Global and Planetary Change 100 (2013) 295–307

Contents lists available at SciVerse ScienceDirect

Global and Planetary Change

journal homepage: www.elsevier.com/locate/gloplacha

Warming and wetting signals emerging from analysis of changes in climate extreme
indices over South America
María de los Milagros Skansi a,⁎, Manola Brunet b, c, Javier Sigró b, Enric Aguilar b,
Juan Andrés Arevalo Groening d, Oscar J. Bentancur e, Yaruska Rosa Castellón Geier f,
Ruth Leonor Correa Amaya g, Homero Jácome h, Andrea Malheiros Ramos i, j, Clara Oria Rojas k,
Alejandro Max Pasten l, Sukarni Sallons Mitro m, Claudia Villaroel Jiménez n, Rodney Martínez o,
Lisa V. Alexander p, P.D. Jones c, q
a
Departamento Climatología, Servicio Meteorológico Nacional, 25 de Mayo 658 (C1002ABN), Ciudad Autónoma de Buenos Aires, Argentina
b
Centre for Climate Change, Department of Geography, University Rovira i Virgili, Av. Catalunya, 35, 43071, Tarragona, Spain
c
Climatic Research Unit, University of East Anglia, Norwich, NR4 7TJ, United Kingdom
d
Instituto Nacional de Meteorología e Hidrología (INAMEH), Carretera Hoyo de la Puerta, Parque Tecnológico Sartenejas, Edificio INAMEH. Municipio Baruta,
Estado Miranda, 1080 Venezuela
e
Dpto. Biometría, Estadística y Computación, Facultad de Agronomía, UDELAR, Av. Garzón 780, Montevideo, CP.12900, Uruguay
f
Unidad Climatologia, Institucion: Servicio Nacional de Meteorologia e Hidrologia (SENAMHI), Calle Reyes Ortiz No. 41 (Zona Central), La Paz, Bolivia
g
Grupo de Gestión de Datos y Red Meteorológica, Subdirección de Meteorología, Instituto de Hidrología, Meteorología y Estudios Ambientales, IDEAM,
Carrera 10 No. 20–30 Piso 6, Colombia
h
Dpto. de Climatología, Instituto Nacional de Meteorología e Hidrología (INAMHI), Calle Iñaquito No. N36-14 y Corea, Codigo Postal No. 16-310, Quito, Ecuador
i
Coordenação de Desenvolvimento e Pesquisa (CDP). Instituto Nacional de Meteorologia (INMET). Eixo Monumental, Via S1 Sudoeste. 70680-900, Brasília-DF, Brazil
j
Geophysics Centre of Évora (CGE), University of Évora, Portugal. Rua Romão Ramalho, 59. 7000-671. Évora, Portugal
k
Centro de Prediccion Numerica de la Direccion General de Meteorología, Servicio Nacional de Meteorologia e Hidrologia del Peru, Jr. Cahuide 785, Jesus Maria, Lima, Peru
l
Direccion Nacional de Aeronautica Civil–Dirección de Meteorologia e Hidrología Gerencia de Climatologia e Hidrología, Departamento de Climatología, Cnel.
Francisco Lopez 1080 c/ De La Conquista, Paraguay
m
Meteorological Service Suriname, Magnesiumstraat 41, Paramaribo, Surinam
n
Direccion Meteorológica de Chile, Subdepartamento de Climatología y Meteorología Aplicada, Seccion de Met. Aplicada — Oficina de Estudios, Av. Portales 3450, Estación Central,
Santiago, Chile
o
Centro Internacional para la Investigación del Fenómeno de El Niño (CIIFEN), Escobedo 1204 y 9 de Octubre, P.O. Box 09014237, Guayaquil, Ecuador
p
Climate Change Research Centre and Centre of Excellence for Climate Systems Science, University of New South Wales, Sydney NSW 2052, Australia
q
Center of Excellence for Climate Change Research/Dept of Meteorology, Faculty of Meteorology, Environment and Arid Land Agriculture, King Abdulaziz University, P. O. Box 80234,
Jeddah 21589, Saudi Arabia

a r t i c l e i n f o a b s t r a c t

Article history: Here we show and discuss the results of an assessment of changes in both area-averaged and station-based
Received 30 July 2012 climate extreme indices over South America (SA) for the 1950–2010 and 1969–2009 periods using
Accepted 11 November 2012 high-quality daily maximum and minimum temperature and precipitation series. A weeklong regional work-
Available online 20 November 2012
shop in Guayaquil (Ecuador) provided the opportunity to extend the current picture of changes in climate
extreme indices over SA.
Keywords:
daily temperature and precipitation data
Our results provide evidence of warming and wetting across the whole SA since the mid-20th century on-
quality control wards. Nighttime (minimum) temperature indices show the largest rates of warming (e.g. for tropical nights,
homogenization cold and warm nights), while daytime (maximum) temperature indices also point to warming (e.g. for cold
climate extreme indices days, summer days, the annual lowest daytime temperature), but at lower rates than for minimums. Both
temperature extreme indices change tails of night-time temperatures have warmed by a similar magnitude, with cold days (the annual lowest
precipitation extreme indices change nighttime and daytime temperatures) seeing reductions (increases). Trends are strong and moderate (mod-
ETCCDI erate to weak) for regional-averaged (local) indices, most of them pointing to a less cold SA during the day
South America
and warmer night-time temperatures.
Amazonia

⁎ Corresponding author at: Departamento Climatología, Servicio Meteorológico Nacional, 25 de Mayo 658 (C1002ABN), Ciudad Autónoma de Buenos Aires, Argentina. Tel.: +54
11 51676767x18259/18273; fax: +54 11 51676709.
E-mail addresses: [email protected] (M.M. Skansi), [email protected] (M. Brunet), [email protected] (J. Sigró), [email protected] (E. Aguilar),
[email protected] (J.A. Arevalo Groening), [email protected] (O.J. Bentancur), [email protected] (Y.R. Castellón Geier), [email protected] (R.L. Correa Amaya),
[email protected] (H. Jácome), [email protected] (A. Malheiros Ramos), [email protected] (C. Oria Rojas), [email protected] (A.M. Pasten),
[email protected] (S. Sallons Mitro), [email protected] (C. Villaroel Jiménez), [email protected] (R. Martínez), [email protected] (L.V. Alexander),
[email protected] (P.D. Jones).

0921-8181/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.gloplacha.2012.11.004
296 M.M. Skansi et al. / Global and Planetary Change 100 (2013) 295–307

northeastern Brazil Regionally-averaged precipitation indices show clear wetting and a signature of intensified heavy rain events
western South America over the eastern part of the continent. The annual amounts of rainfall are rising strongly over south-east SA
southeastern South America (26.41 mm/decade) and Amazonia (16.09 mm/decade), but north-east Brazil and the western part of SA have
experienced non-significant decreases. Very wet and extremely days, the annual maximum 5-day and 1-day
precipitation show the largest upward trends, indicating an intensified rainfall signal for SA, particularly over
Amazonia and south-east SA. Local trends for precipitation extreme indices are in general less coherent spatially,
but with more general spatially coherent upward trends in extremely wet days over all SA.
© 2012 Elsevier B.V. All rights reserved.

1. Introduction the network of stations used in global analysis (e.g. Alexander et al.,
2006; Vose et al., 2005 or Brown et al., 2008), are not globally uniform
The study of extreme weather and climate events is a current topic and contain irregular or limited data over northern Latin America and
of higher scientific and societal interest. It is being fuelled by relevant South America as a whole, Africa, parts of Australia, India, the Middle
scientific communities, including various climatological branches that East and Southern Asia, which restricts our ability to estimate changes
assess climate change (e.g. observational, modeling, adaptation and im- in extremes over these regions (Seneviratne et al., 2012: 123).
pact sectors). This issue has been addressed recently by the Intergov- For South America (SA hereafter), some effort has been made to as-
ernmental Panel on Climate Change (IPCC, 2012) in the Special Report sess changes in climate extremes based on temperature and precipita-
on Extremes (SREX). This has provided the most comprehensive global tion station data at the daily scale (for temperature extremes: Vincent
review and assessment on the relation between climate extremes, their et al., 2005; Alexander et al., 2006 and for precipitation extremes:
impacts and the strategies to manage associated perils. Haylock et al., 2006; Khan et al., 2007; Sheffield and Wood, 2008;
A number of issues, however, constrain our current understanding Grimm and Tedeschi, 2009; Dai, 2011; Mo and Berbery, 2011).
and scientific confidence in the observed changes in extremes. Parts of SA have been more intensively explored, such as SE SA for ob-
Among others, availability and accessibility of long-term and high- served changes in temperature extremes (e.g. Rusticucci and Barrucand,
quality climate series at the relevant time scales for assessing extremes 2004; Barrucand et al., 2008; Marengo and Camargo, 2008; Rusticucci
(e.g. daily and sub-daily) is one of the most serious gaps, particularly and Renom, 2008; Marengo et al., 2009; Renom, 2009; Tencer, 2010;
over some regions of the world (often called climate-data-sparse re- Rusticucci, 2012) or for precipitation extremes (e.g. Dufek and Ambrizzi,
gions), such as most of South America. The availability of climate series 2008; Dufek et al., 2008; Marengo et al., 2009; Pscheidt and Grimm,
is also limited temporally, since for most of the world the length of 2009; Penalba and Robledo, 2010; Llano and Penalba, 2011; Teixeira
digitized daily climate series (e.g. for temperature and precipitation) and Satyamurty, 2011). Other sub-regional studies are focused on NE
only goes back in time to the mid-20th century and for many regions Brazil for precipitation extremes (e.g. Santos and Brito, 2007; Silva and
is restricted to the 1970s onwards. In addition, there are also concerns Azevedo, 2008; Santos et al., 2009) and over western SA for temperature
regarding the quality and homogeneity of the available series, which extremes (Falvey and Garreaud, 2009) and for precipitation extremes
could compromise the robustness of assessed changes. Many countries (Dufek et al., 2008).
also restrict access to their higher temporal resolution time-series. In From these studies, there is a clear geographical imbalance in the as-
short, the quality and quantity of accessible climate series still limit sessments of one or another part of SA and in the number of stations
our understanding of the observed changes in climate extremes, partic- employed. Most previous studies have focused on southern SA, with
ularly over data-sparse regions (Trenberth et al., 2007: Appendix 3.B.2). limited studies and data over the northern half of SA. Most analyses
A number of international groups have made major efforts to ad- however point to observed changes in temperature extremes consistent
vance both knowledge of global changes in climate extremes and to pro- with warming when averaged over the whole continent but with re-
mote the recovery and development of climate data (i.e. ensuring data gional variations (Vincent et al., 2005: 5016–5020). However, while ex-
quality and homogeneity) over data-sparse regions. In this regard, the treme indices based on minimum (i.e. night-time) temperature have
ETCCDI1 has largely contributed to this effort by advancing knowledge warmed, those based on maximum (i.e. daytime) temperature show lit-
on changes in climate extremes through the formulation of a suite of tle change or have cooled, particularly over southern SA (Rusticucci,
27 core climate extreme indices calculated from daily temperature 2012, pp. 4–6).
and precipitation data (http://cccma.seos.uvic.ca/ETCCDI/list_27_indices. The scientific confidence in the observed changes over SA, there-
shtml). They have also promoted the analysis and monitoring of extremes fore, ranges from low to medium, depending on the region analyzed
around the world through organizing regional workshops in data-sparse (Seneviratne et al., 2012: Table 3.2, p. 194). There is low confidence
regions that have involved scientists from National Meteorological and in the assessed changes in extremes based on either daily maximum
Hydrological Services (NMHS) as part of ETCCDI's two-pronged approach or minimum temperature data over the northern half of SA, including
(Peterson and Manton, 2008, p. 1266). Amazonia, due to the irregular network in these regions. In the south-
Contributions from the ETCCDI to filling in gaps in data-sparse re- ern half of SA (including NE Brazil, south, SE and west Coast of SA)
gions and enhancing analyses of the global picture of changes in ex- there is medium confidence in the estimated extreme temperature
tremes (Trenberth et al., 2007, based on Alexander et al., 2006) have trends. In the case of heat waves and warm spells, the confidence is
helped to improve knowledge and understanding about how and how low over all SA, including southern SA, due to either insufficient evi-
much climatic extremes are changing under climate change. However, dence or to spatially varying trends.
A similar uncertain picture is apparent when assessing changes in
precipitation extremes over all SA, due to both the scarcity of studies
and spatially incoherent trends in either heavy events (e.g. those de-
1
Joint World Meteorological Organization (WMO) Commission for Climatology fined as daily precipitation >95th or 99th percentiles) or in dryness
(CCl)/World Climate Research Programme (WCRP) project on Climate Variability and (e.g. consecutive dry days — CDD, Palmer Drought Severity Index –
Predictability (CLIVAR)/Joint WMO–Intergovernmental Oceanographic Commission PDSI – indices) reducing the scientific confidence in the estimated
of the United National Educational, Scientific and Cultural Organization (UNESCO)
Technical Commission for Oceanography and Marine Meteorology (JCOMM) Expert
trends. There is medium confidence that there have been increases in
Team on Climate Change Detection and Indices (ETCCDI: http://www.clivar.org/ heavy precipitation events over Amazonia and many parts of NE Brazil,
organization/etccdi). but a few areas in the west (W) coast of SA indicate decreases or mixed
 M.M. Skansi et al. / Global and Planetary Change 100 (2013) 295–307 297

results (Seneviratne et al., 2012: Table 3.2, p. 194, based on Haylock et 2.2. Time-series quality control (QC), homogeneity testing and
al., 2006; Santos and Brito, 2007; Silva and Azevedo, 2008; Santos et homogenization
al., 2009).
The scientific confidence provided by the SREX report for changes in Time-series were quality controlled using RClimDex. This software
dryness in SA is low either due to spatially varying trends or inconsis- was developed by the ETCCDI and is freely available at http://cccma.
tent results among assessments. Slight reductions in the CDD index seos.uvic.ca/ETCCDI/software.shtml. We complemented RClimDex
have been estimated over Amazonia, but with inconsistent spatial with an additional QC procedure and software (also freely available
signals. However the opposite signal (tenuous increases in dryness) at http://www.c3.urv.cat/data1.html along with the user manual
has been found over southern SA and contrasting spatial signals and (Aguilar et al., 2010)). The time-series were subjected to the QC pro-
inconsistencies among studies over NE Brazil and W SA also return cedures by the participants either during or after the workshop. This
low confidence in the estimated trends in dryness (Seneviratne et al., involved identifying and documenting potential non-systematic er-
2012: Table 3.2, p. 194 based on Haylock et al., 2006; Dufek and rors and ensuring that the time-series were reasonably free of gross
Ambrizzi, 2008; Dufek et al., 2008; Sheffield and Wood, 2008; Llano errors, at the same time as ensuring internal, temporal and spatial
and Penalba, 2011; Dai, 2011). consistency of the records.
With the aim of improving both the spatial and temporal cover- RClimDex and the additional QC software provide several graphi-
age of former assessments (e.g. Vincent et al., 2005; Haylock et al., cal and numerical output files which can be used in tandem to further
2006) and enhancing the global picture of changes in climate ex- assess the quality of the data. Graphical output includes monthly and
tremes over SA, a week-long ETCCDI regional workshop was held interannual box-plots, which allow for the identification of outliers
in Guayaquil (Ecuador) in the Centro Internacional para la for RR, TX, TN and DTR series. Assessment can then be made as to
Investigación del Fenomeno El Niño (CIIFEN) in January 2011. This the reliability of data by using expert judgment and by consulting
workshop followed the “recipe” devised by the ETCCDI and enabled original data sources. Details of all the additional numerical and
a more extended assessment (both spatially and temporally) of graphical output is given in Aguilar et al. (2010: pp.4–6).
changes in climate extremes over SA. Once the QC exercise is complete, workshop participants can test
Therefore, this study aims to analyze both local and regional/ their time series for consistency or “homogeneity”. With the time-
sub-regional changes in annual temperature and precipitation series quality controlled at the workshop, the attendees were trained
extremes over SA. This is done by calculating a core set of 27 ETCCDI in the application of two homogeneity tests: (i) RHtestV3 method and
extreme indices from high-quality daily weather data and estimating software (Wang et al., 2010: section 5, see http://cccma.seos.uvic.ca/
trends. ETCCDI/software.shtml) and (ii) an application of the Standard Normal
In the next sections we provide details of the network and indices Homogeneity Test (Alexandersson and Moberg, 1997) run in R and
used for assessing recent changes, the methodology applied for ensur- called RSNHT (available at http://www.c3.urv.cat/data1.html; Aguilar,
ing the quality and homogeneity of the final series employed for esti- 2010). RHtestV3 was used to test homogeneity on the RR series and
mating local and regional annual trends. In addition, we discuss our RSNHT was used not only for testing homogeneity, but also for adjusting
findings and provide insights on the shortcomings that SA countries the TX/TN series at the monthly scale. The homogeneity testing of the RR
are still facing to gain a truly regional picture of the observed changes series was also supported by visual inspection of the indices and looking
in extremes. for unusual behavior. After homogeneity testing of the RR time-series, 28
records were rejected for having too many break points and thus only
2. Data, methods, indices and trend estimation 252 out of the original 280 RR series were used to compute the extreme
indices.
2.1. Rationale for the ETCCDI regional workshop The 261 (262) TX (TN) records were subjected to homogenization
by using the RSNHT software. On average 1.1 (1.2) breaks in TX (TN) se-
In a previous ETCCDI regional workshop held in Brazil (Vincent et al., ries were detected, validated and accounted for by interpolating month-
2005; Haylock et al., 2006) in 2004, a network of 68 (54) temperature ly adjustments returned by the RSNHT to the daily scale following the
(precipitation) time-series covering the period 1960–2000 were ana- procedure recommended by Vincent et al. (2002: pp. 1325–1326).
lyzed, which enabled the first regional assessment on changes in ex-
tremes over SA as a whole to be produced (Rusticucci, 2012: p. 4). 2.3. The network used, extreme indices and trend estimation
The low density of stations employed in both studies supported
the need for enhancing the spatial and temporal availability of The network brought by the attendees to the workshop is shown in
high-quality daily time-series since only a limited assessment of ob- Table A in the supplementary information. This table provides the sta-
served changes could be performed. The ETCCDI regional workshop tion names by country, start and end dates, geographical coordinates,
held in Guayaquil (Ecuador) in January 2011 presented an opportuni- elevation and the variables available at the station. Fig. 1 shows the
ty to extend the network for SA. Participants from mainly NMHS' location of the original station network assessed, depicting the final
in Argentina, Bolivia, Brazil, Chile, Colombia, Ecuador, Paraguay, number of stations used for assessing extremes (circles) and those
Peru, Surinam, Uruguay and Venezuela attended the workshop and rejected (triangles), along with the borders of the fours sub-regions
brought their best, longest and most complete digital daily maximum analyzed as adapted from the SREX report.
temperature (TX), minimum temperature (TN) and precipitation Following QC, homogeneity testing and homogenization, only 188
(RR) time-series for analysis. (252) temperature (precipitation) series were assessed to be suitable
The participants brought daily records for 261 (262) TX (TN) se- for computing extreme indices (those locations shown in Fig. 1);
ries and 280 RR records from a selection of data available in their although for percentile indices only a maximum of 145 stations were
national databanks, representing well distributed stations and cov- considered suitable. The network includes records spanning the whole
ering the main climatic types in their countries. During the work- of the 20th century up to 2010, but with only a few records going
shop itself only about 15% of records could be analyzed, so intense back to the 1900s (e.g. in Surinam for RR records) or back to the
post-workshop analysis was required. Many time-series were dis- 1930s and 1940s (e.g. in Argentina and Colombia for the three variables
covered to contain large data gaps or many missing values that and decades respectively, and Venezuela for RR for the latter decade).
compromised data completeness and their suitability for computing Although about 28% of the stations start in the 1950s, it is not until
extreme indices. This is common among South American national the late 1960s that about 94% of the stations have data (see Table A in
networks. the supplementary data). This latter period allows a larger number of
298 M.M. Skansi et al. / Global and Planetary Change 100 (2013) 295–307

SA, likely the best analyzed region in SA so far). For analysis of precipita-
tion extremes there has been much better coverage in previous analyses
10 AMZ (e.g. Penalba and Robledo, 2010, Fig. 2, p. 534 and Table 1, p. 535–536
over the La Plata Basin; Teixeira and Satyamurty, 2011, Fig. 2, p. 1915
for southern and SE Brazil), but those time-series do not extend to recent
0 years.
NEB
The authors are aware that even with the enhanced spatial and
temporal coverage presented here, the complex topography and
-10 large variety of climates in this continent, make it difficult to compre-
hensively analyze observed changes in climate extremes at lower
spatial scales (e.g. national, sub-regional). Our network is, neverthe-
Latitude

less, large enough to improve the global picture over all SA. This is
-20
particularly the case over the northern half of SA, but also in the west-
ern part of SA. The new study also improves the temporal extent over
all SA.
-30

2.3.1. The ETCCDI extreme indices

-40 As discussed elsewhere (e.g. Klein Tank et al., 2009; Zhang et al.,
2011; Zwiers et al., 2011), the ETCCDI extreme indices were defined
with the aim of both monitoring changes in “moderate” extremes
and for enhancing climate change detection studies given their high
-50
signal-to-noise ratio (Zhang et al., 2011: p. 854).
WSA SSA To compute the ETCCDI extremes requires long, continuous, quality
controlled and homogeneous daily time-series. This requirement is an
-80 -70 -60 -50 -40 -30
Longitude issue in many parts of the world, including SA, where there is a lack of
high-quality daily measurements covering several decades. Therefore,
Fig. 1. Location map showing the network of stations assessed, including those used for indices calculation using RClimDex, a number of conditions have to
(circles) and not used (triangles). Stations with temperature and precipitation data be met. An annual value of an index will not be calculated if there are
are shown in red, while green (yellow) have only precipitation (temperature) series.
more than 15 days missing in a year. In addition, the percentile-based
Solid (empty) circles are those considered homogeneous or homogenized (not homog-
enized temperature records, only precipitation series used). Boxes identify the four SA indices will only be calculated if at least 80% of the data are present in
sub-regions adapted from the SREX Report (see text for details). the reference period. These requirements for data completeness had
an impact on the final number of extreme indices computed from the
homogenized temperature series in this study, since only a maximum
of 170 (145) locations could compute temperature (percentile-based)
indices.
indices to be calculated annually from stations mainly located in
From the 27 core extreme or “moderately extreme” indices de-
the northernmost part of SA where data availability is very limited in
fined by the ETCCDI, we assess 13 (9) temperature (precipitation) in-
previous studies. This justifies the analysis periods used in this paper
dices in this assessment. Table 1 outlines index names, definitions,
i.e. 1969–2009 for the common analysis of local trends and the 1950–
units and the number of stations for which each index has been calcu-
2010 for the regional and sub-regional series of extreme indices
lated for both periods assessed. While the indices chosen can be cal-
assessment.
culated both annually and monthly by RClimDex, in this assessment
A visual comparison between our Fig. 1 and Figure 1 in Vincent et al.
we only consider the annual values.
(2005: p. 5015), Figure 1 in Haylock et al. (2006: p. 1494) and Fig. 3,
Percentile indices are calculated using the 1971–2000 base period,
plots a–f, in Marengo et al. (2009, p. 2248) highlights the improvement
in order that most series could be included in the trend analysis over
in the spatial coverage of the data and indices used in this study. An
the 1950–2010 period. To eliminate possible bias in the trend estima-
enhanced spatial resolution is especially evident over the northern
tion of the percentile-based indices associated with the existing inho-
half of SA, including the whole W SA. The density of stations particularly
mogeneities at the limits of the reference period, the RClimDex
over Brazil, Colombia, Surinam or Venezuela is somewhat better than
software follows the bootstrapping approach proposed by Zhang et
that used in previous studies and in Ecuador, Peru, Chile it is remarkably
al. (2005: pp. 1643–1644).
better. This enables us to expand the analysis to these parts of SA,
which have been highlighted in recent reviews as data-sparse areas
(e.g. Rusticucci, 2012, p.2; Seneviratne et al., 2012: Table 3.2, p. 194). 2.3.2. Linear trend estimation of indices
The improvement in coverage of this study can also be seen Although the RClimDex software also produces trend files for each
over southern SA when compared with previous SA assessments index, in this study the estimated change is explained by a linear trend
(e.g. Vincent et al., 2005; Haylock et al., 2006). Over the western coast fitted over two different periods: (i) a common period (1969–2009)
of SA, both studies have assessed relatively few numbers of records: used for assessing local trends and (ii) the period 1950–2010 for which
20 (16) temperature (precipitation) series. In this study we have ana- the linear trends have been estimated on a regional (all SA) and
lyzed 53 temperature and precipitation series over this sector, which sub-regional scales as defined by SREX (IPCC, 2012: Fig. 3–1, p. 123 and
also expands across Peru, Ecuador and Colombia where previous cover- Appendix 3.A-2. See also Fig. 1).
age was scarce. Trends were calculated annually by adapting Sen's (1968) slope
Other studies focused on parts of SA use similar or reduced coverage estimator, following the method proposed by Zhang et al. (2000) in
than this study (e.g. Falvey and Garreaud, 2009: Table 1; Rosenbluth a study of annual temperature and precipitation change over Canada.
et al., 1997: Table Ia–b pp. 69–70 and Fig. 1, p.71; Rusticucci and This more robust approach for trend estimation was adopted because
Barrucand, 2004: Fig. 1, p. 4110; Barrucand et al., 2008: Fig. 1; the indices have mostly non-Gaussian distributions and also because
Rusticucci and Renom, 2008: Fig. 1 in p. 1084 and pp. 1084–1085; daily data could contain large real outliers that could compromise the
Renom, 2009: p. 13 for temperature extremes over the southern part of results returned by the non-resistant least squares method.
 M.M. Skansi et al. / Global and Planetary Change 100 (2013) 295–307 299

Table 1 tabulated values in Kendall (1955). Trends have only been estimated
Temperature and precipitation indices from the ETCCDI analyzed in this assessment for an index if less than 25% of the annual values were missing. This
with associated definitions and units (for further details see also http://
cccma.seos.uvic.ca/ETCCDI/list_27_indices.shtml), along with the number of stations
has had an impact over the northern part of SA, particularly over Brazil,
for which each index has been computed for both periods: the 1950–2010 for Colombia and Venezuela for temperature indices. This highlights
area-averaged indices and 1969–2009 for station-based indices. All indices are calcu- the need to fill in gaps by recovering, digitizing and reconstructing
lated annually from January to December. Italics indicate those indices that are not long-term and high-quality climate records over these areas.
percentile-based (see Section 2.3.2 for details).
To provide a global picture of the sign and magnitude of the estimat-
ID Index name Indices definitions Units No. of stations ed changes in extreme indices over all SA and over its four sub-regions:
1969– 1950– Amazonia (region 7), the NE of Brazil (region 8), W SA (region 9) and SE
2009 2010 SA (region 10) (see Appendix 3.A-2 of the IPCC, 2012 for coordinates
TXx Highest Annual highest value of daily C deg 133 150
and Fig. 1 for the adapted borders), we have calculated five (one for
Tmax maximum temperature all SA and four for its regions) simple area-averaged indices from the
TNx Highest Annual highest value of daily C deg 136 148 available indices series estimated for each region. As stated, the indices
Tmin minimum temperature trends have been estimated locally for the common period 1969–2009,
TXn Lowest TmaxAnnual lowest value of daily C deg 133 148
which has been selected because it was the time interval that also
maximum temperature
TNn Lowest Tmin Annual lowest value of daily C deg 137 149 returned a larger number of trends with better cover the northernmost
minimum temperature part of SA. They are mainly used for assessing spatial coherency of the
TN10p Cold nights Percentage of days when % 131 144 emerging signals.
TN b 10th percentile from the days To minimize latitudinal/longitudinal and altitudinal effects on trend
1971–2000 reference period
TX10p Cold days Percentage of days when % 127 145
estimation for area-averaged indices, we created anomalies with re-
TXb 10th percentile from the days spect to a 1971–2000 reference period for each station series for all in-
1971–2000 reference period dices that are not based on percentiles and whose units are given in
TN90p Warm Percentage of days when % 131 145 absolute quantities (those shown in italics in Table 1). This makes the
nights TN >90th percentile from the days
assessments between stations more comparable given the rich climatic
1971–2000 reference period
TX90p Warm days Percentage of days when % 128 145 diversity in SA. Also, to adjust the variance bias associated with varying
TX> 90th percentile from the days the sample size in these regionally/sub-regionally averaged series
1971–2000 reference period over time, we applied the approach developed by Osborn et al. (1997:
DTR Daily Annual mean difference between C 121 170 pp. 92–93) to minimize this bias in the indices time-series.
temperature TX and TN Deg
range
SU25 No. summer Annual count of days when Days 126 170 3. Results and discussion
days TX > 25 °C
TR20 No. tropical Annual count of days when Days 100 141 In this section, we describe the results of the analysis carried out for
nights TN > 20 °C
assessing changes in annual temperature and precipitation extremes
WSDI Warm spell Annual count of days with at Days 145 145
duration least 6 consecutive days when over SA, both area-averaged and station-based. First, we examine the
index TX> 90th percentile from the spatially-averaged trends for the extreme indices and, second, we pro-
1971–2000 reference period vide the results for local trends to assess spatial coherency. Next, we dis-
CSDI Cold spell Annual count of days with at Days 145 144 cuss and put our findings in the context of previous assessments.
duration least 6 consecutive days when
index TN b 10th percentile from the
1971–2000 reference period 3.1. Regional and sub-regional trends
Rx1day The highest Annual maximum 1-day mm 169 238
1-day RR precipitation Table 2 (Table 3) shows the trend coefficients estimated, along with
amount
their significance levels, for each TX and TN (RR) area-averaged indices
Rx5day The highest Annual maximum consecutive mm 169 240
5-day RR 5-day precipitation for the 1950–2010 period at both global-all SA- and sub-regional levels.
amount The calculated ± standard errors for the 95% confidence interval of the
SDII Simple daily Annual total precipitation divided mm/ 173 238 extreme indices coefficient trend are given in brackets. Significant sig-
RR intensity by the number of wet days day nals of warming (wetting and intensified rainfall events) are evident
index (defined as
throughout the whole continent (mainly over the eastern part of SA, ex-
precipitation ≥ 1.0 mm) in the
year cluding NE Brazil).
R20 No. of heavy Annual count of days when daily Days 170 240 Over all SA, the TN-based indices record faster rates of warming than
RR days RR ≥20 mm TX-based indices. Strong reductions (increases) are estimated for cold
CDD Consecutive Maximum number of consecutive Days 173 241
nights (warm nights), while cold (warm) days shows moderate
dry days days with daily rainfall b 1 mm
CWD Consecutive Maximum number of consecutive Days 174 242
(weak) downward (upward) trends (significant at the 5% for the latter).
wet days days with daily rainfall ≥ 1 mm Tropical nights (TR20) are also warming at higher rates than summer
R95p Very wet Annual total precipitation when mm 171 241 days (SU25), with TR20 (SU25) recording strong (moderate) upward
days RR >95th percentile from the trends (Table 2). Also the annual coldest night and day and the warmest
1971–2000 reference period
night show a strong tendency toward higher temperatures, while the
R99p Extremely Annual total precipitation when mm 152 239
wet days RR >99th percentile from the warmest day shows no significant change. Reductions in the duration
1971–2000 reference period of cold spells (a proxy for cold waves) are also significant over all SA,
PRCPTOT Wet-days Annual total RR from wet days mm 174 244 but increases in warm spells (a proxy for heat waves) are weak and
annual RR >1 mm
do not reach statistical significance for the continent as a whole. As
amount
the TN-based indices change at higher rates than the TX indices, the an-
nual Diurnal Temperature Range (DTR) shows a moderate downward
Annual trends of all indices are tested for statistical significance at trend over all SA.
the 0.01 (0.05) confidence level for regional and sub-regional averaged Fig. 2 shows annual anomaly series for cold nights and days (Fig. 2a
indices (station-based trends) unless otherwise stated. The 95% confi- and c) and warm nights and days (Fig. 2b and d) averaged over all SA,
dence intervals for trend coefficients have also been estimated from depicting the warming signal over the continent as a whole. Fig. 3
300 M.M. Skansi et al. / Global and Planetary Change 100 (2013) 295–307

Table 2
Annual trends (in days/decade) for the period 1950–2010 for regionally and sub-regionally averaged temperature indices using a robust linear trend estimate along with the±standard
errors in brackets using a 95% confidence interval (see Section 2.3.2 for details). Bold (italic) indicates significance at 0.01 (0.05) levels. AMA stands for Amazonia, NEB for NE Brazil,
WSA for western South America and SESA for SE South America.

Index All SA AMA (region 7) NEB (region 8) WSA (region 9) SESA (region 10)

TXn 0.20 (0.11/0.28) 0.19 (0.09/0.29) 0.30 (0.14/0.45) 0.07 (−0.03/0.16) 0.20 (0.05/0.33)
TXx −0.05 (−0.15/0.06) 0.12 (0.04/0.18) 0.34 (0.23/0.46) 0.11 (−0.01/0.24) −0.14 (−0.27/−0.01)
TNn 0.20 (0.09/0.31) 0.32 (0.24/0.41) 0.56 (0.37/0.73) 0.19 (0.07/0.30) 0.13 (0/0.27)
TNx 0.18 (0.13/0.22) 0.24 (0.17/0.31) 0.34 (0.27/0.40) 0.22 (0.15/0.28) 0.13 (0.06/0.20)
TX10p −0.61 (−0.988/−0.43) −0.85 (−1.28/−0.35) −2.09 (−2.56/−1.62) −0.53 (−1.16/0.13) −0.52 (−0.83/−0.26)
TX90p 0.62 (0.10/1.04) 1.20 (0.67/1.73) 3.05 (1.94/4.09) 1.18 (0.61/1.86) 0.44 (−0.05/0.86)
TN10p −1.77 (−2.11/−1.49) −2.27 (−2.85/−1.81) −4.50 (−5.36/−3.68) −1.60 (−2.02/−1.10) −1.43 (−1.75/−1.13)
TN90p 1.54 (1.17/1.90) 2.28 (1.73/2.87) 4.02 (3.41/4.71) 1.60 (1.07/2.06) 1.30 (0.98/1.58)
SU25 1.26 (0.20/2.18) 1.60 (0.65/2.44) 6.94 (5.07/8.71) 1.46 (0.63/2.37) 0.68 (−0.52/2.14)
TR20 4.68 (3.81/5.61) 1.67 (1.32/2.07) 10.61 (8.61/12.11) 2.86 (2.02/4.04) 2.84 (2.13/3.58)
WSDI 0.09 (−0.44/0.72) 1.05 (0.31/2.13) 1.78 (0.65/3.90) 0.35 (0.05/0.66) −0.25 (−0.77/0.30)
CSDI −1.09 (−1.45/−0.83) −1.92 (−2.94/−1.09) −3.74 (−5.50/−2.12) −0.81 (−1.24/−0.45) −0.64 (−0.84/−0.44)
DTR −0.12 (−0.16/−0.07) 0.40 (0.30/0.56) −0.07 (−0.19/0.06) 0.17 (0.07/0.28) −0.03 (−0.10/0.04)

shows annual time-series for the annual coldest night (Fig. 3a) and wet days (about 36 mm more) and the highest 1-day precipitation
coldest day (Fig. 3b), tropical nights (Fig. 3c) and summer days (5 mm more intense). Weak increases in the number of consecutive
(Fig. 3d). All these indices clearly point to higher temperatures in these rainy days (0.24 days longer) and in the simple daily intensity (about
annual coldest events and an increasing in the number of days recording 0.50 mm more intense wet days) indices cannot explain the strong up-
values exceeding the 20 °C (25 °C) during night-time (daytime). ward trend in the total amount of annual rainfall, which is more likely
On sub-regional scales the largest trends have been estimated over related to the intensification seen in the heaviest events, although we
NE Brazil for cold nights (about 27% days decrease in frequency) and have not explored this further. Also, consecutive dry days are increasing
warm nights (about 24% days increase in frequency), followed by though not significantly, suggesting that a wetter continent might be
Amazonia for both indices by averaging 17 and 23 stations respectively. more likely associated with rainfall intensification rather than with an
Also increases (decreases) in warm (cold) days are remarkable over increment in the frequency of wet days >1 mm. Finally, the number
NE Brazil: 18% (13%) more (less) frequent and Amazonia with 7% (5%) of heavy rain (RR > 20 mm= R20) events also moderately increase
more (less) frequent. Moderate reductions (increments) in cold (warm) over all SA (Table 3).
nights and days are evident over W and SE SA, where a network of 26 Fig. 4 shows the time-varying annual anomalies for total annual rain-
and 79 stations respectively have been used. Tropical nights and summer fall (Fig. 4a), annual wettest consecutive 5-day precipitation (Fig. 4c),
days increase over the four sub-regions with lower rates of change for the very wet days (Fig. 4b) and extremely wet days (Fig. 4d). All of them
latter index. Tropical nights (summer days) change faster over NE Brazil show long-term, steady increases between 1950 and 2010. Annual pre-
and W and SE SA (NE Brazil, Amazonia and W SA). Cold spells show cipitation totals have seen a strong increase up to the mid-1970s
significant reductions over the four sub-regions with the highest rates followed by stagnation in higher amounts until 2010. Fig. 5 shows the
being estimated over NE Brazil (about 22 days shorter at present than annual anomaly series for R20 (Fig. 5a) and the annual wettest 1-day
in the 1950s), while warm spells increase significantly over NE Brazil (Fig. 5b), both recording increases over all SA.
(about 11 days longer), Amazonia and W SA, with the latter being signif- Sub-regional signals show SE SA (103 stations) as the region with the
icant at the 5% level. Finally, DTR increased strongly (slightly) over highest rates of change for annual rainfall (about 158 mm wetter than it
Amazonia (W SA), while weak and non-significant downward trends was in the 1950s), followed by Amazonia (77 stations and about
are estimated over the two remaining regions (Table 2). 97 mm higher), while W SA (NE Brazil) sees moderate (weak) but
Clear and significant wetting and intensified rainfall signals emerge non-significant reductions estimated from a network of 32 and 30 sta-
from the analysis of precipitation extreme indices averaged over all SA tions respectively. Very wet days and extremely wet days show upward
over 1950 to 2010 (Table 3). Annual total precipitation is strongly in- trends that are related to significant increases over Amazonia (SE SA)
creasing when averaged over the whole continent (about 92 mm with about 192 (87) mm higher amounts than in the 1950s. These events
more rain at present than in the 1950s), which is being accompanied are also increasing over W and NE Brazil but they are not statistical signif-
by high rates of upward trends in heavy events. This is particularly evi- icant. The annual wettest day only increases significantly at the 5%
dent for increases in events exceeding the 95th percentile (about level over SE SA (6.24 mm wetter) and Amazonia (4 mm wetter), but
92 mm more intense nowadays than in the 1950s), followed by the over W SA (NE Brazil) the estimated upward (downward) trends are
highest 5-day consecutive rainfall amounts (11 mm more), extremely non-significant. For the annual wettest consecutive 5-day events

Table 3
The same as Table 2, but for precipitation indices.

Index All SA AMA (region 7) NEB (region 8) WSA (region 9) SESA (region 10)

Rx1day 0.86 (0.33/1.37) 0.67 (0.09/1.1.9 −0.52 (−2.14/1.10) 0.63 (−0.31/1.64) 1.04 (0.09/1.97)
Rx5day 1.86 (0.91/2.72) 1.10 (0/2.13) 0.29 (−3.58/3.73) −0.78 (−3.00/1.55) 2.40 (0.73/3.99)
R20 0.23 (0.10/0.36) 0.09 (−0.20/0.35) 0.16 (−0.68/0.79) −0.10 (−0.34/0.11) 0.52 (0.28/0.79)
CDD 0.25 (−0.27/1.00) −0.23 (−1.04/0.62) −5.58 (−8.98/−2.68) 1.12 (−1.01/3.10) 0.41 (−0.34/1.34)
CWD 0.04 (−0.02/0.11) 0.08 (−0.02/0.16) −0.07 (−0.17/0.14) 0.02 (−0.17/0.22) 0.05 (0/0.10)
R95p 15.29 (11.42/19.50) 32.02 (24.02/39.77) 9.86 (−7.00/25.27) 4.55 (−4.77/12.15) 14.49 (7.07/21.15)
R99p 5.95 (3.86/7.52) 10.65 (7.50/13.75) 5.76 (−1.49/12.63) 1.45 (−2.27/5.52) 5.66 (3.11/8.33)
SDII 0.09 (0.03/0.15) −0.06 (−0.11/0) 0.01 (−0.16/0.18) 0 (−0.15/0.15) 0.19 (0.08/0.30)
PRCPTOT 15.40 (5.83/22.71) 16.09 (0.33/31.85) −1.42 (−41.71/40.63) −13.97 (−33.34/3.80) 26.41 (11.71/42.89)
 M.M. Skansi et al. / Global and Planetary Change 100 (2013) 295–307 301

1950

1960

1970

1980

1990

2000

2010

1950

1960

1970

1980

1990

2000

2010
25 25

20
a b
20

% days
% days
15 15

10 10

5 5
TN10p TN90p
0 0

25 25
c d

% days
20 20
% days

15 15

10 10

5 5
TX10p TX90p
0 0
1950

1960

1970

1980

1990

2000

2010

1950

1960

1970

1980

1990

2000

2010
Fig. 2. Annual time-series (1950–2010) of area-averaged temperature indices over all South America for cold nights (Fig. 2a), warm nights (Fig. 2b), cold days (Fig. 2c) and warm
days (Fig. 2d). The indices are smoothed with a 13-year Gaussian filter. See Section 2.3.2 for details on trend estimation and adjustment of variance bias associated with varying
sample size in the area-averaged indices.

significant increases are found over SE SA and Amazonia (the latter signif- found. A similar signal is found for heavy events (RR>20 mm), with a
icant at the 5% level), while over W SA (NE Brazil) they are decreasing moderate upward trend over SE SA (3 days more than in the 1950s),
(increasing) but non-significant. Consecutive wet days (CWD) show while in the other sub-regions except W SA, heavy rainfall events are in-
weak and non-significant upward trends over the four sub-regions except creasing but not significantly (Table 3).
NE of Brazil. CDD, a proxy for dryness, shows significant reductions
over the NE of Brazil (33 days longer than in the 1950s) and Amazonia 3.2. Local trends for temperature and precipitation extreme indices
(although non-significant) and over W and SE SA consecutive dry days
are increasing but not significantly. The SDII index increase only over SE A summary of the sign of the trend and its significance for all the
SA (about 3 mm more intense), while weak but non-significant increases locally estimated indices are given in Table 4. Indices that represent
are seen in the remaining sub-regions except W SA where no change is warming such as cold night reductions, warm night and tropical night
1950

1960

1970

1980

1990

2000

2010

1950

1960

1970

1980

1990

2000

2010

3 3
a b
2 2

1 1
C Deg

C Deg

0 0

-1 -1

-2 -2
TNn TXn
-3 -3

30 30
c d
20 20

10 10
C Deg
C Deg

0 0

-10 -10

-20 -20
TR20 SU25
-30 -30
1950

1960

1970

1980

1990

2000

2010

1950

1960

1970

1980

1990

2000

2010

Fig. 3. Annual anomaly (with respect to the 1971–2000 reference period) between 1950 and 2010 of area-averaged temperature indices for South America and for the annual
lowest night-time (Fig. 3a) and daytime (Fig. 3b) temperatures, tropical nights (Fig. 3c) and summer days (Fig. 3d).
302 M.M. Skansi et al. / Global and Planetary Change 100 (2013) 295–307

1950

1960

1970

1980

1990

2000

2010

1950

1960

1970

1980

1990

2000

2010
mm mm
200 400
a b
100
300

200
-100

PRCPTOT R95p
-200 100

20 150
c d
10
100

50
-10

Rx5day R99p
-20 0
1950

1960

1970

1980

1990

2000

2010
1950

1960

1970

1980

1990

2000

2010

mm mm

Fig. 4. The same as Fig. 3 but for total annual rainfall (Fig. 4a) and annual maximum consecutive 5-day precipitation (Fig. 4c), and the same as Fig. 2 but for very wet days (Fig. 4b)
and extremely wet days (Fig. 4d) indices.

increases have more significant than non-significant local trends. No mostly record increases, with a few spatially incoherent trends in
rain-based index shows more significant than non-significant trends. southern Paraguay and central Argentina (Fig. 7 upper left panel).
Fig. 6 shows local trends for cold nights and days and for warm nights Generalized and significant increases are estimated for summer days,
and days. Both generalized reductions (increases) in the frequency of except for a subset of stations (Fig. 7 bottom left panel). A stronger
cold (warm) nights and days over most of the South American locations warming signal is seen over SE Brazil and over southern SA with a sim-
are evident with a high spatial coherency of the signals, particularly for ilar spatial pattern to that estimated for tropical nights. TNn (upper
cold nights. Downward trends are strong over the northern and western right panel) and TXn indices (bottom right panel) show consistent
parts of SA, while SE SA also sees moderate and weak reductions, some and generalized changes toward higher values of both the annual
of them not significant. A few differences to this are the local non- lowest nighttime and daytime temperatures over all SA, particularly
significant trends seen over eastern Uruguay, south Argentina, Chile, in the northern half of the continent with most being moderate and
Paraguay, Peru and northwest Brazil (Fig. 6 upper left plot). Further re- significant upward trends. This indicates a consistent warming signal
search is required to assess whether this is due to the complex topogra- in warmest nights and days.
phy at these stations or whether homogenization issues could explain Trends (both upward and downward) in precipitation indices at
them. Similar spatial patterns and signals, although weaker, have been station locations are mostly non-significant. This much more contrast-
established for cold days (Fig. 6 upper right panel) with strong negative ing pattern compared to the temperature indices is expected given
trends in the northern part of SA, while the southern part sees mostly that precipitation has higher temporal and spatial variability. This char-
non-significant reductions. Similar patterns (strong increases in the acteristic, however, does not contradict the clear wetting signal that
north and weak in the southern parts of SA) are found for warm nights emerged from our assessment at the regional and sub-regional scales,
(Fig. 6 bottom left panel). Warm days show a general increase over the since averaging across locations increases the signal to noise ratio.
northern part of SA, while the southern part has mostly non-significant Fig. 8 shows rates of change for PRCPTOT (which can be used as base-
(both increasing and decreasing) trends (Fig. 6 bottom right plot). line to place observed changes in other precipitation extreme indices
Fig. 7 shows local trends for tropical nights, summer days and the in context), heavy events (such as the R95p, Rx1day, R99p or Rx5day)
annual lowest night-time and daytime temperatures. Tropical nights and CDD indices.

3 15

2
a b 10

1 5
Days

mm

0 0

-1 -5

-2 -10
R20 Rx1day
-3 -15
1950

1960

1970

1980

1990

2000

1950

1960

1970

1980

1990

2000

2010
2010

Fig. 5. The same as Fig. 3 but for the number of heavy rainy days >20 mm (Fig. 5a) and the annual wettest day (Fig. 5b).
 M.M. Skansi et al. / Global and Planetary Change 100 (2013) 295–307 303

Table 4 nights and summer days (Vincent et al., 2005: Fig. 3b p. 5018 and
Number of stations for which each index has returned significant negative and positive Fig. 2a p. 5017), although some disagreement in a few Argentinean,
trends (at the 5% level), along with non-significant trends for the annual temperature
and precipitation station-based indices over 1969–2009.
Paraguayan and Uruguayan locations can be seen for summer days.
Comparing findings for smaller scales (e.g. for Uruguay from
Index ID Negative Positive Non-significant Rusticucci and Renom, 2008: p. 1086, 1088), similar weak downward
TXx 3 41 89 (upward) trends for cold nights and days (warm nights) to our findings
TNx 1 54 81 are seen, although neither study finds statistical significance. Also our
TXn 0 28 105
findings point to weak and non-significant increases in warm days,
TNn 3 41 93
TN10p 91 1 39 while their assessment indicated decreases. However, the highlighted
TX10p 50 2 75 differences in trends between our study and the previous studies
TN90p 0 83 48 discussed can be explained not only by the different networks used,
TX90p 2 53 73 but also because of the differences in length of periods assessed.
DTR 31 6 84
Also, there is good agreement between our results for local trends
SU25 2 37 89
TR20 2 52 46 in precipitation indices when compared with those from Haylock
WSDI – 19 44 et al. (2006) for all SA, bearing in mind the differences highlighted
CSDI 1 5 10 above on the variety of locations and periods assessed. Most of the
Rx1day 3 6 161
local trends for PRCPTOT, Rx1day, Rx5day, R95p and CDD indices in
Rx5day 4 10 155
SDII 7 16 151 Haylock et al. (2006: Fig. 2, p. 1497) show strong similarities to our
R20 5 16 149 findings. However, for the R99p index, our results indicate general in-
CDD 3 16 154 creases across SA while the previous study showed mixed trends.
CWD 6 6 160 They show similar contrasting spatial patterns with mixed upward
R95p 3 12 156
and downward trends, although differences between statistical sig-
R99p 1 13 138
PRCPTOT 5 11 158 nificance of trends between both assessments must be highlighted,
since the previous study estimated more significant local trends
than those calculated by us.
For PRCPTOT (Fig. 8 upper left panel), both strong to moderate (mod- Other agreement also occurs in the annual evolution of area-
erate to weak) upward (downward) trends, mostly non-significant, are averaged precipitation indices between both assessments. Haylock
evident. In general there are widespread mostly non-significant increases et al., 2006 (Fig. 3 p. 1498 and Fig. 4 p. 1499) show time-series
over SA in very wet days and extremely wet days (Fig. 8 upper middle of R20 and R99 indices for the four quadrants of the continent (NW,
panel and upper right panel respectively). Very wet days have a less spa- NE, SE and SW) and found significant trends over the SE quadrant
tially consistent wetting signal than extremely wet days, since the latter for both indices, while in our assessment both indices show upward
records general upward trends over all SA with a few significant and and significant trends over SE SA, although they are stronger over
moderate trends. Changes in both extreme indices are clear and spatially Amazonia (a region with a non-significant trend in the previous study
coherent indicators of the intensification of the heaviest events over the due to the highly sparse data).
continent.
The annual wettest consecutive 5-day and 1-day precipitation have 4. Summary and outlook
less spatially consistent patterns (Fig. 8 bottom left panel and bottom
middle panel respectively). The dominant signal is a tendency toward In this study we have analyzed changes in temperature and pre-
higher amounts in both indices despite many moderate to low (low to cipitation extremes by using an extended network of daily quality
moderate) and non-significant downward trends in Rx5-day (Rx1-day). controlled, homogeneity tested or homogenized records over South
Finally, local trends for consecutive dry days (Fig. 8 bottom right panel) America covering the 2nd half of the 20th century up to 2010. The
show mostly upward trends, some significant with strong rates of change, time-series assessed, however, present a number of problems.
although downward non-significant trends are also present. Lack of multi-decadal time-series at the daily scale over most of the SA
countries is one issue which is still hampering our knowledge on how
3.3. Discussion and by how much extremes are changing under climate change. Another
serious problem in the continent is climate series completeness, since
Previous studies of South American temperature and precipitation large amounts of missing periods and values are a common characteristic
extremes have used a variety of networks and time periods for their of South American daily data. This has had a negative impact on the com-
assessments. This study can be used to highlight some of the similar- putation of the ETCCDI extreme indices used in our study, especially for
ities and differences with those other assessments where either sim- temperature-based indices in general and percentile-based indices in
ilar timescales or similar networks were used. particular.
Reasonable agreement with our findings for temperature indices Despite WMO Resolution 40 on the free exchange of meteorological
and those reported by Vincent et al. (2005) is found. This is particu- and related historical data, accessibility of long-term and high-quality
larly true for cold and warm nights and to a lesser extent for cold climate records with an appropriate time resolution is a major issue,
days mainly over southern and western SA (see Fig. 3c, 3d p. 5018 since data sharing is restricted due to national policies that preclude
and Fig. 2c p. 5017 in Vincent et al., 2005). However, this cannot data exchange. Access to these data is an essential requirement before
be stated for changes in warm days (Vincent et al., 2005: Fig. 2d, we can confidently detect or predict climate variability and change
p. 5017). Our findings point to strong warming over northern SA (e.g. Brunet and Jones, 2011: pp. 30–34; Thorne et al., 2011: pp. 4–6).
and weak upward trends over southern SA in contrast to the previous Therefore, there is still a need to promote data rescue and data develop-
study which showed mixed signals (both warming and cooling) over ment activities at both the international and national level in order to
all SA, especially over Argentina where we have mainly estimated weak improve observed and projected changes in extremes over SA.
upward trends. For the regionally averaged indices, there is good agree- Even given the above mentioned problems, this assessment has
ment for cold and warm nights (Vincent et al., 2005, Fig. 5c and 5d, improved our knowledge of the spatial and temporal changes in tem-
p. 5020 respectively), while cold and warm days (Fig. 5a and b p. 5020) perature and precipitation extremes over all SA and its regions than
show less vigorous warming than in this study. Good agreement is also was possible in previous studies. Nevertheless, we are aware that
found between both assessments for upward trends in tropical our study and findings are limited both in time and space, since we
304 M.M. Skansi et al. / Global and Planetary Change 100 (2013) 295–307

TN10P TX10P
10 10

0 0

-10 -10
Latitude

Latitude
-20 -20

-30 -30

-40
%days/10 yrs -40 %days/10 yrs
< -6 to -6
-20 < -6 to -6
-20
-6 to -2 -6 to -2
-2 to 0 -2 to 0
-50 0 to 2 0 to 2
2 to 6 -50 2 to 6
6 to
>= 6 20 6 to
>= 6 20

-80 -70 -60 -50 -40 -30 -90 -80 -70 -60 -50 -40
Longitude Longitude

10
TN90P 10
TX90P

0 0

-10 -10
Latitude

Latitude

-20 -20

-30 -30

-40 -40
%days/10 yrs %days/10 yrs
< -6 to -6
-20 < -6 to -6
-20
-6 to -2 -6 to -2
-2 to 0 -2 to 0
-50 0 to 2 -50 0 to 2
2 to 6 2 to 6
6 to
>= 6 20 6 to
>= 6 20

-80 -70 -60 -50 -40 -30 -80 -70 -60 -50 -40 -30
Longitude Longitude

Fig. 6. Local robust trends estimated annually for the 1969–2009 period for cold nights (upper left plot), cold days (upper right panel), warm nights (bottom left panel) and warm days
(bottom right panel), all showing warming. See Section 2.3.2 for details on trend estimation.

do not explore changes at the intra-annual scale (e.g. monthly and becoming substantially less cold, SA has experienced widespread de-
seasonal), or using the whole distribution of data, or provide results creases in cold extremes (e.g. cold nights and days, the annual lowest
at fine spatial scales (e.g. national and sub-national). TN and TX values, cold spells) along with increases, although less
In line with global assessments (Alexander et al., 2006; Donat marked, in warm extremes (e.g. warm days, the annual highest TX
et al., submitted for publication) which found that the world is and TN values or warm spells).
 M.M. Skansi et al. / Global and Planetary Change 100 (2013) 295–307 305

10 TR20 10
TNn

0 0

-10 -10

Latitude
Latitude

-20 -20

-30 -30

-40 °C/10 yrs

-40
days/10 yrs <-1.5
-3 to -1.5
< -6to -6
-10 -1.5 to -0.5
-6 to -2 -0.5 to 0
-2 to 0 0 to 0.5
-50 0 to 2 -50 0.5 to 1.5
2 to 6 >1.5
1.5to 3
>= 6 10
6 to

-80 -70 -60 -50 -40 -30 -80 -70 -60 -50 -40 -30
Longitude Longitude

TXn
10 SU25 10

0 0

-10 -10
Latitude
Latitude

-20 -20

-30 -30

°C/10 yrs
-40 days/10 yrs -40 <-1.5
-3 to -1.5
< -6to -6 -1.5 to -0.5
-10
-0.5 to 0
-6 to -2
0 to 0.5
-2 to 0
0.5 to 1.5
-50 0 to 2 -50 1.51.5
to 3
2 to 6 >=
>= 6 10
6 to

-80 -70 -60 -50 -40 -30 -80 -70 -60 -50 -40 -30
Longitude Longitude

Fig. 7. The same as Fig. 6 but for tropical nights (upper left panel), summer days (bottom left panel), and the annual lowest night time (upper right plot) and daytime (bottom right
plot) temperatures.

SA is also becoming wetter as a whole, with Amazonia and Many of the principal findings on global changes in climate ex-
SE SA leading the increases in the total amount of annual precipi- tremes have resulted from ETCCDI regional workshops. It is hoped
tation. This upward trend seems to be more related to intensifi- that in the near future that up-to-date data availability and accessibil-
cation of heavy rainfall (particularly over Amazonia and SE SA) ity will evolve towards accessible and user-friendly platforms such as
than to increases in the duration or frequency of consecutive wet those produced for the European Climate Assessment and Dataset
days. Significant increases in consecutive dry days also point to this (ECA&D) project for Europe (http://eca.knmi.nl/). There are potential
feature. opportunities to build up a South American portal using the same
306 M.M. Skansi et al. / Global and Planetary Change 100 (2013) 295–307

PRCPTOT R95P 10 R99P

10 10

0 0 0

-10 -10 -10

Latitude

Latitude
Latitude

-20 -20 -20

-30 -30 -30

mm/10 yrs mm/10 yrs mm/10 yrs
<-300
-75 to -75 <-300
-75 to -75
-40 -75 to -25
-40 <-300
-75 to -75 -40 -75 to -25
-75 to -25
-25 to 0 -25 to 0 -25 to 0
0 to 25 0 to 25 0 to 25
25 to 75 25 to 75 25 to 75
-50 75 75
>= to 300 -50 75 75
>= to 300 -50 75 75
>= to 300

-80 -70 -60 -50 -40 -30 -80 -70 -60 -50 -40 -30 -80 -70 -60 -50 -40 -30
Longitude Longitude Longitude

RX5DAY RX1DAY CDD

10 10 10

0 0 0

-10 -10 -10

Latitude

Latitude
Latitude

-20 -20 -20

-30 -30 -30

days/10 yrs
mm/10 yrs mm/10 yrs
< -6to -6
-16
< -6to -6
-20 < -6to -6
-20 -6 to -4
-40 -6 to -2 -40 -6 to -2 -40 -2 to 0
-2 to 0 -2 to 0 0 to 2
0 to 2 0 to 2 2 to 6
2 to 6 2 to 6 6 to6 20
>=
6>=to6 20 -50 6>=to6 20 -50
-50

-80 -70 -60 -50 -40 -30 -80 -70 -60 -50 -40 -30 -80 -70 -60 -50 -40 -30
Longitude Longitude Longitude

Fig. 8. The same as Fig. 6 but for annual total rainfall (upper left panel), very wet days (upper central panel), extremely wet days (upper right panel), annual maximum consecutive
5-day precipitation (bottom left panel), annual maximum 1-day precipitation (bottom central panel) and consecutive dry days (bottom right panel).

technology, software and experience as in ECA&D. This endeavor will CE110001028 and LP100200690. MB is supported by the European
need both adequate support and increased interaction with all South Community's Seventh Framework Programme (FP7/2007–2013)
American NMHSs to ensure adequate data provision to enable a ro- under Grant Agreement 242093 (EURO4M: European Reanalysis
bust climate service to a wide community of users. and Observations for Monitoring). PDJ acknowledges the support of
Finally, the climate series used in this study are accessible from the European Community's Seventh Framework Programme (FP7/
the websites of the Brazilian, Chilean and Venezuelan NMHSs, while 2007–2013) under Grant Agreement 212492 (CLARIS LPB: A Europe–
those belonging to the NMHSs in Bolivia, Colombia, Ecuador and Peru South America Network for Climate Change Assessment and Impact
will be made accessible through the WSACAD (Western South America Studies in La Plata Basin).
Climate Assessment & Dataset) portal once implemented on-line.
Finally, the series from Argentina, Paraguay, Suriname and Uruguay
References
NMHSs are accessible under request from the relevant NMHS.
Supplementary data to this article can be found online at http:// Aguilar, E., 2010. Manual de uso de RSNHT.R (software de homogenización del Center for
Climate Change, URV, Tarragona, España), Centre for Climate Change self-editing,
dx.doi.org/10.1016/j.gloplacha.2012.11.004.
Tarragona. (12 pp. Available at: http://www.c3.urv.cat/data/manual/manual_rsnht.
pdf and accessed on the 20th July 2012).
Aguilar, E., Sigró, J., Brunet, M., 2010. RCLIMDEX con funcionalidades extras de Control de
Acknowledgments Calidad. Manual de Uso, versión 1.0, Centre for Climate Change self-editing, Tarragona.
(20 pp. Available at: http://www.c3.urv.cat/data/manual/Manual_rclimdex_extraQC.r.
pdf and accessed on the 20th July 2012).
We thank to the WMO Education and Training program for
Alexander, L.V., Zhang, X., Peterson, T.C., Caesar, J., Gleason, B., Klein Tank, A.M.G.,
funding the Guayaquil Workshop under the coordination of the Haylock, M., Collins, D., Trewin, B., Rahimzdeh, F., Tagipour, A., Kumar Kolli, R.,
WMO WIS-Data Management Applications Division. The authors are Revadekar, J.V., Griffiths, G., Vincent, L., Stephenson, D.B., Burn, J., Aguilar, E.,
also grateful to CIIFEN and Centre for Climate Change (C3) for their Brunet, M., Taylor, M., New, M., Zhai, P., Rusticucci, M., Vazquez Aguirre, J.L.,
2006. Global observed changes in daily climate extremes of temperature and pre-
scientific guidance and leadership during and post workshop activi- cipitation. Journal of Geophysical Research-Atmospheres 111, D05109 http://
ties. LVA is supported by Australian Research Council grants dx.doi.org/10.1029/2005JD006290.
 M.M. Skansi et al. / Global and Planetary Change 100 (2013) 295–307 307

Alexandersson, H., Moberg, A., 1997. Homogenization of Swedish temperature data. 1. Rusticucci, M., 2012. Observed and simulated variability of extreme temperature events over
Homogeneity test for linear trends. International Journal of Climatology 17, 25–34. South America. Atmospheric Research 106 (2012), 1–17 http://dx.doi.org/10.1016/
Barrucand, M., Rusticucci, M., Vargas, W., 2008. Temperature extremes in the south j.atmosres.2011.11.001.
of South America in relation to Atlantic Ocean surface temperature and Southern Rusticucci, M., Barrucand, M., 2004. Observed trends and changes in temperature
Hemisphere circulation. Journal of Geophysical Research 113, D20111 http:// extremes over Argentina. Journal of Climate 17 (20), 4099–4107.
dx.doi.org/10.1029/2007JD009026. Rusticucci, M., Renom, M., 2008. Variability and trends in indices of quality controlled
Brown, S.J., Caesar, J., Ferro, C.A.T., 2008. Global changes in extreme daily temperature daily temperature extremes in Uruguay. International Journal of Climatology 28 (8),
since 1950. Journal of Geophysical Research-Atmospheres 113, D05115. 1083–1095.
Brunet, M., Jones, P.D., 2011. Emerging data rescue initiatives: bringing historical Santos, C.A., Brito, J.I.B., 2007. Análise dos índices de extremos para o semiárido do
climate data into the 21st century. Climate Research 47, 29–40 http://dx.doi.org/ Brasil e suas relações com TSM e IVDN. Revista Brasileira de Meteorologia 22 (3),
10.3354/cr00960. 303–312.
Dai, A., 2011. Drought under global warming: a review. Wiley Interdisciplinary Santos, C.A.C., Brito, J.I.B., Rao, T.V.R., Meneses, E.A., 2009. Tendências dos Índices de
Reviews: Climate Change 2 (1), 45–65. precipitação no Estado do Ceará. Revista Brasileira de Meteorologia 24 (1), 39–47.
Donat, M.G., Alexander, L.V., Yang, H., Durre, I., Vose, R., Dunn, R. Willett, K., Aguilar, E. Sen, P.K., 1968. Estimates of the regression coefficient based on Kendall's tau. Journal of
Brunet, M., Caesar, J., Hewitson, B., Jack, C., Klein Tank, A.M.G., Kruger, A.C., Maren- the American Statistical Association 63, 1379–1389.
go, J., Peterson, T.C., Renom, M., Oria Rojas, C., Rusticucci, M., Salinger, J., Sekele, S.S., Seneviratne, S.I., Nicholls, N., Easterling, D., Goodess, C.M., Kanae, S., Kossin, J., Luo, Y.,
Srivastava, A.K., Trewin, B., Villarroel, C., Vincent, L.A., Zhai, P., Zhang, X., Kitching, S., Marengo, J., McInnes, K., Rahimi, M., Reichstein, M., Sorteberg, A., Vera, C., Zhang, X.,
submitted for publication. Updated analyses of temperature and precipitation 2012. Changes in climate extremes and their impacts on the natural physical environ-
extreme indices since the beginning of the twentieth century: The HadEX2 dataset. ment. Managing the Risks of Extreme Events and Disasters to Advance Climate
Journal of Geophysical Research — Atmospheres. Change Adaptation. In: Field, C.B., Barros, V., Stocker, T.F., Qin, D., Dokken, D.J.,
Dufek, A.S., Ambrizzi, T., 2008. Precipitation variability in Sao Paulo State, Brazil. Ebi, K.L., Mastrandrea, M.D., Mach, K.J., Plattner, G.-K., Allen, S.K., Tignor, M.,
Theoretical and Applied Climatology 93 (3–4), 167–178. Midgley, P.M. (Eds.), A Special Report of Working Groups I and II of the Intergov-
Dufek, A.S., Ambrizzi, T., da Rocha, R.P., 2008. Are reanalysis data useful for calculating ernmental Panel on Climate Change (IPCC). Cambridge University Press,
climate indices over South America? Trends and Directions in Climate Research: Cambridge, UK, and New York, NY, USA, pp. 109–230.
In: Gimeno, L., Garcia-Herrera, R., Trigo, R.M. (Eds.), Annals of the New York Sheffield, J., Wood, E.F., 2008. Global trends and variability in soil moisture and drought
Academy of Sciences, New York, NY, USA, vol. 1146, pp. 87–104. characteristics, 1950–2000, from observation-driven simulations of the terrestrial
Falvey, M., Garreaud, R.D., 2009. Regional cooling in a warming world: recent temper- hydrologic cycle. Journal of Climate 21 (3), 432–458.
ature trends in the southeast Pacific and along the west coast of subtropical South Silva, A.G., Azevedo, P., 2008. Índices de tendências de Mudanças Climáticas no Estado
America (1979–2006). Journal of Geophysical Research 114, D04102 http:// da Bahia. Engenheiria Ambiental 5, 141–151.
dx.doi.org/10.1029/2008JD010519. Teixeira, M.S., Satyamurty, P., 2011. Trends in the frequency of intense precipitation
Grimm, A.M., Tedeschi, R.G., 2009. ENSO and extreme rainfall events in South America. events in southern and southeastern Brazil during 1960–2004. Journal of Climate
Journal of Climate 22, 1589–1609 http://dx.doi.org/10.1175/2008JCLI2429.1, 2009. 24 http://dx.doi.org/10.1175/2011JCLI3511.1.
Haylock, M.R., Peterson, T.C., Alves, L.M., Ambrizzi, T., Anunciação, Y.M.T., Baez, J., Tencer, B., 2010. Variabilidad de los eventos extremos de temperatura observados y
Barros, V.R., Berlato, M.A., Bidegain, M., Coronel, G., Corradi, V., Garcia, V.J., modelados en el sudeste de Sudamérica, y sus proyecciones ante un escenario de
Grimm, A.M., Karoly, D., Marengo, J.A., Marino, M.B., Moncunill, D.F., Nechet, D., cambio climático, Ph.D. thesis, Universidad de Buenos Aires, Buenos Aires (Argentina),
Quintana, J., Rebello, E., Rusticucci, M., Santos, J.L., Trebejo, I., Vincent, L.A., 2006. 100 pp. Available at: www.digital.bl.fcen.uba.ar and accessed on the 20th July 2012.
Trends in total and extreme South American rainfall 1960–2000 and links with Thorne, P.W., Willett, K.M., Allan, R.J., Bojinski, S., Christy, J.R., Fox, N., Gilbert, S., Jolliffe, I.,
sea surface temperature. Journal of Climate 19, 1490–1512. Kennedy, J.J., Kent, E., Klein Tank, A., Lawrimore, J., Parker, D.E., Rayner, N.,
IPCC, 2012. Managing the risks of extreme events and disasters to advance climate Simmons, A., Song, L., Stott, P.A., Trewin, B., 2011. Guiding the creation of a compre-
change adaptation. In: Field, C.B., Barros, V., Stocker, T.F., Qin, D., Dokken, D.J., hensive surface temperature resource for 21st century climate science. BAMS http://
Ebi, K.L., Mastrandrea, M.D., Mach, K.J., Plattner, G.-K., Allen, S.K., Tignor, M., dx.doi.org/10.1175/2011BAMS3124.1.
Midgley, P.M. (Eds.), A Special Report of Working Groups I and II of the Intergov- Trenberth, K.E., Jones, P.D., Ambenje, P., Bojariu, R., Easterling, D., Klein Tank, A., Parker, D.,
ernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK, Rahimzadeh, F., Renwick, J.A., Rusticucci, M., Solden, B., Zhai, P., 2007. Observations:
and New York, NY, USA (582 pp.). surface and atmospheric climate change. In: Solomon, S., Qin, D., Manning, M.,
Kendall, M.G., 1955. Rank Correlation Methods. Oxford Univ. Press, New York. Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L. (Eds.), Climate Change
Khan, S., Kuhn, G., Ganguly, A.R., Erickson III, D.J., Ostrouchov, G., 2007. Spatio-temporal 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth
variability of daily and weekly precipitation extremes in South America. Water Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge
Resources Research 43, W11424 http://dx.doi.org/10.1029/2006WR005384. University Press, Cambridge, UK, and New York, NY, pp. 235–336.
Klein Tank, A.M.G., Zwiers, F.W., Zhang, X., 2009. Guidelines on Analysis of Extremes Vincent, L.A., Zhang, X., Bonsal, B.R., Hong, W.D., 2002. Homogenization of daily
in a Changing Climate in Support of Informed Decisions for Adaptation, WMO-TD temperatures over Canada. Journal of Climate 15, 1322–1334.
No. 1500/WCDMP-No. 72, Geneva . (52 pp.). Vincent, L.A., Peterson, T.C., Barros, V.R., Marino, M.B., Rusticucci, M., Carrasco, G.,
Llano, M.P., Penalba, O.C., 2011. A climatic analysis of dry sequences in Argentina. Inter- Ramirez, E., Alves, L.M., Ambrizzi, T., Berlato, M.A., Grimm, A.M., Marengo, J.A.,
national Journal of Climatology 31, 504–513 http://dx.doi.org/10.1002/joc.2092. Molion, L., Moncunill, D.F., Rebello, E., Anunciação, Y.M.T., Quintana, J., Santos, J.L.,
Marengo, J.A., Camargo, C.C., 2008. Surface air temperature trends in southern Brazil Baez, J., Coronel, G., Garcia, J., Trebejo, I., Bidegain, M., Haylock, M.R., Karoly, D.,
for 1960–2002. International Journal of Climatology 28, 893–904 http://dx.doi.org/ 2005. Observed trends in indices of daily temperature extremes in South America
10.1002/joc.1584. 1960–2000. Journal of Climate 18, 5011–5023.
Marengo, J.A., Rusticucci, M., Penalba, O., Renom, M., 2009. An intercomparison of observed Vose, R.S., Easterling, D.R., Gleason, B., 2005. Maximum and minimum temperature
and simulated extreme rainfall and temperature events during the last half of the twen- trends for the globe: an update through 2004. Geophysical Research Letters 32,
tieth century: part 2: historical trends. Climatic Change 98 (3–4), 509–529. L23822.
Mo, K.C., Berbery, E.H., 2011. Drought and persistent wet spells over South America Wang, X.L., Chen, H., Wu, Y., Feng, Y., Pu, Q., 2010. New techniques for the detection and
based on observations and the U.S. CLIVAR Drought Experiments. Journal of adjustment of shifts in daily precipitation data series. Journal of Applied Meteorology
Climate 24, 1801–1820 http://dx.doi.org/10.1175/2010JCLI3874.1. and Climatology 49, 2416–2436 http://dx.doi.org/10.1175/2010JAMC2376.1.
Osborn, T.J., Briffa, K.R., Jones, P.D., 1997. Adjusting variance for sample-size in tree-ring Zhang, X., Vincent, L.A., Hogg, W.D., Niitsoo, A., 2000. Temperature and precipitation
chronologies and other regional mean time series. Dendrochronologia 15, 89–99. trends in Canada during the 20th century. Atmosphere-Ocean 38, 395–429.
Penalba, O.C., Robledo, F.A., 2010. Spatial and temporal variability of the frequency of Zhang, X., Hegerl, G., Zwiers, F., Kenyon, J., 2005. Avoiding inhomogeneity in percentile-
extreme daily rainfall regime in the La Plata Basin during the 20th century. Climate based indices of temperature extremes. Journal of Climate 18, 1641–1651.
Change 98 (3–4), 531–550. Zhang, X., Alexander, L.V., Hegerl, G.C., Klein-Tank, A., Peterson, T.C., Trewin, B., Zwiers, F.W.,
Peterson, T.C., Manton, M.J., 2008. Monitoring changes in climate extremes: a tale of 2011. Indices for monitoring changes in extremes based on daily temperature and
international collaboration. Bulletin of the American Meteorological Society 2008, precipitation data. Wiley Interdisciplinary Reviews: Climate Change 2, 851–870
1266–1271 http://dx.doi.org/10.1175/2008BAMS2501.1 (September). http://dx.doi.org/10.1002/wcc.147.
Pscheidt, I., Grimm, A.M., 2009. Frequency of extreme rainfall events in southern Brazil Zwiers, F.W., Alexander, L.V., Hegerl, G.C., Knutson, T.R., Naveau, P., Nicholls, N., Schär, C.,
modulated by interannual and interdecadal variability. International Journal of Seneviratne, S.I., Zhang, X., 2011. Community paper on climate extremes. Challenges
Climatology 29, 1988–2011 http://dx.doi.org/10.1002/joc.1799. in Estimating and Understanding Recent Changes in the Frequency and Intensity
Renom, M., 2009. Temperaturas extremas en Uruguay. Análisis de la variabilidad tem- of Extreme Climate and Weather Events, World Climate Research Programme
poral de baja frecuencia y su relación con la circulación de gran escala, Ph.D. thesis, Open Science Conference, 24–28 October 2011, Denver, CO, USA (45 pp. Available
Universidad de Buenos Aires, Buenos Aires (Argentina), 166 pp. Available at: www. at http://www.isse.ucar.edu/extremevalues/wcrpextr.pdf and accessed on the 20th
digital.bl.fcen.uba.ar and accessed on the 20th July 2012. July 2012).
Rosenbluth, B., Fuenzalida, H.A., Aceituno, P., 1997. Recent temperature variations in
southern South America. International Journal of Climatology 17 (1), 67–85.