Abstract

Increases in greenhouse gas emissions are expected to cause changes both in climatic variability in the Pacific linked to El Niño–Southern Oscillation (ENSO) and in long-term average climate. While mean state and variability changes have been studied separately, much less is known about their combined impact or relative importance. Additionally, studies of projected changes in ENSO have tended to focus on changes in, or adjacent to, the Pacific. Here we examine projected changes in climatic conditions during El Niño years and in ENSO-driven precipitation variability in 36 CMIP5 models. The models are forced according to the RCP8.5 scenario in which there are large, unmitigated increases in greenhouse gas concentrations during the twenty-first century. We examine changes over much of the globe, including 25 widely spread regions defined in the IPCC special report Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation (SREX). We confirm that precipitation variability associated with ENSO is projected to increase in the tropical Pacific, consistent with earlier research. We also find that the enhanced tropical Pacific variability drives ENSO-related variability increases in 19 SREX regions during DJF and in 18 during JJA. This externally forced increase in ENSO-driven precipitation variability around the world is on the order of 15%–20%. An increase of this size, although substantial, is easily masked at the regional level by internally generated multidecadal variability in individual runs. The projected changes in El Niño–driven precipitation variability are typically much smaller than projected changes in both mean state and ENSO neutral conditions in nearly all regions.

1. Introduction

El Niño–Southern Oscillation (ENSO) is a naturally occurring phenomenon centered in the equatorial Pacific (Collins et al. 2010) that modulates climatic conditions in many regions around the world (e.g., Ropelewski and Halpert 1989; Power et al. 1999; Diaz and Markgraf 2000; Christensen et al. 2013; Nurse et al. 2014; Murphy et al. 2014; Poveda et al. 2001; Mo 2010; Brönnimann et al. 2007; Zhang et al. 1996, 1999; Shabbar and Khandekar 1996; Nakagawa et al. 2000; Kripalani and Kulkarni 1997; Nurse et al. 2014; Smith et al. 2013; Lough et al. 2016; Australian Bureau of Meteorology and CSIRO 2011a,b). ENSO is active during two phases, El Niño and La Niña, and inactive during so-called neutral years, when climatic conditions in the equatorial Pacific tend to be near their long-term average. “Business-as-usual” increases in greenhouse gas concentrations are expected to cause future changes to both year-to-year precipitation variability in the Pacific linked to ENSO (Power et al. 2013; Cai et al. 2015; Chung et al. 2014; Chung and Power 2016; Watanabe et al. 2014; Seager et al. 2012; Huang and Xie 2015; Huang 2016), as well as changes in long-term average temperature and precipitation in many locations (e.g., Power et al. 2012; Collins et al. 2010; IPCC 2013, 2014; Lough et al. 2016).

While mean state and variability changes have been studied separately, very little is known about their combined impact or relative importance during future El Niño or La Niña years. Bonfils et al. (2015) recently helped to fill this gap by estimating the contribution of changes to both ENSO-driven variability and mean state changes to December–February (DJF) precipitation in the twenty-first century. They concluded that global warming tends to enhance both El Niño– and La Niña–driven precipitation variability in most locations and that twenty-first-century ENSO-driven anomalies, when combined with projected mean state changes, can, in some regions, cause unprecedented precipitation.

Here we extend this work to include examination of changes during June–August (JJA) and in ENSO-related surface temperature variability. We also assess the robustness of their key results for DJF using an alternative approach, and we identify several major features of the results that have not been identified previously. We will, for example, provide a quantitative estimate of the increase in ENSO-driven variability around the world, compare the size of changes in ENSO-driven variability with the size of changes in both mean state and ENSO neutral conditions, and consider the signal-to-noise ratio and the implications it has for future climatic conditions at a regional level.

The framework we use in this investigation is schematically represented in Fig. 1. In this idealized example precipitation in El Niño years during the twentieth century (E20) tends to differ from precipitation experienced during neutral years (N20), and precipitation during the twenty-first century in El Niño years (E21) is different from precipitation experienced during twenty-first-century neutral years (N21). We refer to these differences (i.e., δE20 = E20N20 and δE21 = E21N21) as El Niño–driven deviations throughout this paper. Note that El Niño–driven deviations differ from El Niño–driven anomalies (i.e., differences between E and the average) in places where the average and N differ.

Fig. 1.

Schematic representation of climate variables (e.g., precipitation in a particular region) for the twentieth (blue line) and twenty-first (red line) centuries. The terms E20, L20, and N20 represent twentieth-century composites of precipitation during El Niño, La Niña, and ENSO-neutral years, respectively; E21, L21, and N21 represent their twenty-first century counterparts. The terms δE20 = E20N20 and δE21 = E21N21 represent the magnitude of El Niño–driven precipitation deviations in the twentieth and twenty-first centuries, respectively, and δL20 = L20N20 and δL21 = L21N21 are their La Niña-year counterparts. Also, ΔE = E21E20 represents the change in precipitation during El Niño years between the twenty-first and twentieth centuries, and ΔL = L21L20 represents the change in precipitation during La Niña years between the twenty-first and twentieth centuries. The change in the El Niño– and La Niña–driven precipitation deviations between the twenty-first and twentieth centuries are represented by ΔδE = δE21δE20 and ΔδL = δL21δL20, respectively. With this framework E21 = E20 + δE20 + ΔδE + ΔN, L21 = L20 + δL20 + ΔδL + ΔN, ΔE = E21E20 = ΔδE + ΔN, ΔL = L21L20 = ΔδL + ΔN, and the change in ENSO-driven precipitation variability is Δ(EL) = (E21L21) − (E20L20). Finally, ΔN = N21N20, which is the difference between the twentieth- and twenty-first century composites during ENSO-neutral years.

Fig. 1.

Schematic representation of climate variables (e.g., precipitation in a particular region) for the twentieth (blue line) and twenty-first (red line) centuries. The terms E20, L20, and N20 represent twentieth-century composites of precipitation during El Niño, La Niña, and ENSO-neutral years, respectively; E21, L21, and N21 represent their twenty-first century counterparts. The terms δE20 = E20N20 and δE21 = E21N21 represent the magnitude of El Niño–driven precipitation deviations in the twentieth and twenty-first centuries, respectively, and δL20 = L20N20 and δL21 = L21N21 are their La Niña-year counterparts. Also, ΔE = E21E20 represents the change in precipitation during El Niño years between the twenty-first and twentieth centuries, and ΔL = L21L20 represents the change in precipitation during La Niña years between the twenty-first and twentieth centuries. The change in the El Niño– and La Niña–driven precipitation deviations between the twenty-first and twentieth centuries are represented by ΔδE = δE21δE20 and ΔδL = δL21δL20, respectively. With this framework E21 = E20 + δE20 + ΔδE + ΔN, L21 = L20 + δL20 + ΔδL + ΔN, ΔE = E21E20 = ΔδE + ΔN, ΔL = L21L20 = ΔδL + ΔN, and the change in ENSO-driven precipitation variability is Δ(EL) = (E21L21) − (E20L20). Finally, ΔN = N21N20, which is the difference between the twentieth- and twenty-first century composites during ENSO-neutral years.

In the schematic, the late twenty-first-century El Niño–driven deviation (δE21 = E21N21) differs from its twentieth-century counterpart (δE20 = E20N20), so that ∆δE = δE21δE20 ≠ 0, and precipitation during ENSO neutral years during the twenty-first century (N21) is different from its twentieth-century counterpart (N20). In this paper the symbol Δ is used to indicate changes that occur in response to external forcing. Thus ΔδE, for example, represents the projected change in El Niño–driven deviations. Under this simple framework precipitation during future El Niño years can be expressed as E21 = N20 + δE20 + ∆δE + ∆N, where ∆N = N21N20. Similarly, precipitation during La Niña years in the twenty-first century can be expressed as L21 = N20 + δL20 + ∆δL+∆N = L20 + ∆δL + ∆N, where ∆δL = δL21δL20, and where δL21 = L21N21 and δL20 = L20N20 refer to precipitation deviations during La Niña years in the twenty-first and twentieth centuries, respectively. These symbols, and the relationships between them, are summarized in Table 1.

Table 1.

Symbols used and the interrelationships.

Symbols used and the interrelationships.
Symbols used and the interrelationships.

We use output from CMIP5 models to estimate climatic conditions during ENSO events (we use the phrases ENSO events or years to refer to El Niño and La Niña events or years collectively) in the latter half of the twenty-first century. We examine changes during future El Niño years relative to neutral conditions during the twentieth century. This is a useful perspective, as it provides information on how future conditions will change relative to what has been experienced in the past. We also quantify the contribution of ∆δE and ∆N to the changes. This enables us to provide a much clearer picture of the role of El Niño in future climate than has been provided to date. We will find, for example, that in some regions δE20, ∆δE, and ∆N reinforce each other, producing precipitation extremes that are far greater than those typically experienced during El Niño events in the past.

Much of the discussion above is focused on climatic conditions during future El Niño events, relative to conditions experienced during the twentieth century. It is also useful to know how future climate conditions will vary from year to year in association with ENSO. Previous studies have examined this issue; however, they have tended to focus on changes in (see references above), or adjacent to, the Pacific (e.g., Kug et al. 2010; Meehl and Teng 2007). Here we examine changes to ENSO-driven variability over much of the globe, measured using ∆(EL) = ∆E − ∆L = (∆δE + ∆N) − (∆δL + ∆N) = ∆δE − ∆δL.

We examine changes between 45°S and 45°N, where most people live, and in 25 regions around the world identified previously in the IPCC special report Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation (SREX) (see Fig. 2 and Table 2; IPCC 2012). We consider regions in the Pacific, the Indian Ocean, North America, Central America and Mexico, “small islands” and the Caribbean, South America, Asia, southern Europe and the Mediterranean, Africa, and Australasia.

Fig. 2.

The 21 land-based regions used in this investigation (see Table 2 for abbreviations): North America (western North America, central North America, eastern North America), Central America/Mexico, small islands/Caribbean, South America (the Amazon, western South America, Northeast Brazil, southeastern South America), Asia (western Asia, Tibetan Plateau, East Asia, Southeast Asia), southern Europe/Mediterranean, Africa (Sahara, West Africa, East Africa, southern Africa), and Australasia (North Australia, southern Australia/New Zealand). The four oceanic regions (not shown) are the equatorial Pacific (180°–220°E, 5°S–5°N), southern tropical Pacific (180°–220°E, 15°–5°S), northern tropical Pacific (180°–220°E, 5°–15°N), and Indian Ocean (55°–95°E, 25°S–°N).

Fig. 2.

The 21 land-based regions used in this investigation (see Table 2 for abbreviations): North America (western North America, central North America, eastern North America), Central America/Mexico, small islands/Caribbean, South America (the Amazon, western South America, Northeast Brazil, southeastern South America), Asia (western Asia, Tibetan Plateau, East Asia, Southeast Asia), southern Europe/Mediterranean, Africa (Sahara, West Africa, East Africa, southern Africa), and Australasia (North Australia, southern Australia/New Zealand). The four oceanic regions (not shown) are the equatorial Pacific (180°–220°E, 5°S–5°N), southern tropical Pacific (180°–220°E, 15°–5°S), northern tropical Pacific (180°–220°E, 5°–15°N), and Indian Ocean (55°–95°E, 25°S–°N).

Table 2.

Regions examined (see Fig. 2).

Regions examined (see Fig. 2).
Regions examined (see Fig. 2).

This paper is organized as follows. The models and methods used are described in section 2. This section also defines the 25 regions in which projected changes are analyzed. Model variability is briefly assessed in section 3. The results are provided in section 4, the reasons why ENSO teleconnections change (together with related issues) are discussed in section 5, and the main conclusions are presented in section 6.

2. Methods

a. Models and forcing scenario

Thirty-six models (see Table S1 in the online supplemental material) from the CMIP5 archive (Taylor et al. 2012) forced using the RCP8.5 scenario are used. The details of this business-as-usual scenario are provided elsewhere (Moss et al. 2010; van Vuuren et al. 2011). All coupled climate models and the observations were regridded to a 1.5° latitude × 1.5° longitude grid prior to analysis.

b. Motivation for the framework used

Differences between El Niño–driven deviations (i.e., EN) and El Niño–driven anomalies (i.e., EM, where M is the average), can occur where M and N differ. Differences between M and N can arise in regions where the impact of El Niño on M outweighs the impact of La Niña years (e.g., in the central eastern Pacific; Chung and Power 2016). Consequently EM underestimates the full impact of El Niño on precipitation in such regions. Similarly, the impact of La Niña on M outweighs the impact of El Niño in some regions (e.g., over parts of Australia; Power et al. 2006; Chung and Power 2017), and so LM underestimates the full impact of La Niña on precipitation in such regions. In both cases the impact of El Niño and La Niña on precipitation is therefore more accurately represented by EN (=δE) and LN (=δL), respectively.

In addition, the impact of global warming on precipitation depends upon the phase of ENSO in some regions (Chung and Power 2017). We therefore adopt a framework using N as the reference value rather than the average M, as this allows us to unambiguously quantify the impact of ENSO during the three phases of ENSO (i.e., neutral years, El Niño, and La Niña), and it avoids the ambiguity that can arise from asymmetric contributions to M from El Niño and La Niña. This framework differs from that used by Bonfils et al. (2015), as they use the long-term average as their primary reference. We will examine differences that arise between the two methods in section 4.

Bonfils et al. (2015) also employ an analysis method that is different from ours in several ways. For example, they project model precipitation variability onto the observed pattern of precipitation variability, whereas we use modeled patterns without projecting them onto observed variability. They also use a different method to isolate the ENSO variability. They remove the global average temperature and detrend, whereas we use spectral filtering. As it is important to know if these choices have a significant impact on the results, we will compare key results we obtain with the results obtained by Bonfils et al. (2015) for the period they focused on (i.e., DJF).

c. Classification of El Niño, neutral, and La Niña years

The method used to define El Niño, neutral, and La Niña years is the same as that used by Power et al. (2017a). A spectral filter was first used to eliminate climate variability and changes with periods longer than 13 years. Empirical orthogonal function (EOF) analysis (Lorenz 1956) was used to extract the first ENSO pattern in the resulting interannual surface temperature of the observations and every model. The influence of the first EOF on interannual variability in area-averaged averaged sea surface temperature (SST) in the Niño-3.4 region (5°N–5°S, 170°–120°W) was then calculated for the observations and each model. The magnitude and sign of the resulting time series in each model, N3.4*(t) say, was then used to classify years as El Niño, La Niña, or neutral years using N3.4* > 0.8σ20, N3.4* < −0.8σ20 or −0.8 σ20 ≤ N3.4* ≤ +0.8σ20, respectively. Here σ20 is the twentieth-century value of the standard deviation of N3.4*(t) in each model and t is time. The EOF analysis was performed on June–December averages of surface air temperature.

The results presented in this paper are based on the periods 1950–99 and 2050–99, after removing the first and last seven years from each period. These years were removed to avoid possible near-end issues associated with the spectral filtering used to calculate the EOFs. Spectral filtering was only used to define ENSO years. The composites presented are based on unfiltered data.

d. Regions

As mentioned in the introduction, special attention is given to changes in the 25 regions, comprising 21 continental regions (see Fig. 2) and four oceanic regions. The regions are listed in Table 2. The four oceanic regions are the equatorial Pacific (180°–220°E, 5°S–5°N), the southern tropical Pacific (180°–220°E, 15°–5°S), the northern tropical Pacific (180°–220°E, 5°–15°N), and the Indian Ocean (55°–95°E, 25°S–0°).

e. Stippling and stars in plots

Stippling in the maps and stars in the bar plots depicting model results indicate that two-thirds or more of the models agree on the sign of the change. We will refer to this as a robust result.

f. Observational data

We use three observational datasets, two for precipitation [GPCC (Schneider et al. 2015) for land regions and CMAP (Xie and Arkin 1997) for ocean regions] and one for sea surface temperature (HadISST; Rayner et al. 2003).

g. Significance test used in Fig. 3

The assessment of statistical significance of the ENSO signal in each region in Fig. 3 begins by identifying all of the El Niño and La Niña years in the observational surface temperature record using the method outlined above. The observed precipitation that occurs during the El Niño and La Niña years is then identified and ranked in each region. The average rank of the precipitation during El Niño years is then calculated. For the regions over land, the data extend from 1950 to 2012, during which time there are a total of 13 El Niño years and 17 La Niña years. This gives a total of 30 ENSO years and corresponding ranks from 1 to 30. For the regions over ocean the data extend from 1979 to 2012, giving 6 El Niño years and 9 La Niña years and ranks from 1 to 15.

Fig. 3.

Observed time series of JJA precipitation in the 25 regions. El Niño years are indicated with red circles and La Niña years with blue circles. Precipitation was averaged over each region. EOF analysis (as described in the text) was used to identify both El Niño and La Niña years. The percentages represent the proportion of Monte Carlo simulations (time series based on randomly selecting the data) that did not exhibit the ENSO signals as large as that present in the observations. The higher the percentage, the more robust the signal. See section 2 for further details. Data: GPCC (Schneider et al. 2015) for land regions, CMAP (Xie and Arkin 1997) for ocean regions, and Rayner et al. (2003) for SST data used to calculate the EOFs employed to categorize El Niño and La Niña years.

Fig. 3.

Observed time series of JJA precipitation in the 25 regions. El Niño years are indicated with red circles and La Niña years with blue circles. Precipitation was averaged over each region. EOF analysis (as described in the text) was used to identify both El Niño and La Niña years. The percentages represent the proportion of Monte Carlo simulations (time series based on randomly selecting the data) that did not exhibit the ENSO signals as large as that present in the observations. The higher the percentage, the more robust the signal. See section 2 for further details. Data: GPCC (Schneider et al. 2015) for land regions, CMAP (Xie and Arkin 1997) for ocean regions, and Rayner et al. (2003) for SST data used to calculate the EOFs employed to categorize El Niño and La Niña years.

Monte Carlo simulations are then used to produce time series of artificial data of length 30 for land regions and 15 for ocean regions. The simulated time series are then ranked. The average rank of the first 13 years of the artificial time series (for land regions) or 6 years (for ocean regions) is then calculated and compared with the average rank of the observational data in each region. A total of 100 000 simulations are conducted. The percentage given in each subplot indicates the proportion of Monte Carlo simulations with average ranks that are less extreme than the corresponding observational value for El Niño years. The higher the percentage, the less likely it is that the observed rank can be achieved by chance, and the higher the statistical significance of the observed ENSO signal.

3. Model assessment

ENSO modulates precipitation in many of the 25 regions in both the observations and models (Figs. 3 and 4). While the ability of climate models to simulate ENSO has improved (Flato et al. 2013; Power et al. 2013; Bellenger et al. 2014), the simulation of ENSO in the latest generation of climate models is still not perfect (Flato et al. 2013; Power et al. 2013; Bellenger et al. 2014), and this is true for the difference between El Niño and La Niña precipitation deviations (i.e., E20L20; Fig. 4). It is reassuring to note, however, that the models do have a degree of skill in simulating ENSO. This is reflected in the fact that the sign of E20L20 is the same in the observations and the multimodel mean (MMM; Fig. 4) in 23 of the 25 regions during JJA. The equivalent figure for DJF (not shown) is 20. The equivalent figures for El Niño deviations (not shown) alone are 21 out of 25 for both JJA and DJF.

Fig. 4.

Histograms representing the difference between composites of JJA precipitation during El Niño and La Niña years (i.e., EL), for observations (solid) and MMM (striped). Observational data used as in Fig. 3. Observational and modeled composites have the same sign in 23 of the 25 regions. A star above observational composites indicates that the five or more of the six El Niño events that occur during the period analyzed have precipitation deviations of the same sign. A star above modeled composites indicates that the sign of EL is the same in two-thirds or more of the 36 models used (i.e., 24 or more models). This corresponds to a statistical significance level of approximately 95% under the assumption of model independence. Because models exhibit codependencies, the corresponding statistical level is likely overestimated. See Power et al. (2012) for further details.

Fig. 4.

Histograms representing the difference between composites of JJA precipitation during El Niño and La Niña years (i.e., EL), for observations (solid) and MMM (striped). Observational data used as in Fig. 3. Observational and modeled composites have the same sign in 23 of the 25 regions. A star above observational composites indicates that the five or more of the six El Niño events that occur during the period analyzed have precipitation deviations of the same sign. A star above modeled composites indicates that the sign of EL is the same in two-thirds or more of the 36 models used (i.e., 24 or more models). This corresponds to a statistical significance level of approximately 95% under the assumption of model independence. Because models exhibit codependencies, the corresponding statistical level is likely overestimated. See Power et al. (2012) for further details.

4. Results

Projected changes in climatic conditions during future El Niño years

1) Comparisons between ∆δE and E20N20 (=δE20)

During JJA (Fig. 5; see also Table S2) the magnitude of ∆N exceeds the magnitude of ∆δE, (i.e., |∆N| > |∆δE|), in 24 out of 25 regions. In fact in nearly all regions |∆N| ≫ |∆δE|. The exceptions are southeastern South America, the southern tropical Pacific, and western North America.

Fig. 5.

Histograms representing regional averages of JJA MMM precipitation (mm day−1). From left to right in the histograms (see Table 1 for variable descriptions): ΔN = N21N20, (light gray bars); ΔδE = δE21δE20 (dark gray bars); δE20 = E20N20 (blue bars); and E21N20 = δE20 + ΔδE + ΔN (red bars). Black stars indicate that two-thirds or more of the models agree on the sign of the bar. Red stars indicate that two-thirds or more of the models agree on the sign of the difference between the values represented by the red and blue bars (=E21E20 = ΔδE + ΔN).

Fig. 5.

Histograms representing regional averages of JJA MMM precipitation (mm day−1). From left to right in the histograms (see Table 1 for variable descriptions): ΔN = N21N20, (light gray bars); ΔδE = δE21δE20 (dark gray bars); δE20 = E20N20 (blue bars); and E21N20 = δE20 + ΔδE + ΔN (red bars). Black stars indicate that two-thirds or more of the models agree on the sign of the bar. Red stars indicate that two-thirds or more of the models agree on the sign of the difference between the values represented by the red and blue bars (=E21E20 = ΔδE + ΔN).

Twentieth-century El Niño–driven precipitation deviations (δE20) during JJA (Fig. 5; see Table S2) are reinforced robustly (i.e., in two-thirds or more of models) by ∆δE (i.e., δE20 and ∆δE have the same sign) in the equatorial Pacific. As variability in this region has a major impact on variability via teleconnections to many other regions, we might expect to see widespread robust changes elsewhere. However, reinforcement during JJA is only evident in 14 other regions. Furthermore changes in δE are only robust in four regions outside the equatorial Pacific.

El Niño–driven precipitation deviations during DJF (Fig. 6; Table S3) are reinforced by ∆δE in 16 regions, although the intensification is only robust in three regions.

Fig. 6.

As in Fig. 5, but for DJF.

Fig. 6.

As in Fig. 5, but for DJF.

2) Comparisons between E21N20 and E20N20 (=δE20)

Precipitation during twenty-first-century El Niño events differs more from twentieth-century neutral conditions than does precipitation during twentieth-century El Niño events (Fig. 5; Table S2) in 20 regions. That is, |E21N20| > |δE20| in 20 regions. However, the El Niño–driven deviations are only reinforced by ∆N + ∆δE in 13 of these regions (i.e., the sign of E21N20 is the same as the sign of δE20 in 13 regions). In the other seven regions there is a change of sign between δE20 and E21N20. The regions exhibiting robust changes are identified in Fig. 5 using red stars.

Figure 6 and Table S3 indicate that DJF precipitation during El Niño years becomes more extreme (i.e., |E21N20| > |δE20|) in 18 of the 25 regions, although the El Niño–driven deviations are reinforced by ∆N + ∆δE in only 14 of these regions. In other words, the sign of E21N20 is the same as δE20 in only 14 of these 18 regions. The most robust changes occur in the equatorial Pacific, the southern tropical Pacific, western North America, eastern North America, Central America/Mexico, the West Coast of South America, southeastern South America, western Asia, the Tibetan Plateau, East Asia, Southeast Asia, southern Europe/Mediterranean, West Africa, and East Africa.

3) Maps

Maps of δE20, E21N20, and other key quantities during JJA are presented in Fig. 7. The spatial structure of the simulated E21N20 (Fig. 7c) is similar in many respects to the simulated δE20 (Fig. 7b) but is much larger in magnitude in many regions. This includes much more intense precipitation in the equatorial Pacific and much greater drying over equatorial Southeast Asia, the eastern equatorial Indian Ocean, and Mexico and Central America. The reinforcement of δE20 in much of the equatorial Indo-Pacific arises because both ∆N (Fig. 7d) and ∆δE (Fig. 7e) tend to reinforce the twentieth-century El Niño–driven deviations (δE20; Fig. 7b).

Fig. 7.

Precipitation (mm day−1) maps for JJA showing (a) observational and (b) MMM values of δE20, and MMM values of (c) E21N20 (= δE20 + ΔδE + ΔN), (d) ΔN, and (e) ΔδE. Stippling in (a) indicates that five or more of the six El Niño events that occur during the period analyzed have precipitation deviations of the same sign. Stippling in the remaining panels indicates that the sign of the change is the same in two-thirds or more of the 36 models used (i.e., 24 or more models). Observational data are from CMAP.

Fig. 7.

Precipitation (mm day−1) maps for JJA showing (a) observational and (b) MMM values of δE20, and MMM values of (c) E21N20 (= δE20 + ΔδE + ΔN), (d) ΔN, and (e) ΔδE. Stippling in (a) indicates that five or more of the six El Niño events that occur during the period analyzed have precipitation deviations of the same sign. Stippling in the remaining panels indicates that the sign of the change is the same in two-thirds or more of the 36 models used (i.e., 24 or more models). Observational data are from CMAP.

The changes evident in ∆δE—which measure change in El Niño’s contribution to variability about a changing mean state (i.e., the “volatility”)—are consistent with previous research that focused on changes in ENSO-driven precipitation anomalies relative to a changed background state (Power et al. 2013, 2017a; Chung et al. 2014; Chung and Power 2015, 2016). These studies showed that the intensification of El Niño–driven anomalies primarily arises in the Indo-Pacific from a nonlinear interaction between background warming and largely unchanged El Niño–driven SST anomalies.

During JJA in the twentieth century, El Niño increases precipitation over the central and eastern Pacific (Figs. 7a,b, which show δE20 in the observations and models, respectively). Both the projected changes to both the ENSO-neutral state (∆N) and El Niño deviations (∆δE) reinforce the existing El Niño–driven deviations (δE20) in the equatorial Pacific at this time of the year (i.e., JJA); however, the reinforcement from ∆N is much stronger.

This analysis reveals for the first time that changes to precipitation in future El Niño years during JJA are, relative to twentieth-century neutral conditions, primarily determined by changes in background climate rather than changes in El Niño–driven variability. This is also true for El Niño years during DJF (Fig. 8). It is also evident from comparing Figs. 7 and 8 that ∆δE tends to be greater over the equatorial Pacific during JJA than during DJF. This contrasts with ∆N (and the change in the mean; not shown) which is much larger during JJA than it is during DJF.

Fig. 8.

As in Fig. 7, but for DJF.

Fig. 8.

As in Fig. 7, but for DJF.

4) Comparison with Bonfils et al. (2015)

It is of interest to compare the results with those of Bonfils et al. (2015), who examined changes during DJF, as we have done in Fig. 8. As noted previously, Bonfils et al. (2015) used different techniques to estimate quantities related to ENSO-driven variability. Despite these differences, our results are very similar to those obtained by Bonfils et al. (2015; see their Fig. 4). This indicates that the main results are not sensitive to the details of either analysis method.

The results are similar in part because the projected change in the average [the reference used by Bonfils et al. (2015)] is very similar to the projected change in N (the reference value we use) nearly everywhere (Fig. 9). Given that we also used several different approaches in our analysis to the methods used by Bonfils et al. (2015), we can confirm that the main conclusions of both studies do not depend upon the details of the analysis method. It also indicates that ∆N (i.e., the change in precipitation during ENSO neutral years) can be regarded as a very good approximation to mean state change.

Fig. 9.

Precipitation (mm day−1) maps showing JJA MMM values of (a) ΔN, (b) ΔM, and (c) ΔN − ΔM. Here M is the individual model mean, and ΔM = M21M20. Stippling indicates that the sign of the change is the same in two-thirds or more of the 36 models used (i.e., 24 or more models).

Fig. 9.

Precipitation (mm day−1) maps showing JJA MMM values of (a) ΔN, (b) ΔM, and (c) ΔN − ΔM. Here M is the individual model mean, and ΔM = M21M20. Stippling indicates that the sign of the change is the same in two-thirds or more of the 36 models used (i.e., 24 or more models).

5) Precipitation during El Niño years

The reinforcement of some important features of El Niño precipitation anomaly patterns is very marked. For example, the prominent dipole in the equatorial Indo-Pacific in δE20 (Fig. 7) with drying over equatorial southeast Asia and the eastern Indian Ocean and enhanced precipitation over most of the equatorial Pacific, becomes over 200% more intense. Drying during El Niño becomes well over 200% more intense over central America/Mexico (cf. E21N20 and δE20 in Fig. 7; i.e., cf. Figs. 7b and 7c).

As illustrated in Table S2, in some other regions ∆N has the opposite sign to δE20, and this can reduce the magnitude or even change the sign of precipitation relative to N20 (i.e., of E21N20). In East Asia, for example, precipitation during JJA declines in the twentieth century (i.e., δE20 < 0) are replaced by precipitation increases in the twenty-first century (E21N20 > 0; Fig. 5; Table S2).

These results reveal that in many regions, projected changes in precipitation during ENSO neutral years, in mean state precipitation, and in El Niño–driven precipitation deviations can lead to precipitation amounts during El Niño years that are very different from what has typically been experienced in El Niño years to date (see, e.g., Figs. 4, 68a,b, and 9a,b).

6) Surface temperature

The preceding sentence is true for surface temperature everywhere (Fig. 10), indicating that climatic conditions during future El Niño events—including both temperature and in many places precipitation—will be very different to what has been experienced in the past over the entire surface of Earth (we have examined 45°S–45°N, but we know that surface temperature also increases at higher latitudes). Figures 58 and 10 indicate, however, that this is largely due to ΔN, not ΔδE.

Fig. 10.

As in Fig. 5, but for surface temperature (i.e., histograms representing regional averages of JJA MMM surface temperature).

Fig. 10.

As in Fig. 5, but for surface temperature (i.e., histograms representing regional averages of JJA MMM surface temperature).

Figure 10 also shows that there is a small but robust reduction in the magnitude of El Niño–driven surface temperature deviations in the equatorial Pacific. This will act to partially offset the enhancement of the El Niño–driven precipitation anomalies arising from nonlinearity (Power et al. 2017a).

7) Projected changes in ENSO-driven variability

So far we have examined climatic conditions during future El Niño years relative to N20 and changes in the magnitude of El Niño–driven precipitation and temperature deviations. In this subsection we extend the analysis by examining changes in ENSO-driven variability between twenty-first-century El Niño years and twenty-first-century La Niña years. This variability is an important feature of future climate, as the larger this variability is, the greater the impact the variability will tend to have on managed and natural systems.

A simple and useful measure of ENSO-driven variability is EL (i.e., the difference between precipitation in El Niño and La Niña years) (see Fig. 1 and Table 1). The projected changes in EL are depicted in Fig. 11 (JJA) and Fig. 12 (DJF) and Tables S4 and S5. Note that EL is projected to increase in 18 of the 25 regions during JJA and in 19 regions during DJF. This is consistent with previous studies showing that El Niño–driven (Chung and Power 2015, 2016) and La Niña–driven (Chung and Power 2014, 2016) precipitation anomalies in the Pacific will increase and that precipitation variability generally over the Pacific will tend to increase (e.g., Power et al. 2017a). These results are collectively significant at a high level. It should be noted, however, that the changes in EL are only robust in four regions during JJA and three regions during DJF. All robust changes during JJA and DJF represent an intensification of the twentieth-century EL signal.

Fig. 11.

Histograms representing regional averages of JJA MMM El Niño (δE20 and δE21; blue and red bars on the left side of each panel) and La Niña (δL20 and δL21; two middle bars) precipitation deviations (mm day−1) and ENSO-driven precipitation variability (mm day−1; E20L20 and E21L21; bars on the right side of each panel), in the twentieth and twenty-first centuries.

Fig. 11.

Histograms representing regional averages of JJA MMM El Niño (δE20 and δE21; blue and red bars on the left side of each panel) and La Niña (δL20 and δL21; two middle bars) precipitation deviations (mm day−1) and ENSO-driven precipitation variability (mm day−1; E20L20 and E21L21; bars on the right side of each panel), in the twentieth and twenty-first centuries.

Fig. 12.

As in Fig. 11, but for DJF.

Fig. 12.

As in Fig. 11, but for DJF.

Of the 15 regions with robust EL signals during JJA in the twentieth century, 12 regions have signals that intensify during the twenty-first century, although only four of the increases are robust. None exhibits robust weakening. Of the 20 regions with robust EL signals during DJF in the twentieth century, 17 regions have signals that intensify during the twenty-first century, although only three of the increases are robust. None of these 20 regions exhibits robust weakening.

If we restrict attention to regions beyond the equatorial Pacific with a robust ENSO signal in the twentieth century (see Tables S4 and S5), then there is a 21% median increase in ENSO-driven precipitation variability in JJA in the 14 regions (with increases in 12 of the 14 regions, a result that is statistically significant at the 99.9% level) and a 16% median increase in DJF across 19 regions (with increases in 14 regions, with statistical significance > 99% level).

The results suggest that the robust increase in the magnitude of variability in the tropical Pacific appears strong enough to produce a 15%–20% increase in variability in many regions beyond the tropical Pacific. The results for the regions taken collectively are highly robust in the sense that it extremely unlikely to see such widespread agreement on the sign of change in so many regions by chance. However, when the regions are examined individually, the projected changes are only robust in a very limited number of regions. These last two results (i.e., a collective significance that is very high, but with low statistical significance in individual models at the region level) are consistent with the presence of a pervasive signal (i.e., a widespread increase in ENSO-driven variability) that has a low signal-to-noise ratio at the regional scale. Thus the projected increase in ENSO-driven variability is easily masked in individual models on a regional scale by naturally occurring, internally generated multidecadal variability. Internal variability can occur as random contrasts in ENSO activity on multidecadal time scales, as well as in internal processes not related to ENSO that also influence climate in the regions examined.

8) Changes in (EL)/N

We have seen that N exhibits increases and decreases across the globe, while EL tends to increase in magnitude. It is also of interest to know if E − L tends to increase if it is expressed as a fraction of N. The answer is yes. In JJA, 14 of the 15 regions that have a robust twentieth-century EL signal exhibit an increase in (EL)/N from the late twentieth century to the late twenty-first century, a result that is statistically significant above the 99.99% level. In DJF (EL)/N increases in 13 of the 20 regions that have a robust twentieth-century EL signal (statistically significant above the 94% level). The median percentage increases in (EL)/N are 36% (JJA) and 10% (DJF).

9) |ΔN/N| compared with Δ(EL)/(E − L)

As noted above, |∆N| tends to be much larger than |∆δE|. However, the size of the fractional changes in N and EL show the reverse relationship. In other words, |∆N/N| tends to be much smaller than |∆(EL)/(EL)| in regions where there are robust twentieth-century signals in EL. In such regions the median values of 100|∆N/N| are 7% (JJA) and 6% (DJF), whereas the median values of 100|∆(EL)/(EL)| are 23% (JJA) and 30% (DJF). These findings are consistent with the magnitude of percent increases in projections of precipitation extremes tending to be greater than the magnitude of percent changes in average precipitation (see, e.g., Whetton et al. 2015).

5. Discussion

The primary driver of ENSO teleconnections is the impact that ENSO has on the intensity or position of convective precipitation in the tropical Pacific. Projected changes in ENSO teleconnections can therefore arise if the impact of ENSO on convective precipitation in the tropical Pacific changes. This is in fact what tends to occur in both CMIP3 and CMIP5 models. These models tend to project an intensification of ENSO-driven precipitation in the central-eastern Pacific (Kug et al. 2010; Power et al. 2013; Chung et al. 2014; Chung and Power 2016; Watanabe et al. 2014; Huang and Xie 2015; Huang 2016), and the center of maximum variability tends to shift east near the equator (Kug et al. 2010; Power et al. 2013; Bayr et al. 2014).

The intensification of ENSO-driven convective precipitation in the equatorial Pacific alone would be expected to increase the intensity of ENSO teleconnections, and the eastward shift in the center of action for ENSO has been shown to drive an eastward shift in teleconnections to North America (Kug et al. 2010; Meehl and Teng 2007).

The intensification of ENSO teleconnections in both DJF (Bonfils et al. 2015; section 4) and JJA (section 4) is consistent with the intensification of ENSO-driven variability in the tropical Pacific. However, we saw that while the intensification is collectively significant in the sense that it is evident in, for example, 19 regions of the 25 regions we examined, the intensification was only robust (i.e., evident in two-thirds or more of models) in a small number of regions. The simplest explanation for these results is that the intensification of ENSO teleconnections occurs in response to the enhanced ENSO-driven variability in the tropical Pacific, but the remote enhancement is generally small relative to the magnitude of internal multidecadal variability and is therefore easily swamped by internal multidecadal variability in individual model runs.

Our results are also consistent with those of Perry et al. (2017), who showed that the spatial extent of statistically significant ENSO-driven precipitation and temperature variability is projected to increase in CMIP5 models. This too is expected in response to enhanced ENSO-driven variability in convective precipitation in the tropical Pacific.

ENSO teleconnections might also be modified to some extent because the mean state through which ENSO signals are communicated is also projected to change (see, e.g., IPCC 2014). However, we are not aware of any study that has examined the impact of mean state changes on ENSO teleconnections. This would be a useful topic for further investigation.

6. Conclusions

Precipitation during future El Niño events is a function of 1) current levels of El Niño–driven precipitation variability and projected changes in both 2) El Niño–driven precipitation variability and 3) precipitation in neutral years, as illustrated in Fig. 1. Here we examined the combined impact of these three factors on precipitation during El Niño years around the world during June–August (JJA) and December–February (DJF) under the RCP8.5 scenario in 36 CMIP5 models. We found that projected changes in precipitation during El Niño years are, in general, primarily determined by a balance between factors 1 and 3, with factor 2 typically making only a relatively small contribution.

Precipitation experienced during El Niño events in the latter half of the twenty-first century, relative to twentieth-century neutral conditions, varies markedly from place to place. For example, over the Mediterranean, Northeast Brazil, Central America, parts of Southeast Asia, and southern Australia, twenty-first-century El Niño events tend to be drier than twentieth-century El Niño events. Over East Asia and the Tibetan Plateau, on the other hand, precipitation during El Niño years tends to be higher. Over southern Africa and the Caribbean islands precipitation during El Niño years tends to be lower.

We also found the following:

  • ENSO-driven precipitation variability is projected to increase in 18 of the 25 regions examined during JJA and in 19 of the regions during DJF.

  • Regions with a robust ENSO signal in the twentieth century (i.e., statistically significant correlations between ENSO and regional precipitation) exhibit a median 21% increase in ENSO-driven precipitation variability (with increases in 13 out of 15 regions) in JJA and a median 16% increase in DJF (20 regions, with increases in 15 regions).

  • The multimodel results for the regions, taken collectively, are highly robust, and the DJF result is consistent with the findings of Bonfils et al. (2015).

  • When the regions are examined individually, however, the projected changes in EL (i.e., the difference between precipitation composites in El Niño and La Niña years; see Fig. 1 and Table 1 for definitions of variables used and the relationships between them) are only robust in a very limited number of regions.

  • The last four results are consistent with there being an externally forced increase in ENSO-driven precipitation around the world on the order of 15%–20%. This represents an increase that, although substantial, is small enough to be easily masked at the regional level in individual runs by sampling error arising from internal variability on multidecadal time scales. This internal variability can take the form of random multidecadal changes in the magnitude of ENSO-driven convective precipitation in the tropical Pacific, as well as internal multidecadal variability unrelated to ENSO that influences the regions.

  • As the real world is analogous to an individual model “realization,” it too could exhibit internal multidecadal variability in the late twenty-first century that masks (or reinforces) this externally forced signal. We can say, however, that the external forcing increases the likelihood that the variability in the real world will be intensified under high-emissions scenarios.

  • The magnitude of ENSO-driven precipitation variability, expressed as a fraction of neutral precipitation {i.e., ∆[(EL)/N]}, tends to increase in regions with robust twentieth-century ENSO signals. The increases are on the order of 36% (JJA) and 10% (DJF).

  • In regions that have robust twentieth-century signals in ENSO variability, fractional changes in ENSO-driven variability tend to be much larger than fractional changes in ENSO neutral precipitation in the same regions [i.e., |∆(EL)/(EL)| ≫ |∆N/N|].

  • Projected changes in mean state and neutral precipitation are very similar over the globe. This helps to explain the agreement between the results we obtained for DJF and the results presented by Bonfils et al. (2015) for the same season. This agreement arises despite the fact that the methods used are different. This increases confidence in both the results that Bonfils et al. (2015) present and those presented here.

  • Projected mean state changes in surface temperature are everywhere much larger than corresponding projected changes in El Niño–driven surface temperature variability.

Together, these results indicate that the latter half of the twenty-first century under the RCP8.5 scenario is (conservatively) more likely than not to see enhanced ENSO precipitation variability in many locations about mean state and ENSO neutral conditions that are generally very different from those experienced during the twentieth century.

Caveats and further research

Note that CMIP5 models still exhibit shortcomings in their ability to simulate ENSO and recent multidecadal changes in Pacific climate (e.g., England et al. 2014; Kociuba and Power 2015; Power et al. 2017a). This lowers confidence in the results, even if there is a high degree of agreement among models. It will therefore be useful to examine the key issues raised in this paper in the next generation of climate models, especially if they are better able to simulate ENSO and recent climate change in the Pacific (e.g., Kociuba and Power 2015; Power et al. 2017b). Additional useful areas for future research include

  • explaining why |∆N| tends to be much larger than |∆δE|, and why |∆N/N| tends to be much smaller than |∆(EL)/(EL)|, and

  • clarifying the role of mean state changes in causing changes to teleconnections.

Acknowledgments

This work is supported by the Earth System and Climate Change Hub of the Australian Government’s National Environmental Science Program. We acknowledge the World Climate Research Programme’s Working Group on Coupled Modelling, which is responsible for CMIP, and we thank the climate modeling groups 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 of software infrastructure in partnership with the Global Organization for Earth System Science Portals. We wish to thank Geert Jan van Oldenborgh (KNMI) for the code used to define the regions, as well as anonymous reviewers and Hanh Nguyen for providing comments that helped to improve the manuscript.

REFERENCES

REFERENCES
Australian Bureau of Meteorology and CSIRO
,
2011a
: Regional Overview. Vol. 1, Climate Change in the Pacific: Scientific Assessment and New Research, CSIRO, 257 pp.
Australian Bureau of Meteorology and CSIRO
,
2011b
: Country Reports. Vol. 2, Climate Change in the Pacific: Scientific Assessment and New Research, CSIRO, 273 pp.
Bayr
,
T.
,
D.
Dommenget
,
T.
Martin
, and
S.
Power
,
2014
:
The eastward shift of the Walker circulation in response to global warming and its relationship to ENSO variability
.
Climate Dyn.
,
43
,
2747
2763
, https://doi.org/10.1007/s00382-014-2091-y.
Bellenger
,
H.
,
E.
Guilyardi
,
J.
Leloup
,
M.
Lengaigne
, and
J.
Vialard
,
2014
:
ENSO representation in climate models: From CMIP3 to CMIP5
.
Climate Dyn.
,
42
,
1999
2018
, https://doi.org/10.1007/s00382-013-1783-z.
Bonfils
,
C. J. W.
,
B. D.
Santer
,
T. J.
Phillips
,
K.
Marvel
,
L. R.
Leung
,
C.
Doutriaux
, and
A.
Capotondi
,
2015
:
Relative contributions of mean-state shifts and ENSO-driven variability to precipitation changes in a warming climate
.
J. Climate
,
28
,
9997
10 013
, https://doi.org/10.1175/JCLI-D-15-0341.1.
Brönnimann
,
S.
,
E.
Xoplaki
,
C.
Casty
,
A.
Pauling
, and
J.
Luterbacher
,
2007
:
ENSO influence on Europe during the last centuries
.
Climate Dyn.
,
28
,
181
197
, https://doi.org/10.1007/s00382-006-0175-z.
Cai
,
W.
, and Coauthors
,
2015
:
ENSO and greenhouse warming
.
Nat. Climate Change
,
5
,
849
859
, https://doi.org/10.1038/nclimate2743.
Christensen
,
J. H.
, and Coauthors
,
2013
: Climate phenomena and their relevance for future regional climate change. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 1217–1308.
Chung
,
C. T. Y.
, and
S. B.
Power
,
2014
:
Precipitation response to La Niña and global warming in the Indo-Pacific
.
Climate Dyn.
,
43
,
3293
3307
, https://doi.org/10.1007/s00382-014-2105-9.
Chung
,
C. T. Y.
, and
S. B.
Power
,
2015
:
Modelled rainfall response to strong El Niño sea surface temperature anomalies in the tropical Pacific
.
J. Climate
,
28
,
3133
3151
, https://doi.org/10.1175/JCLI-D-14-00610.1.
Chung
,
C. T. Y.
, and
S. B.
Power
,
2016
:
Modelled impact of global warming on ENSO-driven precipitation changes in the tropical Pacific
.
Climate Dyn.
,
47
,
1303
1323
, https://doi.org/10.1007/s00382-015-2902-9.
Chung
,
C. T. Y.
, and
S. B.
Power
,
2017
:
The non-linear impact of El Niño, La Niña and the Southern Oscillation on seasonal and regional Australian precipitation
.
J. South. Hemisphere Earth Syst. Sci.
,
67
,
25
45
, https://doi.org/10.22499/3.6701.003.
Chung
,
C. T. Y.
,
S. B.
Power
,
J. M.
Arblaster
,
H. A.
Rashid
, and
G. L.
Roff
,
2014
:
Nonlinear precipitation response to El Niño and global warming in the Indo-Pacific
.
Climate Dyn.
,
42
,
1837
1856
, https://doi.org/10.1007/s00382-013-1892-8.
Collins
,
M.
, and Coauthors
,
2010
:
The impact of global warming on the tropical Pacific Ocean and El Niño
.
Nat. Geosci.
,
3
,
391
397
, https://doi.org/10.1038/ngeo868.
Diaz
,
H. F.
, and
V.
Markgraf
, Eds.,
2000
:
El Niño and the Southern Oscillation: Multiscale Variability and Global and Regional Impacts.
Cambridge University Press, 497 pp.
England
,
M. H.
, and Coauthors
,
2014
:
Recent intensification of wind-driven circulation in the Pacific and the ongoing warming hiatus
.
Nat. Climate Change
,
4
,
222
227
, https://doi.org/10.1038/nclimate2106.
Flato
,
G.
, and Coauthors
,
2013
: Evaluation of climate models. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 741–866.
Huang
,
P.
,
2016
:
Time-varying response of ENSO-induced tropical Pacific rainfall to global warming in CMIP5 models. Part I: Multimodel ensemble results
.
J. Climate
,
29
,
5763
5778
, https://doi.org/10.1175/JCLI-D-16-0058.1.
Huang
,
P.
, and
S.-P.
Xie
,
2015
:
Mechanisms of change in ENSO-induced tropical Pacific rainfall variability in a warming climate
.
Nat. Geosci.
,
8
,
922
926
, https://doi.org/10.1038/ngeo2571.
IPCC
,
2012
: Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation. Cambridge University Press, 582 pp., http://www.ipcc.ch/pdf/special-reports/srex/SREX_Full_Report.pdf.
IPCC
,
2013
: Summary for policymakers. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 1–29.
IPCC
,
2014
: Climate Change 2014: Synthesis Report. Cambridge University Press, 151 pp., http://www.ipcc.ch/pdf/assessment-report/ar5/syr/SYR_AR5_FINAL_full_wcover.pdf.
Kociuba
,
G.
, and
S. B.
Power
,
2015
:
Inability of CMIP5 models to simulate recent strengthening of the Walker circulation: Implications for projections
.
J. Climate
,
28
,
20
35
, https://doi.org/10.1175/JCLI-D-13-00752.1.
Kripalani
,
R. H.
, and
A.
Kulkarni
,
1997
:
Climatic impact of El Niño/La Niña on the Indian monsoon: A new perspective
.
Weather
,
52
,
39
46
, https://doi.org/10.1002/j.1477-8696.1997.tb06267.x.
Kug
,
J.-S.
,
S.-I.
An
,
Y.-G.
Ham
, and
I.-S.
Kang
,
2010
:
Changes in El Niño and La Niña teleconnections over North Pacific–America in the global warming simulations
.
Theor. Appl. Climatol.
,
100
,
275
282
, https://doi.org/10.1007/s00704-009-0183-0.
Lorenz
,
E. N.
,
1956
: Empirical orthogonal functions and statistical weather prediction. Massachusetts Institute of Technology Science Rep. 1, 49 pp.
Lough
,
J.
, and Coauthors
,
2016
: Observed and projected changes in surface climate of tropical Pacific Islands. Vulnerability of Pacific Island Agriculture and Forestry to Climate Change, M. Taylor, A. McGregor, and B. Dawson, Eds., Pacific Community, 47–102.
Meehl
,
G. A.
, and
H.
Teng
,
2007
:
Multi-model changes in El Niño teleconnections over North America in a future warmer climate
.
Climate Dyn.
,
29
,
779
790
, https://doi.org/10.1007/s00382-007-0268-3.
Mo
,
K. C.
,
2010
:
Interdecadal modulation of the impact of ENSO on precipitation and temperature over the United States
.
J. Climate
,
23
,
3639
3656
, https://doi.org/10.1175/2010JCLI3553.1.
Moss
,
R. H.
, and Coauthors
,
2010
:
The next generation of scenarios for climate change research and assessment
.
Nature
,
463
,
747
756
, https://doi.org/10.1038/nature08823.
Murphy
,
B. F.
,
S. B.
Power
, and
S.
McGree
,
2014
:
The varied impacts of El Niño–Southern Oscillation on Pacific island climate
.
J. Climate
,
27
,
4015
4036
, https://doi.org/10.1175/JCLI-D-13-00130.1.
Nakagawa
,
M.
, and Coauthors
,
2000
:
Impact of severe drought associated with the 1997–1998 El Niño in a tropical forest in Sarawak
.
J. Trop. Ecol.
,
16
,
355
367
.
Nurse
,
L. A.
, and Coauthors
,
2014
: Small islands. Climate Change 2014: Impacts, Adaptation, and Vulnerability. Part B: Regional Aspects, V. R. Barros et al., Eds., Cambridge University Press, 1613–1654.
Perry
,
S. J.
,
S.
McGregor
,
A.
Sen Gupta
, and
M. H.
England
,
2017
:
Future changes to El Niño–Southern Oscillation temperature and precipitation teleconnections
.
Geophys. Res. Lett.
,
44
,
10 608
10 616
, https://doi.org/10.1002/2017GL074509.
Poveda
,
G.
,
A.
Jaramillo
,
M. M.
Gil
,
N.
Quiceno
, and
R.
Mantilla
,
2001
:
Seasonality in ENSO-related precipitation, river discharges, soil moisture, and vegetation index in Colombia
.
Water Resour. Res.
,
37
,
2169
2178
, https://doi.org/10.1029/2000WR900395.
Power
,
S.
,
T.
Casey
,
C.
Folland
,
A.
Colman
, and
V.
Mehta
,
1999
:
Interdecadal modulation of the impact of ENSO on Australia
.
Climate Dyn.
,
15
,
319
324
, https://doi.org/10.1007/s003820050284.
Power
,
S.
,
M.
Haylock
,
R.
Colman
, and
X.
Wang
,
2006
:
The predictability of interdecadal changes in ENSO activity and ENSO teleconnections
.
J. Climate
,
8
,
2161
2180
, https://doi.org/10.1175/JCLI3868.1.
Power
,
S.
,
F.
Delage
,
R.
Colman
, and
A.
Moise
,
2012
:
Consensus of twenty-first-century rainfall projections in climate models more widespread than previously thought
.
J. Climate
,
25
,
3792
3809
, https://doi.org/10.1175/JCLI-D-11-00354.1.
Power
,
S.
,
F.
Delage
,
C. T. Y.
Chung
,
G.
Kociuba
, and
K.
Keay
,
2013
:
Robust twenty-first-century projections of El Niño and related precipitation variability
.
Nature
,
502
,
541
545
, https://doi.org/10.1038/nature12580.
Power
,
S.
,
F.
Delage
,
C. T. Y.
Chung
,
H.
Ye
, and
B. F.
Murphy
,
2017a
:
Humans have already increased the risk of major disruptions to Pacific rainfall
.
Nat. Commun.
,
8
,
14368
, https://doi.org/10.1038/ncomms14368.
Power
,
S.
,
F.
Delage
,
G.
Wang
,
I.
Smith
, and
G.
Kocuiba
,
2017b
:
Apparent limitations in the ability of CMIP5 climate models to simulate recent multi-decadal change in Pacific surface temperature: Implications for global temperature projections
.
Climate Dyn.
,
49
,
53
69
, https://doi.org/10.1007/s00382-016-3326-x.
Rayner
,
N. A.
,
D. E.
Parker
,
E. B.
Horton
,
C. K.
Folland
,
L. V.
Alexander
,
D. P.
Rowell
,
E. C.
Kent
, and
A.
Kaplan
,
2003
:
Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century
.
J. Geophys. Res.
,
108
,
4407
, https://doi.org/10.1029/2002JD002670.
Ropelewski
,
C. F.
, and
M. S.
Halpert
,
1989
:
Precipitation patterns associated with the high index phase of the Southern Oscillation
.
J. Climate
,
2
,
268
284
, https://doi.org/10.1175/1520-0442(1989)002<0268:PPAWTH>2.0.CO;2.
Schneider
,
U.
,
A.
Becker
,
P.
Finger
,
A.
Meyer-Christoffer
,
B.
Rudolf
, and
M.
Ziese
,
2015
: GPCC full data reanalysis at 1.0°: Monthly land-surface precipitation from rain-gauges built on GTS-based and historic data, version 7.0. Global Precipitation Climatology Centre, https://doi.org/10.5676/DWD_GPCC/FD_M_V7_100.
Seager
,
R.
,
N.
Naik
, and
L.
Vogel
,
2012
:
Does global warming cause intensified interannual hydroclimate variability?
J. Climate
,
25
,
3355
3372
, https://doi.org/10.1175/JCLI-D-11-00363.1.
Shabbar
,
A.
, and
M.
Khandekar
,
1996
:
The impact of El Niño–Southern Oscillation on the temperature field over Canada: Research note
.
Atmos.–Ocean
,
34
,
401
416
, https://doi.org/10.1080/07055900.1996.9649570.
Smith
,
I.
,
A.
Moise
,
K.
Inape
,
B.
Murphy
,
R.
Colman
,
S. B.
Power
, and
C. T. Y.
Chung
,
2013
:
ENSO-related rainfall changes over the New Guinea region
.
J. Geophys. Res. Atmos.
,
118
,
10 665
10 675
, https://doi.org/10.1002/jgrd.50818.
Taylor
,
K. E.
,
R. J.
Stouffer
, and
G. A.
Meehl
,
2012
:
An overview of the CMIP5 and the experiment design
.
Bull. Amer. Meteor. Soc.
,
93
,
485
498
, https://doi.org/10.1175/BAMS-D-11-00094.1.
van Vuuren
,
D. P.
, and Coauthors
,
2011
:
The representative concentration pathways: An overview
.
Climatic Change
,
109
,
5
31
, https://doi.org/10.1007/s10584-011-0148-z.
Watanabe
,
M.
,
Y.
Kamae
, and
M.
Kimoto
,
2014
:
Robust increase of the equatorial Pacific rainfall and its variability in a warmed climate
.
Geophys. Res. Lett.
,
41
,
3227
3232
, https://doi.org/10.1002/2014GL059692.
Whetton
,
P.
, and Coauthors
, Eds.,
2015
: Climate change in Australia: Projections for Australia’s NRM regions. CSIRO and Bureau of Meteorology Tech. Rep., 216 pp.
Xie
,
P.
, and
P. A.
Arkin
,
1997
:
Global precipitation: A 17-year monthly analysis based on gauge observations, satellite estimates, and numerical model outputs
.
Bull. Amer. Meteor. Soc.
,
78
,
2539
2558
, https://doi.org/10.1175/1520-0477(1997)078<2539:GPAYMA>2.0.CO;2.
Zhang
,
R.
,
A.
Sumi
, and
M.
Kimoto
,
1996
:
Impact of El Niño on the East Asian monsoon: A diagnostic study of the ’86/87 and ’91/92 events
.
J. Meteor. Soc. Japan
,
74
,
49
62
, https://doi.org/10.2151/jmsj1965.74.1_49.
Zhang
,
R.
,
A.
Sumi
, and
M.
Kimoto
,
1999
:
A diagnostic study of the impact of El Niño on the precipitation in China
.
Adv. Atmos. Sci.
,
16
,
229
241
, https://doi.org/10.1007/BF02973084.

Footnotes

Supplemental information related to this paper is available at the Journals Online website: https://doi.org/10.1175/JCLI-D-18-0138.s1.

© 2018 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Supplemental Material