Changes in ENSO Characteristics in Model Simulations with Considerably Altered Background Climate States

Thea Siuts aGEOMAR Helmholtz Centre for Ocean Research Kiel, Kiel, Germany

Search for other papers by Thea Siuts in
Current site
Google Scholar
PubMed
Close
,
Tobias Bayr aGEOMAR Helmholtz Centre for Ocean Research Kiel, Kiel, Germany

Search for other papers by Tobias Bayr in
Current site
Google Scholar
PubMed
Close
, and
Joke F. Lübbecke aGEOMAR Helmholtz Centre for Ocean Research Kiel, Kiel, Germany
bChristian-Albrechts-University Kiel, Germany

Search for other papers by Joke F. Lübbecke in
Current site
Google Scholar
PubMed
Close
https://orcid.org/0000-0002-7839-3284
Free access

Abstract

Changes in the background climate are known to affect El Niño–Southern Oscillation (ENSO) by altering feedbacks that control ENSO’s characteristics. Here, the sensitivity of ENSO variability to the background climate is investigated by utilizing two Community Earth System Model, version 1 (CESM1), simulations in which the solar constant is altered by ±25 W m−2. The resulting stable warm and cold climate mean state simulations differ in terms of ENSO amplitude, frequency, diversity, asymmetry, and seasonality. In the warm run, ENSO reveals a larger amplitude and occurs at higher frequencies relative to the cold and control runs as well as observations. The warm run also features more eastern Pacific El Niños, an increased asymmetry, and a stronger seasonal phase locking. These changes are linked to changes in the mean state via the amplifying and damping feedbacks. In the warm run, a shallower mean thermocline results in a stronger subsurface–surface coupling, whereas the cold run reveals reduced ENSO variability due to a reduced Bjerknes feedback in accordance with a deeper mean thermocline and enhanced surface wind stress. A strong zonal advective and upwelling feedback further contribute to the large ENSO amplitude in the run with a warmer mean state. In the cold run, ENSO events are partly forced by anomalous shortwave radiation. However, in light of the large temperature contrast between the simulations of up to 6 K in the tropical Pacific, the relatively small changes in ENSO variability highlight the robustness of ENSO dynamics under vastly different climate mean states.

© 2022 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Joke F. Lübbecke, jluebbecke@geomar.de

Abstract

Changes in the background climate are known to affect El Niño–Southern Oscillation (ENSO) by altering feedbacks that control ENSO’s characteristics. Here, the sensitivity of ENSO variability to the background climate is investigated by utilizing two Community Earth System Model, version 1 (CESM1), simulations in which the solar constant is altered by ±25 W m−2. The resulting stable warm and cold climate mean state simulations differ in terms of ENSO amplitude, frequency, diversity, asymmetry, and seasonality. In the warm run, ENSO reveals a larger amplitude and occurs at higher frequencies relative to the cold and control runs as well as observations. The warm run also features more eastern Pacific El Niños, an increased asymmetry, and a stronger seasonal phase locking. These changes are linked to changes in the mean state via the amplifying and damping feedbacks. In the warm run, a shallower mean thermocline results in a stronger subsurface–surface coupling, whereas the cold run reveals reduced ENSO variability due to a reduced Bjerknes feedback in accordance with a deeper mean thermocline and enhanced surface wind stress. A strong zonal advective and upwelling feedback further contribute to the large ENSO amplitude in the run with a warmer mean state. In the cold run, ENSO events are partly forced by anomalous shortwave radiation. However, in light of the large temperature contrast between the simulations of up to 6 K in the tropical Pacific, the relatively small changes in ENSO variability highlight the robustness of ENSO dynamics under vastly different climate mean states.

© 2022 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Joke F. Lübbecke, jluebbecke@geomar.de

1. Introduction

El Niño–Southern Oscillation (ENSO) is the dominant mode of climate variability on interannual time scales with environmental and societal consequences felt worldwide (McPhaden et al. 2020). Understanding how ENSO might change in the future is thus of high relevance. ENSO dynamics and characteristics have been shown to strongly depend on the tropical Pacific oceanic and atmospheric mean state, which affects the feedbacks governing the growth and damping of El Niño and La Niña events (e.g., Collins et al. 2010; Cai et al. 2015). However, as many compensating feedbacks act at the same time it is difficult to determine the net effect of future warming, and changes in ENSO properties due to mean state changes are the subject of an ongoing debate. In particular, even though models participating in the sixth phase of the Coupled Model Intercomparison Project (CMIP6) have arrived at some consensus on changes in ENSO characteristics, there still is considerable uncertainty on how ENSO amplitude and frequency may change in the future as marked by large differences across models (van Oldenborgh et al. 2005; Merryfield 2006; Guilyardi 2006; Yeh et al. 2009; Latif and Keenlyside 2009; Collins et al. 2010; DiNezio et al. 2012; Stevenson 2012; Santoso et al. 2013; Bellenger et al. 2014; Taschetto et al. 2014; Kim et al. 2014b; Cai et al. 2015; Chen et al. 2017; Beobide-Arsuaga et al. 2021; Hayashi et al. 2020; Fredriksen et al. 2020; Stevenson et al. 2021). Differences between studies might also arise depending on whether a transient climate response or a stabilized new quasi equilibrium is analyzed (Kim et al. 2014a; Callahan et al. 2021). In this study we aim to provide a detailed analysis of changes in the positive and negative ENSO feedbacks in climate model simulations with extremely different background states that are well stabilized in the new equilibrium.

The feedbacks governing ENSO involve ocean dynamics and ocean–atmosphere interactions. A positive sea surface temperature (SST) anomaly in the eastern equatorial Pacific Ocean during El Niño weakens the prevailing easterly trade winds, which further amplifies the SST anomaly through the positive Bjerknes feedback. This occurs through a relaxed equatorial thermocline slope that results in the upwelling of warmer waters. Decreased trade winds also lead to weaker surface currents and less upwelling of cold subsurface water, which increase the warm SST anomaly through the zonal advective and upwelling feedbacks (Timmermann et al. 2018). Damping of the SST anomalies is on average provided by the mean meridional currents and the mean upwelling (dynamical damping) as well as by ocean–atmosphere heat fluxes, mainly the latent heat flux (LH) and the shortwave radiation (SW). Warmer SSTs increase the evaporation and thus lead to a negative LH feedback. Further, the eastward shift of the rising branch of the Walker circulation during El Niño increases the cloud cover over the central and eastern equatorial Pacific, thus leading to a negative SW feedback (Lloyd et al. 2009; Bayr et al. 2018). Anomalies are the inverse for La Niña events. While the positive feedbacks dominate during boreal fall and winter, causing the SST anomalies to grow, the negative feedbacks are strongest during the following early spring, resulting in the decay of an El Niño event (Tziperman et al. 1998; Wengel et al. 2018; Bayr et al. 2021). This seasonality of the amplifying and damping feedbacks leads to the phase locking of El Niño and La Niña events to boreal winter (Wengel et al. 2018; Timmermann et al. 2018).

The tropical Pacific mean state can be altered on the one hand by internal variability (e.g., Wittenberg et al. 2014; Lübbecke and McPhaden 2014) and on the other hand by external forcing like in the glacial periods (e.g., DiNezio et al. 2011; Brown et al. 2020) or the current and future global warming due to increased greenhouse gas emissions (e.g., Cai et al. 2015; Brown et al. 2020; McPhaden et al. 2020; Stevenson et al. 2012b). In general, an increase of ENSO activity is expected from increased vertical stratification and a shallower mean thermocline in the eastern tropical Pacific that would lead to enhanced sensitivity of SSTs to ENSO related wind stress forcing (Timmermann 1999; Zhang et al. 2008) and due to weaker equatorial surface currents in response to a slowdown of the Walker circulation (Cai et al. 2015). On the other hand, the projected decreased zonal temperature gradient and increased thermodynamic damping could lead to a decrease in ENSO amplitude (Collins et al. 2010; Cai et al. 2015). There are, however, conflicting results; for example, Kohyama et al. (2018) relate high thermal stratification to weaker ENSO amplitude.

Paleo-ENSO research suggests that El Niño and La Niña events have occurred in every epoch even though their behavior and properties may vary. Fossil coral-based records from Indonesia (Hughen et al. 1999) and Papua New Guinea (Tudhope et al. 2001) indicate that ENSO has been active for at least 130 000 years, and there is evidence for climate ENSO connections spanning back to the ice ages. These records do suggest that ENSO tends to be weaker during glacial periods, which appear to be associated with a more La Niña–like mean state in the tropical Pacific (Ford et al. 2015; Bush and Philander 1998; Zhu et al. 2017). As for the Holocene, coral records suggest highly variable ENSO activity over the past 7000 years but with particularly high variance in the last 50 years (Cobb et al. 2013; Grothe et al. 2019), suggesting an intensification of ENSO under anthropogenic climate change.

In this study, we want to examine the influence of different background climate conditions on ENSO variability using two long simulations of the CESM1 climate model in an approach similar to that of Stevenson et al. (2012a), who investigated 800-yr-long CCSM3.5 simulations with CO2 stabilized at different levels to exclude transient effects. In the simulations considered here, vastly different mean states are generated by changing the solar forcing corresponding to an increase or a reduction of solar energy equivalent to a doubling or halving of the CO2 concentration. The simulations can thus be regarded as extreme cases of mean state changes caused by external forcing. Attributing observed or projected changes in ENSO characteristics to changes in the background state is complicated by large internally generated variability and by the fact that variations on interannual time scales can themselves change the mean state so that a causal relationship can be hard to establish (Fedorov et al. 2020). Further, mean state changes in the observational record and, depending on the scenario and considered time frame, to a certain degree also in future projections are rather small and subject to multidecadal variability and trends. This complicates the interpretation of statistical analysis as only a limited number of ENSO events occur under a certain mean state. Our approach meets the challenge of relatively small and variable background changes by imposing a strong external forcing that results in large changes of the background mean state, and by analyzing a long period of 360 years that is taken after the climate has reached a new equilibrium. It thereby provides an additional perspective on the question of the mean state dependence of ENSO variability. After comparing the mean climate in warm, cold, and control CESM1 simulations with observations, the ENSO characteristics and variability are analyzed in a detailed feedback analysis and addressing the relative roles of dynamical and thermodynamical ENSO forcing. Using climate simulations that are several hundred years long that were run into new quasi equilibria, this study aims to contribute to the understanding of ENSO’s sensitivity to climate background conditions, which is essential in assessing the ENSO response to global warming.

The paper is structured as follows: In section 2 we describe the model and data used in this study. Section 3 shows the difference in the mean states between the warm, cold, and control runs, and the changes in variability are investigated in section 4. The positive and negative feedbacks are analyzed in section 5, and the main results are summarized and discussed in section 6.

2. Data and methods

a. CESM1–CAM5

In this study we use output from the Community Earth System Model, version 1.2.2, which consists of the Community Atmosphere Model, version 5 (CAM5.3), with a horizontal resolution of 1.9° × 2.5° and 30 vertical levels (Stolpe et al. 2019; Neale et al. 2012) coupled to an interactive ocean based on the Los Alamos National Laboratory (LANL) Parallel Ocean Program, version 2 (POP2; Smith et al. 2010), with nominal 1° horizontal resolution and 60 vertical levels varying from 10 m near the surface to 250 m at depth (Hurrell et al. 2013). CESM1–CAM5 consists of fully coupled atmosphere, ocean, land, and sea ice components (Kay et al. 2015) but does not contain an interactive ice sheet model and all experiments are performed without dynamic vegetation (Stolpe et al. 2019).

The simulations analyzed here have been described in detail by Stolpe et al. (2019). In these model experiments, CESM1 has run into new quasi equilibria with an externally imposed perturbation in the radiative energy budget of Earth’s climate system. The solar constant is adjusted by ±25 W m−2 from a long control integration with preindustrial atmospheric composition. Increasing or decreasing the solar constant leads to a radiative forcing equivalent to the one caused by a doubling or halving, respectively, of the preindustrial atmospheric CO2 concentration (Stolpe et al. 2019). The simulations with the solar constant altered by ±25 W m−2 are further referred to as the “warm run” and “cold run,” respectively. For this study, we use 360 years from the end of the control as well as the cold and warm runs corresponding to the model years 2000–2359.

Previous studies using CESM1 and NCAR’s Community Climate System Model (CCSM) have shown that these models realistically simulate ENSO and are well suited to investigate changes in its variability (e.g., under global warming); although ENSO asymmetry is underestimated (Zhang et al. 2017) this is less the case than in most CMIP models (Hayashi et al. 2020). As an example, Zheng et al. (2018) analyzed the CESM large ensemble under the representative concentration pathway (RCP) 8.5 scenario and found an increase in ENSO amplitude that is however subject to a large uncertainty resulting from high internal variability.

b. Observational data and reanalysis products

To compare the three CESM1 simulations with observational data and reanalysis products, we use observed SSTs for the time period 1958–2017 from HadISST (Rayner et al. 2003), subsurface potential temperature data for the same period from HadEN4.2.0 (Good et al. 2013), ocean velocities from SODA 2.0.4 reanalysis (Carton and Giese 2008), and zonal wind stress and heat fluxes for the period 1979–2017 from ERA-Interim reanalysis (Simmons et al. 2007).

c. Definition of El Niño and La Niña events

El Niño and La Niña events are defined as time periods for which the SST anomaly of the Niño-3.4 region exceeds the standard deviation (multiplied by −1 for La Niña) for at least three consecutive months. The month which shows the largest SST anomalies is referred to as the peak month of the event. The Niño-3 region is defined according to Trenberth (1997) as 90°–150°W, 5°S–5°N; the Niño-3.4 region is defined as 120°–170°W, 5°S–5°N; the Niño-4 region is defined as 160°E–150°W, 5°S–5°N.

d. Contributions of thermodynamics and ocean dynamics to the SST change

The relative contributions of thermodynamics and ocean dynamics to the SST change during El Niño growth is estimated by a method proposed and applied by Bayr et al. (2021): The net heat flux Qnet is integrated over the 6 months preceding the maximum of each El Niño and La Niña event:
dSSTtd=1cpρt=6t=0QnetHdt,
where cp = 4000 J kg−1 K−1 is the specific heat capacity at constant pressure of seawater, ρ = 1024 kg m−3 is the average density of seawater, H is the temporally and spatially varying mixed layer depth (MLD) in meters, and t is time in months. The MLD is defined as the depth at which the ocean temperature has decreased by 0.2 K relative to the surface. We normalize the SST change (dSST) and the SST change caused by the heat fluxes (dSSTtd) by dSST at the peak time (t = 0), yielding an SST change per kelvin of warming and enabling comparison of the individual experiments irrespective of their amplitude. The integration is also performed separately for the SW, longwave (LW), sensible heat (SH), and LH fluxes so as to estimate their contributions to the SST tendency. Because the observed SST change is the balance of the SST changes due to ocean dynamics (dSSToc) and net heat flux (dSSTtd), we can estimate the contribution of the ocean dynamics from the results above as
dSSTocdSSTdSSTtd.
In this way, it is possible to quantify the contributions of thermodynamics and wind-driven ocean dynamics during ENSO growth and to assess if ENSO is dynamics-driven and damped by the heat fluxes, as in observations, or also partly heat flux–driven, as can be found in many climate models with a strong equatorial cold tongue. Note that this method does not account for subgrid or high-frequency variability and depends on the definition of the MLD. As shown in Bayr et al. (2019), however, it is able to describe the contribution of thermodynamics and wind-driven ocean dynamics to ENSO growth very reasonably when we compare this method with more sophisticated methods such as the Bjerknes stability index (Jin et al. 2006).

To estimate the uncertainty in the SST change by the heat fluxes, we apply a bootstrapping approach in which we randomly choose 1000 times only two-thirds of the El Niño and La Niña events and calculate the SST change. The uncertainty is shown as error bars, indicating the 90% quantile of the estimated values.

3. Mean state differences

Before comparing the ENSO characteristics in the different model simulations, we take a look at the mean states. Figure 1 shows the SST, wind field, and thermocline depth difference in the tropical Pacific between the last 360 years of the warm and cold simulation. By altering the solar forcing, the background climate conditions in the tropical Pacific Ocean change dramatically. SSTs in the warm run are on average about 4°–7°C higher than in the cold run with the largest differences found in the equatorial cold tongue region accompanied by weaker northeasterly and southeasterly trade winds (Fig. 1a, Table 1). These mean state changes are consistent with the ones reported from simulations with increasing CO2 levels (e.g., Xie et al. 2010; Stevenson et al. 2012a). To compare the SST patterns in the simulations independent of the mean climate background state, the SST relative to the tropical Pacific (120°E–80°W, 20°N–20°S) is calculated as discussed by Izumo et al. (2020) and shown in Figs. 2a–d. The relative SST patterns are in general similar to the observations in all runs with the western Pacific warm pool and eastern equatorial cold tongue due to wind-driven upwelling of colder subsurface waters. Relative warm pool SST is, however, slightly higher in the cold run, the cold tongue extends farther to the west, and lower temperatures prevail off the Peruvian coast. The warmer western Pacific warm pool together with the colder equatorial cold tongue results in a stronger zonal SST gradient in the cold run.

Fig. 1.
Fig. 1.

Tropical Pacific mean state difference between last 360 years of warm and cold simulation of (a) SST and wind stress and (b) thermocline depth calculated as the depth of the maximum vertical temperature gradient.

Citation: Journal of Climate 36, 2; 10.1175/JCLI-D-21-1004.1

Fig. 2.
Fig. 2.

Tropical Pacific mean state of (left) the cold run, (left center) the control simulation, (right center) the warm run, and (right) observations for (a)–(d) SST relative to the tropical Pacific (shading) and surface wind stress (arrows), (e)–(h) zonal ocean surface velocities, (i)–(l) vertical wind at 500-hPa height, and (m)–(p) subsurface potential temperature (shading) and thermocline depth calculated as depth of the strongest vertical temperature gradient (solid gray lines) and as 18°, 23°, and 20°C isotherms (dotted green lines) for the cold, warm run, and control run and observations, respectively.

Citation: Journal of Climate 36, 2; 10.1175/JCLI-D-21-1004.1

Table 1

Comparison of the warm, control, and cold runs, showing differences in the mean state, the variability, and the feedback strengths. Feedbacks indicate the regression strengths of the three Bjerknes feedback components as well as the zonal advective and upwelling feedbacks with the zonal and vertical velocity anomalies taken during El Niño. Here, TD is thermocline depth and TDA is TD anomaly.

Table 1

The surface wind stress patterns (black arrows in Figs. 2a–d) all show the northeasterly and southeasterly trades between 20°N and 20°S converging in the western equatorial Pacific in the warm pool region. The total wind speed is highest in the cold run and weakest in the warm run, in agreement with the difference in the zonal SST gradient (Table 1). The zonal ocean surface velocities (Figs. 2e–h) reveal the wind-driven westward South Equatorial Current (SEC) with cores slightly north and south of the equator, the westward North Equatorial Current (NEC) centered at about 12°N and the eastward North Equatorial Countercurrent (NECC) located at about 5°N. The SEC, which is driven by the easterly winds and contributes to the western Pacific warm pool by transporting warm surface waters into the western equatorial Pacific, is the strongest westward flowing current in all runs. Consistent with the wind speed, all these zonal ocean surface currents are strongest in the cold run and weakest in the warm run.

Following Bony et al. (1997), the vertical wind at 500-hPa height as shown in Figs. 2i–l is a good proxy for the position and strength of the Walker circulation. The Walker circulation is stronger in the cold run and the rising branch is located farther toward the west relative to the control and the warm run.

The thermocline depth is calculated in two different ways: as the depth of the strongest vertical temperature gradient (solid gray line in Figs. 2m–p) and as the depth of an isotherm (dashed green line). A comparison with the depth of the strongest vertical temperature gradient shows that the 18°, 20°, and 23°C isotherms are a good measure for the thermocline depth in the cold, control and warm run, respectively. In the following we will use the maximum vertical temperature gradient when the different simulations are compared with respect to their thermocline depth and z18, z20, and z23, respectively, for the individual feedback calculations in section 5. The equatorial thermocline is deeper in the western Pacific and shoals toward the east. The comparison between the cold and the warm run shows that the mean thermocline is deeper in the cold run, in particular in the western basin (Fig. 1b). This results in a steeper zonal thermocline slope in the cold run (Table 1). In the warm run, the mean thermocline is shallower and the slope in the eastern Pacific flatter.

In summary, the cold run consistently exhibits a stronger zonal SST gradient, stronger wind stress, stronger ocean surface currents, and a steeper thermocline tilt than the warm run. It is noteworthy that the control simulation lies about centrally in between the warm and cold mean state run for all variables considered here, suggesting that the response of the tropical Pacific mean state to the altered forcing might be approximately linear.

4. Differences in variability and ENSO characteristics

In this section, we compare the interannual variability of selected variables associated with ENSO between the control, warm, and cold runs as well as observations. Time series of Niño-3.4 SST anomalies (Fig. 3) reveal distinct differences between the simulations. The variability is clearly highest in the warm run with SST anomalies frequently exceeding ±2.5°C. Such strong events occur much more rarely in the cold run. This difference is reflected in the standard deviation averaged over the Niño-3.4 region, which amounts to 0.85°, 1.30°, and 1.55°C in the cold, control, and warm run, respectively, as compared with 0.80°C in observations. While about 17 events in 100 years match the definition for El Niño events on average in the control simulation, similar to observations, only about 12 events occur in 100 years in the cold run, but about 26 events take place in the warm run. The ENSO frequency in the warm run lies thus within a broad range peaking at periods of 3–4 years, which is close to present-day observations, while in the cold run no such clear maximum is visible in the spectrum (Fig. 4b). The spatial patterns of the SST variability (Figs. 5a–d) show the highest values in the eastern equatorial region, implying strong variability in the SSTs of the equatorial cold tongue that is associated with ENSO. Relative to observations, the region of highest variability extends farther west in the model simulations—in particular, in the cold run. The variability is markedly weaker in the cold run and enhanced in the warm run relative to the control simulation.

Fig. 3.
Fig. 3.

Time series of Niño-3.4 SST anomalies as well as their standard deviation for the (a) warm run, (b) control run, and (c) cold run. Shading indicates the values exceeding ±1 standard deviation.

Citation: Journal of Climate 36, 2; 10.1175/JCLI-D-21-1004.1

Fig. 4.
Fig. 4.

(a) Standard deviation of SST in the Niño-3.4 region for each calendar month, normalized by the annual mean standard deviation, for the observations (green), cold run (blue), warm run (orange), and control run (black). The error bars are indicated by triangles and mark the uncertainty on a 90% confidence level, estimated by a bootstrapping approach as described in the methods section. (b) The spectrum of SST in the Niño-3.4 region.

Citation: Journal of Climate 36, 2; 10.1175/JCLI-D-21-1004.1

Fig. 5.
Fig. 5.

Tropical Pacific variability in (left) the cold run, (left center) the control simulation, (right center) the warm run, and (right) observations, depicted as the standard deviation of (a)–(d) SST (shading) and zonal surface wind stress (contour lines), (e)–(h) zonal ocean surface velocities, (i)–(l) vertical wind at 500-hPa height, and (m)–(p) subsurface potential temperature.

Citation: Journal of Climate 36, 2; 10.1175/JCLI-D-21-1004.1

The zonal ocean velocities show high variability in the equatorial Pacific in all three runs (Figs. 5e–g). The variability is distinctly higher in the warm run and more pronounced in the western Pacific with the SEC varying by up to 50 cm s−1. The vertical winds at 500-hPa height show higher variability in the warm run than in the cold run at the equator (Fig. 5k), whereas the variability is higher in the cold run at 10°N (Fig. 5i). In the cold run, the reduced variability in the Niño-3 region (Fig. 5i) collocates with the subsidence region in the mean state (Fig. 2i).

Observed SST variability in the Niño-3.4 region depends strongly on the season as ENSO events are tightly phase locked to the seasonal cycle: they grow during boreal autumn, peak in winter, and disappear during late winter and spring (Rasmusson and Carpenter 1982). To compare the seasonal phase locking of ENSO between the three model runs and with observations, the normalized standard deviation is calculated as a function of calendar months (Fig. 4a). To estimate the uncertainty of the seasonal variation of the SST variability we apply a bootstrapping approach, randomly choosing 1000 times two-thirds of the years. The triangles indicate the 90% quantile of the estimated values. In observations, the Niño-3.4 SST anomalies tend to peak near the end of the calendar year, with a clear standard deviation maximum in boreal winter and a minimum in early summer. The control run well reproduces the observed seasonal variation, while the cold run exhibits some variability also during boreal summer and thus a weaker seasonal phase locking. In the warm run, the highest variability is found in November and February as compared with the December maximum in observations.

Spatial patterns of the simulated El Niño and La Niña events are presented as composites of normalized SST and surface wind stress anomalies centered about the peak month of the event (Fig. 6). During El Niño (Figs. 6a–d) the trade winds weaken along the equator, and the central and eastern equatorial Pacific warms. At the onset phase of the El Niño event (not shown) strong westerly wind anomalies act against the mean wind in all three runs, consistent with observations. During the peak phase, the SSTs tend to decrease slightly in the western equatorial Pacific and increase significantly in the east accompanied by a weakening of the winds. The positive SST anomalies extend westward from the South American coast and the anomalous winds converge on the positive SST anomalies in the eastern equatorial Pacific, which is known as the Gill–Matsuno-type response to heating in the eastern equatorial Pacific (Matsuno 1966; Gill 1980). Although the general spatial patterns of the El Niño composites look similar in all simulations and observations, there are notable differences in the amplitude and longitudinal extent of the positive SST anomalies: The El Niño amplitude is larger in the warm run than in the cold run, control simulation, and observations. In the cold run, the positive SST anomaly reaches farther west, up to 150°E, while the warm anomalies in the warm run, the control run, and observations are limited to east of 160°–165°E. This pattern difference is reflected in the number of central Pacific (CP) versus eastern Pacific (EP) El Niño events. Relative to the control and warm simulation, more CP events take place in the cold run whereas the warm run has the highest percentage of EP events (Table 1). EP events are here defined by a positive trans-Niño index (TNI; i.e., the normalized SST difference between the Niño-1+2 and Niño-4 region) and CP events by a negative TNI (Trenberth and Stepaniak 2001). The surface wind stress anomalies are most intense in the warm run.

Fig. 6.
Fig. 6.

Composites of SST anomalies (color shading) and surface wind stress anomaly (arrows) normalized by the standard deviation for (left) the cold run, (left center) the control run, (right center) the warm run, and (right) observations for (a)–(d) El Niño events and (e)–(h) La Niña events and (i)–(l) their difference. Statistically insignificant SSTA regions are shaded gray at the 5% level as based on a two-sided Student’s t test.

Citation: Journal of Climate 36, 2; 10.1175/JCLI-D-21-1004.1

For La Niña, the highest SST anomalies are found in the central-to-eastern equatorial Pacific in all simulations, in agreement with observations. In contrast to El Niño, the composite of La Niña events shows a slightly higher amplitude of SST anomalies in the cold run (Figs. 6e–h). The La Niña events are also slightly more broadscale in the cold run with anomalies extending farther off the equator as well as to the west. In combination with the difference in the El Niño patterns this leads to a reduced asymmetry between the ENSO warm and cold phase in the cold run. In the observations, this asymmetry between stronger, more eastward El Niño and weaker, more westward La Niña events is clearly visible in the difference between the composites (Fig. 6l). While not as pronounced as in the observations, it also shows up in the control and warm run (Figs. 6j,k) but is much reduced in the cold run (Fig. 6i). This visual impression is confirmed by the asymmetry parameter α, which is by far lowest in the cold run (Table 1); α is calculated by a quadratic fit on the principal component (PC) time series of the first and second empirical orthogonal functions of tropical Pacific SST anomalies, which is PC2(t) = α[PC1(t)]2 + βPC1(t) + γ (Dommenget et al. 2013; Cai et al. 2020).

5. ENSO feedbacks

To understand the differences in ENSO variability between the simulations, we calculate the individual components of the Bjerknes feedback. We then specifically look at the role of the upwelling and zonal advective feedback and investigate the heat flux damping.

a. Bjerknes feedback

To investigate potential differences in the Bjerknes feedback, a cross-correlation analysis is performed for its individual components. In the case of an amplifying feedback, the cross-correlation function will be rather symmetric and of the same sign for both positive and negative lags. In the case of a weak or damping feedback between the parameters, the cross-correlation function will be antisymmetric or asymmetric (Keenlyside and Latif 2007).

The first part of the Bjerknes feedback (i.e., the relationship between western basin zonal wind stress and eastern basin SST anomalies) is presented in Fig. 7a by the lead–lag correlation between the Niño-4 zonal wind stress anomaly and the Niño-3 SST anomaly. The bell shape of the lead–lag correlation indicates a reinforcing feedback; that is, the western basin surface zonal wind anomalies force the eastern central basin SST anomalies, which then in turn reinforce the wind anomalies. The highest correlation is found for zonal winds preceding the SST anomalies by two months in the cold run and in observations and by one month in the warm run as well as in the control run. The correlation is highest in the warm and lowest in the cold run and also the regression coefficients of 1.06 × 10−2 N m−2 K−1 (warm), 1.01 × 10−2 N m−2 K−1 (control), and 0.75 × 10−2 N m−2 K−1 (cold) indicate that the wind–SST feedback is strongest in the warm run (Table 1). This difference can be explained by the mean state location of the cold tongue. A cold tongue that is located as far west as found in the cold run (Fig. 2a) shifts the deep convection farther to the west (Fig. 2i) and thus reduces the response of the surface wind stress to SST anomalies (Bayr et al. 2020). In addition, the overall colder temperatures reduce the variability of the convection (Fig. 5i), as it is harder to reach the threshold of deep convection in a colder climate.

Fig. 7.
Fig. 7.

Lead–lag correlations: (a) Niño-4 zonal surface wind stress vs Niño-3 SST anomaly; (b) Niño-4 zonal surface wind stress vs Niño-3 thermocline depth anomaly; (c) Niño-3 SST anomaly vs Niño-3 thermocline depth anomaly. The thermocline depth is calculated as the depth of the isotherms (z18, z20, and z23).

Citation: Journal of Climate 36, 2; 10.1175/JCLI-D-21-1004.1

The second element of the Bjerknes feedback is the link between the western basin zonal wind anomaly (Niño-4 τx anomalies) and the eastern basin thermocline depth anomalies (Niño-3 TDA), depicted in Fig. 7b. Consistent with Keenlyside and Latif (2007), the highest correlations are found for surface zonal wind anomalies preceding the eastern thermocline depth variations by about one month in observations. This is also true for the cold run while the peak is broader around zero lag for both the warm run and the control simulation. Correlations are highest in the warm and lowest in the cold run with the control simulation in between. The regression coefficients (Table 1) show, however, a stronger thermocline response to a given wind stress anomaly in the cold run relative to the warm and control runs.

The connection between the zonal wind stress anomalies and the thermocline depth variations is provided by equatorial waves. An eastward-propagating downwelling Kelvin wave is generated by the wind anomalies and depresses the thermocline in the east leading to reduced upwelling and mixing of cooler waters (Timmermann et al. 2018). Hovmöller composites of the thermocline depth anomalies (not shown) reveal eastward propagation structures of a deepening thermocline similar to observations in the warm and the control run while the propagation of the signal is less evident in the cold run.

The third element of the positive Bjerknes feedback is the coupling between the eastern Pacific thermocline depth anomalies (Niño-3 TDA) and the eastern SST anomalies (Niño-3 SSTA) presented in Fig. 7c. The highest correlations are found for thermocline depth variations preceding the SST anomalies by two months in the cold run and the control run and by three months in the warm run. In observations the thermocline depth leads by 1 month, so the temporal offset is longer in the all-model simulations. In general, correlations as well as regression coefficients (Table 1) are higher in the warm run and observations than in the cold run, indicating a stronger subsurface–surface coupling. This might be related to the depth of the climatological thermocline, which is shallower in the eastern Pacific in the warm run and in observations than in the cold run, allowing for a more direct coupling. However, correlation coefficients in the cold run might additionally be lower due to the smaller ENSO amplitude, enhancing the relative role of noise.

In summary, all three components of the Bjerknes feedback are present in the model runs with different mean states but the total feedback appears to be stronger in the warm run, consistent with the enhanced ENSO amplitude found in this simulation. While the thermocline depth response to wind stress is actually larger in the cold run, both the wind response to SST anomalies and the relation between thermocline depth and SST anomalies are stronger in the warm run. The strengths of the individual feedback components can be linked back to the differences in the mean state. The westward extension of the cold tongue and the associated westward shift of the atmospheric deep convection in the cold run weakens the wind stress response to SST anomalies, and the deeper mean thermocline in the cold run inhibits the subsurface–surface coupling.

b. Upwelling feedback

As shown in the previous section, the subsurface–surface coupling is enhanced in the warm run. In this context, local upwelling anomalies that act on the vertical temperature gradient (i.e., wdT/dz) can play an important role as part of the so-called Ekman or upwelling feedback. It has been suggested that a weaker mean stratification in the upper ocean decreases the sensitivity of upwelling to wind forcing, hence decreases the thermocline–subsurface temperature coupling (Xiang et al. 2011; Kim et al. 2014b).

To calculate wdT/dz, we estimated vertical ocean velocities and their anomalies from the divergence of the horizontal surface current field in the Niño-3 region. During an El Niño event, the upwelling is strongly reduced as a response to weaker local winds. The vertical velocity anomalies are found to be largest in the warm run, consistent with the strong surface wind anomalies (Fig. 6c) and the high variability in the surface currents (Fig. 5g). While they are similar in the control run, they are almost two orders of magnitude smaller in the cold run. Potential reasons for that might be the weaker ENSO-related wind variability as well as the more westward location of these wind anomalies in the cold run, which could result in downwelling favorable wind anomalies within the Niño-3 box. The vertical velocity anomalies act on the vertical temperature gradient which is comparatively similar between the runs but strongest in the warm run as well (Fig. 2o), resulting in the largest upwelling feedback contribution in this run (Table 1).

c. Zonal advective feedback

Zonal ocean velocity anomalies are eastward in all three simulations during El Niño events (Fig. 8) corresponding to a weakening of the westward SEC. This weakening of the surface currents is due to positive SST anomalies in the eastern equatorial basin weakening the equatorial surface winds (Fig. 7a). In addition, wind anomalies in the western central Pacific flatten the thermocline along the equator, which also contributes to a weakening of the SEC (Wang and McPhaden 2000; Kessler 2006; McPhaden and Yu 1999). An eastward current anomaly transports warm western Pacific waters eastward, leading to an amplification of the cold tongue warming. Already four months before the peak of the event all runs show a positive eastward zonal velocity anomaly along the equator (Fig. 8). This eastward velocity anomaly acts against the mean zonal current and intensifies with time—in particular, in the warm run. At the peak month of the El Niño event, the eastward velocity anomaly is strongest in the warm run at the equator and extends over the entire Pacific basin (not shown). Although the mean surface currents are distinctly weaker in the warm run, the anomalies during El Niño are much higher and even exceed the mean SEC, hence the surface current reverses and flows eastward during an El Niño event. These zonal current anomalies then act on the zonal SST gradient. Even though this gradient is weakest in the warm run, the combined effect is dominated by the large differences in the current anomalies and udT/dx over the Niño-3.4 region is thus largest in the warm run (Table 1). During La Niña events, the SEC strengthens with the largest anomaly also found in the warm run. The current anomaly is, however, much weaker than during El Niño (Fig. 8), likely contributing to the asymmetry between the ENSO warm and cold phases.

Fig. 8.
Fig. 8.

Zonal ocean surface velocity anomalies during ENSO events at 155°W averaged over 2°N–2°S. Solid lines represent La Niña events, and dash–dotted lines are El Niño events.

Citation: Journal of Climate 36, 2; 10.1175/JCLI-D-21-1004.1

Our results show that the zonal advective feedback has the largest amplitude in the warm run. The weaker trade winds (Fig. 2c) in accordance with weaker ocean velocities (Fig. 2g) in the warm run require only a moderate westward wind forcing to reduce the westward flowing ocean surface currents. Less intense surface currents imply an eastward transport of the mean negative zonal SST gradient, that is, with warmer waters in the west, forcing the eastern Pacific cold tongue to warm (Kessler 2006). In contrast, in the cold run, strong zonal surface winds (Fig. 2a) and ocean velocities (Fig. 2e) prevail. Thus, the impact of westerly wind anomalies on the mean flow is reduced, yielding a comparably weaker zonal advective feedback. In conclusion, in the warm run the zonal advective feedback is likely an important contributor to the enhanced ENSO amplitude, together with the changes in the upwelling and Bjerknes feedbacks described above.

d. Atmospheric heat flux feedback

While the Bjerknes, Ekman, and zonal advective feedback cause the growth of ENSO events, ocean–atmosphere heat fluxes typically damp ENSO related SST anomalies in the central-eastern equatorial Pacific (Jin 1997; Wang and McPhaden 2000). The shortwave and latent heat flux feedbacks are the largest contributors to the damping of net heat flux Qnet (Lloyd et al. 2009). A comparison of the evolution of Qnet, as well as the SW and LH anomalies during both El Niño and La Niña events in the control, warm, and cold runs with atmospheric reanalysis products (Fig. 9), reveals a number of differences that can be related to the different atmospheric mean states. Whereas in the control run the strongest convective response is located in the Niño-4 region as in observations (Figs. 9b,l), it is shifted to the west in the cold run and to the east in the warm run (Figs. 9g,q). This can be explained by the mean state position of the rising branch of the Walker circulation, which is in comparison with the control simulation located farther to the west in the cold run because of the stronger cold tongue and farther to the east in the warm run due to the weaker cold tongue (Fig. 2). The position of the convective response in turn determines the position of the SW feedback, which is also shifted to the west in the cold run and to the east in the warm run in comparison with the control experiment and observations (Figs. 9d,i,n,s). As a result, a positive SW feedback arises in the Niño-3 region in the cold run, which can be observed in many climate models with a strong cold tongue (Lloyd et al. 2009, 2012; Bayr et al. 2018, 2020). This can be explained by enhanced low stratus cloud cover in the Niño-3 region due to the stronger cold tongue, which dissolves when the SSTs rise during El Niño events (Lloyd et al. 2009; Ying and Huang 2016; Stevenson et al. 2021). In agreement with observations, the LH feedback is strongest in the Niño-3 region in all simulations, but the amplitude is strongest in the warm run (Figs. 9e,j,o,t). Therefore, Qnet is very different in the three runs: The control run shows a similar pattern over the Niño-3 and Niño-4 region as in observations, but with overall weaker amplitudes, while in the cold run the Qnet damping is split up into a western and an eastern center and in the warm run the center of the pattern is located in the Niño-3 region (Figs. 9c,h,m,r). These differences can be explained by the position of the SW feedback.

Fig. 9.
Fig. 9.

Composite Hovmöller diagrams of all El Niño and La Niña events together (normalized by Niño-3.4 SST, so that both types of events have the same sign in the anomaly pattern) in the equatorial Pacific (averaged between 5°S and 5°N), with 5-month running mean Niño-3.4 index > 0.5 standard deviations as the selection criterion, for (a)–(e) observations/reanalysis data, (f)–(j) the CESM −25 W m−2 experiment, (k)–(o) the CESM control experiment, and (p)–(t) the CESM +25 W m−2 experiment for (left) equatorial SST, (left center) vertical wind at 500 hPa (omega; negative upward), (center) net heat flux Qnet, (right center) net shortwave radiation, and (right) latent heat flux. All variables are normalized with mean Niño-3.4 SST 3 months before and after the maximum of the events and are centered in time on the month of the maximum of the ENSO events (lag 0). All heat fluxes are defined as positive downward. Note the different color bar range for each heat flux. The dashed lines mark the Niño-3 and Niño-4 regions and the maximum of the ENSO events in time. The outlined box indicates the combined Niño-3.4 and Niño-3 region, for which the contribution of the thermodynamics to the ENSO dynamics is further investigated.

Citation: Journal of Climate 36, 2; 10.1175/JCLI-D-21-1004.1

ENSO dynamics result from an interplay of ocean dynamics and thermodynamics (Jin et al. 2006; Wengel et al. 2018). We can calculate the contribution of the thermodynamics to ENSO related SST change by integrating the heat fluxes during the ENSO growth phase as proposed by Bayr et al. (2021). As we know the absolute SST change, we can estimate the contribution of ocean dynamics to the SST change during ENSO events from the residual, as described in the methods section. We integrate the heat fluxes for observations and the three CESM runs over the combined Niño-3 and Niño-3.4 region (Fig. 10). To get an absolute SST change of +1 K during an ENSO event in observations, a SST change by ocean dynamics of +3.2 K is needed, as the remainder (−2.2 K) is damped away by the atmospheric heat fluxes, mostly by the SW feedback (−0.9 K) and the LH feedback (−1.2 K). The estimates from the CESM experiments suggest that ocean dynamics plays a more important role in the warm run than in the cold run for the ENSO-related SST change (Fig. 10), consistent with the results from the feedback analysis above. In the cold run, the SW feedback acts as an additional forcing (+0.5 K K−1) and is therefore partly compensating the weaker ocean dynamics. ENSO in the cold run thus presents a hybrid of ocean-driven and shortwave-driven ENSO dynamics, which is quite different from observed ENSO dynamics, where the SW feedback acts clearly as a damping (Bayr et al. 2019). This compensating behavior is known from CMIP models with a strong cold bias in the Cold Tongue region (Guilyardi et al. 2009; Bayr et al. 2019; Planton et al. 2021). In the control experiment the SST change by the SW feedback is close to zero (0.0 K K−1), while in the warm run it is a clear damping (−0.6 K K−1). This shift of the SW feedback from a damping in the warm run to a forcing in the cold run can be explained by the more pronounced cold tongue that shifts the Walker circulation to the west and the ITCZ to the north. This in turn leads in the mean state to more low-level stratiform clouds in the Niño-3 region that dissolve when SST increases (Lloyd et al. 2009). The SST damping by the LH feedback is also slightly weaker in the cold run (−0.4 K K−1) than in the control run (−0.6 K K−1) and the warm run (−0.8 K K−1) and therefore also contributes to the overall weaker Qnet damping in the cold run. The LW feedback (−0.2 K K−1) is a bit stronger in the cold run than in the control run (−0.1 K K−1) and is close to zero in the warm run (0.0 K K−1) but is overall much weaker than the SW and LH feedback.

Fig. 10.
Fig. 10.

Average SST change in the combined Niño-3 and Niño-3.4 region at the maximum of all El Niño and La Niña events in observations and the three CESM experiments, due to ocean dynamics (blue bar) and net heat flux (gray bar), further divided into shortwave radiation (yellow bar), longwave radiation (red bar), sensible heat flux (cyan bar), and latent heat flux (green bar). All values are normalized per 1 K total SST change in the combined Niño-3 and Niño-3.4 region. The error bars indicate the 90% confidence interval estimated by a bootstrapping approach, as described in the methods section.

Citation: Journal of Climate 36, 2; 10.1175/JCLI-D-21-1004.1

We note that the cold run also shows a clear westward propagation of the SST anomalies (Fig. 9f), which is common for partly shortwave-driven ENSO dynamics (Dommenget 2010; Bayr et al. 2018, 2019). In the control run there is no longitudinal propagation visible and an eastward propagation can be seen in the warm run (Figs. 9k,p). The weaker phase locking and asymmetry of ENSO is also common for partly shortwave-driven ENSO in climate models (Bayr et al. 2021). In summary we see a continuum of ENSO dynamics in these three experiments, with a clear ocean dynamics-driven ENSO in the warm run on the one side and a hybrid of ocean dynamics-driven and shortwave-driven ENSO in the cold run on the other side. This highlights that even though ENSO dynamics seem to be quite robust over different climate mean states, substantial differences occur in the interplay of ocean dynamics and thermodynamics.

6. Summary and discussion

In this study, we examined the mean state dependence of ENSO using two CESM1 simulations that were run into a new quasi equilibrium with vastly altered background climate states, similar to Stevenson et al. (2012a). Mean SSTs in the eastern equatorial Pacific differ by about 6 K between the warm and the cold run (Fig. 1a). As a result of the maximum difference in the cold tongue region, the warm run presents a weaker mean zonal temperature gradient than the cold run, accompanied by a weaker mean zonal wind stress, a weaker Walker circulation, and weaker zonal ocean surface velocities (Fig. 2). The thermocline in the warm run is shallower and flatter while the vertical stratification is enhanced in the upper ocean (Figs. 2m–o). The mean state changes in the simulation with an increased solar constant generally agree with multimodel mean state changes under greenhouse gas–induced warming such as described by Vecchi and Soden (2007), Collins et al. (2010), and Cai et al. (2015), although the response can differ substantially between individual models and simulations.

In response to the mean state differences, increased ENSO variability is found in the warm run with more frequent El Niño and La Niña events that occur with a higher amplitude. Compared to the cold run, the warm run also features a stronger asymmetry between El Niño and La Niña events, a stronger seasonal phase locking and more EP El Niños. In an analysis of the main feedbacks that are known to govern the growth and damping of ENSO events, we find a stronger wind response to SST anomalies, stronger zonal current anomalies and stronger subsurface–surface coupling in the warm run, consistent with the enhanced ENSO variability. The Qnet damping is increased as well but not enough to counteract the enhanced positive feedbacks.

Many previous studies have linked changes in the tropical Pacific mean state to variations in ENSO variability, partly with contradicting conclusions. A flatter, shallower mean thermocline as found in the warm run has been related to an enhanced SST response to thermocline depth anomalies in Collins et al. (2010) and Philip and van Oldenborgh (2006), which is consistent with our results, while other studies have related increased zonal wind stress and a steeper thermocline tilt to a stronger thermocline feedback (Borlace et al. 2013). We also do not find a stronger thermocline response to wind forcing in the warm run as suggested by some studies. Hu et al. (2013) argued that the thermocline depth and slope are relevant by setting the strength of wind sensitivity to SST variations but that their relation is nonlinear: a thermocline slope that is either too large or too small along the equator could suppress ENSO variability. Warmer SSTs in the Niño-3 region may also contribute to the stronger ENSO variability, as extreme El Niños, during which the ITCZ swings toward the equator, happen more frequently in the warm run, consistent with Stevenson et al. (2021). Because of the large zonal current anomalies, the zonal advective feedback is another important contributor to the enhanced ENSO amplitude in the warm run, consistent with DiNezio et al. (2012). The stronger zonal advective feedback might also contribute to the more pronounced seasonality of ENSO in the warm run, according to a recent study by Chen and Jin (2022), who found that biases in the seasonal phase locking in climate models can be related to a too weak zonal advective feedback contribution to the seasonal SST modulation.

Paleo-ENSO research suggests that ENSO has always been an active climate mode and was operating even during glacial and interglacial times in the distant past. This study compares two climate model simulations with very different background states caused by changes in the solar constant to investigate the relevance of the mean climate for ENSO variability under new equilibrium conditions. Noticeable, even though the mean state is changed dramatically both simulations present ENSO variability that is not fundamentally different from that of the present day. Consistent with paleostudies addressing the LGM, the ENSO variability tends to be weaker in colder climates (Tudhope et al. 2001; Ford et al. 2015; Zhu et al. 2017; Bush and Philander 1998; Wolff et al. 2011; Leduc et al. 2009), associated with strong easterly trades and subsequently a deeper thermocline (Zhu et al. 2017; Fedorov and Philander 2001; Clement et al. 2000; Wang and An 2002). The weaker variability with a less pronounced ENSO asymmetry and phase looking can also be explained by the hybrid of ocean dynamics-driven and shortwave-driven ENSO dynamics, which is a result of the stronger cold tongue and the associated westward shift of the rising branch of the Walker circulation and northward shift of the Hadley cell (Dommenget 2010; Dommenget et al. 2014; Bayr et al. 2018, 2019, 2021). In contrast, in a warmer climate, the ENSO variability increases, indicating that ENSO strength under global warming could also be enhanced. Key climate parameters affecting ENSO variability are the depth of the thermocline in the eastern equatorial Pacific, the strength of the zonal SST gradient along the equator, and the intensity of surface winds near the equator. Besides adding to the findings on the mean state dependence of ENSO, our results highlight the robustness of ENSO dynamics among a vast range of different climate mean state. The robustness of ENSO across different climate states has previously been discussed, for example, by Manucharyan and Fedorov (2014), who conducted sensitivity climate model experiments with varying zonal SST gradients and found consistent ENSO signals also resulting from a compensation of feedbacks, in their case mainly increased upwelling feedback counteracted by increased thermal damping. Our study further suggests that compensating effects between the positive and negative feedbacks lead to a broad continuum of ENSO dynamics, from ocean dynamics-driven ENSO like in present-day observations to a hybrid of ocean dynamics-driven and shortwave-driven ENSO dynamics under strong cold tongue mean states.

Acknowledgments.

The authors thank Dr. Iselin Medhaug and the group of the Institute for Atmospheric and Climate Science, ETH Zurich, for providing the CESM1 model output and thank the anonymous reviewers for their constructive comments. We acknowledge ECMWF and Hadley Center for providing the reanalysis and observation dataset. This work was supported by the Deutsche Forschungsgemeinschaft (DFG) project “Influence of Model Bias on ENSO Projections of the 21st Century” through Grant 429334714.

Data availability statement.

The CESM1 model output used in this study is available online (https://data.iac.ethz.ch/Siuts_etal_2021_ENSOColdWarmMeanstate). HadISST and HadEN4.2.0 can be downloaded from https://www.metoffice.gov.uk/hadobs/, and ERA-Interim reanalysis data are available from https://apps.ecmwf.int/datasets/.

REFERENCES

  • Bayr, T., M. Latif, D. Dommenget, C. Wengel, J. Harlaß, and W. Park, 2018: Mean-state dependence of ENSO atmospheric feedbacks in climate models. Climate Dyn., 50, 31713194, https://doi.org/10.1007/s00382-017-3799-2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bayr, T., C. Wengel, M. Latif, D. Dommenget, J. Lübbecke, and W. Park, 2019: Error compensation of ENSO atmospheric feedbacks in climate models and its influence on simulated ENSO dynamics. Climate Dyn., 53, 155172, https://doi.org/10.1007/s00382-018-4575-7.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bayr, T., D. Dommenget, and M. Latif, 2020: Walker circulation controls ENSO atmospheric feedbacks in uncoupled and coupled climate model simulations. Climate Dyn., 54, 28312846, https://doi.org/10.1007/s00382-020-05152-2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bayr, T., A. Drews, M. Latif, and J. Lübbecke, 2021: The interplay of thermodynamics and ocean dynamics during ENSO growth phase. Climate Dyn., 56, 16811697, https://doi.org/10.1007/s00382-020-05552-4.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bellenger, H., E. Guilyardi, J. Leloup, M. Lengaigne, and J. Vialard, 2014: ENSO representation in climate models: From CMIP3 to CMIP5. Climate Dyn., 42, 19992018, https://doi.org/10.1007/s00382-013-1783-z.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Beobide-Arsuaga, G., T. Bayr, A. Reintges, and M. Latif, 2021: Uncertainty of ENSO-amplitude projections in CMIP5 and CMIP6 models. Climate Dyn., 56, 38753888, https://doi.org/10.1007/s00382-021-05673-4.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bony, S., K. M. Lau, and Y. C. Sud, 1997: Sea surface temperature and large-scale circulation influences on tropical greenhouse effect cloud radiative forcing. J. Climate, 10, 20552077, https://doi.org/10.1175/1520-0442(1997)010<2055:SSTALS>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Borlace, S., W. Cai, and A. Santoso, 2013: Multidecadal ENSO amplitude variability in a 1000-yr simulation of a coupled global climate model: Implications for observed ENSO variability. J. Climate, 26, 93999407, https://doi.org/10.1175/JCLI-D-13-00281.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brown, J. R., and Coauthors, 2020: Comparison of past and future simulations of ENSO in CMIP5/PMIP3 and CMIP6/PMIP4 models. Climate Past, 16, 17771805, https://doi.org/10.5194/cp-16-1777-2020.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bush, A. B., and S. G. H. Philander, 1998: The role of ocean–atmosphere interactions in tropical cooling during the Last Glacial Maximum. Science, 279, 13411344, https://doi.org/10.1126/science.279.5355.1341.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cai, W., and Coauthors, 2015: ENSO and greenhouse warming. Nat. Climate Change, 5, 849859, https://doi.org/10.1038/nclimate2743.

  • Cai, W., B. Ng, T. Geng, L. Wu, A. Santoso, and M. J. McPhaden, 2020: Butterfly effect and a self-modulating El Niño response to global warming. Nature, 585, 6873, https://doi.org/10.1038/s41586-020-2641-x.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Callahan, C. W., C. Chen, M. Regenstein, J. Bloch-Johnson, S. Yang, and E. J. Moyer, 2021: Robust decrease in El Niño/Southern Oscillation amplitude under long-term warming. Nat. Climate Change, 11, 752757, https://doi.org/10.1038/s41558-021-01099-2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Carton, J. A., and B. S. Giese, 2008: A reanalysis of ocean climate using simple ocean data assimilation (SODA). Mon. Wea. Rev., 136, 29993017, https://doi.org/10.1175/2007MWR1978.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chen, C., M. A. Cane, A. T. Wittenberg, and D. Chen, 2017: ENSO in the CMIP5 simulations: Life cycles, diversity, and responses to climate change. J. Climate, 30, 775801, https://doi.org/10.1175/JCLI-D-15-0901.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chen, H., and F.-F. Jin, 2022: Dynamics of ENSO phase-locking and its biases in climate models. Geophys. Res. Lett., 49, e2021GL097603, https://doi.org/10.1029/2021GL097603.

    • Search Google Scholar
    • Export Citation
  • Clement, A. C., R. Seager, and M. A. Cane, 2000: Suppression of El Niño during the mid-Holocene by changes in the Earth’s orbit. Paleoceanography, 15, 731737, https://doi.org/10.1029/1999PA000466.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cobb, K. M., N. Westphal, H. R. Sayani, J. T. Watson, E. D. Lorenzo, H. Cheng, R. L. Edwards, and C. D. Charles, 2013: Highly variable El Niño–Southern Oscillation throughout the Holocene. Science, 339, 6770, https://doi.org/10.1126/science.1228246.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Collins, M., and Coauthors, 2010: The impact of global warming on the tropical Pacific Ocean and El Niño. Nat. Geosci., 3, 391397, https://doi.org/10.1038/ngeo868.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • DiNezio, P. N., A. Clement, G. A. Vecchi, B. Soden, A. J. Broccoli, B. L. Otto-Bliesner, and P. Braconnot, 2011: The response of the Walker circulation to Last Glacial Maximum forcing: Implications for detection in proxies. Paleoceanography, 26, PA3217, https://doi.org/10.1029/2010PA002083.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • DiNezio, P. N., B. P. Kirtman, A. C. Clement, S.-K. Lee, G. A. Vecchi, and A. Wittenberg, 2012: Mean climate controls on the simulated response of ENSO to increasing greenhouse gases. J. Climate, 25, 73997420, https://doi.org/10.1175/JCLI-D-11-00494.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dommenget, D., 2010: The slab ocean El Niño. Geophys. Res. Lett., 37, L20701, https://doi.org/10.1029/2010GL044888.

  • Dommenget, D., T. Bayr, and C. Frauen, 2013: Analysis of the non-linearity in the pattern and time evolution of El Niño Southern Oscillation. Climate Dyn., 40, 28252847, https://doi.org/10.1007/s00382-012-1475-0.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dommenget, D., S. Haase, T. Bayr, and C. Frauen, 2014: Analysis of the slab ocean El Niño atmospheric feedbacks in observed and simulated ENSO dynamics. Climate Dyn., 42, 31873205, https://doi.org/10.1007/s00382-014-2057-0.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fedorov, A. V., and S. G. Philander, 2001: A stability analysis of tropical ocean–atmosphere interactions: Bridging measurements and theory for El Niño. J. Climate, 14, 30863101, https://doi.org/10.1175/1520-0442(2001)014<3086:ASAOTO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fedorov, A. V., S. Hu, A. T. Wittenberg, A. F. Z. Levine, and C. Deser, 2020: ENSO low-frequency modulation and mean state interactions. El Nino Southern Oscillation in a Changing Climate. Geophys. Monogr., Vol. 253, Amer. Geophys. Union, 173–198, https://doi.org/10.1002/9781119548164.ch8.

    • Crossref
    • Export Citation
  • Ford, H. L., A. C. Ravelo, and P. J. Polissar, 2015: Reduced El Niño–Southern Oscillation during the Last Glacial Maximum. Science, 347, 255258, https://doi.org/10.1126/science.1258437.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fredriksen, H., J. Berner, A. C. Subramanian, and A. Capotondi, 2020: How does El Niño–Southern Oscillation change under global warming? A first look at CMIP6. Geophys. Res. Lett., 47, e2020GL090640, https://doi.org/10.1029/2020GL090640.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gill, A. E., 1980: Some simple solutions for heat-induced tropical circulation. Quart. J. Roy. Meteor. Soc., 106, 447462, https://doi.org/10.1002/qj.49710644905.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Good, S. A., M. J. Martin, and N. A. Rayner, 2013: EN4: Quality controlled ocean temperature and salinity profiles and monthly objective analyses with uncertainty estimates. J. Geophys. Res. Oceans, 118, 67046716, https://doi.org/10.1002/2013JC009067.

    • Search Google Scholar
    • Export Citation
  • Grothe, P. R., and Coauthors, 2019: Enhanced El Niño–Southern Oscillation variability in recent decades. Geophys. Res. Lett., 46, e2019GL083906, https://doi.org/10.1029/2019GL083906.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Guilyardi, E., 2006: El Niño–mean state–seasonal cycle interactions in a multi-model ensemble. Climate Dyn., 26, 329348, https://doi.org/10.1007/s00382-005-0084-6.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Guilyardi, E., A. T. Wittenberg, A. Fedorov, M. Collins, C. Wang, A. Capotondi, G. J. van Oldenborgh, and T. Stockdale, 2009: Understanding El Niño in ocean–atmosphere general circulation models: Progress and challenges. Bull. Amer. Meteor. Soc., 90, 325340, https://doi.org/10.1175/2008BAMS2387.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hayashi, M., F.-F. Jin, and M. F. Stuecker, 2020: Dynamics for El Niño–La Niña asymmetry constrain equatorial-Pacific warming pattern. Nat. Commun., 11, 4230, https://doi.org/10.1038/s41467-020-17983-y.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hu, Z. Z., A. Kumar, H. L. Ren, H. Wang, M. L’Heureux, and F.-F. Jin, 2013: Weakened interannual variability in the tropical Pacific Ocean since 2000. J. Climate, 26, 26012613, https://doi.org/10.1175/JCLI-D-12-00265.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hughen, K. A., D. P. Schrag, S. B. Jacobsen, and W. Hantoro, 1999: El Niño during the last interglacial period recorded by a fossil coral from Indonesia. Geophys. Res. Lett., 26, 31293132, https://doi.org/10.1029/1999GL006062.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hurrell, J. W., and Coauthors, 2013: The Community Earth System Model: A framework for collaborative research. Bull. Amer. Meteor. Soc., 94, 13391360, https://doi.org/10.1175/BAMS-D-12-00121.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Izumo, T., J. Vialard, M. Lengaigne, and I. Suresh, 2020: Relevance of relative sea surface temperature for tropical rainfall interannual variability. Geophys. Res. Lett., 47, e2019GL086182, https://doi.org/10.1029/2019GL086182.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jin, F.-F., 1997: An equatorial ocean recharge paradigm for ENSO. Part II: A stripped-down coupled model. J. Atmos. Sci., 54, 830847, https://doi.org/10.1175/1520-0469(1997)054<0830:AEORPF>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jin, F.-F., S. T. Kim, and L. Bejarano, 2006: A coupled-stability index for ENSO. Geophys. Res. Lett., 33, L23708, https://doi.org/10.1029/2006GL027221.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kay, J. E., and Coauthors, 2015: The Community Earth System Model (CESM) Large Ensemble Project: A community resource for studying climate change in the presence of internal climate variability. Bull. Amer. Meteor. Soc., 96, 13331349, https://doi.org/10.1175/BAMS-D-13-00255.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Keenlyside, N. S., and M. Latif, 2007: Understanding equatorial Atlantic interannual variability. J. Climate, 20, 131142, https://doi.org/10.1175/JCLI3992.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kessler, W. S., 2006: The circulation of the eastern tropical Pacific: A review. Prog. Oceanogr., 69, 181217, https://doi.org/10.1016/j.pocean.2006.03.009.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kim, S. T., W. Cai, F.-F. Jin, A. Santoso, L. Wu, E. Guilyardi, and S. I. An, 2014a: Response of El Niño sea surface temperature variability to greenhouse warming. Nat. Climate Change, 4, 786790, https://doi.org/10.1038/nclimate2326.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kim, S. T., W. Cai, F.-F. Jin, and J. Y. Yu, 2014b: ENSO stability in coupled climate models and its association with mean state. Climate Dyn., 42, 33133321, https://doi.org/10.1007/s00382-013-1833-6.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kohyama, T., D. L. Hartmann, and D. S. Battisti, 2018: Weakening of nonlinear ENSO under global warming. Geophys. Res. Lett., 45, 85578567, https://doi.org/10.1029/2018GL079085.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Latif, M., and N. S. Keenlyside, 2009: El Niño/Southern Oscillation response to global warming. Proc. Natl. Acad. Sci. USA, 106, 20 57820 583, https://doi.org/10.1073/pnas.0710860105.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Leduc, G., L. Vidal, O. Cartapanis, and E. Bard, 2009: Modes of eastern equatorial Pacific thermocline variability: Implications for ENSO dynamics over the last glacial period. Paleoceanography, 24, PA3202, https://doi.org/10.1029/2008PA001701.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lloyd, J., E. Guilyardi, H. Weller, and J. M. Slingo, 2009: The role of atmosphere feedbacks during ENSO in the CMIP3 models. Atmos. Sci. Lett., 10, 249254, https://doi.org/10.1002/asl.227.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lloyd, J., E. Guilyardi, and H. Weller, 2012: The role of atmosphere feedbacks during ENSO in the CMIP3 models. Part III: The shortwave flux feedback. J. Climate, 25, 42754293, https://doi.org/10.1175/JCLI-D-11-00178.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lübbecke, J. F., and M. J. McPhaden, 2014: Assessing the twenty-first-century shift in ENSO variability in terms of the Bjerknes stability index. J. Climate, 27, 25772587, https://doi.org/10.1175/JCLI-D-13-00438.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Manucharyan, G. E., and A. V. Fedorov, 2014: Robust ENSO across a wide range of climates. J. Climate, 27, 58365850, https://doi.org/10.1175/JCLI-D-13-00759.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Matsuno, T., 1966: Quasi-geostrophic motions in the equatorial area. J. Meteor. Soc. Japan, 44, 2543, https://doi.org/10.2151/jmsj1965.44.1_25.

  • McPhaden, M. J., and X. Yu, 1999: Equatorial waves and the 1997–98 El Niño. Geophys. Res. Lett., 26, 29612964, https://doi.org/10.1029/1999GL004901.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McPhaden, M. J., A. Santoso, and W. Cai, Eds., 2020: El Niño Southern Oscillation in a Changing Climate. Geophys. Monogr., Vol. 253, Amer. Geophys. Union, 528 pp.

    • Crossref
    • Export Citation
  • Merryfield, W. J., 2006: Changes to ENSO under CO2 doubling in a multimodel ensemble. J. Climate, 19, 40094027, https://doi.org/10.1175/JCLI3834.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Neale, R. B., C.-C. Chen, P. H. Lauritzen, D. L. Williamson, A. J. Conley, A. K. Smith, M. Mills, and H. Morrison, 2012: Description of the NCAR Community Atmosphere Model (CAM 5.0). NCAR Tech. Note NCAR/TN-486+STR, 289 pp.

    • Crossref
    • Export Citation
  • Philip, S. Y., and G. J. van Oldenborgh, 2006: Shifts in ENSO coupling processes under global warming. Geophys. Res. Lett., 33, L11704, https://doi.org/10.1029/2006GL026196.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Planton, Y. Y., and Coauthors, 2021: Evaluating climate models with the CLIVAR 2020 ENSO metrics package. Bull. Amer. Meteor. Soc., 102, E193E217, https://doi.org/10.1175/BAMS-D-19-0337.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rasmusson, E. M., and T. H. Carpenter, 1982: Variations in tropical sea surface temperature and surface wind fields associated with the Southern Oscillation/El Niño. Mon. Wea. Rev., 110, 354384, https://doi.org/10.1175/1520-0493(1982)110<0354:VITSST>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 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.

    • Search Google Scholar
    • Export Citation
  • Santoso, A., S. McGregor, F. F. Jin, W. Cai, M. H. England, S. I. An, M. J. McPhaden, and E. Guilyardi, 2013: Late-twentieth-century emergence of the El Niño propagation asymmetry and future projections. Nature, 504, 126130, https://doi.org/10.1038/nature12683.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Simmons, A., S. Uppala, D. Dee, and S. Kobayashi, 2007: ERA-Interim: New ECMWF reanalysis products from 1989 onwards. ECMWF Newsletter, No. 110, ECMWF, Reading, United Kingdom, 25–35, https://www.ecmwf.int/sites/default/files/elibrary/2006/14615-newsletter-no110-winter-200607.pdf.

    • Crossref
    • Export Citation
  • Smith, R., and Coauthors, 2010: The Parallel Ocean Program (POP) reference manual: Ocean component of the Community Climate System Model (CCSM) and Community Earth System Model (CESM). Los Alamos National Laboratory Rep. LAUR-01853, 141 pp., https://www.cesm.ucar.edu/models/cesm1.0/pop2/doc/sci/POPRefManual.pdf.

    • Crossref
    • Export Citation
  • Stevenson, S., 2012: Significant changes to ENSO strength and impacts in the twenty-first century: Results from CMIP5. Geophys. Res. Lett., 39, L17703, https://doi.org/10.1029/2012GL052759.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stevenson, S., B. Fox-Kemper, and M. Jochum, 2012a: Understanding the ENSO–CO2 link using stabilized climate simulations. J. Climate, 25, 79177936, https://doi.org/10.1175/JCLI-D-11-00546.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stevenson, S., B. Fox-Kemper, M. Jochum, R. Neale, C. Deser, and G. Meehl, 2012b: Will there be a significant change to El Niño in the twenty-first century? J. Climate, 25, 21292145, https://doi.org/10.1175/JCLI-D-11-00252.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stevenson, S., A. T. Wittenberg, J. Fasullo, S. Coats, and B. Otto-Bliesner, 2021: Understanding diverse model projections of future extreme El Niño. J. Climate, 34, 449464, https://doi.org/10.1175/JCLI-D-19-0969.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Stolpe, M. B., I. Medhaug, U. Beyerle, and R. Knutti, 2019: Weak dependence of future global mean warming on the background climate state. Climate Dyn., 53, 50795099, https://doi.org/10.1007/s00382-019-04849-3.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Taschetto, A. S., A. S. Gupta, N. C. Jourdain, A. Santoso, C. C. Ummenhofer, and M. H. England, 2014: Cold tongue and warm pool ENSO events in CMIP5: Mean state and future projections. J. Climate, 27, 28612885, https://doi.org/10.1175/JCLI-D-13-00437.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Timmermann, A., 1999: Detecting the nonstationary response of ENSO to greenhouse warming. J. Atmos. Sci., 56, 23132325, https://doi.org/10.1175/1520-0469(1999)056<2313:DTNROE>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Timmermann, A., and Coauthors, 2018: El Niño–Southern oscillation complexity. Nature, 559, 535545, https://doi.org/10.1038/s41586-018-0252-6.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., 1997: The definition of El Niño. Bull. Amer. Meteor. Soc., 78, 27712777, https://doi.org/10.1175/1520-0477(1997)078<2771:TDOENO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., and D. P. Stepaniak, 2001: Indices of El Niño evolution. J. Climate, 14, 16971701, https://doi.org/10.1175/1520-0442(2001)014<1697:LIOENO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tudhope, A. W., and Coauthors, 2001: Variability in the El Niño–Southern Oscillation through a glacial–interglacial cycle. Science, 291, 15111517, https://doi.org/10.1126/science.1057969.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tziperman, E., M. A. Cane, S. E. Zebiak, Y. Xue, and B. Blumenthal, 1998: Locking of El Niño’s peak time to the end of the calendar year in the delayed oscillator picture of ENSO. J. Climate, 11, 21912199, https://doi.org/10.1175/1520-0442(1998)011<2191:LOENOS>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • van Oldenborgh, G. J., S. Y. Philip, and M. Collins, 2005: El Niño in a changing climate: A multi-model study. Ocean Sci., 1, 8195, https://doi.org/10.5194/os-1-81-2005.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Vecchi, G. A.</