Clim Dyn DOI 10.1007/s00382-014-2196-3

Projected impact of twenty-first century ENSO changes on rainfall over Central America and northwest South America from CMIP5 AOGCMs Daniel F. Steinhoff • Andrew J. Monaghan Martyn P. Clark

•

Received: 9 October 2013 / Accepted: 20 May 2014 Ó Springer-Verlag Berlin Heidelberg 2014

Abstract Due to the importance that the El Nin˜oSouthern Oscillation (ENSO) has on rainfall over the tropical Americas, future changes in ENSO characteristics and teleconnections are important for regional hydroclimate. Projected changes to the ENSO mean state and characteristics, and the resulting impacts on rainfall anomalies over Central America, Colombia, and Ecuador during the twenty-first century are explored for several forcing scenarios using a suite of coupled atmosphere– ocean global climate models (AOGCMs) from the fifth phase of the Coupled Model Intercomparison Project (CMIP5). Mean-state warming of eastern tropical Pacific sea surface temperatures, drying of Central America and northern Colombia, and wetting of southwest Colombia and Ecuador are consistent with previous studies that used earlier versions of the AOGCMs. Current and projected future characteristics of ENSO (frequency, duration, amplitude) show a wide range of values across the various AOGCMs. The magnitude of ENSO-related rainfall anomalies are currently underestimated by most of the models, but the model ensembles generally simulate the correct sign of the anomalies across the seasons around the peak ENSO effects. While the models capture the broad present-day ENSO-related rainfall anomalies, there is not a clear sense of projected future changes in the precipitation anomalies. Keywords ENSO
CMIP5
Precipitation
Teleconnections

D. F. Steinhoff (&)
A. J. Monaghan
M. P. Clark National Center for Atmospheric Research, P.O. Box 3000, Boulder, CO 80307-3000, USA e-mail: [email protected]

1 Introduction The El Nin˜o-Southern Oscillation (ENSO)—an ocean– atmosphere oscillation manifested in shifting sea surface temperatures across the tropical Pacific Ocean—has a profound impact on the regional hydroclimate of Central America and northwestern South America, particularly at interannual timescales. El Nin˜o events generally lead to persistently drier conditions throughout much of the region, whereas La Nin˜a events lead to wet spells (e.g., Hastenrath 1976; Yasunari 1987; Aceituno 1988; Rogers 1988; Ropelewski and Halpert 1987, 1989, 1996; Poveda and Mesa 1997; Gianninni et al. 2000; Waylen and Poveda 2002; Lyon and Barnston 2005; Poveda et al. 2006 and references therein; Ropelewski and Bell 2008; Grimm and Tedeschi 2009; Karnauskas and Busalacchi 2009; Grimm 2011; Poveda et al. 2011; McGlone and Vuille 2012; Hoyos et al. 2013). However, over Central America ENSOrelated rainfall anomalies reverse during the early rainy season (boreal late spring and early summer) following the mature ENSO phase (e.g., Waylen et al. 1996; Gianninni et al. 2000). Regional precipitation extremes are especially sensitive to ENSO fluctuations, and in turn have important societal impacts due to associated drought and flooding (e.g., Lyon 2003; Lyon and Barnston 2005; Grimm and Tedeschi 2009; Poveda et al. 2011; Chou et al. 2012). The remote forcing of ENSO on the region is due to shifting atmospheric teleconnection patterns that arise in response to quasi-periodic, evolving near-surface ocean temperature anomalies that result from ENSO variability (Grimm and Ambrizzi 2009). This teleconnection is represented by two distinct mechanisms (Giannini et al. 2001; Chiang et al. 2002). The first mechanism is an anomalous Walker circulation setup by the rearrangement of convection over the eastern tropical Pacific (Saravanan and Chang

123

D. F. Steinhoff et al.

2000; Chiang et al. 2000, 2002). During El Nin˜o (La Nin˜a) events, anomalous warm (cold) tropospheric temperatures spread throughout the tropics to stabilize (destabilize) the troposphere and suppress (enhance) convection. The second mechanism is related to the meridional SST gradient over the tropical Atlantic and its effect on positioning of the ITCZ (Hastenrath and Greischar 1993; Curtis and Hastenrath 1995; Enfield and Mayer 1997; Giannini et al. 2001; Chen and Taylor 2002; Gu and Adler 2006). During an El Nin˜o event, trade winds over the tropical North Atlantic weaken, resulting in weakened surface fluxes and warming SSTs. The weakened trade winds are caused in part by the ENSO-related Pacific-North American (PNA, Horel and Wallace 1981) pattern and the effect on geopotential height gradients over the tropical North Atlantic (Nobre and Shukla 1996). The warming SSTs throughout the mature phase of the ENSO event result in a northwarddisplaced Atlantic ITCZ during the early rainy season, and reversal of ENSO-related rainfall anomalies. Thus for the tropical Americas, the tropical North Atlantic SSTs are understood to primarily influence the early rainy season, whereas eastern tropical Pacific SSTs directly come into play in the later rainy season (e.g., Taylor et al. 2002; Mo and Berbery 2011; Wu and Kirtman 2011). ENSO-related displacement of the Hadley circulation over the tropical Americas has also been implicated towards rainfall variability over northern South America (Rasmusson and Mo 1993; Poveda and Mesa 1997). ENSO-related rainfall effects change for western Pacific (or ‘‘Modoki’’) El Nin˜o events, with weak anomalies during boreal spring generally reversing in sign from those of eastern Pacific El Nin˜o events during early summer, and extending through the following boreal autumn (Gouirand et al. 2013). Besides large-scale circulation patterns, ENSO also affects regional-scale climate processes. The Caribbean Low-Level Jet (CLLJ, Amador 1998; Martin and Schumacher 2011), a localized intensification of the easterly trade winds between northern South America and the Greater Antilles, is enhanced during boreal summer of El Nin˜o events. This reduces precipitation over the Caribbean through moisture flux divergence (e.g., Wang 2007), but increases precipitation along portions of the Caribbean coast of Central America due to orographic enhancement and convergence near the jet exit (Waylen et al. 1996). The westerly CHOCO (Chorro del Occidente Colombiano) jet (e.g., Poveda and Mesa 2000; Mapes et al. 2003; Poveda et al. 2006) near 5°N over the far eastern Pacific Ocean just west of Colombia, and convergence with the CLLJ, result in western Colombia being one of the rainiest places on earth (Poveda et al. 2014). During El Nin˜o events, the SST gradient in the eastern tropical Pacific Ocean is weaker, leading to a weaker CHOCO jet, weaker moisture transport, and reduced convective storm activity (Grimm and

123

Tedeschi 2009). ENSO impacts on regional hydroclimate also extend to a chain of processes involving land–atmosphere feedbacks through perturbations in precipitation, soil moisture, and evapotranspiration (Poveda and Mesa 1997), affecting the rate of moisture recycling on the eastern (Amazonian) side of the Andes (Poveda et al. 2011). Changes in ENSO characteristics—such as its mean state, amplitude, frequency, and the location of its maximum anomalies—are expected to affect the teleconnection patterns described above and subsequently precipitation over northwestern South America and Central America (e.g., Stevenson et al. 2012a). Potential changes to ENSO for a variety of future greenhouse gas (GHG) forcing scenarios have been evaluated from the third phase of the Coupled Model Intercomparison Project (CMIP3), which supported the Fourth Assessment Report (AR4) of the Intergovernmental Panel on Climate Change (IPCC, IPCC 2007). Projected changes in ENSO amplitude from previous studies are mixed, with a wide range of model behaviors (van Oldenborgh et al. 2005; Guilyardi 2006; Meehl et al. 2006; Merryfield 2006; Yeh et al. 2006). Changes are attributed to either thermocline effects or the width of the ENSO zonal wind response. In terms of changes to the frequency of ENSO occurrence, Merryfield (2006) and An et al. (2008) identify a decreasing ENSO period in a future warming climate. However, Guilyardi (2006) finds no evidence of changes in ENSO periodicity among 23 CMIP3 AOGCMs. Cai et al. (2014) estimate a doubling of the occurrence of extreme El Nin˜o events, resulting from enhanced warming of equatorial SSTs (see Sect. 4). Even though there is no consensus of AOGCM projections of future changes in ENSO characteristics such as amplitude or period under increasing GHG scenarios, the mean ENSO background state in the tropical Pacific is projected to change towards weaker trade winds, enhanced warming of SSTs nearest the equator, a shoaling thermocline, and steeper thermal gradients across the thermocline (e.g., Collins et al. 2010). It is not clear whether the lack of consensus on future ENSO characteristic changes is due to model deficiencies or because the changes to ENSO characteristics will likely be small. Still, robust changes to ENSO-related rainfall variability are expected for the tropical Pacific (and presumably other regions as well), despite uncertainty in ENSO characteristics (Power et al. 2013). Regardless of whether the ENSO amplitude and period change, it is possible that ENSO teleconnections may be altered due to change in the mean ENSO state (e.g., Stevenson et al. 2012a, b) and due to remote climate changes in regions influenced by ENSO (Vecchi and Wittenberg 2010). If teleconnection patterns associated with ENSO are altered, precipitation characteristics may change in Central America and northwestern South America where ENSO is a key climatic modulator (Sheffield and Wood 2008;

Projected impact of twenty-first century ENSO

Seager et al. 2012). In this paper we employ an ensemble of recently-completed coupled AOGCM experiments that were run as part of the CMIP5 (Taylor et al. 2012) to (1) examine changes in the mean state and characteristics of ENSO and (2) assess how changes in ENSO may impact the hydroclimate of northwestern South America and Central America. The CMIP5 experiments support the Fifth Assessment Report (AR5) of the Intergovernmental IPCC. Yeh et al. (2012) note differences in tropical Pacific SST trends between simulations from CMIP5 and the earlier third phase of CMIP (CMIP3; Meehl et al. 2007),

resulting in differences in ENSO characteristics between the two model sets. Therefore, updates to ENSO characteristics and associated rainfall teleconnections with the CMIP5 models are warranted.

2 Models, data, and methods Monthly output from 15 CMIP5 AOGCMs (Taylor et al. 2012) that support the IPCC AR5 is used in this study. The model output is obtained from the Earth System Grid—

Table 1 List of models used in this study Model

Institution

Grid Spacing # Vertical Levels

# Ens. Members

BCC-CSM1.1

Beijing Climate Center, China Meteorological Administration

T42

3,1,1,1

National Center for Atmospheric Research

0.9°Lat, 1.25°Lon

CCSM4

26 6,6,6,6

26 CESM1-CAM5

National Center for Atmospheric Research

0.9°Lat, 1.25°Lon

3,3,3,3

30 CSIRO-Mk3.6.0

FGOALS-s2 GFDL-ESM2G

Commonwealth Scientific and Industrial Research Organisation in collaboration with the Queensland Climate Change Centre of Excellence

T63

LASG, Institute of Atmospheric Physics, Chinese Academy of Sciences

1.66°Lat, 2.81°Lon

Geophysical Fluid Dynamics Laboratory

2.0°Lat, 2.5°Lon

10,10,10,10

18 3,1,1,1

26 1,1,1,1

24 GFDL-ESM2M

Geophysical Fluid Dynamics Laboratory

2.0°Lat, 2.5°Lon

1,1,1,1

24 GISS-E2-R

NASA Goddard Institute for Space Studies

2.0°Lat, 2.5°Lon

6,1,1,1

40 HadGEM2-ES

Met Office Hadley Centre

1.25°Lat, 1.875°Lon

4,4,4,4

38 IPSL-CM5A-LR

Institut Pierre-Simon Laplace

1.875°Lat, 3.75°Lon

6,4,1,4

39 MIROC5

MIROC-ESM

MIROC-ESM-CHEM

MRI-CGCM3

Atmosphere and Ocean Research Institute (The University of Tokyo), National Institute for Environmental Studies, and Japan Agency for Marine-Earth Science and Technology

T85

Japan Agency for Marine-Earth Science and Technology, Atmosphere and Ocean Research Institute (The University of Tokyo), and National Institute for Environmental Studies

T42

Japan Agency for Marine-Earth Science and Technology, Atmosphere and Ocean Research Institute (The University of Tokyo), and National Institute for Environmental Studies

T42

Meteorological Research Institute

T159

5,3,3,3

40

3,1,1,1

80

1,1,1,1

80

5,1,1,1

48 NorESM1-M

Norwegian Climate Centre

0.9°Lat, 1.25°Lon

3,1,1,1

26 Number of ensemble members in fourth column refer to the number of realizations used in the Historical, RCP2.6, RCP6.0, and RCP8.5 simulations

123

D. F. Steinhoff et al.

Program for Climate Model Diagnosis and Intercomparison (ESG-PCMDI) gateway at Lawrence Livermore National Laboratory, http://pcmdi3.llnl.gov/esgcet/home.htm. Information about each of the models is presented in Table 1. The models chosen have output available for the four simulations used in this study. The first is the ‘‘Historical’’ simulation, forced by observed atmospheric composition changes (both natural and anthropogenic), covering the period 1861–2005 in all of the models. Twenty-first century conditions are represented by three additional simulations, called ‘‘Representative Concentration Pathways’’ (RCPs), that are denoted by the globally-averaged top-ofthe-atmosphere radiative imbalance (in W m-2) in 2100 (Moss et al. 2010). In terms of CO2 concentrations, as previous IPCC reports are based on the Special Report on Emissions scenarios (SRES), RCP2.6 is well below B1, RCP6.0 is slightly above A1B, and RCP8.5 exceeds A2. RCP4.5 was excluded to reduce computational cost and for clarity of explanation. RCP2.6 and RCP8.5 cover the full envelope of prediction, and we then chose one ‘‘middle’’ RCP scenario (RCP6.0). RCP simulations overlap from 2006 to 2100 in all models used. Even though strong cases are made for using stabilized forcing in studying ENSO (e.g., Trenberth and Hurrell 1994; Guilyardi 2006; Stevenson et al. 2012a, b), we choose to use transient forcing simulations, representing realistic conditions, as we are most interested in the changes in rainfall over Central America and northwestern South America that may occur in the forthcoming decades of the twenty-first century. For the multi-model ensemble, model output is remapped to the domain with the coarsest grid spacing (T42). All available realizations (ensemble members) are used, and results are calculated for each realization before being averaged for plotting purposes. ENSO characteristics are ˜ O 3.4 index analyzed using SST anomalies over the NIN region (e.g., Trenberth 1997, 5°S–5°N, 170°W–120°W). El Nin˜o (La Nin˜a) events are defined similarly to Meehl et al. ˜ O3.4 region (2006), as SST anomalies over the NIN exceeding ? (-) 1 standard deviation in magnitude over the DJF average using 2–7 year bandpass filtered data (a Fourier filter is used). We use DJF as the basis for defining ENSO events, since it is the peak time period of ENSO amplitude (Fig. 1). Hereafter we denote this period as DJF1, referring to boreal winter of year 1, with seasons before DJF1 in year 0, and seasons after in year 1. Statistical significance of ENSO event amplitude and duration differences between RCP and historical simulations are computed using permutation tests, sampling without replacement. For ENSO amplitude, we sample over the pooled bandpass filtered time series 10,000 times. For models with multiple realizations, the time series from all realizations are concatenated to provide a larger sample. For ENSO duration, we sample over the duration values for

123

˜ O3.4 SST 1901–2009 anomalies for each a El Fig. 1 HadISST NIN Nin˜o and b La Nin˜a event (red lines) and the composite event (blue line). Seasonal delineations denoted in text. El Nin˜o years (December year 0) include 1902, 1914, 1918, 1925, 1930, 1940, 1957, 1965, 1972, 1976, 1982, 1986, 1991, and 1997. La Nin˜a years include 1903, 1916, 1938, 1942, 1949, 1955, 1966, 1970, 1973, 1983, 1984, 1988, 1995, 1998, and 2007

sets of discrete ENSO events as defined above. For ENSO period, statistical significance of the spectral maximum within the 2–7 year frequency range is estimated against a lag-1 red noise process. Statistical significance of ENSOrelated rainfall anomalies against the corresponding model climatology is assessed using the bootstrap method (with replacement; 10,000 samples) by randomly resampling 20 seasons (assuming approximately 20 ENSO events of each phase during the full time periods). Several observational datasets are used for model verification and reference. The Hadley Centre Sea Ice and Sea

Projected impact of twenty-first century ENSO

characteristics. The NASA Modern Era RetrospectiveAnalysis for Research and Application (MERRA), the second reanalysis produced by NASA’s Global Modeling and Assimilation Office (GMAO) (Rienecker et al. 2011), is used to assess contemporary circulation fields at monthly timescales. Output is available from 1979 to 2011 at 0.5° latitude by 0.67° longitude grid spacing.

3 Aspects of regional climate

Fig. 2 a Annual average rainfall (mm) from CRU TS3.1 dataset over 1979–2009 period. b Monthly average rainfall (mm) over areas outlined in a

Surface Temperature data set (HadISST) dataset includes global monthly 1° gridded SST from 1870 through 2011 (Rayner et al. 2003). The University of East Anglia Climate Research Unit (CRU) TS 3.1 monthly precipitation dataset (Mitchell and Jones 2005), available at 0.5° grid spacing over land from 1901 to 2009, is used for rainfall verification and to characterize general rainfall

Figure 2a shows the annual average rainfall over Central America, Colombia, Ecuador, and surrounding areas for the 1979–2009 period. The Caribbean coast of Central America receives ample rainfall due to the moisture-laden easterly trade winds. Colombia comprises several distinct rainfall regimes, with the Pacific coast being one of the rainiest places on earth (‘‘Lloro´’’, Choco´ region, Poveda and Mesa 2000). The existence of this feature results from low-level moisture convergence by the CHOCO jet, wind shear aloft, orographic lifting over the western Andes, and mesoscale convective system (MCS) development. Rainfall amounts diminish over the Andes, then increase over the Amazon basin of eastern Colombia (e.g., Horel et al. 1989). Just to the south, over Ecuador and northwestern Peru, is one of the driest regions on Earth, associated with the Eastern Pacific cold tongue. Figure 2b shows the annual cycle of precipitation averaged over portions of Central America, Colombia, and Ecuador (areas shown in Fig. 2a). There are clear distinctions of wet (May–October) and dry (November–April) seasons over Central America. The decrease in rainfall during boreal summer results from the mid-summer drought (MSD; e.g., Magan˜a et al. 1999; Small et al. 2007). The annual cycle is similar over much of Colombia, peaking in May through October, with slightly drier conditions over boreal summer (associated with the MSD) and a dry season over boreal winter (although there are strong regional differences, see Poveda et al. 2011). To the south over Ecuador, rainfall peaks in March and April, when the ITCZ is making its annual migration northward through the region, with relatively dry conditions the rest of the year. There is substantial variability in monthly and sub-monthly rainfall within the averaging boxes of Fig. 2a. The Andes separate distinct precipitation regimes over Colombia and Ecuador, as shown in Fig. 2a, with strong diurnal variability at very local scales (Poveda et al. 2005). Regionalscale variability also occurs across mountainous regions between the Caribbean and Pacific coasts of Central America (e.g., Waylen et al. 1996; Poveda et al. 2006). Earlier, it was mentioned that the low-level circulation (easterly trade winds over the Caribbean, westerly CHOCO jet over western Colombia) are important for low-level

123

D. F. Steinhoff et al. Fig. 3 MERRA 1979–2011 average 925 hPa wind vectors (colored by wind speed, m s-1) for a DJF, b MAM, c JJA, and d SON. Reference vector in lower right refers to 10 m s-1

moisture advection and rainfall. Figure 3 shows the seasonal evolution of the 925 hPa winds from MERRA, 1979–2011. The easterly trades winds, including the Caribbean Low-level Jet (CLLJ) north of Colombia (Amador 1998), are an important moisture source for the Caribbean flank of Central America (Dura´n-Quesada et al. 2010; Fig. 2a). Maximum wind speeds occur during DJF and JJA, and minima occur in MAM and SON. The CHOCO jet occurs just off the Pacific coast of Colombia, is thermally-driven by the meridional SST gradient over the eastern Pacific, and enhanced by the westerly cross-equatorial austral trade winds (Poveda and Mesa 2000; Hastenrath 2002). The seasonal cycle of the CHOCO jet, and the associated moisture convergence over western Colombia, largely explains the wetter conditions during the boreal autumn wet season compared to the boreal spring wet season over western Colombia.

123

4 Projected changes of ENSO To analyze mean-state ENSO changes we show the multimodel average mean-state changes of SST over the Pacific basin for each RCP simulation from the Historical simulation in Fig. 4. Mean-state warming occurs over the entire tropical Pacific, unanimous in all models making up the multi-model ensemble, and enhanced in equatorial regions, as found in previous studies that employed CMIP3 AOGCMs (e.g., Liu et al. 2005; Vecchi and Soden 2007; Mu¨ller and Roeckner 2008; Gastineau et al. 2009; Xie et al. 2010). Tropical Pacific SSTs increase from 1 to 1.5 °C for the RCP2.6 simulations to over 2 °C for the RCP8.5 simulations. Warming maxima are concentrated over the eastern Pacific in all three RCP simulations, again similar to results from the CMIP3 models (e.g., DiNezio et al. 2009). The decreased zonal SST gradient across the Pacific

Projected impact of twenty-first century ENSO Fig. 4 a Multi-model ensemble average RCP2.6 minus Historical SST (°C). Box ˜ O3.4 index region indicates NIN for reference. b Same as a except for RCP6.0. c Same as a except for RCP8.5

results in weaker easterly trade winds and a weaker Walker circulation (e.g., Held and Soden 2006; Vecchi et al. 2006; Zhang and Song 2006; Power and Smith 2007, Stevenson et al. 2012b). Also, differential warming in the tropical Northern Hemisphere results in a northward shift of the Hadley cell (Seager et al. 2010; Herceg Bulic et al. 2012). Reduced warming over the tropical North Atlantic, compared with the global tropical mean warming, is evident for the RCP8.5 simulation. This differential warming has been implicated in drying much of the tropical Americas during boreal spring and summer in future climate change scenarios (Vecchi and Soden 2007; Leloup and Clement 2009; Rauscher et al. 2011). Using the definition of ENSO events provided in Sect. 2, the periods between ENSO events (i.e., the frequency of events, represented by the maximum spectral power from each AOGCM and each RCP scenario) are shown in Fig. 5. A summary of changes and estimates of statistical significance for each model are shown in Table 2. For AOGCMs

Fig. 5 Maximum spectral power in the 2–7 year band, representing the characteristic period between ENSO events for each model, for Historical simulations (abscissa) and RCP simulations (ordinate, color-coded). Observed HadISST value indicated by ‘‘H.’’ Statistical significance for each model indicated in Table 2

123

D. F. Steinhoff et al. ˜ O 3.4) period, Table 2 Future projected changes in ENSO (NIN amplitude (standard deviation), and event duration for each model used in this study Model

Period

Amplitude

Duration

BCC-CSM1.1

H*, D*, I*, I*

D, D**, I

EN: D, D, I

CCSM4

H*, D*, D*, D*

D**, D**, D**

EN: D, D, D LN: D, D, D**

CESM1-CAM5

D*, D*, D*

I**, I**, I**

EN: D, D, D*

LN: I, I, I**

LN: I, I, I** CSIRO-Mk3.6.0

D, D, D

I**, I*, I**

FGOALS-s2

D, D, D

I**, D**, I**

EN: D, D, D LN: D, D, D EN: D, I, I LN: D, I, I

GFDL-ESM2G

D*, D, D*

I, I**, I**

EN: D*, D, D

GFDL-ESM2M

H*, D*, D*, D*

D**, D**, D

EN: I, I, D

GISS-E2-R

H*, D*, D*, D*

D**, D**, D**

HadGEM2-ES

I, I, I

D, I**, I

EN: D, D, D

IPSL-CM5A-LR

I, D*, D

D**, D**, D**

EN: D, D, D

MIROC5

H*, D*, D*, D*

I**, I**, I**

EN: I, D, I

MIROC-ESM

D, D, D

D**, I, I**

MIROC-ESMCHEM

I, I, I

D**, D**, I

EN: D, D, D* LN: I, D, D

MRI-CGCM3

H*, D*, D*, D*

I**, I**, I**

EN: D*, I, D

NorESM1-M

H*, D*, I*, I

D**, I**, D**

EN: D, D, I

LN: I, D, D

˜ O 3.4 region for Historical simuFig. 6 Standard deviation of NIN lations (abscissa) and RCP simulations (ordinate, color-coded). Observed HadISST value indicated by ‘‘H.’’ Statistical significance for each model indicated in Table 2

LN: D, D, D EN: D, D, I LN: D**, D, D LN: D**, D, I LN: D, D, D LN: I, I, I EN: I, D, D* LN: D, I, I

LN: D, D, D LN: D*, D**, I

Values in each column represent RCP2.6, RCP6.0, and RCP8.5, respectively. ‘‘I’’ indicates increase and ‘‘D’’ indicates decrease. For period column, ‘‘H*’’ indicates historical simulation, and ‘‘*’’ represents a statistically significant maximum spectral power at the 95 % level relative to a lag-1 red noise process. For amplitude column, ‘‘*’’ and ‘‘**’’ indicate statistical significance of the difference from the historical simulation at the 95 and 99 % levels, respectively, using 10,000 sample permutation test. For duration column, ‘‘EN’’ indicates El Nin˜o, ‘‘LN’’ represents La Nin˜a, ‘‘*’’ and ‘‘**’’ indicate statistical significance of the differences from the historical simulation at the 90 and 95 % levels, respectively, using 10,000 sample permutation test

with multiple ensemble members for a given scenario, periods are calculated separately for each realization and then averaged. Note that the maximum spectral power is dependent upon the time series filtering process, and is insensitive to multi-modal spectra. There is a wide range of

123

Fig. 7 a Duration of El Nin˜o events for each model, for Historical simulations (abscissa) and RCP simulations (ordinate, color-coded). Observed HadISST value indicated by ‘‘H.’’ Same as a except for La Nin˜a. Statistical significance for each model indicated in Table 2

Projected impact of twenty-first century ENSO

Fig. 8 Mean-state annual average changes in rainfall for each RCP simulation minus the Historical simulation. Stippling indicates that 10 of 15 models have changes of the same sign

ENSO periods across the models, from about 3–7 years, centered around the observed period with maximum spectral power of about 5 years. From the distribution of the points in Fig. 5 and the listing of statistically significant changes in Table 2, the majority of models predict a shorter ENSO period in the future. Previous studies using CMIP3 and earlier AOGCMs have also indicated a tendency for decreased period of ENSO events in future climate simulations (Timmermann et al. 1999; Merryfield 2006; Meehl et al. 2006; An et al. 2008). A similar analysis is done for ENSO amplitude, repre˜ O3.4 SST anomsented by the standard deviation of NIN alies, in Fig. 6. A summary of changes and estimates of statistical significance for each model are shown in Table 2. For the contemporary period, most models underestimate ENSO amplitude compared to observations, although notably the Community Climate System Model Version 4 (CCSM4) has a stronger-than-observed amplitude (see Deser et al. 2012) that tends to decrease under strong GHG forcing (Stevenson et al. 2012a). While there are strong statistically significant changes between historical and RCP simulations for most models, there is no consensus among models on future changes to ENSO amplitude, as found by Stevenson (2012) and Watanabe et al. (2012). As noted in Sect. 1, there is no clear agreement from previous studies of CMIP3 AOGCMs on the sign of ENSO amplitude changes with increased greenhouse gas forcing, and therefore the CMIP5 AOGCM results for ENSO amplitude are not particularly different from those in the CMIP3 models. To estimate ENSO event duration, we define event ˜ O3.4 SST anomalies remain length as each month the NIN above (El Nin˜o) or below (La Nin˜a) 0.5 standard deviations from the mean (similar to the definition for duration used

by Deser et al. 2012). Figure 7 shows both El Nin˜o and La Nin˜a event duration for all models. A summary of changes and estimates of statistical significance for each model are shown in Table 2. For El Nin˜o, event duration is centered around the observed duration of 9.5 months for the Historical simulations, and only shortens considerably for RCP8.5, perhaps in accordance with the decreasing ENSO return period shown in Fig. 5. For La Nin˜a, the event duration in the Historical simulations (*10–11 months) is shorter than observed (*12 months), as noted by Okumura and Deser (2010). However, there is not a clear model consensus (and little statistical significance) for future changes in duration.

5 Projected changes of rainfall Due to the mean-state changes of tropical Pacific SSTs, it is expected that the mean-state rainfall over Central America and northwestern South America will change as well. Figure 8 shows the mean-state differences between each RCP simulation and the Historical simulation for the multimodel ensemble. Northern Colombia, Venezuela, and Central America generally have decreasing annual precipitation with increasing RCP. In contrast, Ecuador and southwestern Colombia have increasing annual precipitation in future simulations. These changes are consistent with numerous other studies, many utilizing downscaled regional climate models, some of which employed CMIP3 AOGCMs (Giorgi 2006; Neelin et al. 2006; Vera et al. 2006; Boulanger et al. 2007; Sheffield and Wood 2008; Urrutia and Vuille 2009; Marengo et al. 2010; Karmalkar et al. 2011; Rauscher et al. 2011; Herceg Bulic´ et al. 2012; Power et al. 2012; Hidalgo et al. 2013). The changes are

123

D. F. Steinhoff et al.

Fig. 9 Same as Fig. 8, except for a MAM, b JJA, c SON, and d DJF

123

Projected impact of twenty-first century ENSO Fig. 10 CRU TS3.1 JJA0 1901–2009 rainfall anomaly composites for a El Nin˜o, b La Nin˜a, c El Nin˜o - La Nin˜a, and d El Nin˜o ? La Nin˜a, using events listed in Fig. 1. Dashed outlines in a and b refer to averaging areas used in Fig. 16. Stippled areas in a and b represent anomalies significant at the 95 % level from a 10,000 sample bootstrap test

attributed to a southward shift of the ITCZ (Rauscher et al. 2008; Karmalkar et al. 2011), a weaker and southwarddisplaced southeast Pacific subtropical anticyclone (Urrutia and Vuille 2009; Marengo et al. 2012), and changes in surface-atmosphere feedbacks from changes in soil moisture and precipitation recycling (e.g., Poveda and Mesa 1997; Poveda et al. 2011). On a seasonal basis (Fig. 9), the Central American drying is concentrated in boreal spring and summer (Rauscher et al. 2011), but becomes yearround for RCP8.5. Rainfall increases over the far eastern Pacific and over much of Ecuador year-round, while wetting over Colombia is greatest in boreal winter (So¨rensson et al. 2010). With these mean-state future projections in mind, it is of interest to explore how ENSO teleconnections to regional rainfall may change in the future. We review the contemporary ENSO-rainfall associations cited in the introduction by showing CRU TS3.1 observed rainfall anomalies from

El Nin˜o and La Nin˜a event composites for JJA0, DJF1, and JJA1, followed by the Historical simulation model ensemble average event composite anomalies and RCP8.5 projected change. In JJA0, El Nin˜o drying and La Nin˜a wetting occur over much of the region, except for weak drying during both phases over portions of Panama and Costa Rica, and small differences over Ecuador (Figs 10a, b). The phase differences are largely linear (see Hoerling et al. 1997; Fig. 10c), except for the Caribbean flank of Central America and the Pacific coast of western Colombia (Fig. 10d), which are directly affected by the easterly trade winds and the CHOCO jet, respectively. The model ensemble robustly represents the El Nin˜o (La Nin˜a) drying (wetting) over Central America and western Colombia, with opposite-signed changes just offshore of Ecuador (Fig. 11a, b). However, regional-scale details from Fig. 10, particularly over Central America, are not captured. Changes to the model anomaly composites for RCP8.5

123

D. F. Steinhoff et al. Fig. 11 a JJA0 model ensemble average rainfall anomaly composites for a Historical El Nin˜o, b Historical La Nin˜a, c RCP 8.5 minus Historical El Nin˜o, and d RCP 8.5 minus Historical La Nin˜a. Stippling indicates that 10 of 15 models have changes of the same sign

(Fig. 11c, d) show some evidence of enhanced ENSOrelated rainfall anomalies over portions of Central America and west of Colombia and Ecuador. ENSO-related rainfall anomalies are similar during DJF1 (Fig. 12), except the Caribbean coast of Costa Rica now dries during El Nin˜o (Fig. 12a), associated with weakening of the easterly trade winds (e.g., Poveda et al. 2006). Drying over the Colombian Pacific Lowlands is consistent with the influence that ENSO has on the CHOCO jet (Poveda and Mesa 2000; Poveda et al. 2006; Grimm and Tedeschi 2009). The reduced meridional SST gradient during El Nin˜o conditions weakens the CHOCO jet, reducing moisture transport inland and reducing MCS occurrences. Wetter conditions are found over coastal Ecuador, as the ITCZ shifts south during El Nin˜o conditions. Rainfall anomalies during La Nin˜a largely mirror those of El Nin˜o (Fig. 12b), indicated in Fig. 12c by the

123

linear component of the ENSO variability. The nonlinear component in Fig. 12d indicates that Ecuador is less affected by La Nin˜a than El Nin˜o. The model ensemble correctly represents the El Nin˜o (La Nin˜a) drying (wetting) over most of the plotted domain, including the enhancement along the western Colombian coast, and wetting (drying) over coastal Ecuador (Fig. 13a, b). Model consensus on RCP8.5 projected changes (Fig. 13c, d) are again weak overall, except for enhancement of ENSO-related anomalies over Guatemala and Belize. A complex pattern arises during JJA1 (Fig. 14a,b), where ENSO-related rainfall anomalies generally reverse sign from DJF1 over much of Central America, Colombia, Venezuela, and the Caribbean (e.g., Gianninni et al. 2000). Portions of Costa Rica and Panama are again an exception, which may relate to some JJA1 seasons being JJA0 seasons of the opposite phase. Western Colombia rainfall is only

Projected impact of twenty-first century ENSO Fig. 12 Same as Fig. 10 except for DJF1

affected by La Nin˜a (Fig. 14c, d), and anomalies over Ecuador are small. The model ensemble anomalies generally capture the reversal in sign from DJF1 of the rainfall anomalies over portions of Central America (Fig. 15a, b). Some evidence of an enhancement of these patterns for RCP8.5 is seen (Fig. 15c, d), with wetting (drying) over Nicaragua, Honduras, and western Colombia for El Nin˜o (La Nin˜a). The ENSO event composite rainfall anomalies across the suite of climate models for the Historical and RCP simulations, averaged over land regions of Colombia, Central America, and Ecuador (see dashed boxes in Fig. 10a, b for averaging regions), are shown in Fig. 16 for JJA0, SON0, DJF1, MAM1, and JJA1. The box and whiskers plot represents the distribution of rainfall anomalies across the individual model ensembles, while the black dots represent each individual realization from each model. The values from the CRU TS3.1 dataset (land only)

from Figs. 10, 12, and 14 are shown by a purple asterisk. Direction of changes and statistical significance of the ENSO DJF1 rainfall anomalies for each model (against its own climatology) are shown in Table 3. For El Nin˜o over Colombia (Fig. 16a) in JJA0, SON0, and DJF1, there are not strong differences between historical and RCP simulations, but most of the statistically significant anomalies in Table 3 are negative. The distribution of models clearly underestimates the observed rainfall anomalies for JJA0 and DJF1, although it is much closer for SON0. Observed anomalies become slightly positive for JJA1, which the model distribution captures, although earlier than observed in MAM1. The results for La Nin˜a (Fig. 16b) generally mirror those for El Nin˜o. The model distribution shows positive rainfall anomalies through all seasons that are smaller than those observed. While it is encouraging that the model distribution captures the sign of the rainfall anomalies during El Nin˜o and La Nin˜a, there are still a

123

D. F. Steinhoff et al.

Fig. 13 Same as Fig. 11 except for DJF1

substantial number of model realizations showing anomalies of opposite sign, pointing to deficiencies in simulating ENSO, regional rainfall, or both (Vera and Silvestri 2009). The model distribution captures the swing from negative to positive El Nin˜o rainfall anomalies over Central America during year 1 (Fig. 16c), and is well-constrained to the small observed changes during the dry season in DJF1 and MAM1. The observed La Nin˜a positive rainfall anomalies (Fig. 16d) are well-simulated by the model distribution, and models with statistically significant DJF1 anomalies (Table 3) suggest that positive anomalies will continue into the future. Weak anomalies are observed seasonally over Ecuador during El Nin˜o (Fig. 16e). During JJA0, the driest part of the year in this region, observed and simulated anomalies

123

are small, but increase through SON0 to DJF1 and MAM1. However, there is little consensus in the model distribution to the observed positive anomalies, and the range of model responses becomes very large. Clearly, the rainfall variability of this region is poorly simulated in many of the models, likely from the sharp and opposing gradients of precipitation anomalies over this region, especially for El Nin˜o. Statistically significant anomalies in Table 3 tend to be negative and remain so in the future. Observed La Nin˜a anomalies over this region (Fig. 13b) tend to be small, and while the ensemble distribution captures this, there are wide-ranging values for the anomalies, with statistically significant changes approximately split between the sign of the anomalies. For all three regions, there is not a clear sense of the future changes of ENSO rainfall anomalies

Projected impact of twenty-first century ENSO Fig. 14 Same as Fig. 10 except for JJA1

across the model distribution, with varying signs and statistical significances between models and even between different RCP scenarios of the same model (Table 3).

6 Conclusions In this study we have explored changes to the ENSO mean state, period, amplitude, and duration, and associated rainfall changes over Central America and northwestern South America in future climate projections from the CMIP5 multi-model AOGCM dataset. As previously mentioned, ENSO strongly affects rainfall over the region through both large-scale teleconnection patterns (i.e., Walker circulation and tropical Atlantic SSTs) and regional-scale processes (i.e., CLLJ, CHOCO jet, and land– atmosphere feedbacks). Quantifying future projections of not only mean-state rainfall, but also ENSO-related rainfall

anomalies, is an important task due to the effects that ENSO has on extreme rainfall and drought events. For the mean-state, there is robust warming of SSTs over the tropical Pacific, drying over Central America and northern Colombia and Venezuela, and wetter conditions over southwestern Colombia into Ecuador and northern Peru. The analysis of projected changes to ENSO characteristics yields a wide range of results among the model output used. While a majority of models show an increase in ENSO frequency with greater greenhouse gas concentration in future climate, there is no clear consensus for changes to event duration or event amplitude. An increase in ENSO event frequency would correspondingly increase the potential for extreme rainfall events in regions with strong ENSO-rainfall teleconnections, including the Caribbean coast of Central America and western Colombia. Changes to regional rainfall from composite ENSO events, like ENSO itself, show a wide range of responses

123

D. F. Steinhoff et al. Fig. 15 Same as Fig. 11 except for JJA1

among the various model simulations of future climate. ENSO-related rainfall anomalies are generally underestimated by the models. While the models capture the broad rainfall anomalies in the observations, there is not agreement on the changes in ENSO-related rainfall anomalies in future simulations. Note that changes to the ‘‘flavor’’ of ENSO, like increased incidences of El Nin˜o-Modoki (e.g., Yeh et al. 2009; Choi et al. 2011), were beyond the scope of this study, because submonthly temporal resolution is required to properly distinguish canonical and Modoki events (Karnauskas 2013). However, the El Nin˜o-Modoki rainfall patterns described in the introduction (Gouirand et al. 2013) would be expected to become more prevalent. We do not explicitly study extreme ENSO events, from which the atmospheric response over the eastern tropical Pacific and associated teleconnections change nonlinearly from moderate ENSO events, and which are robustly

123

predicted to double in frequency over the twenty-first century (Cai et al. 2014). We also do not assess changes to the land–atmosphere feedbacks, which are integral components of the ENSO-rainfall teleconnections (e.g., Poveda and Mesa 1997). Such future changes would result primarily through changes in vegetation, like deforestation (e.g., Poveda et al. 2011). Even though there are well-documented problems in correctly simulating ENSO and its effects on tropical Americas rainfall, the fact that vastly different ENSOrelated rainfall effects occur over the rather short distance between Central America and Ecuador, and Colombia being in a ‘‘transition’’ zone, points to the importance of properly simulating ENSO and tropical rainfall anomalies on climate time scales. Future research may be best directed at statistically and dynamically downscaling AOGCMs that have the best representations of ENSO.

Projected impact of twenty-first century ENSO

Fig. 16 Rainfall anomalies over land areas of a, b Colombia region, c, d Central America region, and e, f Ecuador region associated with a, c, e El Nin˜o and b, d, f La Nin˜a during JJA0, SON0, DJF1, MAM1, and JJA1. Box and whisker plots represent the ensemble of model

averages across all realizations. Box represents 25th to 75th percentiles, and whiskers represent minimum and maximum values. Black dots represent each realization from all of the models. Purple asterisks represent CRU TS3.1 anomalies

123

D. F. Steinhoff et al. Table 3 Sign of the precipitation changes from climatological average for composite DJF1 El Nin˜o (EN) and La Nin˜a (LN) for each model Model

Colombia

Ecuador

Central America

BCC-CSM1.1

EN: D*, D, D**, D

EN: W, W**, W, W

EN: D, W, D**, D

LN: W**, W*, W**, W

LN: D*, D**, D**, D**

LN: W*, D, W**, W*

CCSM4

EN: D, W, W, W

EN: W, W, W, W

EN: D**, D**, D**, D**

LN: W, D, W, D

LN: D*, D**, D**, D**

LN: W**, W**, W**, W**

CESM1-CAM5

EN:D**, D**, D**, D**

EN: W, D, D, W

EN: D**, D**, D**, D**

LN:W*, W, W**, W*

LN: D, D, D*, D

LN: W**, W**, W**, W**

CSIRO-Mk3.6.0

EN: D, W, W, D

EN:W, D, W, W

EN: W, W, W, D

LN:D, W, W, W

LN:D, D, D, D

LN: D, W, W, W

FGOALS-s2

EN: D**, D**, D**, D**

EN: D*, D**, D**, D*

EN:D*, D**, D**, D**

LN: W**, W, W**, W**

LN: W, W**, W, W

LN: W*, D, W**, W**

GFDL-ESM2G

EN: D*, D**, D*, D** LN: W**, W**, W*, W**

EN: D**, D**, D**, D** LN:W**, W**, W**, W**

EN: D, D**, D, D* LN: W*, W**, W**, W*

GFDL-ESM2M

EN: D**, D**, D**, D**

EN: D**, D**, D**, D**

EN: D*, D**, D**, D**

LN:W**, W**, W**, W**

LN: W**, W**, W**, W**

LN: W**, W, W*, W** EN: D, D, D, D

GISS-E2-R

EN: D, D*, W, W

EN: D, W, D, D

LN: W, W, W, W*

LN: W, D, W**, W*

LN: W*, W, D, W**

HadGEM2-ES

EN: D, D, D**, D

EN: W, W*, W, W*

EN: D, D, D**, D

LN:W*, W, W*, W

LN: D, D, D, D

LN: W*, W, W*, W

IPSL-CM5A-LR

EN: W, W, W, W

EN: W, D, W, D

EN: D, D, D, D

LN: D, W, W, D

LN: D, W, D, D

LN: W, W, W*, W

MIROC5

EN: D**, D**, D**, D**

EN: W**, W**, W**, W**

EN: D*, D**, D**, D**

LN: W**, W, W**, W**

LN: D**, D**, D**, D**

LN: W**, W*, W**, W**

MIROC-ESM

EN: D, D, D, D*

EN: D, D, D, D

EN: D, W, D, D**

LN: D, D, W, W

LN: W, W, D, W

LN: D, D, D*, D

MIROC-ESM-CHEM

EN: D*, D, D*, W

EN: W, W, W**, W*

EN: D, D, D*, D

LN: W**, W, D, W

LN: D, W, W, D

LN: W**, W, D, D

MRI-CGCM3

EN: D**, D, W, D** LN: W**, W*, W**, W**

EN: W, W**, W*, W** LN: D*, D**, D**, D**

EN: D*, D, D, D** LN: W**, W*, W, W**

NorESM1-M

EN: W, W, W*, D

EN: D**, D, D, D*

EN: D**, D, D**, D**

LN: D, D, D*, W

LN: W*, W**, W, W**

LN: W**, W**, W**, W**

Values in each column represent Historical, RCP2.6, RCP6.0, and RCP8.5, respectively. ‘‘W’’ indicates wetter and ‘‘D’’ indicates drier. ‘‘*’’ and ‘‘**’’ indicate statistical significance of the anomalies relative to the corresponding model climatology at the 95 and 99 % levels, respectively, using 10,000 sample bootstrap test

Then, the complex topography of the Andes and Central American mountain ranges, and the related mesoscale features that are important for the regional hydrology, can be adequately resolved to provide improved estimates of rainfall. Such improvements are key to properly resolving extreme rainfall events that have strong societal impacts. Acknowledgments This work was funded by the Inter-American Development Bank through an Interagency Agreement with the National Science Foundation. The National Center for Atmospheric Research is funded in part by the National Science Foundation. We acknowledge the World Climate Research Programme’s Working Group on Coupled Modelling, which is responsible for CMIP, and the climate modeling groups (listed in Table 1 of this paper) for producing and making available their model output. For CMIP the U.S. Department of Energy’s Program for Climate Model Diagnosis and Intercomparison provides coordinating support and led development

123

of software infrastructure in partnership with the Global Organization for Earth System Science Portals. Helpful comments and suggestions we provided from three anonymous reviewers.

References Aceituno P (1988) On the functioning of the Southern Oscillation in the South American sector. Part I: surface climate. Mon Weather Rev 116:505–524 Amador JA (1998) A climatic feature of tropical Americas: the trade wind easterly jet. Top Meteorol Oceanogr 5:91–102 An SI, Kug JS, Ham YG, Kang IS (2008) Successive modulation of ENSO to the future greenhouse warming. J Clim 21:3–21. doi:10.1175/2007JCLI1500.1 Boulanger JP, Martinez F, Segura EC (2007) Projection of future climate change conditions using IPCC simulations, neural networks and Bayesian statistics. Part 2: precipitation mean

Projected impact of twenty-first century ENSO state and seasonal cycle in South America. Clim Dyn 28:255–271. doi:10.1007/s00382-006-0182-0 Cai W, Borlace S, Lengaigne M, van Rensch P, Collins M, Vecchi G, Timmermann A, Santoso A, McPhaden MJ, Wu L, England MH, Wang G, Guilyardi E, Jin F-F (2014) Increasing frequency of extreme El Nin˜o events due to greenhouse warming. Nature Clim Change 4:111–116. doi:10.1038/NCLIMATE2100 Chen AA, Taylor MA (2002) Investigating the link between early season Caribbean rainfall and the El Nin˜o ? 1 year. Int J Climatol 22:87–106. doi:10.1002/joc.711 Chiang JCH, Kushnir Y, Zebiak SE (2000) Interdecadal changes in eastern Pacific ITCZ variability and its influence on the Atlantic ITCZ. Geophys Res Lett 27:3687–3690 Chiang JCH, Kushnir Y, Giannini A (2002) Deconstructing Atlantic Intertropical Convergence Zone variability: influence of the local cross-equatorial sea surface temperature gradient and remote forcing from the eastern equatorial Pacific. J Geophys Res 107:D14004. doi:10.1029/2000JD000307 Choi J, An S-I, Kug J-S, Yeh S-W (2011) The role of mean state on changes in El Nin˜o’s flavor. Clim Dyn 37:1205–1215. doi:10. 1007/s00382-010-0912-1 Chou C, Chen CA, Tan PH, Chen KT (2012) Mechanisms for global warming impacts on precipitation frequency and intensity. J Clim 25:3291–3306. doi:10.1175/JCLI-D-11-00239.1 Collins M et al (2010) The impact of global warming on the tropical Pacific Ocean and El Nin˜o. Nature Geosci 3:391–397. doi:10. 1038/ngeo868 Curtis S, Hastenrath S (1995) Forcing of anomalous sea surface temperature evolution in the tropical Atlantic during Pacific warm events. J Geophys Res 100(C8):15835–15847 Deser C, Phillips AS, Tomas RA, Okumura Y, Alexander MA, Capotondi A, Scott JD, Kwon Y-O, Ohba M (2012) ENSO and pacific decadal variability in community climate system model version 4. J Clim 25:2622–2651. doi:10.1175/JCLI-D-11-00301. 1 DiNezio PN, Clement AC, Vecchi GA, Soden BJ, Kirtman BP, Lee SK (2009) Climate response of the equatorial Pacific to global warming. J Clim 22:4873–4892. doi:10.1175/2009JCLI2982.1 Dura´n-Quesada AM, Gimeno L, Amador JA, Nieto R (2010) Moisture sources for Central America: identification of moisture sources using a Lagrangian analysis technique. J Geophys Res 115:D05103. doi:10.1029/2009JD012455 Enfield DB, Mayer DA (1997) Tropical Atlantic sea surface temperature variability and its relation to El Nin˜o-Southern Oscillation. J Geophys Res 102(C1):929–945 Gastineau G, Li L, Le Treut H (2009) The Hadley and Walker circulation changes in global warming conditions described by idealized atmospheric simulations. J Clim 22:3993–4013 Giannini A, Chiang JCH, Cane MA, Kushnir Y, Seager R (2001) The ENSO teleconnection to the tropical Atlantic Ocean: contributions of the remote and local SSTs to rainfall variability in the tropical Americas. J Clim 14:4530–4544 Gianninni A, Kushnir AY, Cane MA (2000) Interannual variability of Caribbean rainfall, ENSO, and the Atlantic Ocean. J Clim 13:297–311 Giorgi F (2006) Climate change hot spots. Geophys Res Lett 33:L08707. doi:10.1029/2006GL025734 Gouirand I, Moron V, Hu Z-Z, Jha B (2013) Influence of the warm pool and cold tongue El Nin˜os on the following Caribbean rainy season rainfall. Clim Dyn. doi:10.1007/s00382-013-1753-5 Grimm AM (2011) Interannual climate variability in South America: impacts on seasonal precipitation, extreme events, and possible effects of climate change. Stoch Environ Res Risk Assess 25:537–554. doi:10.1007/s00477-010-0420-1 Grimm AM, Ambrizzi T (2009) Teleconnections into South America from the tropics and extratropics on interannual and

intraseasonal timescales. Dev Paleoenviron Res 14:159–191. doi:10.1007/978-90-481-2672-9_7 Grimm AM, Tedeschi RG (2009) ENSO and extreme rainfall events in South America. J Clim 22:1589–1609. doi:10.1175/ 2008JCLI2429.1 Gu G, Adler RF (2006) Interannual rainfall variability in the tropical Atlantic region. J Geophys Res 111:D02106. doi:10.1029/ 2005JD005944 Guilyardi E (2006) El Nin˜o–mean state–seasonal cycle interactions in a multi-model ensemble. Clim Dyn 26:329–348. doi:10.1007/ s00382-005-0084-6 Hastenrath S (1976) Variations in low-latitude circulation and extreme climatic events in the tropical Americas. J Atmos Sci 33:202–215 Hastenrath S (2002) The intertropical convergence zone of the eastern Pacific revisited. Int J Climatol 22:347–356 Hastenrath S, Greischar L (1993) Circulation mechanisms related to northeast Brazil rainfall anomalies. J Geophys Res 98(D3):5093–5102 Held IM, Soden BJ (2006) Robust responses of the hydrological cycle to global warming. J Clim 19:5686–5699. doi:10.1175/JCLI3990.1 Herceg Bulic´ I, Brankovic´ Cˇ, Kucharski F (2012) Winter ENSO teleconnections in a warmer climate. Clim Dyn 38:1593–1613. doi:10.1007/s00382-010-0987-8 Hidalgo HG, Amador JA, Alfaro EJ, Quesada B (2013) Hydrological climate change projections for Central America. J Hydrol 495:94–112. doi:10.1016/j.hydrol.2013.05.004 Hoerling MP, Kumar A, Zhong M (1997) El Nin˜o, La Nin˜a, and the nonlinearity of their teleconnections. J Clim 10:1769–1786 Horel JD, Wallace JM (1981) Planetary-scale atmospheric phenomena associated with the Southern Oscillation. Mon Weather Rev 109:813–829 Horel JD, Hahmann AN, Geisler JE (1989) An investigation of the annual cycle of convective activity over the tropical Americas. J Clim 2:1388–1403 Hoyos I, Baquero-Bernal A, Jacob D, Rodrı´guez BA (2013) Variability of extreme events in the Colombian Pacific and Caribbean catchment basins. Clim Dyn 40:1985–2003. doi:10. 1007/s00382-012-1487-9 IPCC (2007) Climate change 2007: the physical science basis. In: Solomon S, Qin D, Manning M, Chen Z, Marquis M, Averyt KB, Tignor M, Miller HL (eds) Contribution of working group I to the fourth assessment report of the intergovernmental panel on climate change. Cambridge University Press, Cambridge Karmalkar AV, Bradley RS, Diaz HF (2011) Climate change in Central America and Mexico: regional climate model validation and climate change projections. Clim Dyn 37:605–629. doi:10. 1007/s00382-011-1099-9 Karnauskas KB (2013) Can we distinguish canonical El Nin˜o from Modoki? Geophys Res Lett. doi:10.1002/grl.51007 Karnauskas KB, Busalacchi AJ (2009) The role of SST in the East Pacific warm pool in the interannual variability of Central American rainfall. J Clim 22:2605–2623 Leloup J, Clement A (2009) Why is there a minimum in projected warming in the tropical North Atlantic Ocean? Geophys Res Lett 36:L14802. doi:10.1029/2009GL038609 Liu Z, Vavrus S, He F, Wen N, Zhong Y (2005) Rethinking tropical ocean response to global warming: the enhanced equatorial warming. J Clim 18:4684–4700 Lyon B (2003) Enhanced seasonal rainfall in northern Venezuela and the extreme events of December 1999. J Clim 16:2302–2306 Lyon B, Barnston AG (2005) ENSO and the spatial extent of interannual precipitation extremes in tropical land areas. J Clim 18:5095–5109 Magan˜a V, Amador J, Medina S (1999) The midsummer drought over Mexico and Central America. J Clim 12:1577–1588

123

D. F. Steinhoff et al. Mapes BE, Warner TT, Xu M, Negri AJ (2003) Diurnal patterns of rainfall in northwestern South America. Part I: observations and context. Mon Weather Rev 131:799–812 Marengo JA, Ambrizzi T, da Rocha RP, Alves LM, Cuadra SV, Valverde MC, Torres RR, Santos DC, Ferraz SET (2010) Future change of climate in South America in the late twenty-first century: intercomparison of scenarios from three regional climate models. Clim Dyn 35:1073–1097. doi:10.1007/s00382009-0721-6 Marengo JA, Chan Chou S, Kay G, Alves LM, Pesquero JF, Soares WR, Santos DC, Lyra AA, Sueiro G, Betts R, Chagas DJ, Gomes JL, Bustamante JF, Tavares P (2012) Development of regional future climate change scenarios in South America using the Eta CPTEC/HadCM3 climate change projections: climatology and regional analyses for the Amazon, Sa˜o Francisco and the Parana´ River basins. Clim Dyn 38:1829–1848. doi:10.1007/s00382-0111155-5 Martin ER, Schumacher C (2011) The Caribbean low-level jet and its relationship with precipitation in IPCC AR4 models. J Clim 24:5935–5950. doi:10.1175/JCLI-D-11-00134.1 McGlone D, Vuille M (2012) The associations between El Nin˜oSouthern Oscillation and tropical South American climate in a regional climate model. J Geophys Res 117:D06105. doi:10. 1029/2011JD017066 Meehl GA, Teng H, Branstator G (2006) Future changes of El Nin˜o in two global coupled climate models. Clim Dyn 26:549–566. doi:10.1007/s00382-005-0098-0 Meehl GA, Covey C, Delworth T, Latif M, McAvaney B, Mitchell JFB, Stouffer RFJ, Taylor KE (2007) The WCRP CMIP3 multimodel dataset: a new era in climate change research. Bull Am Meteorol Soc 88:1383–1394. doi:10.1175/BAMS-88-9-1383 Merryfield WJ (2006) Changes to ENSO under CO2 doubling in a multimodel ensemble. J Clim 19:4009–4027. doi:10.1175/ JCLI3834 Mitchell TD, Jones PD (2005) An improved method of constructing a database of monthly climate observations and associated highresolution grids. Int J Climatol 25:693–712. doi:10.1002/joc. 1181 Mo KC, Berbery EH (2011) Drought and persistent wet spells over South America based on observations and the U.S. CLIVAR drought experiments. J Clim 24:1801–1820. doi:10.1175/ 2010JCLI3874.1 Moss RH et al (2010) The next generation of scenarios for climate change research and assessment. Nature 463:747–756. doi:10. 1038/nature08823 Mu¨ller WA, Roeckner E (2008) ENSO teleconnections in projections of future climate in ECHAM5/MPI-OM. Clim Dyn 31:533–549. doi:10.1007/s00382-007-0357-3 Neelin JD, Mu¨nnich M, Su H, Meyerson JE, Holloway CE (2006) Tropical drying trends in global warming models and observations. Proc Natl Acad Sci 103:6110–6115 Nobre P, Shukla J (1996) Variations of sea surface temperature, wind stress, and rainfall over the tropical Atlantic and South America. J Clim 9:2464–2479 Okumura YM, Deser C (2010) Asymmetry in the duration of El Nin˜o and La Nin˜a. J Clim 23:5826–5843 Poveda G, Mesa OJ (1997) Feedbacks between hydrological processes in tropical South America and large-scale oceanicatmospheric phenomena. J Clim 10:2690–2702 Poveda G, Mesa OJ (2000) On the existence of Lloro´ (the rainiest locality on Earth): enhanced ocean-land-atmosphere interaction by a low-level jet. Geophys Res Lett 27:1675–1678 Poveda G, Mesa OJ, Salazar LF, Arias PA, Moreno HA, Vieira SC, Agudelo PA, Toro VG, Alvaraz JF (2005) The diurnal cycle of precipitation in the tropical Andes of Colombia. Mon Wea Rev 133:228–240

123

Poveda G, Waylen PR, Pulwarty RS (2006) Annual and inter-annual variability of the present climate in northern South America and southern Mesoamerica. Palaeogeogr Palaeoclim Palaeoecol 234:3–27. doi:10.1016/j.palaeo.2005.10.031 ´ lvarez DM, Rueda O ´ A (2011) Hydro-climatic variability Poveda G, A over the Andes of Colombia associated with ENSO: a review of climatic processes and their impact on one of the Earth’s most important biodiversity hotspots. Clim Dyn 36:2233–2249. doi:10.1007/s00382-010-0931-y Poveda G, Jaramillo L, Vallejo LF (2014) Seasonal precipitation patterns along pathways of South American low-level jets and aerial rivers. Water Resour Res 50:1–21. doi:10.1002/ 2013WR014087 Power SB, Smith IN (2007) Weakening of the Walker circulation and apparent dominance of El Nin˜o both reach record levels, but has ENSO really changed? Geophys Res Lett 34:L18702. doi:10. 1029/2007GL030854 Power SB, Delage F, Colman R, Moise A (2012) Consensus on twenty-first-century rainfall projections in climate models more widespread than previously thought. J Clim 25:3792–3809. doi:10.1175/JCLI-D-11-00354.1 Power S, Delage F, Chung C, Kociuba G, Keay K (2013) Robust twenty-first-century projections of El Nin˜o and related precipitation variability. Nature. doi:10.1038/nature12580 Rasmusson EM, Mo K (1993) Linkages between 200-mb tropical and extratropical circulation anomalies during the 1986–1989 ENSO cycle. J Clim 6:595–616 Rauscher SA, Giorgi F, Diffenbaugh NS, Seth A (2008) Extension and intensification of the Meso-American mid-summer drought in the twenty-first century. Clim Dyn 31:551–571. doi:10.1007/ s00382-007-0359-1 Rauscher SA, Kucharski F, Enfield DB (2011) The role of regional SST warming variations in the drying of Meso-America in future climate projections. J Clim 24:2003–2016. doi:10.1175/2010JCLI3536.1 Rayner NA, Parker DE, Horton EB, Folland CK, Alexander LV, Rowell DP, Kent EC, Kaplan A (2003) Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J Geophys Res 108:D144407. doi:10.1029/2002JD002670 Rienecker MM et al (2011) MERRA: NASA’s modern-era retrospective analysis for research and applications. J Clim 24:3624–3648. doi:10.1175/JCLI-D-11-00015.1 Rogers JC (1988) Precipitation variability over the Caribbean and tropical Americas associated with the Southern Oscillation. J Clim 1:172–182 Ropelewski CF, Bell MA (2008) Shifts in the statistics of daily rainfall in South America conditional on ENSO phase. J Clim 21:849–865. doi:10.1175/2007JCLI1617.1 Ropelewski CF, Halpert MS (1987) Global and regional scale precipitation patterns associated with the El Nin˜o/Southern Oscillation. Mon Weather Rev 115:1606–1626 Ropelewski CF, Halpert MS (1989) Precipitation patterns associated with the high index phase of the Southern Oscillation. J Clim 2:268–284 Ropelewski CF, Halpert MS (1996) Quantifying Southern Oscillation-precipitation relationships. J Clim 9:1043–1059 Saravanan R, Chang P (2000) Interaction between tropical Atlantic variability and El Nin˜o-Southern Oscillation. J Clim 13:2177–2194 Seager R, Naik N, Vecchi GA (2010) Thermodynamic and dynamic mechanisms for large-scale changes in the hydrological cycle in response to global warming. J Clim 23:4651–4668. doi:10.1175/ 2010JCLI3655.1 Seager R, Naik N, Vogel L (2012) Does global warming cause intensified interannual hydroclimate variability? J Clim 25:3355–3372. doi:10.1175/JCLI-D-11-00363.1

Projected impact of twenty-first century ENSO Sheffield J, Wood EF (2008) Projected changes in drought occurrence under future global warming from multi-model, multi-scenario, IPCC AR4 simulations. Clim Dyn 31:79–105. doi:10.1007/ s00382-007-0340-z Small RJO, de Szoeke SP, Xie SP (2007) The Central American midsummer drought: regional aspects and large-scale forcing. J Clim 20:4853–4873. doi:10.1175/JCLI4261.1 So¨rensson AA, Mene´ndez CG, Ruscica R, Alexander P, Samuelsson P, Wille´n U (2010) Projected precipitation changes in South America: a dynamical downscaling within CLARIS. Meteorol Z 19:347–355. doi:10.1127/0941-2948/2010/0467 Stevenson S (2012) Significant changes to ENSO strength and impacts in the twenty-first century: results from CMIP5. Geophys Res Lett 39:L17703. doi:10.1029/2012GL052759 Stevenson S, Fox-Kemper B, Jochum M, Neale R, Deser C, Meehl GA (2012a) Will there be a significant change to El Nin˜o in the twenty-first century? J Clim 25:2129–2145. doi:10.1175/JCLID-11-00252.1 Stevenson S, Fox-Kemper B, Jochum M (2012b) Understanding the ENSO–CO2 link using stabilized climate simulations. J Clim 25:7917–7936. doi:10.1175/JCLI-D-11-00546.1 Taylor MA, Enfield DB, Chen AA (2002) Influence of the tropical Atlantic versus the tropical Pacific on Caribbean rainfall. J Geophys Res 107:C93127. doi:10.1029/2001JC001097 Taylor KE, Stouffer RJ, Meehl GA (2012) An overview of CMIP5 and the experiment design. Bull Am Meteorol Soc 93:485–498. doi:10.1175/BAMS-D-11-00094.1 Timmermann AJ, Oberhuber J, Bacher A, Esch M, Latif M, Roeckner E (1999) Increased El Nin˜o frequency in a climate model forced by future greenhouse warming. Nature 398:694–697. doi:10. 1038/19505 Trenberth KE (1997) The definition of El Nin˜o. Bull Am Meteorol Soc 78:2771–2777 Trenberth KE, Hurrell JW (1994) Decadal atmosphere-ocean variations in the Pacific. Clim Dyn 9:303–319 Urrutia R, Vuille M (2009) Climate change projections for the tropical Andes using a regional climate model: temperature and precipitation simulations for the end of the 21st century. J Geophys Res 114:D02108. doi:10.1029/2008JD011021 van Oldenborgh GJ, Philip SY, Collins M (2005) El Nin˜o in a changing climate: a multi-model study. Ocean Sci 1:81–95 Vecchi GA, Soden BJ (2007) Global warming and the weakening of the tropical circulation. J Clim 20:4316–4340 Vecchi GA, Wittenberg AT (2010) El Nin˜o and our future climate: where do we stand? Interdiscip Rev Clim Change 1:260–270. doi:10.1002/wcc.33

Vecchi GA, Soden BJ, Wittenberg AT, Held IM, Leetmaa A, Harrison MJ (2006) Weakening of tropical Pacific atmospheric circulation due to anthropogenic forcing. Nature 441:73–76. doi:10.1038/nature04744 Vera C, Silvestri G (2009) Precipitation interannual variability in South America from the WCRP-CMIP3 multi-model dataset. Clim Dyn 32:1003–1014. doi:10.1007/s00382-009-0534-7 Vera C, Silvestri G, Liebmann B, Gonza´lez P (2006) Climate change scenarios for seasonal precipitation in South America from IPCC-AR4 models. Geophys Res Lett 33:L13707. doi:10.1029/ 2006GL025759 Wang C (2007) Variability of the Caribbean low-level jet and its relations to climate. Clim Dyn 29:411–422. doi:10.1007/s00382007-0243-z Watanabe M, Kug J-S, Jin F-F, Collins M, Ohba M, Wittenberg AT (2012) Uncertainty in the ENSO amplitude change from the past to the future. Geophys Res Lett 39:L20703. doi:10.1029/ 2012GL053305 Waylen PR, Poveda G (2002) El Nin˜o-Southern Oscillation and aspects of western South America hydro-climatology. Hydrol Process 16:1247–1260 Waylen PR, Caviedes CN, Quesada ME (1996) Interannual variability of monthly precipitation in Costa Rica. J Clim 9:2606–2613 Wu R, Kirtman BP (2011) Caribbean Sea rainfall variability during the rainy season and relationship to the equatorial Pacific and tropical Atlantic SST. Clim Dyn 37:1533–1550. doi:10.1007/ s00382-010-0927-7 Xie SP, Deser C, Vecchi GA, Ma J, Teng H, Wittenberg AT (2010) Global warming pattern formation: sea surface temperature and rainfall. J Clim 23:966–986 Yasunari T (1987) Global structure of the El Nin˜o/Southern Oscillation. Part I: El Nin˜o composites. J Met Soc Jpn 65:67–80 Yeh S-W, Park Y-G, Kirtman BP (2006) ENSO amplitude changes in climate change commitment to atmospheric CO2 doubling. Geophys Res Lett 33:L13711. doi:10.1029/2005GL025653 Yeh S-W, Kug J-S, Dewitte B, Kwon M-H, Kirtman BP, Jin F-F (2009) El Nin˜o in a changing climate. Nature 461:511–515. doi:10.1038/nature08316 Yeh S-W, Ham Y-G, Lee J-Y (2012) Changes in the Tropical Pacific SST trend from CMIP3 to CMIP5 and its implication of ENSO. J Clim 25:7764–7771. doi:10.1175/JCLI-D-12-00304.1 Zhang M, Song H (2006) Evidence of deceleration of atmospheric vertical overturning circulation over the tropical Pacific. Geophys Res Lett 33:L12701. doi:10.1029/2006GL025942

123