Regional climate change scenarios in the Brazilian Pantanal watershed

Abstract

In the Brazilian Pantanal, hydrometeorological conditions exhibit a large interannual variability. This variability includes the seasonality of floods and droughts which can be related to land surface processes and to El Niño/La Niña. Based on regional climate change projections derived from the Eta-HadGEM2 ES models with 20 km latitude-longitude resolution for the RCP8.5 for 2071-2100, it is expected that there will be an annual mean warming of up to or above 5-7°C and a 30% reduction in rainfall by the end of the 21st century. As a consequence of higher temperatures and reduced rainfall, an increased water deficit would be expected, particularly in the central and eastern parts of the basin during spring and summer, which could affect the pulse of the Paraguay River. While the changes projected by the Eta-HadGEM2 ES are consistent with the changes produced by the CMIP5 models for the same scenario and time slice, we can affirm that changes in the hydrology of the Pantanal are uncertain, because in a comparison of CMIP5 and Eta-HadGEM2 ES model projections, some show increases in rainfall and in the discharges of the Paraguay Basin, while others show reductions.

Full text

CLIMATE RESEARCH
Clim Res
Vol. ■ ■
doi: 10.3354/cr01324
Published ■ ■
1. INTRODUCTION
The Pantanal region in South America, one of
world’s largest wetlands, is located in a large
floodplain in the center of the upper Paraguay
River basin. The basin has an area of around
360 000 km
2
, of which the Pantanal occupies about
140 000 km
2
, comprising one of world’s largest
wetlands. During the summer rainy season
(November−March), the rivers overflow their banks
and flood the adjacent lowlands, inundating as
much as 70% of the floodplain by July and
forming shallow lakes and in numerable swamps
and marshes and leaving island-like areas of
higher ground. Large sectors of the Pantanal flood-
plain are submerged from 4 to 8 mo each year by
water depths ranging from a few cm to more than
2 m. During the drier winter season (April−Sep-
tember), the rivers withdraw to their banks, but
the lowlands are only partially drained. The water
leaves via the Paraguay River and eventually into
the Paraná River, leaving behind grasslands that
support grazing animals.
Human activities in the region such as navigation,
cattle ranching, and farming, are strongly regulated
by this hydrologic regime. For instance, the available
land for cattle ranching and farming is dependent on
the extent of the inundation during each wet season.
Several human activities, such as agriculture, cattle
ranching, dam building, and other changes in hy -
draulic conditions, are threatening the Pantanal’s
ecological balance (Tucci & Clarke 1998, Hamilton
1999, 2002, Da Silva & Girard 2004).
The strategic ecological and economic importance
of the Pantanal region has motivated the develop-
ment of international and interdisciplinary projects.
© Inter-Research 2015 · www.int-res.com*Corresponding author: jose.marengo@cemaden.gov.br
C 1324 31 Jul 2015 CE: VK TS: LJ PP: CF Marengo JA, Alves LM, Torres RR
Regional climate change scenarios in the Brazilian
Pantanal watershed
Jose A. Marengo
1,
*
, Lincoln M. Alves
2
, Roger R. Torres
3
1
National Center for Monitoring and Early Warning of Natural Disasters CEMADEN, São Paulo, Brazil
2
National Institute for Space Research INPE, São Paulo, Brazil
3
Federal University of Itajuba UNIFEI, Itajuba, Minas Gerais, Brazil
ABSTRACT: In the Brazilian Pantanal, hydrometeorological conditions exhibit a large interannual
variability. This variability includes the seasonality of floods and droughts which can be related to
land surface processes and to El Niño/La Niña. Based on regional climate change projections
derived from the Eta-HadGEM2 ES models with 20 km latitude–longitude resolution for the
RCP8.5 for 2071−2100, it is expected that there will be an annual mean warming of up to or above
5−7°C and a 30% reduction in rainfall by the end of the 21
st
century. As a consequence of higher
temperatures and reduced rainfall, an increased water deficit would be expected, particularly in
the central and eastern parts of the basin during spring and summer, which could affect the pulse
of the Paraguay River. While the changes projected by the Eta-HadGEM2 ES are consistent with
the changes produced by the CMIP5 models for the same scenario and time slice, we can affirm
that changes in the hydrology of the Pantanal are uncertain, because in a comparison of CMIP5
and Eta-HadGEM2 ES model projections, some show increases in rainfall and in the discharges of
the Paraguay Basin, while others show reductions.
KEY WORDS: Climate change · Pantanal · Hydrology · River levels · Rainfall
Resale or republication not permitted without written consent of the publisher
Contribution to the Theme Section ‘Xxxxx’
PROOF ONLY
Not for redistribution

Clim Res ■ ■
Two of the most important are the CLARIS LPB
(Boulanger et al. 2011) and SINERGIA (Girard et al.
2014), which are dedicated to the investigation of
impacts of natural climate variability and anthro-
pogenic changes in the La Plata basin, and to the
discussion of strategies for adaptation to these
changes.
The Pantanal functions as a large reservoir that
stores water from the surrounding plateaus during
the rainy season and then delivers it slowly to the
lower sections of the Paraguay River, delaying for
almost 6 mo the maximum flows to the Parana River,
thereby minimizing downstream flooding. As a re -
sult, any significant change in the rainfall pattern is
likely to have major impacts on the local ecology and
socioeconomic relations. The peak flood season for
the lower Paraná River basin is 2 to 3 mo earlier than
the flood season for the upper Paraguay River. With-
out the Pantanal, the 2 flood seasons would be simul-
taneous, with severe consequences for downstream
inhabitants.
Therefore, climate change could have severe im -
pacts on the hydrology of the Pantanal (Ioris et al.
2014), with serious socioecological consequences
(Bergier 2013). Furthermore, changes in climate
conditions in the region may also cause significant
disturbances in the functioning of the ecosystem,
mostly by altering precipitation and evapotranspira-
tion rates, which in turn may affect river flow
regime and floodplain inundation dynamics. The
impacts of climate change may even amplify and
worsen undesirable consequences of some human
interventions on hydrologic conditions of the basin
(Bravo et al. 2014).
This paper focuses on estimating the effects of cli-
mate change on climate and hydrology of the region
up to the end of the 21
st
century. For this we review
some of the model projections of climate generated
from the IPCC AR4 as derived by Bravo et al. (2014),
and the IPCC AR5 as derived by Torres & Marengo
(2013) and Marengo et al. (2014). With this back-
ground, we assess the regional climate change pro-
jections for the Pantanal region as generated from
the dynamic downscaling of the HadGEM2-ES
global IPCC AR5 model using the Eta regional model
run with a resolution of 20 km latitude–longitude
generated by Chou et al. (2014a,b). The changes in
rainfall, temperature, and water balance (precipita-
tion minus evaporation, P−E) are considered in order
to investigate comprehensively the possible impacts
of climate change over the Pantanal region in this
study. This paper is one of the contributions of the
CLARIS LPB project.
2. CLIMATE AND HYDROLOGY FEATURES
IN THE PANTANAL
The Pantanal is a semi-arid region. The rainy sea-
son begins in October and ends in April, bringing
monthly precipitation ranging from approximately
100 to 300 mm. In the dry period, monthly precipi -
tation ranges from 0 to 100 mm, with lower year-
to-year variability than in the rainy period. During
the years 1968−2000, annual average precipitation
ranged from 920 to 1540 mm, with a mean value of
1320 mm (Bravo et al. 2014). Rainfall shows inter -
annual variability with higher or lower rainwater
amounts that have caused either severe floods or pro-
nounced dry seasons that influence the flooding.
Large-scale climate phenomena such as El Niño, La
Niña, or the variability linked to the Atlantic Ocean
as well as regional-scale water balance, soil wetness,
and soil moisture storage influence the seasonality of
floods and droughts in the Pantanal (Bergier 2010).
Regarding long-term rainfall variability, Marcuzzo
et al. (2010) and Cardoso & Marcuzzo (2010) ana-
lyzed the monthly trends in precipitation from
1977−2006 from 12 rain gauge stations in the Brazil-
ian Pantanal and noted a small decrease in precipita-
tion with a pronounced inter-annual variability. The
La Plata Basin in general has seen an increase in
annual precipitation in the last 40 yr, on the order of
10% over most of the region, but in some places it
has reached 30% or more (Castañeda & Barros 1994).
Furthermore, the role of land surface processes in
the hydrology of the Pantanal was highlighted by
Clarke (2005). Collischonn et al. (2001) assessed the
flow records from the Paraguay, Paraná, Negro, and
Uruguay Rivers in the La Plata Basin, and rainfall
records from other parts of South America, and found
strong evidence of changes in the runoff regime of
the La Plata basin during the last 40 yr, not all of
which can be attributed to land-use change.
Analysis of the 105 yr record of annual flood peaks
of the Paraguay River at Ladário, downstream of the
Pantanal wetland, shows strong evidence of a serial
correlation between flood peaks in successive years,
and the sequence of annual flood peaks is well-
represented by a lag-1 autoregression (Clarke 2005).
Clarke (2005) found that no direct relation could be
established between Ladário flood peaks and the lim-
ited records of Pantanal rainfall and also showed that
there appears not to be any relation between Ladário
flood peaks and El Niño/La Niña events. Fig. 1 shows
that there is no consistency in the signals from El Niño
or La Niña and peaks or lows in the Ladário records.
The large volume of natural storage within a drainage
2

Marengo et al.: Regional climate change scenarios in Brazil
basin has a marked effect on its regime of hydrologic
extremes. This is somewhat res ponsible for the lower
climate predictability in the region, since soil moisture
and natural water storage are still not well represented
in climate models (Marengo et al. 2003).
Alho & Silva (2012) showed that during the period
from 1962/1963−1972/1973 the Pantanal was un -
usually dry, as shown in Fig. 1. This was then fol-
lowed by a long wet period, which lasted until at
least 2000. During the 11 yr dry period, there were no
records of flooding, except for 1965−1966, when the
observed water level was 16 cm above the 0 mark
(82.4 m in relation to sea level), and the majority of
the flooding peaks were between 1 and 2 m. Since
1973/1974, the Pantanal has been experiencing a
period of inundations, with the flood gauge register-
ing its peak at Ladário in 1988, with a 6.64 m reading.
This was considered the Pantanal’s greatest-ever in -
undation. This wet period, which has already lasted
for 38 yr, is the longest recorded for the region and is
different from the previous periods in that the
phreatic water (the upper surface of the soil, which
forms the water table) rises throughout the entire
year, which results in lower intra-annual and inter-
annual variation in droughts and floods. Further-
more, the position of the average river watermark
has remained between 3 and 4 m and the minimum
mark between 1 and 2 m for the greater part of each
year (Marengo et al. 2014)
In 2006, the Ladário station registered a level of
5.40 m, which was considered by Soares et al. (2008)
as the highest flood peak since 1997, but according to
Gonçalves et al. (2011), the year with the most mem-
orable floods due to high Paraguay River levels was
1988. During the record flow in 1988, about 95% of
the Pantanal plain was flooded. During the outflow of
2007, the Ladário station registered one of the lowest
minimum levels of the last 34 yr, viz. 88 cm on 3
November 2007. A dry period between 2010 and
2012 showed a water level of 85 cm; however, the
lowest on record was in 1964−1973 with 75 cm
(Fig. 1). According to Fantin-Cruz et al. (2011), the
highest flood was that of 1995, when the floodplain
was flooded to a mean depth of 2.56 m. The median
flood event (return period 2 yr) produced a mean
flood depth 1.80 m and lasted 119 d.
3. DESCRIPTION OF GLOBAL AND REGIONAL
MODELS USED TO DERIVE REGIONAL CLIMATE
CHANGE PROJECTIONS FOR THE PANTANAL
For the purposes of this study, we used the regional
climate change scenarios derived from the down -
scaling of the HadGEM2-ES (Collins et al. 2011)
global model via the Eta regional model, run at a hor-
izontal resolution of 20 km, as generated by Chou et
al. (2014a,b). Global climate models are the major
3
Fig. 1. Time series of the Rio Paraguay water levels (mm) at Ladario/Paraguay. Source: Agencia Nacional de Aguas, Brazil
(www.ana.gov.br). The occurrences of El Niño and La Niña are indicated by grey and yellow bars, respectively
(www.cptec.inpe.br)

Clim Res ■ ■
tool used to provide information on climate change
under different greenhouse gas emission scenarios;
however, the grid sizes of these models are about 200
to 100 km. Local features, such as topography, river
basins, and coastlines, may not be sufficiently cap-
tured in the simulations carried out by those global
models. The regional climate models play the impor-
tant role of downscaling the global climate simula-
tions to smaller grid sizes in the area of interest
where the impact studies can be carried out.
The following review by Chou et al. (2014a,b)
describes the global and regional models used, the
emission scenario Representative Concentration
Pathways (RCPs) used, and the projections from the
HadGEM2-ES global and the Eta-HadGEM2-ES
regional climate projections. The HadGEM2-ES is a
global climate model of earth system category devel-
oped by the Hadley Centre (Collins et al. 2011, Mar-
tin et al. 2011). The latitude–longitude resolution is
about 1.875 × 1.275°, and there are 38 levels in the
atmosphere. It has a dynamic vegetation scheme
with carbon cycle representation. A list of major
characteristics and references for this model can be
found in Table 9.A.1 in Chou et al. (2014a).
The Eta model has been adapted to run for long-
term integrations (Pesquero et al. 2010, Chou et al.
2012, Marengo et al. 2012). The dynamics of the
model are developed in the eta vertical coordinate
(Mesinger 1984), which is the most suitable to
operate in regions of steep orography such as the
Andes Cordillera in South and Central America.
The model updates the equivalent CO
2
concentra-
tion every 3 yr. Vegetation greenness varies
monthly, but the type of vegetation is kept the
same during the integration period. The model
does not have ocean dynamics. The sea surface
temperature is taken from each global model
output and is updated daily in the regional Eta
model. Initial soil moisture and soil temperature
come from the respective GCMs. Update of soil
conditions follows the NOAH land surface scheme
(Ek et al. 2003). Lateral boundaries are updated
with global model state variables at 6 h intervals.
The regional model resolution is approximately
20 km in the horizontal and 38 layers in the
vertical. The top of the model is at 25 hPa.
The model domain encompasses most of South
America and Central America and part of the adja-
cent oceans, but our analyses will be focused on the
Pantanal area. In this work, the version of the Eta
model updated by Mesinger et al. (2012) is adapted
for climate change studies and it is applied to de -
velop impact and vulnerability studies (Resende et al.
2011, Rodrigues et al. 2011, Matos et al. 2012) and for
the Brazilian Third National Communication to the
United Nations Framework on Climate Change Con-
vention (UNFCCC).
The emission scenarios used for the Eta-HadGEM2
ES runs were the RCP scenarios RCP8.5 and 4.5,
which correspond to the range from pessimistic to
optimistic (Van Vuuren et al. 2012). The RCP8.5
represents a larger radiative forcing and is similar
to a high-emission scenario (SRES A1 or A2), while
RCP4.5 represents an intermediate scenario, similar
to SRES A1B. The global climate models used in the
IPCC AR5, in general, have shown improvement
over the previous models used in IPCC AR4, in par-
ticular the simulations of precipitation over the tropi-
cal areas. The Eta-HasGEM2 ES model was run for
RCP4.5 and 8.5, for the time slices 2011−2040,
2041−2070, and 2071−2100, with the baseline of
1961−1990. However, in this study we focus on the
RCP8.5-based projections.
In the following, we show a review of simulations
and future climate change projections from the Eta-
HadGEM2 ES for South America and the La Plata
basin, followed by a review of projections from the
CMIP3 and CMIP5 models for the Pantanal region.
3.1. Present climate from the Eta-HadGEM2 ES
Chou et al. (2014a) indicated that the Eta-
HadGEM2 simulations of the present (1961−1990)
reproduce the general regional-scale climatological
features over the South American continent. How-
ever, temperature is generally underestimated in the
Eta simulations as shown in the spatial distribution
and the mean annual cycle, and the simulations also
show a large negative bias in precipitation in north-
ern Brazil during summer, with a positive bias in the
southern part of Amazonia in winter. Some error fea-
tures are inherited from the global model HadGEM2-
ES, such as the double precipitation band of the
ITCZ. Model bias over mountainous areas, either for
temperature or precipitation, is still uncertain due to
the scarcity of observations over those areas.
The amplitude of the annual cycle of precipitation
simulated by the Eta-HadGEM2 ES model is gener-
ally smaller than the respective driver global model
precipitation, especially during the austral summer-
time rainy season (DJF). This causes the Eta simula-
tions to reproduce a better annual cycle of precipita-
tion in the central and northeastern regions of Brazil,
where global model precipitation is excessive during
the rainy period.
4

Marengo et al.: Regional climate change scenarios in Brazil
In the Amazon region, precipitation is under -
estimated by the Eta-HadGEM2 ES and by the global
HadGEM2-ES models. Despite the errors shown in
the evaluation of the simulated present climate by
Chou et al. (2014a), the nested simulations contain
the major features of the South American clim -
atology.
3.2. Future climate from the Eta-HadGEM2 ES
For the future, Chou et al. (2014b) assessed the
4 sets of downscaling simulations based on the Eta
regional model forced by the HadGEM2-ES for the
2 RCP scenarios, 8.5 and 4.5. Focusing on the RCP8.5
scenario, there is a projected warming in the central
part of Brazil. In austral summer, there is a reduction
of precipitation in the central part of Brazil and in
northeastern Brazil, while there is an increase in the
southeastern part of the continent toward the end of
the century. In austral winter, a precipitation de -
crease is found in the northern part of South America
and in most of Central America. A major change is
the reduction in precipitation in southeastern Brazil
toward the end of the century. The northern part of
northeastern Brazil shows negative rainfall anoma -
lies in the RCP8.5 Eta-HadGEM2 ES scenario. The
frequency distributions of temperature and precipi -
tation show the inclusion of extreme high values as
the time slices advance toward the end of the 21
st
century.
In the La Plata Basin (LPB), a warming of about
3−5°C is projected during summer and winter by the
end of the century, while increases in precipitation of
about 2−4 mm d
−1
are projected for the LPB region in
summer. In winter this increase is about 1−3 mm d
−1
but is concentrated mostly over the coastal region.
Events of extreme heavy rainfall become more fre-
quent in southeastern South America, and in 2071−
2100, a reduction in consecutive dry days is detected.
This increase in rainfall extremes is consistent with a
wetter LPB by the end of the century, mainly in sum-
mer (Chou et al. 2014b).
4. ASSESSMENT OF GLOBAL AND
REGIONAL CLIMATE CHANGE PROJECTIONS
FOR THE PANTANAL BASED ON IPCC AR4
AND AR5 MODELS: A REVIEW
We use the following studies as the main refer-
ences for the climate change projections in the region
based on the ensemble of global climate models:
Bravo et al. (2014) for the CMIP3 IPCC AR4 global
models, and Torres & Marengo (2013), Marengo et al.
(2014), and Kirtman et al. (2013) for the CMIP5. The
CMIP3 models used the A2 and B2 emission scenar-
ios while the CMIP5 models used the RCP4.5 and 8.5.
Climate changes simulated in the CMIP3 and CMIP5
ensembles are not directly comparable because of
the differences in prescribed forcing agents (e.g. CO
2
and aerosols) between the SRES and RCP scenarios.
Furthermore, the models may respond differently to
a specific radiative forcing due to different model-
specific climate sensitivities. However, based on the
underlying radiative forcing (or CO
2
concentrations),
one can compare projected changes in the tempera-
ture and precipitation indices and provide an esti-
mate of uncertainty related to the different emission
scenarios.
4.1. Ensemble of CMIP3 models
For the A2 scenario in terms of areal average val-
ues over the basin, and considering the mean values
among all the CMIP3 models, the monthly tempera-
ture anomalies projected for scenario A2 for 2010−
2040 (Bravo et al. 2014) ranged from +0.88°C in Feb-
ruary up to +1.48°C in October. A pattern of positive
anomalies was very clear for the projections through-
out the entire year, despite the dispersion among
results of the models. For each month, the difference
among results of the 20 CMIP3 models was relatively
large, with more discrepancies in February, March,
and April, when both temperature increases and
decreases are projected, and in September and Octo-
ber, where there was the largest range of projected
temperature increases (from +0.29 to +3.3°C in Sep-
tember, and from +0.66 to +6.07°C in October).
The precipitation anomalies show large dispersion,
being projected as either an increase or decrease in
precipitation rates. Projected precipitation anomalies
considering areal average values over the basin
show that there are large anomalies predominating,
but with quite a large dispersion among the results of
the 20 AOGCMs. The largest anomalies projected for
the months of the dry season (JJA) are due to the rel-
atively low precipitation rates during these months.
Bravo et al. (2014) indicated that the dispersion
among the AOGCM results is considerably larger
during the low rainfall period as well as during the
rainy period. Projected anomalies of precipitation for
January (Scenario A2), the wettest month, present a
mean value of +3.5% and the interval defined by
mean ± SD is from

8.2% to 15.3%, with a maximum
5

Clim Res ■ ■
projected anomaly of 44.6% and a minimum of

11.4%. The values for the long-time horizon present
even more dispersion among the AOGCMs. For
instance, for January Scenario A2, the mean pro-
jected anomaly is +10.7% and the mean ± SD inter-
val is between

23.9% and 44.5%, while maximum
and minimum anomalies are

33.1% and 130.3%,
respectively.
4.2. Ensemble of CMIP5 models
In the following, we make an assessment of projec-
tions of changes in summer and wintertime tempera-
tures and rainfall from the CMIP5 models focusing on
the RCP8.5 only, because changes in the RCP4.5
show the same tendency but with lower magnitude.
Projected warming in the RCP8.5 varies between 4
and 7°C for the Pantanal, and the increases in air
temperature are more noticeable in both summer
(DJF) and fall (MAM) seasons, reaching up to 6°C in
2100, and varying between 3.5 and 9°C among mod-
els. In the near term (up to 2040), the warming could
reach 2−3°C and by 2070, it may reach 4−5°C.
Agreement among the CMIP5 models is shown in
Fig. 2C−F. For rainfall changes in the summer season
(DJF), 70% of the models show rainfall increases in
the northern and central parts of the basin, while in
6
Fig. 2. Annual cycle of (A) precipitation (mm d
−1
) and (B) temperature (°C) for the Brazilian Pantanal region as derived from
the IPCC AR5 models. Thick blue/black lines represent the CRU 1961−1990 observed climatology/mean of historical simula-
tions (simulation of the present); thick red lines represent the ensemble mean of the IPCC AR5 models for the 2071−2100
RCP8.5 scenario; thin red lines represent the individual model projections. (C−F) Agreement (%) among models in the change
of rainfall at the seasonal level. Agreement is represented as the percentage of models that show a similar tendency (direction
of change, not magnitude), as indicated in the color scale

Marengo et al.: Regional climate change scenarios in Brazil
the rest of the year, between 60 and 80% of the mod-
els show rainfall reductions. During the austral win-
ter (JJA), the model ensemble suggests rainfall re -
ductions between 60 and 70%. The tendency is
strong in the second half of the 21
st
century, but the
scatter among model members is extremely large,
suggesting high uncertainty in rainfall projections,
particularly for the dry season.
In sum, global models from both CMIP3 and
CMIP5 project warming for the medium- and long-
term horizons until 2100 that can reach up to 4°C or
more, while changes in rainfall during the summer
peak and during the winter dry season are uncertain,
as shown by the large intermodel divergence. Fur-
thermore, while the model ensemble shows total
rainfall as expected to decrease both in summer and
winter, and the possibility of increased soil moisture
deficiency, it is hard to make conclusions about pro-
jected changes in the flood pulse in the Pantanal
region for the future.
Over the La Plata basin, for both CMIP3 and
CMIP5 datasets, Marengo et al. (2014) showed an in -
crease in temperature throughout the basin and a
slight increase in precipitation located mainly in the
Brazilian state of Rio Grande do Sul, Uruguay, and
northeastern Argentina. The projections of increases
in temperature in the CMIP3 ensemble vary from 2°C
(SRES B2) to 3.5°C (SRES A2), while in CMIP5 this in -
crease can vary from 1.5°C (RCP2.6) to 5°C (RCP8.5).
Regarding precipitation, despite the enormous un -
certainties related to this variable, there is good
agreement among the CMIP3 and CMIP5 models for
a small increase of precipitation of 0.1−0.3 mm d
−1
,
mainly in austral summer.
5. REGIONAL CLIMATE CHANGE
PROJECTIONS FOR THE PANTANAL REGION
ETA-HADGEM2 ES
Climate change in regions that are naturally
stressed by low water availability will intensify eco-
logical impacts on aquatic systems (Roland et al.
2012). Since the Pantanal functions as a gigantic
flood regulation system for the Paraguay River water-
shed, alterations in rainfall can significantly affect
the system’s capacity to retain and control flood
events. This section is based on the projections from
the Eta-HadGEM2 ES at annual time scales for the
RCP8.5 until 2100; seasonal changes were also
assessed but are not shown here.
Projected annual mean warming (Fig. 3) in the
region varies from 2.5−3.5°C in 2011−2040, up to
above 5−7°C in 2071−2100, in agreement with the
projections by the CMIP5 models. As for precipita-
tion (Fig. 4), the Eta-HadGEM2 ES projects rainfall
reductions of the order of 10−20% in 2010−2040, and
of 30% by 2071−2100, with a strong variability
among time slices particularly in summer (not
shown). Projected changes from the Eta regional
model nested on the MIROC5 global model as shown
by Chou et al. (2014b) for the Pantanal also show
temperature increases and rainfall decreases during
summer and winter. These projections are consistent
with those derived from the Eta-HadCM3 for the A1B
scenario that was run with a resolution of 40 km
(Marengo et al. 2012).
Changes in P−E are analyzed to investigate the
possible impacts of climate change on water resource
conditions in the region (Fig. 5). P−E changes suggest
that the region will become drier in the future. There
is a broad signal that the magnitudes of the P−E will
decrease, with a mean value of approximately 40%
during the 21
st
century, particularly in the central and
eastern part of the basin, in agreement with negative
anomalies shown by Kirtman et al. (2013) based on
the ensemble of the CMIP5 models for the region.
Thus, we conclude that the P−E changes up to 2071
are consistent with higher temperatures and rainfall
reductions by the end of the 21
st
century, which sug-
gests a reduction in the water budget, expressed as
percentages (Fig. 5). This is also consistent with
Chou et al. (2014b), who, for the Eta-HadGEM2-ES
for 2100, detected reductions in consecutive wet days
and increases in consecutive dry days that are indica-
tors of dry spells and droughts for the region.
Monthly mean temperature anomalies projected for
the RCP8.5 scenario are given in Fig. 6. Clearly, there
is a small yearly cycle, but a much larger spread of
temperatures in the winter months, i.e. from May to
July, is also evident. There is also some indication that
the extremes (maximum and minimum values) are
changing with the season, and there is a clear in crease
in the yearly median value over time. The range of the
entire series is approximately +6°C in July to 8.5°C in
December for the end of the 21
st
century.
Fig. 7 shows projected precipitation anomalies con-
sidering areal average values over the basin. The
mean and range of values over the 3 time slices
appear to remain more or less the same, except for
the extreme values. The range of these series is quite
small compared to that of the monthly temperatures.
In most months (summer and spring), the simulated
extreme changes are more pronounced for both
maxi mum and minimum values. Overall, the rainfall
projections for the Pantanal show negative anomalies
7

Clim Res ■ ■
all year long, at a rate of approximately −1 mm d
−1
.
In sum, for the Pantanal region, future climate-
change scenarios predict rising temperatures and
alterations in seasonal and interannual weather ex -
tremes (including droughts, heat waves, and floods).
As pointed out by Roland et al. (2012), these changes
would favor harmful cyanobacterial blooms in eutro -
phic waters and enhanced vertical stratification of
aquatic ecosystems.
8
Fig. 3. Projected change in annual average temperature (°C) for (A) 2011−2040, (B,E,H) 2041−2070, and (C,F,I) 2071−2100,
relative to the reference 1961−1990, under the RCP8.5 scenario. Color scale is located below the panel

Marengo et al.: Regional climate change scenarios in Brazil
6. SUMMARY AND CONCLUSIONS
Hydrometeorological conditions in the Brazilian
Pantanal, which is recognized as a sensitive eco -
system due to its climate and other geographical fea-
tures, exhibit a large interannual variability, which
seems to be independent of El Niño/La Niña. The
river records at Ladario show decadal-scale varia-
tions, with a dry period through most of the 1960s
extending to the early 1970s.
Based on the regional climate change projections
derived from the Eta-HadGEM2 ES regional models
9
Fig. 4. As in Fig. 3, but for precipitation change (%)

Clim Res ■ ■
with 20 km latitude–longitude resolution for the
RCP8.5 for 2010−2100, future climate projections
show a mean annual warming varying from 2.5−
3.5°C in 2011−2040 up to above 5−7°C for 2071−
2100. For precipitation, the model projects mean
annual rainfall reductions on the order of 10−20% in
2010−2040, and 30% by 2071−2100. However, in
general we noted a reduction in rainfall by 2100, with
a strong variability among time slices, particularly in
summer. The CMIP5 models show an agreement of
10
Fig. 5. As in Fig. 3, but for precipitation−evaporation (indicator of the water balance) (%)

Marengo et al.: Regional climate change scenarios in Brazil
at least 70% of the models toward drier conditions in
the basin from fall to spring.
Regarding the water balance in the region, with
higher temperatures and rainfall reduction by the
end of the 21
st
century, a reduction would be ex -
pected in the water budget, expressed as P−E
(Fig. 4). The changes are more pronounced in the
central and eastern side of the basin and during
spring and summer and would affect the pulse of the
Paraguay River. They are consistent with the pro-
jected changes using the same Eta regional model
nested in the MIROC5 global model at annual and
seasonal levels shown by Chou et al. (2014b). How-
ever, as Bravo et al. (2014) suggested, changes in the
hydrology of the Pantanal are uncertain because of
land surface processes that may not be well repre-
sented by climate models. While most of the model
projections from regional and global models show
increases in rainfall and in the discharges of the
Paraguay Basin in austral summer, they also show
reductions during the rest of the year, with a possible
late rainfall onset and shorter rainy season.
Projections of the impacts of climate change on
wetlands, including effects in the Pantanal and its
watershed, may be still too divergent; these un -
certainties may make it insufficient for describing
climate change impacts in this ecoregion. However,
our results can allow the identification of research
11
Fig. 6. Monthly projected air temperature anomalies (°C)
over the Brazilian Pantanal area for RCP8.5: mean, mini-
mum, and maximum values from the Eta/HadGEM2-ES, for
(A) 2010−2040, (B) 2041−2070, and (C) 2071−2100. The box
represents the mean ± SD
Fig. 7. As in Fig. 6, but for rainfall anomalies (mm d
−1
)

Clim Res ■ ■
that needs to be carried out on climate and environ-
mental change, as well as on the design of strategies
that reduce the vulnerability of the watershed in the
face of climate change. Therefore, while there is the
potential for very large impacts on the hydrology, the
models are not yet able to give us useful information
on rainfall changes due to uncertainty in the rainfall
projections.
Knowledge of severe floods and droughts, which
characterize natural disasters, is fundamental for
wildlife management and nature conservation for the
Pantanal. In addition, human activities are also af -
fected, since cattle ranching and ecotourism are eco-
nomically important in the region; therefore, when
seasons with unusual floods or droughts occur, areas
with human settlements are impacted.
Lastly, as indicated by Petry et al. (2011), the devel-
opment of an ecological risk assessment is the first
step in understanding the Pantanal’s vulnerability to
climate change, beginning with identification and
assessment of existing stressors (i.e. non-climate
stressors); thus climate change projections would be
useful to identify which existing stressors will be
most important in the future, and also where and
how these stressors will occur. This makes it possible
to design and implement effective adaptation
actions.
Acknowledgements. The research leading to the results
reported here received funding from the European Commu-
nity’s Seventh Framework Programme (FP7/2007-2013)
under Grant Agreement no. 212492 (CLARIS LPB – A
Europe–South America Network for Climate Change As -
sessment and Impact Studies in La Plata Basin), Rede-
CLIMA, the National Institute of Science and Technology
(INCT) for Climate Change funded by CNPq Grant no.
573797/2008-0 and FAPESP Grant no. 57719-9, and the
CNPq-IRD Project Mudancas, variabilidade e tendencias do
clima no passado, Presente e future e desastres naturais nas
Regioes tropicais e Subtropicals do Brasil: observacoes e
Modelagem (PRIMO), Ref: 590172/2011-5.
LITERATURE CITED
Alho CJR, Silva JSV (2012) Effects of severe floods and
droughts on wildlife of the Pantanal wetland (Brazil) — a
review. Animals 2: 591−610
Bergier I (2010) River level sensitivity to SOI and NAO in
Pantanal and Amazonia. Proceedings 3
rd
Simpósio de
Geotecnologias no Pantanal. Embrapa Informática Agro -
pecuária/INPE, Cáceres, p 25−34
Bergier I (2013) Effects of highland land-use over lowlands
of the Brazilian Pantanal. Sci Total Environ 463–464:
1060−1066
Boulanger JP, Schlindwein S, Gentile E (2011) Metamorpho-
sis of the CLARIS LPB European project: from a mecha-
nistic to a systemic approach. CLIVAR Exchanges (Spec
Issue LPB) 57: 7−10
Bravo JM, Collischonn W, Paz AR, Allasia D, Domecq F
(2014) Impact of projected climate change on hydrologic
regime of the Upper Paraguay River basin. Clim Change
127: 27−41
Cardoso MRD, Marcuzzo FFN (2010) Mapeamento de três
decênios da precipitação pluviométrica total e sazonal do
bioma Pantanal. Proceedings 3
rd
Simpósio de Geotecno -
logias no Pantanal. Embrapa Informática Agro pecuária/
INPE, Cáceres, MT, Brazil, p 84−94
Castañeda E, Barros V (1994) Las tendencias de la precip-
itación en el Cono Sur de América al este de los Andes.
Meteorológica 19: 23−32
Chou SC, Marengo JA, Lyra AA, Sueiro G and others (2012)
Downscaling of South America present climate driven by
4-member HadCM3 runs. Clim Dyn 38: 635−653
Chou SC, Lyra A, Mourao C, Derezcynski C and others
(2014a) Evaluation of the Eta simulations nested in three
global climate models. Am J Clim Change 3: 438−454
Chou SC, Lyra A, Mourao C, Dereczynski C and others
(2014b) Assessment of climate change over South Amer-
ica under RCP 4.5 and 8.5 downscaling scenarios. Am J
Clim Change 3: 512−525
Clarke RT (2005) The relation between interannual storage
and frequency of droughts, with particular reference to
the Pantanal Wetland of South America. Geophys Res
Lett 32: L05402
Collins WJ, Bellouin N, Doutriaux-Boucher M, Gedney N,
Halloran P, Hinton T (2011) Development and evaluation
of an Earth-System model − HadGEM2. Geosci Model
Dev 4: 1051−1075
Collischonn W, Tucci CEM, Clarke RT (2001) Further evi-
dence of changes in the hydrological regime of the River
Paraguay: part of a wider phenomenon of climate
change? J Hydrol (Amst) 245: 218−238
Da Silva CJ, Girard P (2004) New challenges in the manage-
ment of the Brazilian Pantanal and catchment area. Wetl
Ecol Manag 12: 553−561
Ek MB, Mitchell KE, Lin Y, Rogers E and others (2003)
Implementation of Noah land surface model advances in
the National Centers for Environmental Prediction oper-
ational mesoscale Eta model. J Geophys Res 108: 8851
Fantin-Cruz I, Pedrollo O, Castro NMR, Girard P, Zeilhofer
P, Hamilton SK (2011) Historical reconstruction of flood-
plain inundation in the Pantanal (Brazil) using neural
networks. J Hydrol (Amst) 399: 376−384
Girard P, Boulanger JP, Hutton C (2014) Challenges of cli-
mate change in tropical basins: vulnerability of eco-
agrosystems and human populations. Clim Change 127:
1−13
Gonçalves HC, Mercante MA, Santos ET (2011) Hydrologi-
cal cycle. Braz J Biol 71: 241−253
Hamilton SK (1999) Potential effects of a major navigation
project (Paraguay-Paraná Hidrovía) on inundation in the
Pantanal floodplains. Regul Rivers Res Manag 15:
289−299
Hamilton SK (2002) Human impacts on hydrology in the
Pantanal wetland of South America. Water Sci Technol
45: 35−44
Ioris AAR, Irigaray CT, Girard P (2014) Institutional
responses to climate change: opportunities and barriers
for adaptation in the Pantanal and the Upper Paraguay
River Basin. Clim Change 127: 139−151
Kirtman B, Power SB, Adedoyin JA, Boer GJ and others
(2013) Near-term climate change: projections and pre-
12

Marengo et al.: Regional climate change scenarios in Brazil
dictability. In: Stocker TF, Qin D, Plattner GK, Tignor M
and others (eds) Climate change 2013. The physical sci-
ence basis. Contribution of Working Group I to the Fifth
Assessment Report of the Intergovernmental Panel on
Climate Change. Cambridge University Press, Cam-
bridge, p
Marcuzzo FFN, Faria TG, Cardoso MRD, Melo DCR (2010)
Chuvas no Pantanal brasileiro: análise histórica e
tendência futura. Proceedings 3
rd
Simpósio de Geotecno -
logias no Pantanal. Embrapa Informática Agro pecuária/
INPE, Cacéres, p 170−180
Marengo JA, Cavalcanti IFA, Satyamurty P, Trosnikov I and
others (2003) Assessment of regional seasonal rainfall
predictability using the CPTEC/COLA atmospheric
GCM. Clim Dyn 21: 459−475
Marengo JA, Chou SC, Kay G, Alves LM and others (2012)
Development of regional future climate change scenar-
ios in South America using the Eta CPTEC/HadCM3 cli-
mate change projections: climatology and regional
analyses for the Amazon, São Francisco and the Parana
River Basins. Clim Dyn 38: 1829−1848
Marengo JA, Sampaio G, Alves L (2014) Climate change
scenarios in the Pantanal. In: Bergier I, Assine ML (eds)
Dynamics of the Pantanal wetland in South America.
Springer, The Netherlands (in press)
Martin GM, Bellouin N, Collins WJ, Culverwell ID and oth-
ers (2011) The HadGEM2 family of Met Office Unified
Model Climate Configurations. Geosci Model Dev 4:
723−757
Matos AC, Dereczynski CP, Chou SC, Palmeira R (2012)
Investigação do comportamento dos ciclones no clima
futuro a partir do modelo regional Eta-HadCM3. Pro-
ceedings XVII Congresso Brasileiro de Meteorologia,
Gramado, p 1−5
Mesinger F (1984) A blocking technique for representation
of mountains in atmospheric models. Riv Meteorol Aero-
naut 44: 195−202
Mesinger F, Chou SC, Gomes JL, Dusan J and others (2012)
An upgraded version of the Eta model. Meteorol Atmos
Phys 116: 63−79
Pesquero JF, Chou SC, Nobre CA, Marengo JA (2010) Cli-
mate downscaling over South America for 1961–1970
using the Eta model. Theor Appl Climatol 99: 75−93
Petry P, Rodrigues ST, Neto MRB, Matsumoto MH and oth-
ers (2011) Ecological risk assessment for the Paraguay
River Basin: Argentina, Bolivia, Brazil, and Paraguay.
The Nature Conservancy Brazil and WWF-Brazil,
Brasilia
Resende NC, Giarolla A, Rodrigues DC, Tavares P, Chou SC
(2011) Ocorrência da doença ferrugem-do-café (Hemi -
leia vastatrix) em algumas regiões de São Paulo, baseada
nas projeções climáticas do modelo Eta/CPTEC (Cenário
A1B-IPCC/SRES). Proceedings XVII Congresso Brasilei -
ro de Agrometeorologia, Guarapari, p 1−3
Rodrigues DC, Tavares P, Giarolla A, Chou SC, Resende
NC, De Camargo MBP (2011) Estimativa da ocorrência
de temperatura máxima maior que 34°C durante o flo-
rescimento e maturação do cafeeiro baseado no modelo
Eta/CPTEC 40km (cenário A1B). Proceedings XVII
Congresso Brasileiro de Agrometeorologia, Guarapari,
p 54−57
Roland F, Huszar VLM, Farjalla VF, Enrich-Prast A, Amado
AM, Ometto JPHB (2012) Climate change in Brazil: per-
spective on the biogeochemistry of inland waters. Braz J
Biol 72: 709−722
Soares MTS, Soriano BMA, Santos SA, Abreu UGP, Bergier
I, Pellegrin LA (2008) Monitoramento do comportamento
do rio Paraguai no Pantanal Sul-Mato-Grossense −
2007/2008. Embrapa Pantanal, Corumbá
Torres RR, Marengo JA (2013) Uncertainty assessments of
climate change projections over South America. Theor
Appl Climatol 112: 253−272
Tucci CEM, Clarke RT (1998) Environmental issues in the la
Plata basin. Int J Water Resour Dev 14: 157−173
Van Vuuren DP, Bayer LB, Chuwah C, Ganzeveld L and oth-
ers (2012) A comprehensive view on climate change:
coupling of earth system and integrated assessment
models. Environ Res Lett 7: 024012
13
Editorial responsibility: Enrique Sanchez,
Toledo, Spain
Submitted: February 2, 2015; Accepted: June 29, 2015
Proofs received from author(s): •, 2015