Four sets of downscaling simulations based on the Eta Regional Climate Model forced by two global climate models, the HadGEM2-ES and the MIROC5, and two RCP scenarios—8.5 and 4.5, have been carried out. The objective of this work is to assess the climate change over South America based on the Eta simulations. The future changes are shown in timeslices of 30 years: 2011-2040; 2041-2070 and 2071-2100. The climate change response of the Eta simulations nested in HadGEM2-ES is larger than the Eta nested in MIROC5. Major warming area is located in the central part of Brazil. In austral summer, the reduction of precipitation in the central part and the increase in the southeastern part of the continent are common changes in these simulations, while the EtaHadGEM2-ES intensifies the decrease of precipitation in central Brazil, the Eta-MIROC5 expands the area of increase of precipitation in southern Brazil toward the end of the century. In austral winter, precipitation decrease is found in the northern part of South America and in most of Central America, whereas the reduction in southeastern South America is limited to near coastal region. The time series of temperatures show that warming trends are larger in the Eta-HadGEM2-ES than in the Eta-MIROC5 simulations. Heavier precipitation rates are projected in the Central-South of Brazil toward the end of the century. Increase in the length of consecutive dry days (CDD) in Northeast of Brazil and the decrease of consecutive wet days (CWD) in the Amazon region are common features in these simulations.
Assessment of the impacts of climate change in various socio-economic sectors is an activity necessary for planning at long range. The impacts and vulnerability are local scale issues that require more detailed climate information. Global Climate Models (GCMs) are the major tool to provide information of climate change under different greenhouse gases emission scenarios. However, the grid sizes of these models are about 200 - 100 km. Local features, such as topography, river basins, and coastlines may not be sufficiently captured in the simulations carried out by those GCMs. The regional climate models (RCMs) play the important role of downscaling the global climate simulations to smaller grid sizes in the area of interest where the impact studies can be carried out. The RCMs simulations with grid sizes of about a few tens of kilometers are a compromise between resolution and the long-term integration.
In South America, some RCM runs have adopted the Special Report on Emission Scenarios-SRES [
The Brazilian Second National Communication to the United Nations Framework Convention on Climate Change (UNFCCC) was mostly based on the simulations using the SRES scenarios based on the IPCC 4th Assessment Report (AR4). The recently issued IPCC 5th Assessment Report (AR5) is based on the Representative Concentration Pathway scenarios, RCP 8.5, 6, 4.5, and 2.6 Wm−2 radiative forcing scenarios, which correspond to the range from pessimist to the optimistic emission scenarios. The GCMs used in the AR5, in general, have shown improvement over the previous GCMs used in AR4, in particular the simulations of precipitation over the tropical areas [
In support of the Brazilian Third National Communication to the UNFCCC and of various studies of strategic themes, four sets of downscaling simulations based on the Eta RCM forced by two RCP scenarios―RCP 8.5 and RCP 4.5, and two GCMs, the HadGEM2-ES and the MIROC5, have been carried out. Evaluations of the nested regional simulations of the present climate against observations have been performed in a companion paper [
The objective of this work is to assess the climate change over South America based on the Eta model simulations nested in two GCMs and two RCPs scenarios. The assessment is carried out in terms of the annual cycle, frequency distribution, and some climatic extreme indicators.
In the AR4 [
Two GCMs are adopted for downscaling: the HadGEM2-ES and MIROC5. The HadGEM2-ES is a global climate model of earth system category developed by the Hadley Centre [
Regional Eta model has been adapted to run for long-term integrations [
Results are shown in future timeslices of 30 years: 2011-2040, 2041-2070 and 2071-2100. Climate changes are assessed with respect to the baseline period, 1961-1990, considered as the present climate. The assessment is primarily based on temperature and precipitation displayed in seasonal mean changes, annual cycle, time series, frequency distribution and climatic extreme indicators. Discussions for the three major Brazilian regions, North (NO), Northeast (NE), and the Central-South (CS) regions (
・ 2-m temperature
The mean change of 2-m temperature projected by the Eta-HadGEM and Eta-MIROC5 simulations, for the three timeslices: 2011-2040, 2041-2070 and 2071-2100, and for the two emission scenarios: RCP 4.5 and RCP 8.5 are shown in
・ Precipitation
The mean change of precipitation projected by the Eta-HadGEM and Eta-MIROC5 simulations, for the three timeslices: 2011-2040, 2041-2070, and 2071-2100, and for the two emission scenarios: RCP 4.5 and RCP 8.5 are shown in
In winter, in JJA, with the migration of ITCZ band toward northern latitudes (
・ 2-m temperature
The annual cycle of temperature is shown for RCP 4.5 (Figures 6(a)-(c)) and RCP 8.5 (Figures 6(d)-(f)) for three regions in Brazil: NO, NE and CS. The present climate simulation is also plotted for comparison of the changes. Despite some lags of phase of 1 month between the Eta-HadGEM and Eta-MIROC5, the amplitude of the annual cyclesis approximately maintained along the year in each timeslice. Therefore, the increase in temperature from the present to the each future timeslice is kept about the same magnitude along the year, but with small increase in the amplitude of the temperature cycle in the months around October and November. Larger warming occurs in RCP8.5 as expected in higher equivalent CO2 concentrations. The Eta-HadGEM produces warming values larger than Eta-MIROC5 in all three regions and periods. Larger warming is found in NO region, in RCP 8.5, which can reach about 8˚C on average, whereas the smallest warming is found in NE region in RCP 4.5 with warming of about 1.5˚C. In RCP 4.5, the largest changes in temperature occur between the present and the first period, 2011-2040, whereas in the RCP 8.5 the largest warming occurs between 2041-2070 and 2071-2100. No clear change of phase of the annual cycle is suggested in the projections of temperature for these regions.
・ Precipitation
The annual cycle of precipitation shows that the Eta simulations driven by MIROC5 produces more precipitation than the Eta driven by HadGEM2-ES during the rainy season, and generally less during the dry season, therefore the Eta-MIROC5 exhibits a larger amplitude of the annual cycle of precipitation for both RCPs in all three regions (
The time series of 2-m temperature simulated by the Eta-HadGEM and Eta-MIROC5, from 1961 until 2100, for DJF and JJA, and mean over the three regions (NO, NE and CS) are shown in
regions is larger in Eta-MIROC5 than in the Eta-HadGEM simulations. Clearly, the interannual variability of temperature increases in the future.
The time series of precipitation, for the two emission scenarios, in DJF and JJA, are shown in
The frequency distributions of daily mean temperatures for the three regions are shown in
The frequency distributions of daily-accumulated precipitation for the three regions are shown in
MIROC5 simulations produce more events of heavier precipitation than the Eta-HadGEM. In the RCP 4.5, little impact is noticed on the frequency distribution of Eta-HadGEM precipitation rates in the tropical regions of
NO and NE, but in the subtropical CS region, the Eta-HadGEM simulations exhibit more precipitation impacts with increase of frequency of heavy rains toward the end of the century. In the RCP8.5, the changes in the frequency distribution of precipitation rates are much more evident in all three regions. Heavier precipitation rates are attained in the first and middle timeslices, 2011-2040 and 2041-2070, in NE and CS regions. Despite the general predominant reduction of precipitation in the three regions shown during the rainy season, heavy precipitation events become more frequent in the future timeslices.
Four climatic extreme indicators based on daily precipitation are calculated for the downscaling projections: the total annual precipitation (PRCPTOT), the amount of precipitation from days that exceeded the 95th percentile of daily precipitation (R95p), the annual maximum consecutive dry days (CDD), and the annual maximum consec-
utive wet days (CWD). These indicators can suggest the changes in the characteristics of the total daily precipitation. Differences between the mean indicator at each future timeslice and mean indicator in the present climate timeslice (1961-1990) are shown in
The composite of the lower range of the change of PRCPTOT indicator shows a decrease in annual rainfall during the 21st century in most of the Amazon region and Southeast of Brazil. In those regions, the total annual precipitation can reduce in about 700 mm, with respect to the present climate by the end of the century. An increase in the total annual rainfall (blue areas) occurs over the southern part of South America and reaches anincrease over 500 mm by the end of the century. This result agrees with [
The change of heavy precipitation (R95p) indicator shows a small reduction in the amount of extreme precipitation rates over part of the Amazon, in central and Southeast of Brazil in all three timeslices. This reduction in the accumulated precipitation events is due to the increase toward the end of the century of the value of the 95th percentile of the daily precipitation rate. On the other hand, the increase of R95p indicator with respect to present climate occurs over SESA. The change grows larger toward the end of the 21st century. These changes of a drier climate in the North and Northeast of Brazil and a wetter climate in SESA are similar to the changes found by [
The composite of the upper range of changes of CDD indicator highlights an increase of consecutive dry days mainly over the Northeast of Brazil, where a semi-arid area is already present. This drier condition intensifies toward the end of the century. This region is also identified as a hotspot region in terms of water resources [
The Eta RCM is applied to generate four downscaling simulations of climate change. Two GCMs, the HadGEM2-ES and the MIROC5, and two AR5 IPCC scenarios, RCP 4.5 and RCP 8.5, force the simulations. The objective of this work is to assess the climate change over South America reproduced by these downscaling simulations.
Warming is projected in the entire continent, with larger amplitude in the Eta forced by HadGEM2-ES RCP 8.5 scenario. The warming starts in the central and southeastern Brazil and progresses strongly toward the northern part of the continent. Major change in precipitation is the reduction in Southeast of Brazil. Both Eta simulations driven by HadGEM2-ES and Eta driven by MIROC5 intensify the reduction of precipitation toward the end of the century. The region between the South and Southeast of Brazil exhibits the most mixed signs of precipitation changes. The northern part of Northeast of Brazil shows positive precipitation change with respect to 1961-1990 period in three simulations, but negative change in the RCP 8.5 Eta-HadGEM scenario. The frequency distributions of temperature and precipitation show the inclusion of extreme higher values as the timeslices advance toward the end of the 21st century. Events of extreme heavy rainfall become more frequent in the southeastern South America area and a reduction of heavy rainfall rate is found in the area of annual total precipitation reduction.
These simulations attempt to contribute to the assessment of impacts of climate change in different economic sectors. Generally, socio-economic impacts have local or regional scale, which makes the downscaling technique appropriate for the studies. The pessimistic and optimistic RCP scenarios and the use of two global models producing nested simulations of different error behaviors, attempt to include more possibilities and uncertainties in the assessment of the impacts of climate change. Previous impact studies over the region were based on the downscaling of climate change simulated from the same family of global models, HadAM3 and HadCM3 [
The authors thank: the Brazilian Ministry of Science, Technology, and Innovation for supporting the work through Global Environmental Facility funding (UNDP BRA/10/G32), the Secretariat for Strategic Affairs of the presidency of Brazil for additional funding, Martin Juckes from the British Atmospheric Data Centre for making available HadGEM2-ES dataset, and Seita Emori and Tokuta Yokohata from the Japanese National Institute for Environmental Studies for making available the MIROC5 dataset. Sin Chan Chou thanks the Brazilian National Council for Scientific and Technological Development (CNPq) for the grants 308035/2013-5 and 400792/ 2012-5.