Climate Extremes Indices in The CMIP5 Multimodel Ensemble Part 2.future Climate Projections
Climate Extremes Indices in The CMIP5 Multimodel Ensemble Part 2.future Climate Projections
Climate Extremes Indices in The CMIP5 Multimodel Ensemble Part 2.future Climate Projections
50188, 2013
[1] This study provides an overview of projected changes in climate extremes indices
defined by the Expert Team on Climate Change Detection and Indices (ETCCDI). The
temperature- and precipitation-based indices are computed with a consistent methodology
for climate change simulations using different emission scenarios in the Coupled Model
Intercomparison Project Phase 3 (CMIP3) and Phase 5 (CMIP5) multimodel ensembles.
We analyze changes in the indices on global and regional scales over the 21st century
relative to the reference period 1981–2000. In general, changes in indices based on daily
minimum temperatures are found to be more pronounced than in indices based on daily
maximum temperatures. Extreme precipitation generally increases faster than total wet-day
precipitation. In regions, such as Australia, Central America, South Africa, and the
Mediterranean, increases in consecutive dry days coincide with decreases in heavy
precipitation days and maximum consecutive 5 day precipitation, which indicates future
intensification of dry conditions. Particularly for the precipitation-based indices, there can
be a wide disagreement about the sign of change between the models in some regions.
Changes in temperature and precipitation indices are most pronounced under RCP8.5, with
projected changes exceeding those discussed in previous studies based on SRES scenarios.
The complete set of indices is made available via the ETCCDI indices archive to encourage
further studies on the various aspects of changes in extremes.
Citation: Sillmann, J., V. V. Kharin, F. W. Zwiers, X. Zhang, and D. Bronaugh (2013), Climate extremes indices in the CMIP5
multimodel ensemble: Part 2. Future climate projections, J. Geophys. Res. Atmos., 118, 2473–2493, doi:10.1002/jgrd.50188.
2473
SILLMANN ET AL.: CMIP5 PROJECTIONS OF EXTREMES INDICES
[Taylor et al., 2012] and utilizes a new set of emission scenarios writing, we were able to analyze 19 CMIP5 models (cf.
referred to as Representative Concentration Pathways (RCPs) Table 1), for which daily model output for three RCP
[Moss et al., 2010; van Vuuren et al., 2011b]. The future scenarios (RCP2.6, 4.5 and 8.5) was available. However,
climate simulations of the CMIP3 ensemble were based on the indices for all CMIP5 models on the ESG are being
the emission scenarios described in the IPCC Special Report made available on the EIA (http://www.cccma.ec.gc.ca/
on Emission Scenarios (SRES) [Nakicenovic et al., 2000]. data/climdex/climdex.shtml) for further analysis. We also
[5] Beside the different emission scenarios and models, analyze 11 CMIP3 models (see Table 2), for which the three
differences in the index calculations themselves can also lead SRES emission scenarios considered here (B1, A1B, and
to inconsistencies in the analysis and comparison of changes A2) are available in the PCMDI archive. Note that the
in extremes simulated in the CMIP3 and CMIP5 ensembles. models used in this study differ from the models in Tebaldi
For instance, Tebaldi et al. [2006] used indices that were pro- et al. [2006], who use fewer models, and Orlowsky and
vided as part of CMIP3 by individual model groups using their Seneviratne [2012], who use more models but only for the
own implementations of index calculations, which can lead to SRES A2 scenario.
inconsistencies. Orlowsky and Seneviratne [2012] used a [10] Figure 1 illustrates the evolution of carbon dioxide
larger suite of CMIP3 simulations to calculate a set of indices, (CO2) concentrations as observed in the 20th century
which deviated slightly from the ETCCDI definitions. and prescribed in the 21st century simulations in the
[6] The purpose of this study is thus to document changes SRES and RCP scenarios considered in this study. The
in indices that are calculated in a consistent manner as SRES scenarios are based on storylines assuming different
simulated in the CMIP3 and CMIP5 multimodel ensembles socioeconomic, technological, and political developments
for different emission scenarios. As shown in Rogelj et al. leading to specified changes in emissions that in turn
[2012], the radiative forcing prescribed in the SRES and determine the resulting changes in atmospheric greenhouse
RCP scenarios can lead to different average temperature gas concentrations (e.g., Figure 1) and radiative forcing. At
responses, and we expect that this will also be evident in the end of the 21st century, the CO2 concentrations reach
seasonal and annual temperature and precipitation extremes. about 840 ppm in the SRES A2 scenario, 700 ppm in the
[7] As an essential part of this study, an ETCCDI indices A1B scenario, and 540 ppm in the B1 scenario which
archive (EIA) of indices for the CMIP3 and CMIP5 assumes the most environmentally friendly development
ensembles has been created and is available at http://www. pathway.
cccma.ec.gc.ca/data/climdex/climdex.shtml. The EIA is [11] In contrast to the SRES scenarios, the radiative
described in detail in part 1 of this study [Sillmann et al., forcing trajectories in the RCPs are not associated with
2013], which focuses on the evaluation of the indices in predefined storylines and can reflect various possible
CMIP3 and CMIP5 under present climate conditions. With combinations of economic, technological, demographic,
the exception of a few indices, we show in Sillmann et al. and policy developments [Moss et al., 2010]. The peak-
[2013] that CMIP5 models are generally able to simulate cli- and-decline RCP2.6 scenario is designed to meet the 2 C
mate extremes and their trend patterns as represented by the in- global average warming target compared to pre-industrial
dices in comparison to a gridded observational indices data conditions [van Vuuren et al., 2011a]. It has a peak in the
set. The challenges involved in such an evaluation, given radiative forcing at approximately 3 W/m2 (~400 ppm CO2)
the temporal and spatial resolution of GCMs as well as the before 2100 and then declines to 2.6 W/m2 by the end of the
availability of suitable observational data sets, are also 21st century (~330 ppm CO2, Figure 1). Radiative forcing in
discussed. Part 2 of this study, which is presented here, RCP4.5 peaks at about 4.5 W/m2 (~540 ppm CO2) in year
focuses on the projected changes in the indices based on 2100 [Thomson et al., 2011]. RCP4.5 is comparable to the
CMIP3 and CMIP5 future climate simulations. SRES scenario B1 with similar CO2 concentrations and
[8] The paper is organized as follows. We briefly describe the median temperature increases by 2100 according to Rogelj
multimodel ensembles and featured scenarios in section 2. In et al. [2012]. RCP8.5 assumes a high rate of radiative forcing
section 3, we provide the definitions of indices discussed in increase, peaking at 8.5 W/m2 (~940 ppm CO2) in year 2100
this paper. Our results are presented in section 4 for several [Riahi et al., 2011].
different categories of temperature- and precipitation-based [12] Climate changes simulated in the CMIP3 and CMIP5
indices. A summary of the main findings and concluding ensembles are not directly comparable because of the
remarks are given in section 5. differences in prescribed forcing agents (e.g., CO2 and aerosols)
between the SRES and RCP scenarios as discussed in Rogelj
et al. [2012]. Furthermore, the models may respond
2. Climate Models and Scenarios differently to a specific radiative forcing due to different
[9] We analyze climate simulations of the 20th and 21st model-specific climate sensitivities. However, based on the
century performed by models participating in CMIP3 [Meehl underlying radiative forcing (or CO2 concentrations), one
et al., 2007b] and CMIP5 [Taylor et al., 2012]. While more can compare projected changes in the temperature and
than one realization is available for some models, we precipitation indices and provide an estimate of uncertainty
analyze here only the first ensemble member of each related to the different emission scenarios.
model simulation as a first-order assessment. The CMIP3
and CMIP5 model output is available from the data
archives of the Program for Climate Model Diagnosis 3. Global Climate Extremes Indices
and Intercomparison (PCMDI, http://www-pcmdi.llnl.gov) [13] The indices are based on daily minimum and maximum
and the Earth System Grid data distribution portal of near surface temperature and daily precipitation amounts
(ESG, http://www.earthsystemgrid.org). At the time of (TN, TX, and PR, respectively). Detailed information on the
2474
SILLMANN ET AL.: CMIP5 PROJECTIONS OF EXTREMES INDICES
Table 1. The CMIP5 Models for Which RCP2.6, RCP4.5 and RCP8.5 Simulations Where Available on the ESG (as of September 2012)
for the Time Period 2006–2100a
Model Institution Spatial Resolution (Lon Lat ~ Levels)
1 BCC-CSM1-1 Beijing Climate Center, China Meteorological 128 64L26(T42)
Administration, China
2 BNU-ESM Beijing Normal University, China 128 64L26(T42)
3 CanESM2 Canadian Centre for Climate Modelling and 128 64L35(T63)
Analysis, Canada
4 CCSM4 National Center for Atmospheric Research 288 192L26
(NCAR), USA
5 CNRM-CM5 Centre National de Recherches Meteorologiques, 256 128L31(T127)
Meteo-France, France
6 CSIRO-Mk3-6-0 Australian Commonwealth Scientific and Industrial 192 96L18(T63)
Research Organization, Australia
7 FGOALS-s2 Institute of Atmospheric Physics, Chinese Academy 128 108L26
of Sciences, China
8 GFDL-ESM2G Geophysical Fluid Dynamics Laboratory, USA 144 90L24
9 GFDL-ESM2M Geophysical Fluid Dynamics Laboratory, USA 144 90L24
10 HadGEM2-ES Met Office Hadley Centre, UK 192 145L40
11 IPSL-CM5A-LR Institut Pierre-Simon Laplace, France 96 96L39
12 IPSL-CM5A-MR Institut Pierre-Simon Laplace, France 144 143L39
13 MIROC5 AORI (Atmosphere and Ocean Research Institute), 256 128L40(T85)
NIES (National Institute for Environmental Studies),
JAMSTEC (Japan Agency for Marine-Earth Science
and Technology), Japan
14 MIROC-ESM AORI, NIES, JAMSTEC, Japan 128 64L80(T42)
15 MIROC-ESM-CHEM AORI, NIES, JAMSTEC, Japan 128 64L80(T42)
16 MPI-ESM-LR Max Planck Institute for Meteorology, Germany 192 96L47(T63)
17 MPI-ESM-MR Max Planck Institute for Meteorology, Germany 192 96L95(T63)
18 MRI-CGCM3 Meteorological Research Institute, Japan 320 160L48(T159)
19 NorESM1-M Norwegian Climate Centre, Norway 144 96L26
a
Note that all three RCP simulations of BCC-CSM1-1 as well as the RCP8.5 simulation on HadGEM2-ES were only available until year 2099. The
analysis reported in this paper is based on the first ensemble member of each RCP simulation.
Table 2. The CMIP3 Models for Which All SRES Scenarios (i.e., B1, A1B, and A2) Where Available on the PCMDI Archive
Model Institution Spatial resolution (Lon Lat ~ Levels)
1 cccma-cgcm3 t47 Canadian Centre for Climate Modelling and Analysis, Canada 96 48L32 (T47)
2 cccma-cgcm3 t63 Canadian Centre for Climate Modelling and Analysis, Canada 128 64L32 (T63)
3 cnrm-cm3 Centre National de Recherches Meteorologiques, Meteo-France, 128 64L45 (T63)
France
4 csiro-mk3-0 Australian Commonwealth Scientific and Industrial Research 192 96L18 (T63)
Organization, Australia
5 csiro-mk3-5 Australian Commonwealth Scientific and Industrial Research 192 96L18 (T63)
Organization, Australia
6 gfdl-cm2-0 NOAA/Geophysical Fluid Dynamics Laboratory, USA 144 90L24
7 gfdl-cm2-1 NOAA/Geophysical Fluid Dynamics Laboratory, USA 144 90L24
8 giss-model-e-r NASA/Goddard Institute for Space Studies, USA 72 46L20
9 ipsl-cm4 Institut Pierre-Simon Laplace, France 96 72L19
10 miroc3-2 medres CCSR/NIES/FRCGC, Japan 128 64L20 (T42)
11 mpi-echam5 Max Planck Institute for Meteorology, Germany 192 96L31 (T63)
indices can be found in Alexander et al. [2006], Klein Tank et EIA. The selected indices, as described below, give a com-
al. [2009], and Zhang et al. [2011] and on the ETCCDI prehensive overview of the projected changes in tempera-
website. For the multimodel analysis presented in this paper, ture and precipitation extremes across models and scenar-
all indices are first computed on their native model grids ios. While most indices are defined on an annual basis, a
and then re-gridded to a common 2.5 2.5 grid. While few are also available as monthly statistics or as continuous
daily TN, TX, and PR are only available for two 20 year counts over the total data record and will be specified as
time periods 2046–2065 and 2081–2100 in CMIP3, they such in the description below.
are available for the entire time period from 2006 to 2100 3.1.1. Temperature Indices
for CMIP5. [15] • Absolute Indices
The minimum of TN (TNn) and maximum of TX (TXx)
3.1. Selection of Indices represent the coldest or hottest day of a year, season, or
[14] In this study, we consider a subset of the 27 indi- month, respectively. The annual temperature extremes
ces (see Table 1 in Sillmann et al. [2013]) available in the are used in several studies to express the extreme
2475
SILLMANN ET AL.: CMIP5 PROJECTIONS OF EXTREMES INDICES
2476
SILLMANN ET AL.: CMIP5 PROJECTIONS OF EXTREMES INDICES
[22] Two indices that do not fall in one of the categories which tests whether the multimodel median change is zero.
outlined above are the total wet-day precipitation Changes that are not significant at the 5% significance level
(PRCPTOT) and the simple daily intensity (SDII) indices. are indicated by stippling in the maps of projected changes.
PRCPTOT describes the total annual amount of precipitation We do not consider inter-model agreement in the significance
on wet days defined as days with more than 1 mm of precip- testing and stippling as in IPCC [2007], Tebaldi et al. [2006],
itation. SDII describes the daily precipitation amount aver- and Orlowsky and Seneviratne [2012] and do not use any
aged over all wet days in a year. PRCPTOT and SDII are of the stippling conventions currently under discussion
not necessarily associated with climate extremes but provide [e.g., Tebaldi et al., 2011; Power et al., 2012]. Instead, model
useful information about the relationship between changes agreement or disagreement is illustrated in terms of the
in extreme conditions (e.g., RX5day or R95p) and other interquartile model spread for the 21 subregions for a number
aspects of the distribution of daily precipitation. of indices.
Figure 2. Subregions over land adapted from Giorgi and Francisco [2000, cf. their Table 2] and color-
coded according to continents. Blue, Australia; green, South America; purple, North America; red, Africa;
yellow, Europe; cyan, Asia.
2477
SILLMANN ET AL.: CMIP5 PROJECTIONS OF EXTREMES INDICES
[a] [b]
Minimum of TN (TNn) Maximum of TX (TXx)
8 historical RCP4.5 8 8 historical RCP4.5 8
RCP2.6 RCP8.5 RCP2.6 RCP8.5
6 6 6 6
deg C
deg C
4 4 4 4
2 2 2 2
0 0 0 0
[c] [d]
Frost Days (FD) Tropical Nights (TR)
5 5
historical RCP4.5 historical RCP4.5
RCP2.6 RCP8.5 RCP2.6 RCP8.5
0 0 60 60
−5 −5
40 40
days
days
−10 −10
−15 −15 20 20
−20 −20
0 0
−25 −25
CMIP3 B1 CMIP3 A1B CMIP3 A2 CMIP3 B1 CMIP3 A1B CMIP3 A2
1960 1980 2000 2020 2040 2060 2080 2100 1960 1980 2000 2020 2040 2060 2080 2100
Year Year
Figure 3. Global averages of temperature indices over land as simulated by the CMIP5 ensemble (see
Table 1) for the RCP2.6 (blue), RCP4.5 (green), and RCP8.5 (red) displayed as anomalies from the refer-
ence period 1981–2000. Solid lines indicate the ensemble median and the shading indicates the
interquartile ensemble spread (25th and 75th quantiles). Time series are smoothed with a 20 year running
mean filter. The box-and-whisker plots show the interquartile ensemble spread (box) and outliers
(whiskers) for 11 CMIP3 model simulations of the SRES scenarios A2 (orange), A1B (cyan), and B1
(purple) (see Table 2) globally averaged over the respective future time periods (2046–2065 and 2081–2100)
as anomalies from the CMIP3 reference period 1981–2000.
simulated in CMIP5 for RCP4.5 and RCP8.5. TNn and TXx A modest median increase of 28 days is projected in RCP4.5
increase, respectively, by 4.6 C and 3.5 C in A1B, and by and B1. The smallest median increase in TR (16 days) is pro-
5.6 C and 4.5 C in A2. jected in the RCP2.6 scenario. The patterns in the temporal
[27] A consistent pattern is seen in the evolution of the evolution of threshold indices based on daily maximum
threshold indices based on TN, frost days (FD), and tropical temperature such as ice days (TX < 0 C) or summer days
nights (TR) globally averaged over land (Figures 3c and 3d). (TX > 25 C) (not shown) are similar to those in FD and
In the middle of the 21st century, FD decreases by about TR, respectively, but are less pronounced.
8 days in B1, which lies in the interquartile model range of 4.1.1.2. Spatial and Seasonal Patterns
RCP2.6. FD decreases by 11 and 10 days in A1B and A2, [29] The projected median changes of TNn and TXx simu-
respectively, which is within the interquartile model spread lated in the CMIP5 ensemble are shown in Figure 4 and are sig-
for RCP4.5. By the end of the 21st century, the median nificant across land areas for all three RCPs by the end of the
decreases of FD are 7 days in RCP2.6 and 13 days in 21st century. The spatial patterns of change in TNn and TXx
RCP4.5, with B1 (10 days) being centered between them. are different. In particular, TNn increases more strongly in
A stronger decrease in FD of 16 and 20 days is seen in higher latitudes of the Northern Hemisphere. For RCP2.6,
A1B and A2, respectively, which is still smaller than the TXx changes only moderately over land while stronger
median decrease of 23 days projected in RCP8.5. increases are apparent in TNn, particularly in northern latitudes.
[28] Tropical nights increase by about 18 days in RCP2.6 [30] The greatest changes in TNn, exceeding 12 C, are
and 20 days in B1 in the middle of the 21st century. The simulated in RCP8.5 in such regions as North America,
A1B and A2 median increase of 27 days in TR is within Northern Europe, and North Asia. Presumably, larger
the range of the projected RCP4.5 changes. At the end of changes in TNn in higher latitudes are related to the retreating
the 21st century, the median TR increases most in RCP8.5 snow cover under global warming whereas in the tropics and
(53 days) closely followed by A2 (51 days) and A1B (40 days). the Southern Hemisphere the TNn increases generally remain
2478
SILLMANN ET AL.: CMIP5 PROJECTIONS OF EXTREMES INDICES
Figure 4. The multimodel median of temporally averaged changes in the minimum of TN (TNn, left)
and the maximum of TX (TXx, right) over the time period 2081–2100 displayed as differences (in C)
relative to the reference period (1981–2000) for RCP2.6 (top), RCP4.5 (middle), and RCP8.5 (bottom).
All changes are significant at the 5% significance level.
below 7 C and are comparable with those in TXx. The [32] In contrast to TNn, TXx is projected to warm more
strongest warming in TXx generally occurs in the interior uniformly over the land (Figure 5b). The maximum CMIP5
of the continents, such as in South and North America, median increase of about 6.5 C in RCP8.5 occurs in WNA,
Eastern Europe, north-central Eurasia as well as Australia. CNA, ENA, MED, NEU, CAS, and NAS. The increases in
[31] Regional summaries of the CMIP5 projected changes TXx are generally more pronounced in JJA than DJF in all
in 2081–2100 for the 21 subregions (cf. Figure 2) are RCPs across the northern latitude regions (Figures 5d and
depicted in Figure 5. The strongest median warming in 5f). The Mediterranean (MED) in particular is a region
TNn occurs in Alaska (ALA, 13 C) followed by that in where the TXx warming of about 7 C in JJA for RCP8.5
Greenland (GRL) and Northern Europe (NEU) of about is among the greatest across all subregions and exceeds the
11 C for RCP8.5. Strong warming in TNn can also be seen TNn warming of 6 C. In contrast, TNn and TXx increase
in West, Central, and East North America (WNA, CNA, and only by about 4.5 C in DJF. That is, Mediterranean summer
ENA, respectively), as well as North and Central Asia (NAS extreme temperatures warm more than winter extreme
and CAS, respectively) and Tibet (TIB). In these regions, the temperatures, whereas the opposite is generally true for the
increase in TNn is generally more pronounced in DJF other subregions. The pronounced summer warming in
(Figure 5e), whereas changes in JJA (Figure 5c) are more MED may be related to soil moisture feedbacks [e.g., Hirschi
homogeneous across the regions. et al., 2011] and precipitation deficits [Mueller and
2479
SILLMANN ET AL.: CMIP5 PROJECTIONS OF EXTREMES INDICES
15 15 15 15
deg C
deg C
10 10 10 10
5 5 5 5
0 0 0 0
TIB
AUS
AMZ
SSA
CAM
WNA
CNA
ENA
ALA
GRL
MED
NEU
WAF
EAF
SAF
SAH
SEA
EAS
SAS
CAS
NAS
AUS
AMZ
SSA
CAM
WNA
CNA
ENA
ALA
GRL
MED
NEU
WAF
EAF
SAF
SAH
SEA
EAS
SAS
CAS
TIB
NAS
[c] Minimum of TN (TNn) in 2081−2100 [d] Maximum of TX (TXx) in 2081−2100
JJA JJA
20 RCP2.6 RCP4.5 RCP8.5 20 20 RCP2.6 RCP4.5 RCP8.5 20
15 15 15 15
deg C
deg C
10 10 10 10
5 5 5 5
0 0 0 AUS 0
AMZ
SSA
CAM
WNA
CNA
ENA
ALA
GRL
MED
NEU
WAF
EAF
SAF
SAH
SEA
EAS
SAS
CAS
TIB
NAS
AUS
AMZ
SSA
CAM
WNA
CNA
ENA
ALA
GRL
MED
NEU
WAF
EAF
SAF
SAH
SEA
EAS
SAS
CAS
TIB
NAS
15 15 15 15
deg C
deg C
10 10 10 10
5 5 5 5
0 0 0 0
AUS
AMZ
SSA
CAM
WNA
CNA
ENA
ALA
GRL
MED
NEU
WAF
EAF
SAF
SAH
SEA
EAS
SAS
CAS
TIB
NAS
AUS
AMZ
SSA
CAM
WNA
CNA
ENA
ALA
GRL
MED
NEU
WAF
EAF
SAF
SAH
SEA
EAS
SAS
CAS
TIB
NAS
Figure 5. Projected changes (in C) in annual (ANN), JJA and DJF minima of TN (TNn, left) and
maximum TX (TXx, right) over the time period 2081–2100 as differences relative to the reference period
(1981–2000) for RCP2.6 (blue), RCP4.5 (green), and RCP8.5 (red). Regional mean changes are shown for
each of the 21 subregions (cf. Figure 2). Boxes indicate the interquartile model spread (25th and 75th
quantiles) with the horizontal line indicating the ensemble median and the whiskers showing the extreme
range of the CMIP5 ensemble.
Seneviratne, 2012], which can play an important role in in winter by more than 2 C, even in the lowest forcing
amplifying heat conditions. scenario RCP2.6, in northern regions such as WNA, CNA,
[33] A similar pattern of seasonal and regional changes in ENA, ALA, GRL, NEU, and NAS.
TNn and TXx is seen for RCP4.5 and RCP2.6, albeit less [34] Frost days particularly decrease in western North
pronounced as compared to RCP8.5. Also, the interquartile America, along the Andes and at the southern tip of South
model spread is generally smaller in RCP4.5 and RCP2.6 America as well as in central and northern parts of Europe
compared to RCP8.5. Annual and seasonal median increases and Asia (Figure 6). The decrease is strongest in RCP8.5
in TXx generally do not exceed 2 C in RCP2.6 by the end of with reductions of 80 frost days and more in northern Europe
the century relative to 1981–2000. However, TNn increases and western North America by the end of the 21st century.
2480
SILLMANN ET AL.: CMIP5 PROJECTIONS OF EXTREMES INDICES
Figure 6. Same as Figure 4, but for frost days (FD, left) and tropical nights (TR, right). Stippling
indicates grid points with changes that are not significant at the 5% significance level.
Coastal regions of Antarctica are also projected to experience 4.1.2. Duration Indices
a significant decrease in FD under all three RCPs. [36] Consistent with temperature changes described above,
[35] Tropical nights, based on the fixed 20 C threshold, cold spell duration (CSDI) is projected to decrease and warm
increase most in tropical regions (>100 days), such as those spell duration (WSDI) is projected to increase in all RCPs
south of the Amazon, equatorial and southern Africa, and (Figure 7). The CMIP5 multimodel median changes in these
northern Australia (Figure 6). Changes in TR are also indices are significant everywhere over land. The strongest
relevant for the extra-tropical Northern Hemisphere where increases in WSDI occur in tropical regions and are related
nighttime temperatures are currently well below 20 C. to the magnitude of the change in mean temperature relative
Significant increases in TR are seen in the extra-tropical to the low short-term tropical temperature variability. WSDI
Northern Hemisphere, which are most pronounced in and CSDI are sensitive to the underlying climatological
south-eastern North America, the Mediterranean, and central temperature variability of the respective region [Radinović
Asia. TR increases by as many as 80 days in these regions and Ćurić, 2012], which is small in the tropics and larger in
under RCP8.5, which would mean that almost the entire the extra-tropics. Details in the regional changes of CSDI
summer season will have nighttime temperatures above and WSDI under the RCPs can be found in the supporting in-
20 C. Considering the strong increase in TXx in the MED formation (Figure S1).
and CAS regions as discussed earlier, these regions would [37] The temporal evolution of WSDI and CSDI averaged
face severe heat stress in summer if future climate change over all land regions is also shown in the supporting infor-
follows the path of RCP8.5. mation Figure S1. By the end of the 21st century, the median
2481
SILLMANN ET AL.: CMIP5 PROJECTIONS OF EXTREMES INDICES
Figure 7. The multimodel median of temporally averaged changes in the cold spell (CSDI, left) and
warm spell (WSDI, right) duration index for the period 2081–2100 as differences from the reference
period 1981–2000.
CMIP5 models project a decrease in CSDI by about 3.4 days [39] There is a consistent decrease in cold nights (TN10p)
in RCP2.6, 3.9 days in RCP4.5, and 4.2 days in RCP8.5 and cold days (TX10p) from the late 20th to the 21st century
relative to the 1981–2000 reference period. The projected in all SRES and RCP scenarios (Figures 8a and 8b). The
CMIP3 median decrease in CSDI is somewhat weaker than median decrease is generally more pronounced for TN10p
in the CMIP5 scenarios. WSDI increases strongly under than for TX10p. In RCP2.6, TN10p decreases from about
RCP8.5 with about 167 days (globally average over land) 10% in 1961–1990 to 3% by the end of the 21st century
by year 2100. The median WSDI increase in RCP4.5 and and TX10p decreases to 4%. TN10p decreases further to
RCP2.6 is 75 and 31 days, respectively. 1.5% in RCP4.5 and 0.3% in RCP8.5, whereas TX10p
4.1.3. Percentile Indices decreases to 2% and 0.7% in RCP4.5 and RCP8.5
[38] Projected changes in the percentile indices are shown respectively. In the 2046–2065 period, the median changes
in absolute terms, and not as differences relative to the in SRES B1 are within the RCP2.6 ensemble spread for
reference period as for the other temperature indices. This TN10p and TX10p. For A1B and A2, they are close to the
is because, by construction, the percentile indices represent median RCP4.5 change. By the end of the 21st century,
exceedance rates (in %) relative to the 1961–1990 base the responses for different scenarios diverge further
period, during which they average to approximately 10%, apart, with A2 showing the strongest decrease in TN10p
which will serve as the baseline for future changes. from about 10% to 0.5% and TX10p to 1.3%, which is
2482
SILLMANN ET AL.: CMIP5 PROJECTIONS OF EXTREMES INDICES
Exceedance rate, %
Exceedance rate, %
10 10 10 10
8 8 8 8
6 6 6 6
4 4 4 4
2 2 2 2
CMIP3 B1 CMIP3 A1B CMIP3 A2 CMIP3 B1 CMIP3 A1B CMIP3 A2
0 0 0 0
1960 1980 2000 2020 2040 2060 2080 2100 1960 1980 2000 2020 2040 2060 2080 2100
Year Year
Exceedance rate, %
60 60 60 60
50 50 50 50
40 40 40 40
30 30 30 30
20 20 20 20
1960 1980 2000 2020 2040 2060 2080 2100 1960 1980 2000 2020 2040 2060 2080 2100
Year Year
Figure 8. Same as Figure 3, but for the percentile indices (a) cold nights (TN10p), (b) cold days
(TX10p), (c) warm nights (TN90p), and (d) warm days (TX90p). Changes are displayed as absolute
exceedance rates (in %). By construction the exceedance rate averages to about 10% over the base period
1961–1990.
comparable to the decreases projected under RCP8.5. TN90p and TX90p are from 10% to 63% and 52%,
That is, there will be virtually no cold nights or days as respectively, which is somewhat smaller than in RCP8.5
defined for the 1961–1900 base period under these by 2100.
future projections. The decrease of TN10p and TX10p in [41] The simulated decreases in cold days and nights as
B1 from about 10% to 2% and to 3%, respectively, is less well as the increases in warm days and nights for the three
pronounced than in RCP4.5, whereas TN10p and TX10p RCPs are statistically significant everywhere over land
decrease to 1% and 2%, respectively, in A1B. The latter (Figure 9 for TN10p and TN90p, and supporting informa-
changes lie between the RCP4.5 and RCP8.5 projections. tion Figure S2 for TX10p and TX90p). The changes in the
The interquartile model spread generally becomes percentile indices based on minimum temperature (TN10p
smaller as the projection approaches the zero exceedance and TN90p) are more pronounced than those based on
rate as more and more models simulate fewer and maximum temperature (TX10p and TX90p).
fewer cold nights and days. Consequently, the largest inter- [42] Note that the spatial patterns of change differ from
quartile model spread is seen for the weaker B1 and those for the absolute indices TNn and TXx. The largest
RCP2.6 scenarios. decreases in TN10p and largest increases in TN90p are
[40] Warm nights and days (TN90p and TX90p, projected in tropical regions that are characterized by a small
respectively) show a general increase in the exceedance rate day-to-day temperature variability so that changes in mean
toward the end of the 21st century (Figures 8c and 8d). The temperature are associated with comparatively larger changes
increase is more pronounced for TN90p than for TX90p. in exceedance rates below the 10th and above the 90th
The median increase in TN90p and TX90p in RCP8.5 is percentiles. High northern latitudes are also affected by a
from about 10% in 1961–1990 to 69% and 62% by 2100, strong decrease in TN10p in RCP2.6 and RCP4.5. As
respectively. The smallest increases in TN90p and TX90p, mentioned before, TN10p decreases to near 0% by year
to 31% and 26% respectively, occur in RCP2.6, followed 2100 in most regions under RCP8.5. The smallest changes
by greater respective increases to 44% and 39% in in the percentile indices, in both TN10p and TN90p, are
RCP4.5. In B1, the median change in TN90p and TX90p projected for southern South America.
is 34% and 29%, respectively, which lies between RCP2.6 [43] As pointed out in Klein Tank et al. [2009], it is
and RCP4.5. The A1B median changes are between important to interpret changes in percentile indices on a
RCP4.5 and RCP8.5 with increases in TN90p and TX90p, seasonal basis as these indices are calculated relative to the
respectively, of 56% and 45%. The median A2 increases in annual cycle of the percentile thresholds. The seasonality
2483
SILLMANN ET AL.: CMIP5 PROJECTIONS OF EXTREMES INDICES
Figure 9. The multimodel median of the annual frequency of cold nights (TN10p, left) and warm nights
(TN90p, right) temporally averaged over the period 2081–2100 as absolute values of the exceedance rate
(in %). By construction the exceedance rate averages to about 10% over the base period 1961–1990. Gray
areas indicate values of exact zero percent. All changes are significant at the 5% significance level.
of changes in the percentile indices by the end of the 21st 4.2. Precipitation Indices
century is depicted in the box-and-whisker plots in Figure 10. 4.2.1. Temporal Evolution
TN10p decreases and TN90p increases most in the summer [44] Changes in precipitation indices relative to the
season (see also supporting information, Figure S3 for sea- 1981–2000 reference period are expressed in percentage
sonal aspects of TX10p and TX90p). This effect is more terms. Global land averaged precipitation indices are
pronounced in the northern extra-tropics where temperature projected to increase in the 21st century (Figure 11). Relative
variability is larger than in the tropics. In JJA, for instance, increases in RX5day (Figure 11c), which represents a more
substantial increases in TN90p (from 10% to 85% in extreme aspect of the precipitation distribution, are greater
RCP8.5) occur in such regions as MED, EAS, CAS, and over time than those for PRCPTOT and SDII (Figures 11a
TIB. The increase in TN90p in WNA and CNA is twice as and 11b). In RCP8.5, PRCPTOT and SDII are projected to
large in JJA (~80% exceedance rate) than in DJF (~45% increase by 9% and 12%, respectively, by year 2100, whereas
exceedance rate) under RCP8.5. Regions in the Southern RX5day is projected to increase by 20%. PRCPTOT and SDII
Hemisphere, such as AUS and SSA, also show seasonal show similar median increases of about 3.5% in RCP2.6 and
differences with, for instance, larger increases in TN90p in somewhat stronger median increases of about 6% in
Southern Hemispheric summer (DJF) compared to winter (JJA). RCP4.5, whereas RX5day increases by 6% and 10% in
2484
SILLMANN ET AL.: CMIP5 PROJECTIONS OF EXTREMES INDICES
[a] Cold Nigths (TN10p) in 2081−2100 [b] Warm Nigths (TN90p) in 2081−2100
ANN ANN
12 12 120 120
RCP2.6 RCP4.5 RCP8.5 RCP2.6 RCP4.5 RCP8.5
10 10 100 100
Exceedance rate, %
Exceedance rate, %
8 8 80 80
6 6 60 60
4 4
40 40
2 2
20 20
0 0
AUS
AMZ
SSA
CAM
WNA
CNA
ENA
ALA
GRL
MED
NEU
WAF
EAF
SAF
SAH
SEA
EAS
SAS
CAS
TIB
NAS
AUS
AMZ
SSA
CAM
WNA
CNA
ENA
ALA
GRL
MED
NEU
WAF
EAF
SAF
SAH
SEA
EAS
SAS
CAS
TIB
NAS
[c] Cold Nigths (TN10p) in 2081−2100 [d] Warm Nigths (TN90p) in 2081−2100
JJA JJA
12 12 120 120
RCP2.6 RCP4.5 RCP8.5 RCP2.6 RCP4.5 RCP8.5
10 10 100 100
Exceedance rate, %
Exceedance rate, %
8 8 80 80
6 6 60 60
4 4 40 40
2 2
20 20
0 0 AUS
AMZ
SSA
CAM
WNA
CNA
ENA
ALA
GRL
MED
NEU
WAF
EAF
SAF
SAH
SEA
EAS
SAS
CAS
TIB
NAS
AUS
AMZ
SSA
CAM
WNA
CNA
ENA
ALA
GRL
MED
NEU
WAF
EAF
SAF
SAH
SEA
EAS
SAS
CAS
TIB
NAS
[e] Cold Nigths (TN10p) in 2081−2100 [f] Warm Nigths (TN90p) in 2081−2100
DJF DJF
12 12 120 120
RCP2.6 RCP4.5 RCP8.5 RCP2.6 RCP4.5 RCP8.5
10 10 100 100
Exceedance rate, %
Exceedance rate, %
8 8 80 80
6 6 60 60
4 4 40 40
2 2
20 20
0 0
AUS
AMZ
SSA
CAM
WNA
CNA
ENA
ALA
GRL
MED
NEU
WAF
EAF
SAF
SAH
SEA
EAS
SAS
CAS
TIB
NAS
AUS
AMZ
SSA
CAM
WNA
CNA
ENA
ALA
GRL
MED
NEU
WAF
EAF
SAF
SAH
SEA
EAS
SAS
CAS
TIB
NAS
Figure 10. Projected changes in annual, JJA, and DJF frequency of cold nights (TN10p, left) and warm
nights (TN90p, right) over the time period 2046–2065 (left) and 2081–2100 (right) for RCP2.6 (blue),
RCP4.5 (green), and RCP8.5 (red) as absolute values for the exceedance rate (in %). By construction
the exceedance rate averages to about 10% over the base period 1961–1990. Regional mean changes
are shown for each of the 21 subregions (cf. Figure 2). Boxes indicate the interquartile model spread of
the CMIP5 ensemble with the horizontal line indicating the ensemble median.
RCP2.6 and RCP4.5, respectively. The CMIP5 interquartile [46] Toward the end of the 21st century, the median
model spreads in the three RCPs do not overlap after year CMIP3 projections for PRCPTOT and SDII diverge more
2071 for SDII and RX5day, but remain overlapping for substantially. However, the interquartile model spread also
PRCPTOT throughout the 21st century. increases and thus the ranges continue to overlap for the
[45] The CMIP3 interquartile model spreads in the projection three SRES scenarios. By 2100, the median increase in
of PRCPTOT, RX5day, and SDII overlap for all three SRES PRCPTOT is 4% in B1, which is between those in RCP2.6
scenarios. The changes in PRCPTOT and SDII range below and RCP4.5, 5% in A1B, which is similar to that in
or around the median response in RCP2.6 for the period RCP4.5, and 6% in A2. SDII increases by 4% in B1, similarly
2046–2065. The increase of RX5day under B1 is also to RCP2.6, and by 5% in A1B, which falls between RCP2.6
comparable to that in RCP2.6, while the median A1B and and RCP4.5, as does the SDII increase of 7% in A2. For
A2 increase is similar to the RCP4.5 projections. RX5day, the median change of 8% in B1 is between
2485
SILLMANN ET AL.: CMIP5 PROJECTIONS OF EXTREMES INDICES
Total Wet−day Precipitation (PRCPTOT) Simple Daily Intensity (SDII) Max. 5−day Precipitation (RX5day)
historical RCP4.5 historical RCP4.5 historical RCP4.5
RCP2.6 RCP8.5 RCP2.6 RCP8.5 RCP2.6 RCP8.5
20 20 20 20 20 20
Relative change, %
Relative change, %
Relative change, %
15 15 15 15 15 15
10 10 10 10 10 10
5 5 5 5 5 5
0 0 0 0 0 0
CMIP3 B1 CMIP3 A1B CMIP3 A2 CMIP3 B1 CMIP3 A1B CMIP3 A2 CMIP3 B1 CMIP3 A1B CMIP3 A2
−5 −5 −5 −5 −5 −5
1960 1980 2000 2020 2040 2060 2080 2100 1960 1980 2000 2020 2040 2060 2080 2100 1960 1980 2000 2020 2040 2060 2080 2100
Year Year Year
Figure 11. Same as Figure 3, but for precipitation indices. Changes are displayed relative to the reference
period 1981–2000 (in %).
RCP2.6 and RCP4.5, whereas the A1B and A2 changes of (expressed by R95pT) generally increases in all three RCPs
12% and 16%, respectively, lie between those in RCP4.5 (see supporting information Figure S4, left).
and RCP8.5. In Sillmann et al. [2013], we show that GCMs [50] Regions affected by a decrease in PRCPTOT generally
underestimate observed precipitation magnitudes (e.g., as for coincide with regions where there is a significant increase in
RX5day and SDII), although CMIP5 models show an im- the maximum number of consecutive dry days (CDD) as
provement compared to CMIP3. This model bias can be in depicted in Figure 13 (left). In particular, significant increases
part attributed to the spatial scale mismatch between point- of CDD occur in Central America, the Mediterranean region
estimates of precipitation in observations and grid-box-esti- as well as southern Africa. The simultaneous decrease in
mates in models. This means that downscaling techniques [e. heavy precipitation days (R10mm, Figure 13, right), particular
g., Bürger et al., 2012] should be considered for regional and in RCP8.5,indicates an intensification of meteorological
local assessment of the projected changes in precipitation drought conditions in these regions. On the contrary, in South
extremes. and Southeast Asia, the increases in CDD are combined with
[47] The globally averaged temporal evolution of maximum increases in R10mm and RX5day (see supporting informa-
consecutive dry days (CDD) is not shown here as the temporal tion Figure S4, right) indicating an intensification of both
and spatial variability of this index over global land is very wet and dry seasons in these regions. Significant decreases
large (see also Sillmann et al., [2013]). Changes in CDD in CDD are projected at high northern latitudes, Northeast
expressed in terms of global land averages do not provide a Asia reaching down to the Tibetan Plateau and East Africa
meaningful picture, i.e., trends are small as compared to the as well as Antarctica, which coincide with large increases
overall variability in this index. However, changes in this in R10mm and RX5day in these regions. These regions are
index can be more meaningful for regions where climatic projected to become generally wetter in future. The large
conditions are more uniform as discussed in the following changes (both CDD increases and decreases) in the Sahara
section. are not significant due to high volatility in the lengths of
4.2.2. Spatial and Seasonal Patterns of Changes very long dry spells spanning many years that may occur in
[48] The ratio of extreme precipitation expressed by very this region.
wet days (R95p) to the total wet-day precipitation [51] Figure 14 displays regional summaries of annual and
(PRCPTOT) represents the annual contribution of very wet seasonal changes in extreme precipitation in terms of
days to the total annual wet-day precipitation (also referred RX5day in the 21 subregions. On an annual basis, RX5day
to as R95pT), which is relevant for societal impacts [Alexander generally increases for all three RCPs (Figure 14a). The
et al., 2006]. Therefore, it is useful to examine changes in largest median increase of 30% is projected under RCP8.5
PRCPTOT in relation to those in R95p (Figure 12). By the in South Asia (SAS) followed by increases of 20–30% in
end of the 21st century, PRCPTOT increases significantly over ALA, GRL, NEU, WAF, EAF, SEA, EAS, TIB, and NAS.
large parts of the Northern Hemisphere, East Africa, South and In high northern latitude regions, such as ALA, GRL, and
Southeast Asia as well as Antarctica in all three RCPs relative NAS, and also in the high altitude TIB region, the increase
to the 1981–2000 reference period. The greatest changes in RX5day is less pronounced in JJA (Figure 14b) than in
are projected in high northern and southern latitudes under DJF (Figure 14c) with the strongest increase in NAS of
RCP8.5. Areas of significant projected decreases in about 40% in DJF and 20% in JJA. The opposite is seen in
PRCPTOT include Australia (only significant in RCP2.6) SAS, where summer increases are larger than winter
as well as South Africa, the Mediterranean region, and increases in RX5day. In SEA and EAS, similar increases
Central America (significant in RCP8.5). in RX5day of about 20% in RCP8.5 are projected for JJA
[49] Generally, regional increases or decreases in and DJF. Regions for which the CMIP5 ensemble shows
PRCPTOT coincide with corresponding changes in R95p no or inconsistent changes in JJA, but an increase in DJF
and RX5day (see supporting information Figure S4, right). RX5day are AMZ, SSA, WNA, CNA, NEU, and CAS.
In particular, R95p increases in high northern latitudes, East [52] The smallest annual increase in RX5day is projected
Africa, and Antarctica and decreases in Central America, in MED under all three RCPs. In this region, there are large
South Africa, and the Mediterranean. The contribution of RX5day decreases in summer (JJA), particularly in RCP4.5
very wet days to the annual total wet-day precipitation and RCP8.5, but only a very small increase (5% or less
2486
SILLMANN ET AL.: CMIP5 PROJECTIONS OF EXTREMES INDICES
Figure 12. The multimodel median of temporally averaged total wet-day precipitation (PRCPTOT, left),
very wet day precipitation (R95p, right) over the time period 2081–2100 expressed relative to the
reference period 1981–2000 (in %) for RCP2.6 (top), RCP4.5 (middle), and RCP8.5 (bottom). Stippling
indicates grid points with changes that are not significant at the 5% significance level.
depending on the RCP) in winter (DJF). Regions on the SAH, there is a general disagreement on the sign of seasonal
Southern Hemisphere, such as AUS and SAF, experience changes in the CMIP5 ensemble with a strong outlier
large decreases in RX5day in winter (JJA). In Central (approximately 113% in ANN and JJA). The model agreement
America (CAM), RX5day decreases in JJA for most CMIP5 on the sign of changes for precipitation indices is considered in
models under RCP8.5 and does so consistently under all the next section.
RCPs in DJF for the majority of models. These features 4.2.3. Model Agreement
are not captured in the global maps of annual changes [53] In contrast to the projected changes in the temperature
(see supporting information Figure S4, right), where changes indices, where there is a general agreement on the sign of
in RX5day are not found to be statistically significant in AUS, change independent of the region considered, changes in
SAF, and CAM. The decrease of RX5day in combination with the precipitation indices are less consistent in this regard.
the projected increase in CDD (see Figure 13) suggests severe Figure 15 provides a more detailed regional picture of the
drying of these regions in future climate projections. For extent of model agreement on projected changes in the
regions, such as AMZ, SSA, WNA, CNA, SAH, and CAS, precipitation indices (PRCPTOT, R95p, R10mm, and CDD)
the CMIP5 RX5day projections center around zero in JJA, over the course of the 21st century. We assess the agreement
but show increases in DJF (except SAH). Particularly for on the sign of change in the CMIP5 model ensemble on the
2487
SILLMANN ET AL.: CMIP5 PROJECTIONS OF EXTREMES INDICES
Figure 13. The multimodel median of temporally averaged changes of consecutive dry days (CDD, left)
and heavy precipitation days (R10mm, right) over the time period 2081–2100 for RCP2.6 (top), RCP4.5
(middle), and RCP8.5 (bottom). Changes are displayed as differences (in days) relative to the reference
period (1981–2000). Stippling indicates grid points with changes that are not significant changes at the
5% significance level.
basis of the interquartile model spread (boxes) for each RCP, percentile threshold. It therefore accounts for climatological
which would correspond to an agreement on sign amongst at differences in precipitation between wetter and dryer regions
least 75% of models (referred to as the majority of models in [Klein Tank et al., 2009]. The advantages of such an index
the following), and the full ensemble range (whiskers). become particularly obvious in SAH, where the fixed-
[54] The changes projected for the 2046–2065 period threshold index R10mm does not indicate any changes since
(Figure 15, left column) generally intensify toward the end the 10 mm threshold is rarely exceeded in this region, while
of the century (right column in Figure 15). Changes in R95p does show a change.
R95p, which represents the more extreme aspects of [55] The PRCPTOT and R10mm show increases in similar
precipitation variability as compared to the other three regions with the majority of models also agreeing on the sign
indices, are found to be more consistent among the models of the change. These regions comprise WNA, ENA, ALA,
with most of them generally agreeing on the sign of annual GRL, NEU, and all regions in Asia. In the northern high
increases across most regions, except AUS, CAM, and latitude regions (ALA, GRL, and NAS), as well as in TIB
SAH. Note that R95p is the only precipitation-based index and EAS, this increase in precipitation is accompanied by
discussed in this study that is based on the exceedance of a consistent inter-model agreement on a decrease in CDD.
2488
SILLMANN ET AL.: CMIP5 PROJECTIONS OF EXTREMES INDICES
[a] Max. 5−day Precipitation (RX5day) in 2081−2100 [56] The large outliers in the SAH for the precipitation
ANN indices PRCPTOT, R95p, are due to the large changes
100 RCP2.6 RCP4.5 RCP8.5 100 simulated in the model BNU-ESM (with 173% and 468%,
respectively). The models MIROC5 and FGOALS-s2 form
the extreme outliers in SEA for R10mm with changes of
50 50 24% to 11%, respectively.
[57] The majority of CMIP5 models agree on an increase
in CDD in AUS, AMZ, SSA, CAM, MED, and SAF under
%
2489
SILLMANN ET AL.: CMIP5 PROJECTIONS OF EXTREMES INDICES
[a] Total Wet−day Precipitation (PRCPTOT) in 2046−2065 [b] Total Wet−day Precipitation (PRCPTOT) in 2081−2100
ANN ANN
80 RCP2.6 RCP4.5 RCP8.5 80 80 RCP2.6 RCP4.5 RCP8.5 80
60 60 60 60
40 40 40 40
20 20 20 20
%
%
0 0 0 0
WNA
CAM
MED
CAM
MED
AMZ
CNA
NEU
AMZ
CNA
NEU
ENA
GRL
SAH
CAS
NAS
ENA
GRL
SAH
CAS
NAS
WAF
WAF
SSA
SEA
EAS
SAS
SSA
SEA
EAS
SAS
EAF
SAF
EAF
SAF
ALA
ALA
AUS
AUS
TIB
TIB
[c] Very Wet Days (R95p) in 2046−2065 [d] Very Wet Days (R95p) in 2081−2100
ANN ANN
RCP2.6 RCP4.5 RCP8.5 RCP2.6 RCP4.5 RCP8.5
200 200 200 200
%
%
50 50 50 50
0 0 0 0
WNA
CAM
MED
CAM
MED
AMZ
CNA
NEU
AMZ
CNA
NEU
ENA
GRL
SAH
CAS
NAS
ENA
GRL
SAH
CAS
NAS
WAF
WAF
SSA
SEA
EAS
SAS
SSA
SEA
EAS
SAS
EAF
SAF
EAF
SAF
ALA
ALA
AUS
AUS
TIB
TIB
[e] Heavy Precipitation Days (R10mm) in 2046−2065 [f] Heavy Precipitation Days (R10mm) in 2081−2100
ANN ANN
25 25 25 25
RCP2.6 RCP4.5 RCP8.5 RCP2.6 RCP4.5 RCP8.5
20 20 20 20
15 15 15 15
10 10 10 10
days
days
5 5 5 5
0 0 0 0
−5 −5 −5 −5
WNA
CAM
MED
CAM
MED
AMZ
CNA
NEU
AMZ
CNA
NEU
ENA
GRL
SAH
CAS
NAS
ENA
GRL
SAH
CAS
NAS
WAF
WAF
SSA
SEA
EAS
SAS
SSA
SEA
EAS
SAS
EAF
SAF
EAF
SAF
ALA
ALA
AUS
AUS
TIB
TIB
[g] Consecutive Dry Days (CDD) in 2046−2065 [h] Consecutive Dry Days (CDD) in 2081−2100
ANN ANN
100 100 100 100
RCP2.6 RCP4.5 RCP8.5 RCP2.6 RCP4.5 RCP8.5
80 80 80 80
60 60 60 60
40 40 40 40
days
days
20 20 20 20
0 0 0 0
WNA
CAM
MED
CAM
MED
AMZ
CNA
NEU
AMZ
CNA
NEU
ENA
GRL
SAH
CAS
NAS
ENA
GRL
SAH
CAS
NAS
WAF
WAF
SSA
SEA
EAS
SAS
SSA
SEA
EAS
SAS
EAF
SAF
EAF
SAF
ALA
ALA
AUS
AUS
TIB
TIB
Figure 15. Projected changes in the annual precipitation indices, PRCPTOT, R95p, R10mm, and CDD over two future
time periods 2046–2065 and 2081–2100 for RCP2.6 (blue), RCP4.5 (green), and RCP8.5 (red). Regional mean changes
are shown for each of the 21 subregions (cf. Figure 2). Boxes indicate the interquartile model spread (25th and 75th
quantiles) of the CMIP5 ensemble with the horizontal line indicating the ensemble median. Changes are displayed (in %)
for PRCPTOT and R95p and (in days) for R10mm and CDD relative to the reference period 1981–2000.
2490
SILLMANN ET AL.: CMIP5 PROJECTIONS OF EXTREMES INDICES
increases in high northern latitudes have been associated large impacts, such as alteration of ecosystems and species
with a number of different mechanisms, including reduc- extinction [e.g., Corlett, 2011].
tions in fall/winter ice and snow cover, indirect responses [65] • Extreme precipitation increases proportionally fas-
to decreases in summer ice cover, and increased summer ter than total wet-day precipitation (PRCPTOT). Changes
ocean heating as well as changes in the surface heat fluxes in very wet days (R95p) indicate that extreme precipitation
[e.g., Screen and Simmonds, 2010; Deser et al., 2010; generally increases in most regions, except for regions
Flanner et al., 2011]. In contrast, changes in the seasonal such as Australia, Central America, South Africa, and the
maxima of TX (TXx) are more uniformly distributed Mediterranean region where a precipitation decrease and
over the global land with generally stronger increases in longer dry spells, captured by the consecutive dry days
summer than in winter. Stronger increases in summer TXx (CDD) index, are projected. The Mediterranean region in
also relate to soil-moisture feedbacks as pointed out in particular stands out with an intensification of meteorological
Seneviratne et al. [2006]. drought conditions represented by projected increases in
[61] In agreement with previous studies [e.g., Diffenbaugh CDD that coincide with decreases in indices describing the
and Giorgi, 2012; Fischer and Schaer, 2010; Giorgi and wet part of the precipitation distribution, such as heavy precip-
Lionello, 2008], the projected changes in the indices stand itation days (R10mm). A slight increase in R95p projected for
out in the Mediterranean region indicating a considerable this region further suggest that although dry conditions be-
intensification of heat and water stress in that region. The come more severe, precipitation can be much more extreme
smallest changes in temperature indices are simulated in when it does occur.
southern South America, particularly in Southern Hemispheric [66] • Under RCP2.6, with its modest radiative forcing at
winter (JJA), in accordance with the small projected changes year 2100, annual global averaged changes in temperature
for the mean temperatures in this region [e.g., Meehl et al., extremes are projected to generally remain below 2 C
2007a]. These patterns of climate change intensify with the relative to the reference period 1981–2000. On seasonal and
increasing radiative forcing in the considered scenarios regional scales, however, there are increases in the minimum
confirming results of previous studies [e.g., Cubasch et al., of TN in winter exceeding 3 C, particularly in northern high
1992; Russo and Sterl, 2011]. latitudes.
[62] The most notable findings of our study, complementing [67] • Projected changes in temperature and precipitation
previous studies, include the following: extremes are generally more pronounced in RCP4.5 than in
[63] • The asymmetry in the warming of minimum and B1 simulations, although both have similar amounts of radi-
maximum temperatures as observed in the historical record ative forcing by year 2100. In particular, precipitation
[e.g., Karl et al., 1993; Trenberth et al., 2007] continues extremes under B1 tend to be closer in magnitude to the
and intensifies with increasing radiative forcing in the future range of RCP2.6. However, it also has to be kept in mind that
climate projections. In particular, projected changes in indi- CMIP3 and CMIP5 use different sets of GCMs which poses
ces based on TN (warm and cold nights, frost days and trop- an additional source of uncertainty in making comparisons
ical nights) are more pronounced than in indices based on TX between SRES and RCP projections. Furthermore, to get a
(e.g., warm and cold days, summer and ice days). Note how- better idea of the changes in extremes in relation to the
ever that detection and attribution studies based on TN and amount of radiative forcing, other factors (such as aerosol
TX tend to show that models warm TN less than observed concentrations) that affect climate response would also need
and that they warm TX more than observed over the latter to be assessed.
half of the 20th century [e.g., Zwiers et al., 2011]. This [68] • None of the SRES simulations considered in
means, while the contrast in changes between TN and TX this study or previous studies (i.e., B1, A1B, and A2)
in the models corresponds qualitatively with observations, project changes in temperature and precipitation ex-
the observed contrast tends to be even larger than simulated tremes as pronounced as in RCP8.5, which has the largest
by models. radiative forcing amongst the scenarios we considered.
[64] • The spatial patterns of changes in temperature in- If greenhouse gas emissions continue to rise at the cur-
dices depend on the index type. For example, spatial pat- rent pace or even accelerate, very large changes in
terns of changes in absolute indices (e.g., TXx and TNn) extremes are to be expected as projected in RCP8.5
differ from those in percentile indices. The strongest in- simulations.
crease in TNn is in northern high latitudes, whereas [69] In conclusion, the different categories of indices (e.g.,
changes in TXx are more evenly distributed around the absolute versus percentile indices) provide complementary
globe. The patterns in TNn and TXx reflect changes that information that may be relevant to different applications
are similar to changes in the mean temperature. In contrast, and a careful selection of indices is necessary. Regional
the percentile indices, such as warm and cold nights, adjustments to the index definitions and sector-based
which represent the exceedance rates above or below a definitions (e.g., for agriculture, drainage and reservoir planning,
percentile threshold derived from the 1961–1990 base pe- energy supply, human health, etc.) may be needed to gain
riod, show the highest increase in the tropical regions more information in individual regions for application in
where inter-annual temperature variability is relatively impact and adaptation studies.
small. Therefore, small shifts in the mean of the tempera- [70] Furthermore, future changes in temperature and
ture distribution can lead to larger changes in the excee- precipitation extremes need to be assessed carefully in relation
dance rates than in the high variability extra-tropical to changes in circulation patterns [e.g., van Oldenborgh et al.,
regions. Since ecosystems and human infrastructure in the 2009; Sillmann and Croci-Maspoli, 2009] and other feedback
tropics are adapted to relatively small temperature varia- mechanisms such as snow, soil moisture, and vegetation
tions, small changes in the extremes can have relatively [e.g., Seneviratne et al., 2006; Jaeger and Seneviratne,
2491
SILLMANN ET AL.: CMIP5 PROJECTIONS OF EXTREMES INDICES
2011; Hirschi et al., 2011]. While this was beyond the Cubasch, U., K. Hasselmann, H. Höck, E. Maier-Reimer, U. Mikolajewicz,
B. Santer, and R. Sausen (1992), Time-dependent greenhouse warming
scope of this study, it will however be crucial for a computations with a coupled ocean-atmosphere model, Clim. Dyn., 8,
comprehensive understanding of the future changes simulated 55–69.
for different regions. Deser, C., R. Tomas, M. Alexander, and D. Lawrence (2010), The seasonal
[71] Additional work will also be required to better atmospheric response to projected Arctic sea ice loss in the late twenty-
first century, J. Climate, 23, 333–351, doi:10.1175/2009JCLI3053.1.
quantify the uncertainties in the projected changes in the Diffenbaugh, N. S., and F. Giorgi (2012), Climate change hotspots in the
indices. The model spread and variations in response to the CMIP5 global climate model ensemble, Clim. Chang., 114, 813–822,
different SRES and RCP scenarios are indicative of doi:10.1007/s10584-012-0570-x.
Fischer, E. M., and C. Schaer (2010), Consistent geographical patterns of
uncertainties associated with natural internal variability of changes in high-impact European heatwaves, Nat. Geosci., 3, 398–403,
the climate system, structural, and parametric uncertainty doi:10.1038/ngeo866.
as embodied by the range of climate models considered, Flanner, M. G., K. M. Shell, M. Barlage, D. K. Perovich, and M. A. Tschudi
and future forcing uncertainty. Not considered here, (2011), Radiative forcing and albedo feedback from the Northern
Hemisphere cryosphere between 1979 and 2008, Nat. Geosci., 4,
but equally important, is observational and process knowl- 151–155., doi:10.1038/ngeo1062.
edge uncertainty, as represented by differences between Frich, P., L. Alexander, P. Della-Marta, B. Gleason, M. Haylock, A. Klein
observational and observationally constrained analyses Tank, and T. Peterson (2002), Observed coherent changes in climate
extremes during the second half of the twentieth century, Clim. Res.,
(i.e., reanalyses) and differences between models in repli- 19, 193–212, doi:10.3354/cr019193.
cating the climatological behavior of the indices in the Giorgi, F., and R. Francisco (2000), Uncertainties in regional climate
recent past as documented in Part I of this paper [Sillmann change prediction: A regional analysis of ensemble simulations with the
et al., 2013]. While most models represent extremes rea- HADCM2 coupled AOGCM, Clim. Dyn., 16, 169–182.
Giorgi, F., and P. Lionello (2008), Climate change projections for the
sonably well, including trends in the present climate, it Mediterranean region, Global Planet. Chang., 63, 90–104, doi:10.1016/
should be noted that there are also some indices, which j.gloplacha.2007.09.005.
are not well represented or difficult to evaluate with cur- Gleckler, P. J., K. E. Taylor, and C. Doutriaux (2008), Performance metrics
for climate models, J. Geophys. Res., 113, D06104, doi:10.1029/
rently available observational datasets (e.g., SDII and 2007JD008972.
diurnal temperature range (DTR)). Hirschi, M., S. Seneviratne, V. Alexandrov, F. Boberg, C. Boroneant,
[72] The results presented in this study give a first O. Christensen, H. Formayer, B. Orlowsky, and P. Stepanek (2011),
Observational evidence for soil-moisture impact on hot extremes in
impression of what we can expect from the new CMIP5 southeastern Europe, Nat. Geosci., 4, 17–21, doi:10.1038/ngeo1032.
multimodel ensemble and subsequent studies will no doubt IPCC (2007), Climate Change 2007: The Physical Science Basis.
provide more insights into seasonal and regional changes. Contribution of Working Group I to the Fourth Assessment Report of
As the complete set of indices calculated across the CMIP3 the Intergovernmental Panel on Climate Change, pp. 996, Cambridge
University Press, Cambridge, UK and New York, NY USA.
and CMIP5 ensembles as well as the SRES and RCP IPCC (2012), Managing the Risks of Extreme Events and Disasters to
scenarios will be provided on the ETCCDI indices archive Advance Climate Change Adaptation. A Special Report of Working Groups
(http://www.cccma.ec.gc.ca/data/climdex/climdex.shtml), we I and II of the Intergovernmental Panel on Climate Change, pp. 582,
Cambridge University Press, Cambridge, UK, and New York, NY, USA.
hope to encourage further studies of the changes in climate Jaeger, E. B., and S. I. Seneviratne (2011), Impact of soil moisture-atmosphere
extremes and to facilitate impact studies and the development coupling on European climate extremes and trends in a regional climate
of adaptation strategies by providing the necessary physical model, Clim. Dyn., 36, 1919–1939, doi:10.1007/s00382-010-0780-8.
basis. Karl, T. R., et al. (1993), Asymmetric trends of daily maximum and
minimum temperature, Papers in Natural Resources, Paper 185.
Kharin, V. V., F. W. Zwiers, X. Zhang, and G. C. Hegerl (2007), Changes
in temperature and precipitation extremes in the IPCC ensemble of global
[73] Acknowledgments. We acknowledge the World Climate Research coupled model simulations, J. Climate, 20, 1419–1444, doi:10.1175/
Programme’s Working Group on Coupled Modelling, which is responsible for JCLI4066.1.
CMIP, and we thank the climate modeling groups (listed in Tables 1 and 2 of Kiktev, D., D. M. H. Sexton, L. Alexander, and C. K. Folland (2003),
this paper) for producing and making available their model output. For CMIP, Comparison of modeled and observed trends in indices of daily climate
the U.S. Department of Energy’s Program for Climate Model Diagnosis and extremes, J. Climate, 16, 3560–3571, doi:10.1175/1520-0442(2003)
Intercomparison provides coordinating support and led development of 016<3560:COMAOT>2.0.CO;2.
software infrastructure in partnership with the Global Organization for Earth Klein Tank, A. M. G., F. W. Zwiers, and X. Zhang (2009), Guidelines on
System Science Portals. F.Z. is supported by Australian Research Council analysis of extremes in a changing climate in support of informed
(grant LP100200690) and J.S. is funded by the German Research Foundation decisions for adaptation, Climate data and monitoring WCDMP-No. 72,
(grant Si 1659/1-1). We thank Greg Flato, Chris Derksen, Ross Brown and WMO-TD No. 1500, 56pp.
Elizabeth Bush, and three reviewers for their helpful comments on the Meehl, G., et al. (2007a), Global climate projections, in Climate Change
manuscript. 2007: The Physical Science Basis. Contribution of Working Group I to
the Fourth Assessment Report of the Intergovernmental Panel on Climate
Change, Cambridge University Press, Cambridge, United Kingdom and
References New York, NY, USA, 747–846.
Alexander, L., et al. (2006), Global observed changes in daily climate Meehl, G. A., C. Covey, T. Delworth, M. Latif, B. McAvaney, J. F. B.
extremes of temperature and precipitation, J. Geophys. Res., 111, Mitchell, R. J. Stouffer, and K. E. Taylor (2007b), The WCRP CMIP3
D05109, doi:10.1029/2005JD006290. multi-model dataset: A new era in climate change research, Bull. Am.
Bürger, G., T. Q. Murdock, A. T. Werner, S. R. Sobie, and A. J. Cannon Meteorol. Soc., 88, 1383–1394.
(2012), Downscaling extremes—An intercomparison of multiple statistical Min, S.-K., X. Zhang, F. W. Zwiers, and G. C. Hegerl (2010), Human con-
methods for present climate, J. Climate, 25, 4366–4388, doi:10.1175/ tribution to more-intense precipitation extremes, Nature, 470, 378–381,
JCLI-D-11-00408.1. doi:10.1038/nature09763.
Cardona, O., M. van Aalst, J. Birkmann, M. Fordham, G. McGregor, Morak, S., G. C. Hegerl, and J. Kenyon (2011), Detectable regional changes
R. Perez, R. Pulwarty, E. Schipper, and B. Sinh (2012), Determinants in the number of warm nights, Geophys. Res. Lett., 38, L17703,
of risk: Exposure and vulnerability, in Managing the Risks of Extreme doi:10.1029/2011GL048531.
Events and Disasters to Advance Climate Change Adaptation. A Special Moss, R. H., et al. (2010), The next generation of scenarios for climate
Report of Working Groups I and II of the Intergovernmental Panel on change research and assessment, Nature, 463, 747–756, doi:10.1038/
Climate Change (IPCC), pp. 65–108, Cambridge University Press, nature08823.
Cambridge, UK, and New York, NY, USA. Mueller, B., and S. I. Seneviratne (2012), Hot days induced by precipitation
Corlett, R. T. (2011), Impacts of warming on tropical lowland rainforests, deficits at the global scale, Proc. Natl. Acad. Sci. (USA) , PNAS Early
Trends Ecol. Evol., 26, 606–613, doi:10.1016/j.tree.2011.06.015. Edition, 6 pp. doi:10.1073/pnas.1204330109.
2492
SILLMANN ET AL.: CMIP5 PROJECTIONS OF EXTREMES INDICES
Nakicenovic, N., et al. (2000), Special Report on Emission Scenarios, Tebaldi, C., K. Hayhoe, J. Arblaster, and G. Meehl (2006), Going to
pp. 612, Cambridge University Press, Cambridge, UK, ISBN 0521804930. extremes. An intercomparison of model-simulated historical and future
Orlowsky, B., and S. I. Seneviratne (2012), Global changes in extreme changes in extreme events, Clim. Chang., 79, 185–211, doi:10.1007/
events: Regional and seasonal dimension, Clim. Chang., 110, 669–696, s10584-006-9051-4.
doi:10.1007/s10584-011-0122-9. Tebaldi, C., J. M. Arblaster, and R. Knutti (2011), Mapping model
Patz, J. A., D. Campbell-Lendrum, T. Holloway, and J. A. Foley (2005), agreement on future climate projections, Geophys. Res. Lett., 38,
Impact of regional climate change on human health, Nat. Rev., 438, L23701, doi:10.1029/2011GL049863.
310–317, doi:10.1038/nature04188. Terando, A., K. Keller, and W. E. Easterling (2012), Probabilistic
Power, S. B., F. Delage, R. Colman, and A. Moise (2012), Consensus on projections of agro-climate indices in North America, J. Geophys. Res.,
twenty-first-century rainfall projections in climate models more 117, D08115, doi:10.1029/2012JD017436.
widespread than previously thought, J. Climate, 25, 3792–3809, Thomson, A. M., et al. (2011), RCP4.5: A pathway for stabilization of
doi:10.1175/JCLI-D-11-00354.1. radiative forcing by 2100, Clim. Chang., 109, 77–94, doi:10.1007/
Radinović, D., and M. Ćurić (2012), Criteria for heat and cold wave s10584-011-0151-4.
duration indexes, Theor. Appl. Climatol., 107, 505–510, doi:10.1007/ Trenberth, K., et al. (2007), Observations: Surface and atmospheric climate
s00704-011-0495-8. change, in Climate Change 2007: The Physical Science Basis.
Riahi, K., S. Rao, V. Krey, C. Cho, V. Chirkov, G. Fischer, G. Kindermann, Contribution of Working Group I to the Fourth Assessment Report of
N. Nakicenovic, and P. Rafaj (2011), RCP8.5—A scenario of comparatively the Intergovernmental Panel on Climate Change, Cambridge University
high greenhouse gas emissions, Clim. Chang., 109, 33–57, doi:10.1007/ Press, Cambridge, United Kingdom and New York, NY, USA, 235–336.
s10584-011-0149-y. van Oldenborgh, G. J., S. Drijfhout, A. van Ulden, R. Haarsma, A. Sterl,
Rogelj, J., M. Meinshausen, and R. Knutti (2012), Global warming under C. Severijns, W. Hazeleger, and H. Dijkstra (2009), Western Europe is
old and new scenarios using IPCC climate sensitivity range estimates, warming much faster than expected, Clim. Past, 5, 1–12, doi:10.5194/
Nat. Clim. Chang., 2, 248–253, doi:10.1038/nclimate1385. cp-5-1-2009.
Russo, S., and A. Sterl (2011), Global changes in indices describing moderate van Vuuren, D. P., et al. (2011a), RCP2.6: Exploring the possibility to keep
temperature extremes from the daily output of a climate model, J. Geophys. global mean temperature increase below 2 C, Clim. Chang., 109, 95–116,
Res., 116, D03104, doi:10.1029/2010JD014727. doi:10.1007/s10584-011-0152-3.
Screen, J. A., and I. Simmonds (2010), Increasing fall-winter energy loss van Vuuren, D. P., et al. (2011b), The representative concentration
from the Arctic Ocean and its role in Arctic temperature amplification, pathways: An overview, Clim. Chang., 109, 5–31, doi:10.1007/s10584-
Geophys. Res. Lett., 37, L16707, doi:10.1029/2010GL044136. 011-0148-z.
Seneviratne, S. I., D. Luthi, M. Litschi, and C. Schaer (2006), Land- Weisskopf, M. G., H. A. Anderson, S. Foldy, L. P. Hanrahan, K. Blair,
atmosphere coupling and climate change in Europe, Lett. Nat., 443, T. J. Trk, and P. D. Rumm (2002), Heat wave morbidity and mortality,
205–209, doi:10.1038/nature05095. Milwaukee, Wis, 1999 vs 1995: An improved response?, Am. J. Public
Sillmann, J., and M. Croci-Maspoli (2009), Present and future atmospheric Health, 92, 830–833.
blocking and its impact on European mean and extreme climate, Geophys. Wilks, D. S. (2006), Statistical Methods in the Atmospheric Sciences, 2nd
Res. Lett., 36, L10702, doi:10.1029/2009GL038259. edition, Academic Press, San Diego, 648.
Sillmann, J., and E. Roeckner (2008), Indices for extreme climate events in Zhang, X., G. Hegerl, F. Zwiers, and J. Kenyon (2005), Avoiding
projections of anthropogenic climate change, Clim. Chang., 83–104, inhomogeneity in percentile-based indices of temperature extremes,
doi:10.1007/s10584-007-9308-6. J. Climate, 1641–1651, doi:10.1175/JCLI3366.1.
Sillmann, J., V. V. Kharin, X. Zhang, F. W. Zwiers, and D. Bronaugh, Zhang, X., L. Alexander, G. C. Hegerl, P. Jones, A. K. Tank, T. C. Peterson,
(2013), Climate extremes indices in the CMIP5 multimodel ensemble: B. Trewin, and F. W. Zwiers (2011), Indices for monitoring changes in
Part 1. Model evaluation in the present climate, J. Geophys. Res. Atmos., extremes based on daily temperature and precipitation data, WIREs Clim.
118, 1–18, doi:10.1002/jgrd.50203. Chang., 2, 851–870, doi:10.1002/wcc.147.
Taylor, K. E., R. J. Stouffer, and G. A. Meehl (2012), An overview Zwiers, F., X. Zhang, and Y. Feng (2011), Anthropogenic influence on long
of CMIP5 and the experiment design, Bull. Am. Meteorol. Soc., 93, return period daily temperature extremes at regional scales, J. Climate,
485–498, doi:10.1175/BAMS-D-11-00094.1. 24, 881–892, doi:10.1175/2010JCLI3908.1.
2493