012 and supertyphoon Haiyan
in November 2013, and, in more detail, the Boulder floods of September 2013, all of which were influenced by high sea surface
temperatures that had a discernible human component.

W
eather and climate extremes happen all of the time, weather situation or weather pattern, they are really a description
even in an unchanging climate. Yet there is a justifiably of the event, not a cause. As an explanation, the question should be,
strong sense that some of these extremes are becoming ‘Why did that weather phenomenon behave the way it did?’ In par-
more frequent, and that the main reason is human-induced climate ticular, ‘What influences external to the atmosphere were playing a
change. Indeed, the main way in which climate change is likely to role, and what climate factors were in play?’
affect societies around the world is through changes in extremes. The second kind of attribution relates especially to the objective
As a result, the scientific community faces an increasing demand of assessing the role of human activities, and especially of human-
for regularly updated appraisals of evolving climate conditions and induced climate change, in the event. Perhaps this might be gen-
extreme weather. Such information would be immensely beneficial eralized to a goal of assessing the role of external influences in the
for adaptation planning. climate system on the event in question. So as well as human influ-
The large-scale atmospheric circulation determines where it is ences there may be influences from volcanic eruptions or the Sun.
dry, where it is wet, where it is hot and so on. A recent example Results depend, however, upon how questions are framed5–7.
is documented for the Pacific Northwest 1. Yet as discussed below, The conventional approach to attribution of climate events
in contrast to thermodynamic aspects of climate, forced circulation is described by Stott and colleagues8. Clearly, it is not possible to
changes in climate models can be very non-robust, and physical attribute a single climate extreme event, which by definition is
understanding of the causes of these changes is generally lacking 2. unique and which has a large element of chance in its occurrence,
Therefore, we suggest that separating out the thermodynamic from to a specific cause. Thus, the approach is to characterize the event
dynamic effects may be a fruitful way forwards and result in a differ- and ask (i) whether the likelihood or strength of such events has
ent set of questions to be addressed. In turn, these provide a better changed in the observational record, and (ii) whether this change
basis for communication of climate change to the public. is consistent with the anthropogenic influence as found in one or
more climate models, and thereby assess the ‘fraction of attributable
Complexities of extreme event attribution risk’. The conventional approach has had considerable success with
The climate community has responded to the demand for timely extremes that are strongly governed by thermodynamic aspects of
information by attempting to perform attribution of climate climate change, especially those related to temperature. As the attri-
extremes, partly through the IPCC reports3 but also through bution is unconditional  —  the null hypothesis is that there is no
annual reports offering closer to real-time assessments, the most effect of climate change  — each finding provides another independ-
recent of which, Explaining Extreme Events of 2013 From a Climate ent line of evidence that anthropogenic climate change is affecting
Perspective4, was published by the American Meteorological Society climate extremes.
in September 2014 and reported on several events from 2013. The The conventional approach is severely challenged, however, when
question posed in each case was whether the likelihood or strength it comes to climate extremes that are strongly governed by atmos-
of the event was affected by anthropogenic climate change. pheric circulation, including local aspects of precipitation. The main
More generally, there are perhaps two main kinds of attribu- reason is that changes in the atmospheric circulation related to cli-
tion performed. The first relates the particular extreme event to the mate change are fairly small compared with natural variability, as
associated weather and weather patterns, and this has been a useful has been shown especially by several studies at the National Center
and long-standing activity in climate science. We hear statements for Atmospheric Research using ‘large ensembles’9,10 for 60 years into
such as ‘the drought was caused by a blocking anticyclone’; ‘the the future. Although large changes in atmospheric circulation can be
outbreak of tornadoes was caused by a displaced and active storm readily apparent in a single climate model run, they are not robust
track and jet stream’; ‘the flooding was caused by El Niño’ and so on. and can change considerably in the next run or model. Indeed, what
But because these studies relate the event and phenomenon to the have often been interpreted as differences between models used in

National Center for Atmospheric Research (NCAR), PO Box 3000, Boulder, Colorado 80307, USA. 2Department of Meteorology, University of Reading,
1

Reading RG6 6BB, UK. e-mail: [email protected]

NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION | www.nature.com/natureclimatechange 1

© 2015 Macmillan Publishers Limited. All rights reserved

PERSPECTIVE NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE2657

Global mean sea-level trend August 1993–July 2013 include those documenting the role of anomalous sea surface
temperatures (SSTs) and especially El Niño/Southern Oscillation
(ENSO), the Pacific Decadal Oscillation (PDO)14, the North
Supertyphoon Atlantic Oscillation (NAO) and other so-called modes of variability.
Haiyan In general, it may not be possible to give concrete information about
3−11 November 2013
regional climate change unless one can understand and predict such
Cat.5 modes of variability and their statistics. ENSO may be predictable
for up to 18 months or so, some decadal predictability may exist for
decadal modes for several years15, but, in general, chaotic elements
in the weather and climate system preclude longer-term statements
because these regional patterns are not externally forced.
In most of these cases, the result of these studies will be a descrip-
tion of the large-scale patterns, the anomalous SSTs, and the relation-
ships between the atmospheric circulation, storm tracks, blocking,
–8 –4 0 4 8 12 16 temperatures and precipitation, and perhaps extremes. If consid-
mm yr−1 ered, these studies will undoubtedly conclude that greenhouse-gas
forcing or aerosols played little or no role in the circulation changes,
Figure 1 | Haiyan and sea level. Linear sea-level trends from August 1993 although claims otherwise are sometimes made.
to July 2013 are shown. The global mean is 3.3 mm yr-1, and the track of Instead, with regard to climate change, the questions to be
supertyphoon Haiyan from 3 to 11 November 2013 is indicated in green, answered could be:
with the most intense phase when it was a category 5 (cat. 5) storm
highlighted. Data from AVISO (http://go.nature.com/NPhaEK). • Given the weather pattern, how were the temperatures, precipita-
tion and associated impacts influenced by climate change?
IPCC reports may have arisen at least in part from natural variability. • Given a drought, how was the drying (evapotranspiration)
Hence, forced circulation changes are not well established, and it is enhanced by climate change, and how did that influence the
difficult to detect changes in circulation-related extremes in obser- moisture deficits and dryness of soils, and the wildfire risk? Did
vations because of small signal-to-noise ratios. Thus, the anomalous it lead to a more intense and perhaps longer-lasting drought, as
weather pattern is always the dominant influence for all short-lived is likely 16,17?
events of a week or two, and this is generally true even for years or • Given a flood, where did the moisture come from? Was it
decades. Accordingly, the conventional approach to extreme event enhanced by high ocean temperatures that might have had a cli-
attribution is rather ineffectual in cases that are strongly governed mate change component?
by the changed circulation, with generally an inconclusive outcome. • Given a heat wave, how was that influenced by drought, changes
Even when a detectable anthropogenic influence is found in a model, in precipitation (absence of evaporative cooling from dry land)
the reliability of that finding cannot carry much weight. and extra heat from global warming?
Yet from a risk perspective, it is still important to assess the prob- • Given extreme snow, where did the moisture come from? Was it
able anthropogenic impact on such events. By starting from the related to higher than normal SSTs off the coast or farther afield?
null hypothesis of no climate change, the conventional approach • Given an extreme storm, how was it influenced by anomalous
to extreme-event attribution has to re-establish an anthropogenic SSTs and ocean heat content (OHC), anomalous moisture trans-
influence for each kind of event ab initio, which makes it inherently ports into the storm, and associated rainfall and latent heating?
conservative and prone to ‘type  II errors’ (false negatives) which Was the storm surge worse because of high sea levels?
underestimate the true likelihood of the human influence5,6. Given
that we have considerable confidence in many aspects of climate In other words, given the change in atmospheric circulation that
change3, it seems rather peculiar to ignore that prior knowledge in brought about the event, how did climate change alter its impacts?
our assessment of climate events5. To summarize, therefore, at least in the present state of knowl-
As noted above, the predictability of the dynamics is problem- edge, in our view a more fruitful and robust approach to climate
atic2, and it is mainly through the thermodynamics that reliable extreme-event attribution is to regard the circulation regime or
statements can be made with confidence about the role of climate weather event as a conditional state (whose change in likelihood is
change. In other words, changes in temperatures and temperature not assessed), and ask whether the impact of the particular event was
extremes have a much more robust basis. That carries over to the affected by known changes in the climate system’s thermodynamic
atmospheric moisture amounts through the Clausius–Clapeyron state (for example sea level, sea surface temperature or atmospheric
equation in which the water-holding capacity of the atmosphere moisture content), concerning which there is a reasonably high
goes up exponentially at a rate of about 7% per degree Celsius. level of confidence. Such questions immediately lead to a physi-
Indeed, this is what is observed in actual changes in moisture in the cally based approach that is strongly linked to the event in question.
atmosphere over the oceans and where surface moisture is not lim- Although such questions differ from the conventional ‘frequentist’
ited. But over land, especially in summer, water availability is also a approach, they are perfectly reasonable from a Bayesian perspective,
vital factor 11. Because the main rainfall almost always comes from which can accommodate questions about single events18. Because
moisture convergence in the atmosphere, increased water vapour the questions are posed differently, their answers have a different
in the environment leads to more intense rains and a risk of flood- meaning and focus more on impacts. That is still useful information.
ing 12,13, even if the total amount does not change much. Moreover, it Frequently, the main influences identified external to the atmos-
is during droughts that the extra heat from increased greenhouse gas phere are the changes in SSTs. Of course, the SSTs at any time have a
concentrations accumulates, whereas more generally the presence large natural variability component, but the values are often inflated
of surface moisture adds an evaporative cooling effect. over what they would have been without climate change. The latter
accounts for roughly a 0.6 °C increase since the 1950s globally 3, and
More fruitful scientific questions because this inherently occurs on multidecadal timescales with a
One can ask questions such as: What was the role of internal natural memory through the OHC, it is always present. Small increases in
variability in setting up the pattern? Studies to explore this question atmospheric moisture associated with such an increase in SSTs are

2 NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION | www.nature.com/natureclimatechange

© 2015 Macmillan Publishers Limited. All rights reserved

NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE2657 PERSPECTIVE
30.02
0.8
29.52
Anomalies (°C)

0.4

SSTs (°C)
29.02
0.0
28.52
–0.4
28.02
–0.8

1950 1960 1970 1980 1990 2000 2010

Figure 2 | August SSTs for 12–20° N, 110–100° W, just west of Mexico. The
mean value is 28.92 °C for 1982–1999. The last value is for 2013. Water vapour imagery (6.5/6.7 µm) IR

of the order of 5% (ref. 19). We also assume that the global sea level Figure 3 | Water vapour channel imagery, GOES East 6.5 μm and GOES
rise for the past century of 19 cm (1901–2010)3 reflects the under- West 6.7 μm merged, for 18:45 GMT on 12 September 2013. The image
lying human influence while regional values vary owing to atmos- shows the extensive water vapour and associated activity both west of
pheric circulation variations. Mexico and in the Caribbean Sea and the river of moisture from south of
Baja, Mexico to eastern Colorado. Colors are used to show more intense
Some examples features. Image courtesy of WeatherTAP.com.
The following examples briefly discuss a few events whose attribu-
tion has not been addressed but that received an enormous amount Another example is supertyphoon Haiyan24, which devastated
of media attention. the Philippines in early November 2013.  It is among the largest
Consider the widely publicized event dubbed ‘snowmagge­don’, in and most intense typhoons on record with estimated surface pres-
Washington DC on 5–6 February 201020. Key features included (i) it sure in the eye down to 895  hPa and 1-minute sustained winds
was winter and there was plenty of cold continental air; (ii) there of 315  km  h-1 (gusts to 378  km  h-1) on 7 November 2013 which
was a storm in the right place; and (iii) the unusually high SSTs in makes it the strongest recorded storm ever to strike land (Fig. 1).
the tropi­cal Atlantic Ocean (1.5 °C above normal) led to an excep- It has been called a ‘category 6’ storm25. The OHC and sea level
tional amount of moisture flowing into the storm, which resulted in the region had increased a great deal since 1993 and espe-
in very large amounts of snow. It is this last part that then relates in cially since 1998 in association with the PDO negative phase14,26,
part to anomalous external influences on the atmosphere through so that from 1993 onwards the linear trend in sea level was over
the effects of climate change on SSTs. So although internal variability 16  mm  yr-1 as compared with the global trend of 3.3  mm  yr-1
is also playing a role, the extremes are magnified by climate change. (Fig. 1). Consequently, as the typhoon approached the Philippines,
Superstorm Sandy caused tremendous damage when it it was riding on very high SSTs with very deep support through
made landfall on the New Jersey coast and New York area on the high OHC, and the strong winds and ocean mixing did not
30 October 2012. It began as a hurricane, with peak strengths mak- cause as much cooling as would normally be experienced, probably
ing it a category 3 storm, and it caused substantial impacts in the helping the storm to maintain its tremendous strength25. Moreover,
Caribbean21. But as it moved north it became a hybrid storm before the storm surge was undoubtedly exacerbated considerably by the
making landfall, which greatly increased its overall size. The worst sea levels, which were some 30  cm above 1993 values. Although
problems on the Jersey shore were caused by the strong winds and natural variability through the PDO played a major role, there is
the associated storm surge, leading to extensive flooding. Farther also a global component through increased OHC from the Earth’s
inland, heavy precipitation was also a major problem. Widespread energy imbalance27.
damage from flooding streets, tunnels and subway lines, and power A conclusion, then, is that although climate change is no doubt
outages in and around New York City led to damages exceeding altering the atmospheric circulation, the change is relatively small
2013 US$65  billion22. Fortunately, the storm was well forecast a and can only be discerned from a very large ensemble of model
week in advance, in particular by the European Centre for Medium- runs. That sets the change in odds. But for any event, the particular
Range Weather Forecasts (ECMWF), including the very unusual character of that storm or synoptic situation and natural variability
left hook turn in the track towards the coast. ECMWF has per- rule, while thermodynamic effects increase the impacts.
formed a number of experiments on the performance and behav-
iour of the storm using an ensemble of forecasts 5 days in advance of Boulder floods of September 2013
landfall23. Of note is that they swapped the observed SSTs for clima- One of the recent studies in a special issue4 of the Bulletin of the
tological values that average 1–1.5 °C cooler in a broad strip along American Meteorological Society (BAMS) is of the major floods
the coast. Only small changes occurred to the track of the storm, centred in Boulder in September 2013 (ref. 28). The unfortunate
but the observed SSTs led to a bigger, more intense storm, stronger headline of the news release, which carried over in large letters to
winds and greater precipitation. The average depth of the storm was the front page of the local newspaper in Boulder (Daily Camera,
increased by 7.6 hPa, the wind speeds were increased by 3.6 m s–1 30 September 2014)29, was: “Climate change not to blame for 2013
and the precipitation increased by 35%. Moreover, the storm was Colorado floods”. The paper’s summary was: “The probability for an
riding on sea levels that were higher by about 19  cm because of extreme five-day September rainfall event over northeast Colorado,
global warming. Although perhaps only one-half to one-third of as was observed in early September 2013, has likely decreased due
the SST increase can be blamed on global warming from human to climate change.”
activities, it is readily apparent that the storm surge and associated As noted above, in any weather event like this, the weather situ-
damage was considerably influenced by climate change. It is quite ation is always the main player in the developments, but that in and
possible that the subways and tunnels might not have flooded with- of itself says nothing about the role of climate change. The study
out the warming-induced increases in sea level and in storm inten- did note the importance of having abundant moisture in the region
sity and size, putting a potential price tag of human climate change in order to produce high rainfall amounts. But it did not include
on this storm in the tens of billions of dollars. an assessment as to where the moisture came from. In Denver, the

NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION | www.nature.com/natureclimatechange 3

© 2015 Macmillan Publishers Limited. All rights reserved

PERSPECTIVE NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE2657

Ingrid

Mexico

Manuel

N
200 km
GOES 13 0545Z
11 September 2013
Figure 5 | Tropical storms Manuel and Ingrid. Imagery on 15 September
2013 from Moderate Resolution Imaging Spectroradiometer (MODIS)
on NASA’s Terra satellite. NASA image courtesy of Jeff Schmaltz,
LANCE/EOSDIS MODIS Rapid Response Team.

Experience with simulations of atmospheric rivers at NCAR has

shown that quite high resolution, around a quarter the size of that
used by Hoerling et al.28, is required to perform realistic simulations,
and coarse resolution can lead to discrepancies in the atmospheric
moisture amounts simulated. Hoerling et al.28 are probably correct
in concluding that the set-up of weather systems was rare, and cli-
mate change is not a discernible factor in that respect. But one can-
not realistically attribute an event using a model that does not have
the requisite spatial resolution to represent the relevant dynamics
and is unable to replicate the event in question, and it is desirable
to use more than one model. Moreover, the extremely high SSTs
and record water vapour amounts that accompanied the event
and were instrumental in its evolution probably would not have
GOES 13 0545Z
13 September 2013
occurred without climate change. Even a 10% increase in moisture
in the atmosphere becomes concentrated when focused by topog-
raphy, and further amplified when on the ground as water drains
Figure 4 | Imagery from the 6.5-μm water vapour channel of NOAA’s into channels and rivers.
GOES 13 satellite. Images were taken at 05:45 GMT, on 11 (top) SSTs have been as high in the past in this region west of Mexico
and 13 (bottom) September 2013. The water vapour is in the mid to (Fig.  2), but in previous cases they were part of a much larger-
upper troposphere; the brighter the imagery, the more saturated the scale pattern associated with El Niño events such as in 1997–1998,
air. Colorado is outlined in red. Images courtesy of Axel Graumann, 2004–2005 and 2009–2010. What seems to be unique in 2013 is that
NOAA/NESDIS/NCDC. this was the warmest spot in the western hemisphere and hence
this was the preferred location for low pressure to form and for
three highest amounts of total column water vapour ever recorded low-level wind convergence, which brought large amounts of mois-
for September (since 1956) occurred on 12–13 September 2013 (as ture into the region. Hence there was clearly an internal variability
high as 34 mm). This may not seem huge, but recall that Denver is a component to the patterns of SST, but at the same time, the overall
mile (>1,600 m) above sea level. increase of global SST associated with global warming that occurs
It so happens that the SSTs off the west coast of Mexico, south on multi-decadal timescales was also a factor. In Fig. 2 this warm-
of Baja, west of Guadalajara, were over 30 °C and more than 1 °C ing is especially evident after 1997, when SSTs average >0.4  °C
above normal in August 2013 (see Fig. 2), which made it the hot- above the mean.
test spot for the ocean in the western hemisphere. An incredible
75  mm of total column water vapour was recorded in the atmos- Concluding remarks
phere in that region by NASA satellites. The high SSTs led to the We have suggested four climate events in which anomalously high
large-scale convergence of moisture flowing into the region that was SSTs played a key role in feeding moisture into storms, helping to
siphoned north by a very unusual synoptic situation leading to a intensify the storm and causing heavy rains. Of course, the SSTs
river of atmospheric moisture flowing into Colorado (Figs  3  and at any time have a large natural variability component on top of
4). The heaviest moisture convergence into eastern Colorado coin- the human-induced warming. Associated increases in atmospheric
cided with the times of the plots in Fig. 4, while breaks in the rain moisture are of order 5% (ref. 19), which is magnified as moisture
occurred in between times in some locations (Fig.  3). After that converges into a storm, further magnified as the storm intensi-
river shut off, twin tropical storms formed both sides of Mexico: fies, and concentrated when it runs up against orography and in
Manuel (to the west)30 and Ingrid (to the east)31 (Fig. 5), forming a streamflows. Hence, even though natural variability always domi-
double whammy for Mexico and leading to hundreds of deaths, tens nates in such storms, the research task is to refine the above esti-
of thousands evacuated, tens of thousands of homes damaged and mates and properly account for the human component as at least a
billions of dollars of damage30,31. partial explanation for why extremes are being magnified.

4 NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION | www.nature.com/natureclimatechange

© 2015 Macmillan Publishers Limited. All rights reserved

NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE2657 PERSPECTIVE
There are many other possible examples. For instance, conflicting thermodynamic environment on the extremes and temperatures
results32,33 were found for the Russian heat wave in 2010 that were to and moisture associated with the event.
some extent reconciled34 by recognizing that each study was about Received 10 October 2014; accepted 14 April 2015;
different aspects. One study 32 focused on dynamical aspects whereas published online 22 June 2015
the other 33 was much more about the record high temperatures and
thermodynamic aspects. Indeed, the record high SSTs and associ- References
ated nearby rains in several areas (including the Pakistan floods) 1. Johnstone, J. A. & Mantua, N. J. Atmospheric controls on northeast Pacific
were factors external to the atmosphere35 that altered teleconnec- temperature variability and change, 1900–2012. Proc. Natl Acad. Sci. USA
111, 14360–14365 (2014).
tions into the Russian region. 2. Shepherd, T. G. Atmospheric circulation as a source of uncertainty in climate
Another very recent example is the California drought begin- change projections. Nature Geosci. 7, 703–708 (2014).
ning in 2012. Whereas one study 36 found no significant trends in 3. IPCC Climate Change 2013: The Physical Science Basis (eds Stocker, T. F. et al.)
winter precipitation in recent decades, another 37 pointed out the (Cambridge Univ. Press, 2013).
critical role of the record high annual mean temperatures in com- 4. Herring, S. C., Hoerling, M. P., Peterson T. C. & Stott, P. A. (eds) Explaining
bination with record low annual mean precipitation for 2013 which extreme events of 2013 from a climate perspective. Bull. Am. Meteorol. Soc.
95, S1–S96 (2014).
led to increased evapotranspiration and more intense drought. The 5. Trenberth, K. E. Attribution of climate variations and trends to human
combination of these had impacts on water shortages, vegetation influences and natural variability. WIREs Clim. Change 2, 925–930 (2011).
and agriculture, and increased wildfire risk. The odds of this com- 6. Trenberth, K. E. Framing the way to relate climate extremes to climate change.
bination have increased with human-induced climate change and Climatic Change 115, 283–290 (2012).
anthropogenic warming has increased drought risk37. Again these 7. Wallace, J. M. Weather- and climate-related extreme events:
two studies are consistent with the view that the atmospheric cir- Teachable moments. Eos 93, 120–121 (2012).
8. Stott, P. et al. in Climate Science Serving Society (eds Asrar, G. R. &
culation changes are not the dominant factor as far as the climate Hurrell, J. W.) 307–337 (Springer, 2013).
change aspects are concerned. 9. Deser, C., Phillips, A., Bourdette, V. & Teng, H. Y. Uncertainty in climate
In addition there were 22 articles in the special BAMS issue4. change projections: The role of internal variability. Clim. Dynam.
Over two-thirds focused on aspects of atmospheric circulation and 38, 527–546 (2012).
generally concluded there was little or no influence of anthropo- 10. Deser, C., Phillips, A. S., Alexander, M. A. & Smoliak, B. V. Projecting North
genic climate change. None of the eight studies that dealt with heavy American climate over the next 50 years: Uncertainty due to internal variability.
J. Clim. 27, 2271–2296 (2014).
rains or snows performed an analysis of the moisture budget and 11. Trenberth, K. E. Changes in precipitation with climate change. Clim. Res.
where the abundant water came from. Of the three studies that ana- 47, 123–138 (2011).
lysed the California drought, none dealt with land surface effects 12. Trenberth, K. E. Atmospheric moisture residence times and cycling:
or changes in evapotranspiration associated with climate change. Implications for rainfall rates with climate change. Climatic Change
In contrast, for the Australian heat wave, the role of decreased soil 39, 667–694 (1998).
moisture and drought were noted for their effects on temperatures. 13. Trenberth, K. E., Dai, A., Rasmussen, R. M. & Parsons, D. B. The changing
character of precipitation. Bull. Am. Meteorol. Soc. 84, 1205-1217 (2003).
The summary table in the special issue focused on columns about 14. Trenberth, K. E., Fasullo, J. T., Branstator, G. & Phillips, A. S. Seasonal
anthropogenic influences versus no such influence detected. aspects of the recent pause in surface warming. Nature Clim. Change
There are many other examples of studies that might ask more 4, 911–916 (2014).
targeted questions that would better serve societal needs. Reframing 15. Branstator, G. & Teng, H. Is AMOC more predictable than North Atlantic heat
the null hypothesis and asking different questions would help to content? J. Clim. 27, 3537–3550 (2014).
advance our understanding. Because global warming is real and 16. Seneviratne, S. I. et al. Investigating soil moisture–climate interactions in a
changing climate: A review. Earth Sci. Rev. 99, 125–161 (2010).
present, it is not a question as to whether it is playing a role but 17. Trenberth, K. E. et al. Global warming and changes in drought.
what that role is. A Bayesian approach takes these ‘priors’ appro- Nature Clim. Change 4, 17–22 (2014).
priately into account. We can talk about these effects in terms of 18. Lyons, L., Discovery or fluke: Statistics in particle physics. Phys. Today
changing odds, as many have done. But we can also talk about them 65, 45–51 (2012).
in physical terms. 19. Trenberth, K. E., Fasullo, J. & Smith, L. Trends and variability in column-
To implement the approach suggested here, one needs to be able integrated atmospheric water vapor. Clim. Dynam. 24, 741–758 (2005).
20. February 2010: Snowmageddon, Blizzard of 2010 WINTER: Unprecedented
to simulate the event in question (perhaps with short-term fore- Snowfall Impacts Region (PRESTO, 2010); http://go.nature.com/IBxDJD.
casts, as in the example of superstorm Sandy, or with simulations 21. Blake, E. S., Kimberlain, T. B., Berg, R. J., Cangialosi, J. P. & Beven, J. L. II
whose circulation is constrained in some manner), and then assess Tropical Cyclone Report: Hurricane Sandy Report no. AL182012 (National
the impact of known anthropogenic changes in the climate system’s Hurricane Center, 2013); http://go.nature.com/sjCrrH
thermodynamic state, including changed amounts of water vapour. 22. Hurricane/Post-Tropical Cyclone Sandy, October 22–29, 2012 Service
Only in this way can the necessarily conditional nature of the event Assessment (NOAA, 2013); http://go.nature.com/BqlTxe
23. Magnusson, L. et al. Evaluation of medium-range forecasts for hurricane Sandy.
attribution be implemented. Necessarily this addresses only a subset Mon. Weather Rev. 142, 1962–1981 (2014).
of the possible influences and conditions. But with the underlying 24. Evans, A. D. & Falvey, R. J. Annual Tropical Cyclone Report 2013 (Naval
assumption that the phenomena are largely unchanged, this is per- Maritime Forecast Center/ Joint Typhoon Warning Center, 2013);
haps as much as can be expected for event attribution. After all, a http://go.nature.com/V9JpKu.
single extreme event is not a repeatable thing. 25. Lin, I-I., Pun, I-F. & Lien, C-C. ‘Category-6’ Supertyphoon Haiyan in
The climate is changing: we have a new normal. The environ- global warming hiatus: Contribution from subsurface ocean warming.
Geophys. Res. Lett. 41, 8547–8553 (2014).
ment in which all weather events occur is not what it used to be. 26. Trenberth, K. E. & Fasullo, J. T. An apparent hiatus in global warming?
All storms, without exception, are different. Even if most of them Earth’s Future 1, 19–32 (2013).
look just like the ones we used to have, they are not the same. 27. Trenberth, K. E., Fasullo, J. T. & Balmaseda, M. A. Earth’s energy imbalance.
But we cannot sort out these questions of degree without a large J. Clim. 27, 3129–3144 (2014).
ensemble of model simulations, particularly for events as rare as 28. Hoerling, M. et al. Northeast Colorado extreme rains interpreted in a climate
the Boulder floods, and the kinds of models that can be run in change context. Bull. Am. Meteorol. Soc. 95 (Special issue), S15–S18 (2014).
29. Brennan, C. Boulder researcher: 2013’s flood-triggering rains not
such a way are often incapable of simulating the event in question caused by climate change. Daily Camera (29 September 2014);
and thus lack physical credibility. We argue that under such condi- http://go.nature.com/2y9zuO.
tions it is better for event attribution to focus not on the synop- 30. Pasch, R. J. & Zelinsky, D. A. Hurricane Manuel EP132013 (National Hurricane
tic event, but rather on the influences of the changed large-scale Center, 2014); http://go.nature.com/8ORDI9.

NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION | www.nature.com/natureclimatechange 5

© 2015 Macmillan Publishers Limited. All rights reserved

PERSPECTIVE NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE2657
31. Beven, J. L. II Hurricane Ingrid AL102013 (National Hurricane Center, 2014); Acknowledgements
http://go.nature.com/QzAQuL. The National Center for Atmospheric Research is sponsored by the National Science
32. Dole, R. et al. Was there a basis for anticipating the 2010 Russian heat wave? Foundation. This work is supported by DOE award DE-SC0012711. T.G.S. acknowledges
Geophys. Res. Lett. 38, L06702 (2011). the support provided through the Grantham Chair in Climate Science at the University
33. Rahmstorf, S. & Coumou, D. Increase of extreme events in a warming world, of Reading.
Proc. Natl Acad. Sci. USA 108, 17905–17909 (2011).
34. Otto, F. E. L., Massey, N., van Oldenborgh, G. J., Jones, R. G. & Allen, M. R. Author contributions
Reconciling two approaches to attribution of the 2010 Russian heat wave. K.E.T. led the writing of the paper and conceived of the paper and figures. J.T.F
Geophys. Res. Lett. 39, L04702 (2012). analysed some data and contributed to two figures. All authors contributed to writing
35. Trenberth, K. E. & Fasullo, J. T. Climate extremes and climate change: The the manuscript.
Russian heat wave and other climate extremes of 2010. J. Geophys. Res.
117, D17103 (2012). Additional information
36. Seager, R. et al. Causes and Predictability of the 2011–14 California Drought Reprints and permissions information is available online at www.nature.com/reprints.
(NOAA, 2014); http://go.nature.com/IBpoCA. Correspondence should be addressed to K.E.T.
37. Diffenbaugh, N. S., Swain D. L. & Touma, D. Anthropogenic warming
has increased drought risk in California. Proc. Natl Acad. Sci USA Competing financial interests
112, 3931–3936 (2015). The authors declare no competing financial interests.

6 NATURE CLIMATE CHANGE | ADVANCE ONLINE PUBLICATION | www.nature.com/natureclimatechange

© 2015 Macmillan Publishers Limited. All rights reserved