Abstract

The interdecadal change of the mean state and two types of El Niño was investigated based on the analysis of observational data from 1980 to 2010. It was found that easterly trades and sea surface temperature (SST) gradients across the equatorial Pacific undergo a regime change in 1998/99, with enhanced trades and a significant cooling (warming) over tropical eastern (western) Pacific in the later period. Accompanying this mean state change is more frequent occurrence of central Pacific (CP) El Niño during 1999–2010. The diagnosis of air–sea feedback strength showed that atmospheric precipitation and wind responses to CP El Niño are greater than those to the eastern Pacific (EP) El Niño for given a unit SST anomaly (SSTA) forcing. The oceanic response to the same wind forcing, however, is greater in the EP El Niño than in the CP El Niño. A mixed layer heat budget analysis reveals that zonal advection (thermocline change induced vertical advection) primarily contributes to the CP (EP) El Niño growth.

The role of the mean SST zonal gradient in El Niño selection was investigated through idealized numerical experiments. With the increase of the background zonal SST gradient, the anomalous wind and convection response to a specified EP or CP SSTA shift to the west. Such a difference results in a bifurcation of maximum SSTA tendency, as shown from a simple ocean model. The numerical results support the notion that a shift to the La Niño–like interdecadal mean state is responsible for more frequent occurrence of CP-type El Niño.

1. Introduction

The El Niño–Southern Oscillation (ENSO) is a dominant mode in the tropics and involves strong interannual variation of sea surface temperature (SST) over the equatorial central–eastern Pacific. It has been recognized as the basin-wide oscillation mode that originated from the coupled atmosphere–ocean interaction (Rasmusson and Carpenter 1982; Schopf and Suarez 1988; Battisti and Hirst 1989; Philander 1990; Li 1997; Jin 1997).

Several recent studies (e.g., Larkin and Harrison 2005; Ashok et al. 2007; Yu and Kao 2007; Weng et al. 2009; Kao and Yu 2009; Kug et al. 2009) presented observational evidences that there might be two types of El Niño, with one being central Pacific (CP) El Niño and the other being eastern Pacific (EP) El Niño. The former has a greater SST warming in the central equatorial Pacific, whereas the latter belongs to the canonical (conventional) El Niño and has the largest SST anomaly (SSTA) over the eastern equatorial Pacific. The major physical process controlling the two types of El Niño is different. While the EP El Niño warming is mainly attributed to the thermocline feedback (Battisti and Hirst 1989; Jin 1997; Li 1997), the warming for the CP El Niño is dominated by zonal advection feedback (Kug et al. 2009). The onset of CP El Niño might be linked to the extratropical forcing (Yu et al. 2010; Yu and Kim 2011). The CP-type El Niño sometimes was termed as “dateline El Niño” (Larkin and Harrison 2005), “El Niño Modoki” (Ashok et al. 2007), or “warm pool El Niño” (Kug et al. 2009, 2010). Although involving different definitions, they in general describe a similar SSTA feature. It has been shown that the global teleconnection pattern associated with the CP El Niño differs markedly from that of the EP El Niño (Larkin and Harrison 2005; Ashok et al. 2007; Weng et al. 2009; Kim et al. 2009).

It has been noted that the CP-type El Niño events occur more frequently since 1978 (Ashok et al. 2007; Kug et al. 2009). Ashok et al. (2007) attributed such an El Niño change to the weakening of equatorial easterlies in association with weakened zonal SST gradients. The weakening of the equatorial easterlies led to a flatter thermocline and thus a weaker thermocline feedback in the eastern Pacific. This results in more frequent occurrence of the El Niño Modoki events since 1979. Yeh et al. (2009) noted more frequent occurrence of CP El Niño under anthropogenic global warming. The physical interpretation of such a future El Niño change was similar to Ashok et al. (2007): that is, a weakened Walker circulation under global warming leads to a flatter mean thermocline, which favors more frequent occurrence of CP El Niño events.

In contrast to the argument above, Choi et al. (2011), by analyzing long-term coupled model simulations, found a different relationship between the CP-type El Niño and equatorial trade wind: that is, high occurrence regime of CP-type El Niño is associated with strengthened trade wind and zonal mean SST gradient. Such a model result seems consistent with the recent decadal change of the mean state and El Niño behavior. As shown by Feng et al. (2010) and McPhaden et al. (2011), there is a background state change during the first decade of the twenty-first century with the trade winds becoming stronger, thermocline being shallower, and SST being cooler in the eastern Pacific. Such a background change coincided with more frequent occurrence of CP El Niño (McPhaden et al. 2011).

The controversial mean state–El Niño relationship motivates us to conduct further analyses. In this study, we intend to particularly address the following two questions. What is the fundamental difference between CP and EP El Niño in terms of air–sea feedback strength? How does the mean state affect the El Niño behavior? The rest of the paper is organized as follows: Section 2 gives a brief description of the data and numerical model. Section 3 describes the observed relationship between the interdecadal mean state change and the El Niño structure change. Section 4 analyzes the difference of air–sea feedback processes associated with EP and CP El Niño. In section 5, a mixed layer heat budget analysis is conducted for EP and CP El Niño. In section 6, the impact of the mean state on anomalous atmospheric response and equatorial SSTA tendency is investigated through a set of idealized numerical experiments. Section 7 gives the conclusions and discussion.

2. Data and model

Monthly observational and reanalysis datasets from 1980–2010 were used in this study. The atmospheric data are derived from National Centers for Environmental Prediction–Department of Energy Reanalysis 2 (NCEP–DOE R2) at 2.5° × 2.5° horizontal resolution (Kanamitsu et al. 2002). The precipitation data are from Climate Prediction Center (CPC) Merged Analysis of Precipitation (CMAP) data at 2.5° × 2.5° resolution (Xie and Arkin 1997). The SST is from the Hadley Centre Sea Ice and Sea Surface Temperature dataset (HadISST) with 1° × 1° horizontal resolution (Rayner et al. 2003). The ocean reanalysis data are from the Simple Ocean Data Assimilation data version 2.1 (SODA 2.1). This global ocean reanalysis dataset was created by assimilating observed temperature and salinity data into an eddy permitting Parallel Ocean Program (POP) model forced with 40-yr European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-40) winds. The ocean reanalysis dataset covers a period of 1980–2007 with 0.5° × 0.5° horizontal resolution and 40 levels in the vertical (Carton et al. 2005). The surface heat flux dataset is derived from the objectively analyzed air–sea fluxes version 3 (OAFlux v3) at 1° × 1° horizontal resolution. It was derived based on an objective analysis with use of state-of-the-art bulk flux parameterizations (Yu et al. 2008). For this study, we used the surface latent heat and sensible heat fluxes for the period of 1980–2010 and shortwave radiation, longwave radiation, and net surface heat flux for the period of 1983–2008.

In addition to analyzing the atmosphere and ocean data, we also conducted atmospheric general circulation model (AGCM) experiments. The fifth generation of ECHAM developed at the Max Planck Institute for Meteorology (MPI), ECHAM5 (Roeckner et al. 2003) with a horizontal resolution of T42 and 19 vertical sigma levels, was used. The purpose of the numerical modeling experiments is to examine how the background mean state such as mean zonal SST gradient controls the atmospheric precipitation and wind responses to a specified SSTA. A Cane–Zebiak-type (Cane 1979a; Cane 1979b; Zebiak and Cane 1987) ocean model is further employed to examine how the simulated wind field further affects the SSTA tendency.

3. Interdecadal relationship between the mean state and the El Niño type

Figure 1 shows the linear trends of SST, precipitation, 850-hPa wind, sea surface height (SSH), and surface wind stress fields during 1980–2010. Note that the SST exhibits a significant La Niña–like trend in the tropical Pacific, with a cold (warm) SSTA appearing in the eastern (western) equatorial Pacific. The trend of precipitation and lower-tropospheric wind fields resembles a Gill-type (Gill 1980) response to the eastern Pacific cold SSTA, with suppressed convection and easterly anomalies occurring in the central equatorial Pacific and enhanced convection anomalies in the western Pacific and Maritime Continent. The out-of-phase rainfall pattern between the western and central Pacific arises from both the Gill response to the anomalous SST and the change of the Walker circulation. The trend of the surface wind stress is generally consistent with that of the low-level wind, with a maximum anomaly center located in the central equatorial Pacific. In response to the wind stress change, the SSH trend pattern shows a positive (negative) value in the western (eastern) Pacific. This SSH seesaw pattern indicates that in the past 31 yr the thermocline depth tends to shoal (deepen) in the eastern (western) equatorial Pacific. This is consistent with the recent observed sea level trend across the Pacific basin (Feng et al. 2010; McPhaden et al. 2011).

Fig. 1.

Patterns of linear trends for (a) SST (°C yr−1), (b) precipitation (mm day−1 yr−1) and 850-hPa wind (vector; m s−1 yr−1), and (c) SSH (m yr−1) and surface wind stress (vector; N m−2 yr−1). Boxes E (4°S–4°N, 100°–140°W) and W (4°S–4°N, 120°–160°E) are regions of the largest cooling and warming trends at the equator. Shaded areas and vectors (within 10°S–10°N) are regions that pass the 95% confidence level.

Fig. 1.

Patterns of linear trends for (a) SST (°C yr−1), (b) precipitation (mm day−1 yr−1) and 850-hPa wind (vector; m s−1 yr−1), and (c) SSH (m yr−1) and surface wind stress (vector; N m−2 yr−1). Boxes E (4°S–4°N, 100°–140°W) and W (4°S–4°N, 120°–160°E) are regions of the largest cooling and warming trends at the equator. Shaded areas and vectors (within 10°S–10°N) are regions that pass the 95% confidence level.

To examine whether the trend above is a linear process or involve an interdecadal regime change, we plotted the temporal evolution of the equatorial east–west SST difference field (Fig. 2a). Here, the zonal SST difference field is defined as the difference of the SST averaged at box E and at box W (shown in Fig. 1). (A 31-yr mean value of the east–west SST difference has been removed in Fig. 1.) Note that an interdecadal change of the SST difference appears in 1999. Prior to 1999, the mean SST difference is positive, meaning that the eastern (western) Pacific is relatively warmer (cooler) during this period. After 1999, the mean zonal SST difference is negative, indicating that the eastern (western) Pacific is cooler (warmer).

Fig. 2.

(a) Time series of annual mean SST difference between box E and box W shown in Fig. 1a and (b) time series of the PDO index copied from the University of Washington College of the Environment PDO website (http://jisao.washington.edu/pdo/). The thick black line represents a 3-yr running mean curve. The green line denotes the decadal mean value.

Fig. 2.

(a) Time series of annual mean SST difference between box E and box W shown in Fig. 1a and (b) time series of the PDO index copied from the University of Washington College of the Environment PDO website (http://jisao.washington.edu/pdo/). The thick black line represents a 3-yr running mean curve. The green line denotes the decadal mean value.

The change of the mean zonal SST gradient is consistent with the phase transition of the Pacific decadal oscillation (PDO), which also shows a transition from a warm phase to a cold phase around 1999 (Fig. 2b) (see Northwest Fisheries Science Center website at http://www.nwfsc.noaa.gov/research/divisions/fed/oeip/ca-pdo.cfm). Given the robust interdecadal change signals, in the following we separate all datasets into two periods, 1980–98 [interdecadal period 1 (ID1)] and 1999–2010 [interdecadal period 2 (ID2)].

To show whether such an interdecadal change reflects the trend patterns shown in Fig. 1, we plotted ID2-minus-ID1 patterns for SST, precipitation, 850-hPa wind, SSH, and surface wind stress fields (Fig. 3). As one can see, the gross patterns of Figs. 1 and 3 are quite similar.

Fig. 3.

Difference (ID2 minus ID1) patterns of (a) SST (°C), (b) precipitation (mm day−1) and 850-hPa wind (m s−1), and (c) SSH (m) and surface wind stress (N m−2) fields. Shaded areas and vectors (within 10°S–10°N) are regions that pass the 95% confidence level.

Fig. 3.

Difference (ID2 minus ID1) patterns of (a) SST (°C), (b) precipitation (mm day−1) and 850-hPa wind (m s−1), and (c) SSH (m) and surface wind stress (N m−2) fields. Shaded areas and vectors (within 10°S–10°N) are regions that pass the 95% confidence level.

As the mean state exhibits a marked interdecadal change around 1999, how does the interannual SST variability change in the tropical Pacific? Figures 4a,b show the horizontal distribution of standard deviation of the interannual SSTA averaged from September to the following February each year for the ID1 and ID2 periods, respectively. In ID1, the largest interannual variability appears over the eastern equatorial Pacific, whereas in ID2 the maximum SSTA variability appears over the central equatorial Pacific. The magnitude of the SSTA variability in ID2 is much weaker than that in ID1.

Fig. 4.

(left) Standard deviations of the interannual SSTA averaged from September to the following February during (a) ID1 (1980–98) and (b) ID2 (1999–2010). Boxes A (4°S–4°N, 90°–120°W) and B (4°S–4°N, 150°W–180°) are regions of maximum interannual SSTA center at ID1 and ID2. (right) Longitude–time section of the composite SSTA at the equator (averaged from 3°S to 3°N) for (c) the EP El Niño and (d) the CP El Niño. The contour intervals are 0.5° and 0.2°C for (c) and (d), respectively. The developing phase of the El Niño is denoted by a green box.

Fig. 4.

(left) Standard deviations of the interannual SSTA averaged from September to the following February during (a) ID1 (1980–98) and (b) ID2 (1999–2010). Boxes A (4°S–4°N, 90°–120°W) and B (4°S–4°N, 150°W–180°) are regions of maximum interannual SSTA center at ID1 and ID2. (right) Longitude–time section of the composite SSTA at the equator (averaged from 3°S to 3°N) for (c) the EP El Niño and (d) the CP El Niño. The contour intervals are 0.5° and 0.2°C for (c) and (d), respectively. The developing phase of the El Niño is denoted by a green box.

The time series of SST anomalies averaged over box A (4°S–4°N, 90°–120°W) during ID1 was examined (figure not shown). It was found that two major El Niño episodes (i.e., 1982/83 and 1997/98 events) whose amplitudes exceed one standard deviation stand out. These two events are typical EP El Niño events. Similarly, by examining the time series of the averaged SSTA over box B (4°S–4°N, 150°W–180°) in ID2 (figure not shown), we identify four relatively strong El Niño episodes (2002/03, 2004/05, 2006/07, and 2009/10 events), whose amplitude exceed a half of the SSTA standard deviation during the period. The four events were often referred to as the CP-type El Niños in the previous studies (e.g., Kug et al. 2009).

The current observational analysis reveals an interdecadal relationship between the mean state and the El Niño behavior: that is, in ID1 when an El Niño–like interdecadal mean state occurs, the interannual SST variability is dominated by the EP-type El Niño; in ID2, when a La Niña–like mean state appears, CP-type El Niño events occur more frequently. Thus, a key question that needs to be addressed is, how does the mean state change modulate the El Niño behavior? In the subsequent sections we will first reveal the air–sea feedback difference between the CP- and EP-type El Niño and then examine the role of the mean SST zonal gradient on the preferred longitudinal location of maximum SSTA tendency.

4. Air–sea feedback strength associated with CP and EP El Niño

Figures 4c,d show the time–longitude section of the composite SSTA patterns for the EP and CP El Niño. (The EP El Niño composite is based on two major El Niño events in ID1, whereas the CP El Niño composite is based on four El Niño events in ID2.) The common feature between the EP and CP El Niño is the timing of their peak phases in November–January (NDJ). The difference lies in the developing and decaying features. For the EP El Niño, warm SST anomalies develop earlier in the midspring but decay later after the following spring. After decaying, a cold SSTA appears in the central Pacific. For the CP El Niño, warm SST anomalies develop late in midsummer but decay rather quickly within two seasons. The SST decaying appears firstly in the eastern equatorial Pacific, followed by a decaying in the central Pacific.

The cause of the amplitude difference between EP and CP El Niño is attributed to the difference of air–sea feedback strength during the El Niño developing phase. As seen in Figs. 4c,d, July–October is the period of maximum growth for both the EP and CP El Niños. Thus, in the following we will focus on examining the air–sea feedback intensity in the developing phase (July–October).

To measure how strong the atmosphere responds to the EP and CP El Niños, we rely on a composite analysis. There are two ways to compose the SSTA and associated precipitation and wind patterns. A simple way is to include all monthly mean data in the developing months from July to October. Such a composite SSTA pattern, however, consists of warming in both the central and eastern equatorial Pacific for CP El Niño (figure not shown). To highlight the distinctive spatial patterns between the EP and CP El Niño, we apply a new composite method based on the following criteria: for EP El Niño, the monthly SSTA averaged over box A must exceed one standard deviation and must be 2 times greater than the SSTA averaged over box B; for CP El Niño, the monthly SSTA averaged over box B must be greater than a half of the standard deviation and must be 1.33 times greater than the SSTA averaged over box A. Only when the monthly SSTA satisfies the above criteria was this month selected for the composite analysis. Table 1 lists all months selected for the CP and EP El Niño composite. As one can see, 7 (8) months were selected for the EP (CP) El Niño composite. Although the factors used above are somehow arbitrary, our sensitivity test with a change of the factors from 1.33 to 1.2 and from 2 to 1.5 shows similar composite patterns. By applying this more strict selection methodology, we intend to obtain distinctive atmosphere and ocean patterns associated with the two types of El Niño.

Table 1.

Individual months selected for EP and CP El Niño composites during the developing phase.

Individual months selected for EP and CP El Niño composites during the developing phase.
Individual months selected for EP and CP El Niño composites during the developing phase.

Figures 5a–d shows so-composed SST, precipitation, and 850-hPa surface wind patterns for the EP and CP El Niño, respectively. The composite SSTA pattern exhibits a maximum warming in the eastern Pacific (centered at 100°W) for the EP El Niño. For the CP El Niño, a maximum SSTA anomaly appears in the central Pacific (centered at 160°W), with no warm signal over the eastern equatorial Pacific.

Fig. 5.

Composites of (a),(b) SSTA; (c),(d) precipitation and 850-hPa wind anomalies; (e),(f) SSH and surface wind stress anomalies; and (g),(h) 20°C isothermal depth at the equator (averaged over 3°S–3°N) and the vertical profiles of ocean temperature anomalies averaged from selected months during July–October for (left) the EP El Niño and (right) the CP El Niño. The SST contour intervals are 0.5° and 0.2°C for (a) and (b). The precipitation contour interval and the wind vector scale are 2 mm day−1 and 6 m s−1 for (c) but 1 mm day−1 and 3 m s−1 for (d). The contour intervals and vector scales are 0.05 (0.02) m and 0.05 (0.03) m s−1 in (e) (f). The contour interval is 0.5°C in (g) and (h). For the 20°C isothermal depth, the green line represents the climatology at each of the interdecadal periods and the purple line represents El Niño composites during ID1 and ID2. Shaded areas and vectors are regions that pass the 95% confidence level.

Fig. 5.

Composites of (a),(b) SSTA; (c),(d) precipitation and 850-hPa wind anomalies; (e),(f) SSH and surface wind stress anomalies; and (g),(h) 20°C isothermal depth at the equator (averaged over 3°S–3°N) and the vertical profiles of ocean temperature anomalies averaged from selected months during July–October for (left) the EP El Niño and (right) the CP El Niño. The SST contour intervals are 0.5° and 0.2°C for (a) and (b). The precipitation contour interval and the wind vector scale are 2 mm day−1 and 6 m s−1 for (c) but 1 mm day−1 and 3 m s−1 for (d). The contour intervals and vector scales are 0.05 (0.02) m and 0.05 (0.03) m s−1 in (e) (f). The contour interval is 0.5°C in (g) and (h). For the 20°C isothermal depth, the green line represents the climatology at each of the interdecadal periods and the purple line represents El Niño composites during ID1 and ID2. Shaded areas and vectors are regions that pass the 95% confidence level.

Given the distinctive SSTA patterns between the EP and CP El Niño, we examine the corresponding precipitation and wind patterns associated with the two types of El Niño. In association with the EP El Niño SSTA pattern, maximum precipitation anomalies appears to the east of the dateline, while the maximum westerly anomaly is located near the dateline. In association with the CP El Niño SSTA pattern, the positive precipitation anomaly shifts westward and is located around 160°E. A similar westward shift appears in the zonal wind anomaly field. The amplitude of the precipitation and wind anomalies is smaller in the CP El Niño, as the SSTA amplitude is smaller.

The difference in the large-scale wind stress field causes distinctive ocean responses between the two types of El Niño, as shown by the composite maps of anomalous SSH, 20°C isothermal depth, and ocean temperature (Figs. 5e–h). In the EP El Niño, the SSH is anomalous high (low) over the eastern (western) Pacific. The eastward wind stress anomaly converges onto the high SSH center. In the vertical–longitude cross section, the maximum warm center is located at subsurface around 80-m depth in the eastern Pacific. The maximum subsurface temperature anomaly center is consistent with a great deepening of the ocean thermocline depth in situ.

In the CP El Niño, the magnitude of the SSH anomaly is smaller and its center is shifted toward the central Pacific. Consistent with this feature, the subsurface temperature anomaly is also weaker and shifts westward. The maximum subsurface warming center is located at 140°W around 110-m depth. The thermocline depth anomaly is much weaker compared with that of the EP El Niño.

The gross pattern of the composite atmospheric and oceanic fields for the EP and CP El Niño are in general consistent with previous studies (e.g., Ashok et al. 2007; Kao and Yu 2009; Kug et al. 2009). However, due to a more strict composite methodology applied here, the CP and EP SSTA patterns are more distinguishable.

Figure 6 is a schematic diagram that illustrates the spatial phase relationship among anomalous SST, precipitation, and surface zonal wind for the EP and CP El Niño. While for the EP El Niño the SSTA center is located in the far eastern Pacific, the maximum precipitation and surface wind anomaly centers are located near the dateline. This indicates a large spatial phase difference between the SSTA forcing and the atmospheric response. This differs markedly from the CP El Niño composite in which atmospheric precipitation and wind responses are much closer to the SSTA forcing.

Fig. 6.

Schematic diagram illustrating the zonal phase relationship among the SSTA, 850-hPa wind, and precipitation (or convection) anomalies for the (top) EP and (bottom) CP El Niño.

Fig. 6.

Schematic diagram illustrating the zonal phase relationship among the SSTA, 850-hPa wind, and precipitation (or convection) anomalies for the (top) EP and (bottom) CP El Niño.

A critical difference between the EP and CP El Niño, which has a potential dynamic impact on air–sea coupling, is the longitudinal location of the anomalous zonal wind response. In the EP El Niño, maximum zonal wind anomaly appears in the central Pacific, which can force a zonally asymmetric thermocline depth anomaly response (Li 1997). In the CP El Niño, maximum zonal wind anomaly is located in the western Pacific, which favors a zonal mean thermocline depth anomaly response (Kug et al. 2009). As shown in section 6, the wind difference may cause a longitude-dependent SSTA response.

Besides the distinctive spatial phase relationship, the air–sea feedback strength for the EP and CP El Niño could be different. Following Liu et al. (2011), the SST tendency equation in considering only the Bjerknes dynamic air–sea feedback can be written as

 
formula

where T′ denotes the SSTA, denotes the subsurface ocean temperature anomaly, denotes the climatological mean vertical velocity at the base of the ocean mixed layer, and η is the depth of subsurface temperature layer.

Assuming T′ = δTeσt, the growth rate due to the Bjerknes dynamic air–sea feedback may be determined by

 
formula

where σ denotes the growth rate of the SSTA. The term in the right-hand side of Eq. (2) represents the vertical advection of anomalous subsurface temperature by the mean upwelling velocity. Equation (2) may be rewritten as

 
formula

Equation (3) states that the growth rate or the Bjerknes dynamic air–sea feedback strength depends on the mean vertical velocity, the depth of subsurface temperature layer, and a product of the SST-wind and wind-subsurface temperature coupling coefficients during the El Niño developing phase.

In Eq. (3), R(u, T) measures how strong the atmospheric surface zonal wind responds to a unit (i.e., 1°C) SSTA forcing, whereas R(Te, u) measures how strong the ocean subsurface temperature responds to a unit (i.e., 1 m s−1) zonal wind forcing (via the change of the thermocline depth). To quantitatively determine the feedback strength, we calculated the area-averaged values of anomalous SST, precipitation, 850-hPa surface zonal wind, ocean subsurface temperature , and SSH, averaged over their maximum intensity regions. [Here the anomalous SST, SSH, and subsurface temperature are averaged over box A (B) for EP (CP) El Niño; the anomalous zonal wind is averaged over 4°S–4°N, 170°E–170°W (4°S–4°N, 150°–170°E) for EP (CP) El Niño; and the anomalous precipitation is averaged over 4°S–4°N, 180°–160°W (4°S–4°N, 160°E–180°) for EP (CP) El Niño.] Figures 7a,b show that the amplitudes of the area-averaged precipitation, wind, subsurface temperature, and SSH anomalies in the EP El Niño all exceed those of the CP El Niño. This is not surprising because the EP El Niño SSTA is about 3 times as large as the CP El Niño SSTA amplitude.

Fig. 7.

(top) Amplitude of the anomalous SST (°C), precipitation (mm day−1), 850-hPa zonal wind (m s−1), ocean subsurface temperature (Tsub; °C), and SSH (unit: 0.1 m) for the EP (blue bar) and CP (red bar) El Niño composite. Here, anomalous SST, subsurface temperature, and SSH are averaged over either the box A or box B region; the precipitation anomaly is averaged over (4°S–4°N, 180°–160°W) for EP El Niño and over (4°S–4°N, 160°E–180°) for CP El Niño; the wind anomaly is averaged over (4°S–4°N, 170°E–170°W) for EP El Niño and over (4°S–4°N, 150°–170°E) for CP El Niño; the subsurface temperature anomaly is averaged vertically in the maximum variability region from 40 to 80 m (from 90 to 160 m) in EP (CP) El Niño. (bottom) As in (top), but all values are normalized to (left) a unit SSTA and a unit (right) zonal wind anomaly.

Fig. 7.

(top) Amplitude of the anomalous SST (°C), precipitation (mm day−1), 850-hPa zonal wind (m s−1), ocean subsurface temperature (Tsub; °C), and SSH (unit: 0.1 m) for the EP (blue bar) and CP (red bar) El Niño composite. Here, anomalous SST, subsurface temperature, and SSH are averaged over either the box A or box B region; the precipitation anomaly is averaged over (4°S–4°N, 180°–160°W) for EP El Niño and over (4°S–4°N, 160°E–180°) for CP El Niño; the wind anomaly is averaged over (4°S–4°N, 170°E–170°W) for EP El Niño and over (4°S–4°N, 150°–170°E) for CP El Niño; the subsurface temperature anomaly is averaged vertically in the maximum variability region from 40 to 80 m (from 90 to 160 m) in EP (CP) El Niño. (bottom) As in (top), but all values are normalized to (left) a unit SSTA and a unit (right) zonal wind anomaly.

A more interesting measure of air–sea feedback strength is the atmospheric wind response to a unit SSTA forcing [i.e., R(u, T)] and the oceanic subsurface temperature response to a unit zonal wind anomaly forcing [i.e., R(Te, u)]. Let us first examine the atmospheric feedback coefficient R(u, T). Figure 7c shows the values of the area-averaged precipitation and 850-hPa zonal wind with respect to a unit SSTA forcing for both the EP and CP El Niño. It turns out that the magnitude of the precipitation and wind response to a unit SSTA forcing is greater in CP El Niño than in EP El Niño. The ratio of the wind changes between EP and CP El Niño is 1:1.2. This indicates that the atmospheric response to the SSTA in the central Pacific tends to be stronger than that in the eastern Pacific. Physically, the greater response to the CP El Niño is attributed to warmer background SST in the central Pacific. If the ocean response to a unit wind forcing were same between the CP and EP El Niño, then one would expect a stronger CP El Niño than EP El Niño.

A further examination of the oceanic response to a unit wind forcing shows an opposite result (Fig. 7d). For given the same wind forcing, the response of and SSH anomalies is greater in EP El Niño than in CP El Niño. The ratio of R(Te, u) between EP and CP El Niño is 1:0.6. The physical cause of the different subsurface ocean responses is attributed to the tilting of the mean thermocline. Because it is much shallower in the eastern Pacific, a larger subsurface temperature anomaly can be induced in response to the same wind forcing.

The diagnosis above indicates that, when the atmospheric wind response to a unit SSTA forcing is greater in the central Pacific, the ocean response to a unit wind forcing is stronger in the eastern Pacific. The ratio of the overall Bjerknes air–sea feedback strength between the EP and CP El Niño may be calculated based on Eq. (3). It turns out that the ratio is 1:0.32, with the growth rate of the EP El Niño being 3 times greater than that of the CP El Niño. Although the calculation above does not include other ocean advection and heat flux processes, this ratio reflects well the observed SSTA tendency difference between the EP and CP El Niño during the developing phase.

5. Mixed layer heat budgets associated with EP and CP El Niño

A mixed layer heat budget analysis is performed to assess the relative roles of dynamic and thermodynamic processes in contributing to the development of the two types of El Niño. The mixed layer temperature tendency equation can be written as follows:

 
formula

where T′ denotes the SSTA; T denotes the mixed layer ocean temperature; V is three-dimensional ocean current; = (∂/∂x, ∂/∂y, ∂/∂z) denotes a three-dimensional gradient operator; a bar denotes a climatologic mean variable; a prima represents the anomaly departure from the climatologic mean; −(V · T)′ denotes anomalous temperature advection integrated vertically from the surface to the bottom of the mixed layer and may be separated into linear advection terms and nonlinear advection term −(V′ · T′); Qnet is the summation of the net downward shortwave radiation, surface longwave radiation, surface latent heat, and sensible heat flux at the ocean surface; R represents the residual term; ρo is the density of water; Cw is the specific heat capacity of water; and H denotes the mixed layer depth. Here, a positive heat flux indicates a heating to the ocean, and H is defined as the depth at which the temperature is 0.5°C lower than the sea surface temperature. For simplicity, each term is averaged over box A (B) and a constant mixed layer depth of 50 m (80 m) is used for the EP (CP) El Niño composite.

Figure 8 shows the mixed layer temperature tendency, the ocean advection term, the surface heat flux term, and the summation of the last two terms during the El Niño developing phase. In the EP El Niño composite, the three-dimensional (3D) temperature advection term primarily contributes to the SSTA warming while the heat flux anomaly tends to damp the SSTA. A similar feature appears in the CP El Niño composite but with less magnitude. The summation of the ocean advection term and the surface heat flux term is quite close to the observed SSTA tendency for both EP and CP El Niño composites. This indicates that the heat budget analysis is quite reliable, even though there is uncertainty in the surface heat fluxes and oceanic subgrid processes.

Fig. 8.

Composite SSTA tendency terms (°C month−1) averaged for EP El Niño (left four bars) and for CP El Niño (right four bars). The red bar denotes the observed mixed layer temperature tendency during the developing phase, the blue bar denotes ocean temperature advection term, the purple bar denotes the surface heat flux term, and the green bar denotes the summation of ocean advection and surface heat flux terms.

Fig. 8.

Composite SSTA tendency terms (°C month−1) averaged for EP El Niño (left four bars) and for CP El Niño (right four bars). The red bar denotes the observed mixed layer temperature tendency during the developing phase, the blue bar denotes ocean temperature advection term, the purple bar denotes the surface heat flux term, and the green bar denotes the summation of ocean advection and surface heat flux terms.

The ocean advection terms are further decomposed into nine terms and the results are shown in Fig. 9a. For the EP El Niño, the advection of anomalous temperature gradient by mean upwelling has the largest contribution. As discussed in the previous section, this term reflects the Bjerknes dynamic air–sea feedback and involves both the atmospheric wind response to the SSTA and the ocean thermocline and subsurface temperature responses to the wind forcing. Another large vertical advection term involves the advection of the mean temperature gradient by anomalous upwelling , which also contributes to the EP SSTA growth. For the CP El Niño, however, the largest term is the anomalous zonal advection of mean temperature gradient . The vertical advection is much weaker. Therefore, the heat budget analysis points out different growing mechanisms for EP and CP El Niño.

Fig. 9.

As in Fig. 8, but for (a) individual zonal, vertical, and meridional advection terms during (top) EP and (bottom) CP El Niño developing phases and (b) four anomalous heat flux terms including downward solar radiation (DSRF), net upward longwave radiation (ULRF), latent heat flux (LHF), and sensible heat flux (SHF). Blue and red bars in (b) denote EP and CP El Niño composites. The unit on the vertical axis is °C month−1.

Fig. 9.

As in Fig. 8, but for (a) individual zonal, vertical, and meridional advection terms during (top) EP and (bottom) CP El Niño developing phases and (b) four anomalous heat flux terms including downward solar radiation (DSRF), net upward longwave radiation (ULRF), latent heat flux (LHF), and sensible heat flux (SHF). Blue and red bars in (b) denote EP and CP El Niño composites. The unit on the vertical axis is °C month−1.

In addition to the zonal and vertical advection, the meridional advection of anomalous temperature by mean currents is also large in both types of El Niño. This is because the maximum SSTA is usually located near the equator and poleward mean currents due to the easterly wind forcing appear over the equatorial eastern and central Pacific. As a result, the mean currents advect the maximum SSTA poleward, which broaden the SSTA and contribute to a positive mixed layer temperature tendency (Battisti 1988; Lau et al. 1992).

The net effect of the surface heat flux is generally negative and tends to damp the SSTA in both the EP and CP El Niño. By decomposing the surface heat flux into net downward shortwave radiation, net surface longwave radiation, surface latent heat flux, and surface sensible heat flux anomalies, one may examine the relative roles of each term (Fig. 9b). It is interesting to note that the dominant thermodynamic damping for the EP El Niño comes from the latent heat flux anomaly. This differs markedly from the CP El Niño in which the latent heat flux anomaly is negligible. The cause of such a difference is explained as follows: During the EP El Niño development, the maximum SSTA is located in the eastern Pacific while the anomalous convection center appears in the central Pacific. Easterly anomalies are generated to the east of the convective center, leading to the enhanced surface evaporation over the EP maximum SSTA region. The wind induced latent heat flux increase is further magnified by increased sea–air specific humidity difference because the latter of which depends to a large extent on SSTA (Li and Wang 1994). This is why the latent heat flux anomaly shows the largest negative tendency out of four heat flux terms in the EP El Niño. On the other hand, in the CP El Niño, the SSTA and wind speed effects are largely offset. Whereas a warm SSTA tends to enhance the surface evaporation, the anomalous westerly in the central Pacific (Figs. 5d,f) tends to decrease the wind speed and surface evaporation. As a result, the net effect of the anomalous latent heat flux is small.

To sum up, the heat budget analysis indicates that the vertical (zonal) advection is dominant in the EP (CP) El Niño composite. This is consistent with previous studies (e.g., Kug et al. 2009; Yu et al. 2010). The difference lies in the relative magnitude of each budget term due to more strict case selection. The anomalous latent heat flux tends to damp the SSTA greatly in the EP El Niño, while it is negligible in the CP El Niño.

6. Effect of the background zonal SST gradient on SSTA tendency

Because the tropical mean state and the El Niño behavior experience a coherent interdecadal change since 1999, a key issue that needs to be addressed is how does the mean zonal SST gradient change affect the El Niño behavior. In the following we take a two-step approach. First, with a specified SSTA forcing, we examine how the atmospheric wind and precipitation responds to a specified SSTA under different background zonal SST distributions. Second, we further examine how the ocean responds to the wind changes. Through the idealized atmospheric and ocean model experiments, one may examine the dependence of longitudinal location of the maximum SSTA tendency on the background zonal SST gradient.

In the first step, a series of numerical experiments with ECHAM5 T42 was carried out to examine the tropical atmospheric response to a specified SSTA, given various idealized background mean SST conditions. Define 31-yr (1980–2010) averaged climatological mean SST (from the HadISST data) as T(x,y), the zonal average of T(x,y) as , and the deviation of T(x,y) from as ΔT(x,y). In the control experiments (CTRL), the model was forced by a set of idealized SST distributions that have the formula of , where N was set to be 0.5, 1.0, 1.5, and 2.0 and λ is a function of y and is equal to 1 in the tropics (from 20°S to 20°N), 0 beyond 30°S and 30°N, and a linear transition from 1 to 0 between 20° and 30°N (S). Note that N = 1.0 represents the observed zonal SST gradient in the tropics, whereas N = 0.5 (2) means that the background zonal SST gradient is 50% (2 times) of the observed values. Figure 10 (left) illustrates the zonal distribution of the specified background SST along the equator for N = 0.5, 1.0, 1.5, and 2.0 (experiments N0.5, N1.0, N1.5, and N2.0, respectively). In the sensitivity experiments parallel to N0.5, N1.0, N1.5, and N2.0, the model was forced by an interannual SSTA superposed on the above background mean states. The SSTA has amplitude of 2°C and is symmetric about the equator; the maximum SSTA center is located at 100°W. All the experiments were integrated for 24 months, with solar radiation forcing at top of the atmosphere fixed on 22 March. The anomalous atmospheric response to the SSTA was estimated based on the difference of last 12-month average between the sensitivity and control runs.

Fig. 10.

(left) Specified idealized zonal SST distributions at the equator for each of sensitivity experiments: N1.0 represents the observed zonal SST gradient averaged during the 31 yr. N0.5 (N2.0) represents a case when the mean SST gradient is 50% (200%) of the observed value. The unit of the vertical axis is °C. (right) The pattern of a specified EP SSTA (red contour) and the responses of precipitation and 850-hPa wind anomalies (areas passing the 95% confidence level within 5°S–5°N are plotted) to the specified SSTA based on the ECHAM5 simulations under different background zonal SST gradient conditions (from N0.5 to N2.0).

Fig. 10.

(left) Specified idealized zonal SST distributions at the equator for each of sensitivity experiments: N1.0 represents the observed zonal SST gradient averaged during the 31 yr. N0.5 (N2.0) represents a case when the mean SST gradient is 50% (200%) of the observed value. The unit of the vertical axis is °C. (right) The pattern of a specified EP SSTA (red contour) and the responses of precipitation and 850-hPa wind anomalies (areas passing the 95% confidence level within 5°S–5°N are plotted) to the specified SSTA based on the ECHAM5 simulations under different background zonal SST gradient conditions (from N0.5 to N2.0).

Figure 10 (right) shows anomalous precipitation and 850-hPa wind responses to the SSTA forcing under different zonally asymmetric background SST distributions. In the weak zonal SST gradient (N0.5) run, the maximum precipitation anomaly center appears near 120°W. Associated with the precipitation anomaly are widespread equatorial westerly anomalies extending from the western Pacific to the central–eastern Pacific and weaker easterly anomalies in the far eastern equatorial Pacific. As the background SST gradient becomes greater from N = 1.0 to N = 2.0, one may observe a westward shift of centers of the maximum precipitation and wind anomalies. In the N2.0 run, the maximum precipitation anomaly center is located at 160°W, while the maximum westerly anomalies are confined to the west of 160°W. The numerical experiments demonstrate the important role the background zonal SST gradient plays in determining the longitudinal location of the maximum atmospheric convection and wind response to the SSTA.

It is argued that the background SST state may affect the anomalous atmospheric response through the following mechanisms. First, with the increase of the zonal SST gradient, the mean SST in the eastern Pacific becomes colder (e.g., from 26.2°C in N0.5 run to 23°C in N2.0 run), which prohibits the development of convective activities over the cold tongue region. This leads to the westward shift of the convective center. Second, the stronger zonal SST gradient may accelerate the easterly trade at the equator. The enhanced trades may further advect the anomalous boundary layer convergence and convection westward. Third, the enhanced trade wind due to the increased SST gradient may promote a low-level divergence (convergence) in the eastern (western) Pacific. The moisture change associated with the divergent (convergent) circulation may suppress the anomalous convection in the eastern Pacific and favor the westward shift of the anomalous precipitation center.

In the second step, we examine how the altered convection and wind anomalies due to the change of the background SST gradient further affect the SSTA tendency. To illustrate this effect, a simplified version of the Cane–Zebiak (CZ) coupled ocean–atmosphere model (Cane 1979; Cane 1979b; Zebiak and Cane 1987) is used. By considering anomalous zonal and vertical advection and thermocline variation effects without involving of the surface heat flux, the SSTA tendency equation may be written as

 
formula

where T′ denotes the interannual SSTA, is the climatologic long-term (31 yr) mean SST, is long-term mean upwelling, h (=100 m) and h′ denote mean and anomalous thermocline depth fields, and γ (=0.1 km−1) is a parameter that controls the change of subsurface temperatures due to the thermocline displacement.

The ocean mixed layer current anomaly in Eq. (5) may be determined by the following the surface layer momentum equation:

 
formula

where represents the zonal wind stress anomaly; rs (=day−1 = 86 400−1) is a damping coefficient; us and υs are anomalous zonal and meridional ocean currents in the mixed layer; β is meridional gradient of Coriolis parameter; ρo (=1000 kg m−3) and ρo (=1.25 kg m−3) are densities of water and surface air; y is the distance from the equator; H (=50 m) denotes the mixed layer depth; CD (=1.5 × 10−3) is wind-drag coefficient; denotes the surface wind speed; and u′ stands for the anomalous surface zonal wind.

The anomalous upwelling velocity at the base of the ocean mixed layer ws is determined by the divergence of the surface currents: that is,

 
formula

The anomalous thermocline depth is specified as an idealized cosine function, and its amplitude is determined by maximum zonal wind anomaly at the equator: that is,

 
formula

where Umax denotes the maximum surface zonal wind anomaly at the equator; xo denotes the longitudinal location where zonal wind anomalies converge; L (=30° in longitude) is the quarter of wavelength; α is a constant that controls the strength of h′; and h′ is set to zero beyond the half wavelength (x > L or x < −L).

Figure 11a shows the longitudinal distribution of the SSTA tendency along the equator derived from the simplified CZ model. The wind stress forcing fields used for the calculations are from the N0.5 and N2.0 simulations. For N0.5 run, the maximum SSTA tendency is located in the eastern Pacific near 120°W. This implies that under a weak zonal SST gradient mean state, dynamic air–sea coupling processes favor the growth of the EP-type El Niño. In contrast, at the N2.0 run, the maximum SSTA tendency is weaker and appears over the central equatorial Pacific (near 150°W). This points out that, under a strong zonal SST gradient mean state, dynamic air–sea coupling processes favor the growth of the CP-type El Niño.

Fig. 11.

Longitudinal distributions of the mixed layer temperature tendency due to (a) the ocean dynamic process and (b) the surface latent heat flux effect derived from a CZ-type ocean model with the forcing of the surface wind anomalies from the ECHAM5 N0.5 (solid line) and N2.0 (dashed line) runs. The unit of the vertical axis is °C month−1.

Fig. 11.

Longitudinal distributions of the mixed layer temperature tendency due to (a) the ocean dynamic process and (b) the surface latent heat flux effect derived from a CZ-type ocean model with the forcing of the surface wind anomalies from the ECHAM5 N0.5 (solid line) and N2.0 (dashed line) runs. The unit of the vertical axis is °C month−1.

A further examination of relative roles of the zonal and vertical advection terms indicates that the anomalous vertical advection primarily contributes to the EP SSTA tendency in the N0.5 run. In the N2.0 run, the maximum SSTA tendency in the central Pacific is primarily attributed to the zonal advection (figures not shown). The simple analysis model results seem consistent with Kug et al. (2009) and the heat budget diagnosis result in section 5.

In addition to the ocean advection, the surface latent heat flux process may also contribute to the bifurcation of longitudinal location of maximum SSTA tendencies in the weak and strong SST gradient cases (N0.5 and N2.0 runs). The SSTA tendency equation in the presence of the latent heat flux effect may be written as

 
formula

where ; cw (=4.2 × 103 J Kg−1 K−1) is specific heat capacity of water; Ce (=1.5 × 10−3) is air–sea exchange coefficient; Le (=2.5 × 106 J Kg−1) is latent heat of condensation; is the surface wind speed; qs is the saturation specific humidity at the sea surface, which is a function of SST and sea level pressure; qa is the near-surface air specific humidity, which can be empirically determined from SST, qa = 10−3[(0.972Ts − 8.92)] (Li and Wang 1994); and Ts is the SST.

Figure 11b illustrates this latent heat flux effect. While the zonal bifurcation of the maximum SSTA tendency under the weak and strong SST gradient cases is clearly presented, the longitudinal separation is not as far as that in the presence of the ocean dynamic effect and, more importantly, the magnitude of the resulted SSTA tendency is much weaker. Overall, the SSTA tendency due to either ocean dynamics or surface latent heat flux processes displays a clear longitudinal dependence on the background zonal SST gradient.

Additional experiments were conducted in which the SSTA was placed over the central Pacific (centered at 150°W, as seen in Fig. 12a). It is interesting to note that, for given CP SSTA, the atmospheric response differs under strong and weak background SST gradients. Under a strong gradient, the atmospheric convection and wind responses shift to the west of the maximum SSTA center, while under a weak gradient they are approximately in phase with the maximum SSTA. The so-generated anomalous wind may further affect SSTA tendency, as shown in Fig. 12b. Under the weak background SST gradient, the maximum SSTA tendency appears to the east (near 140°W) of the maximum SSTA center (due to the effect of anomalous downwelling and positive thermocline depth anomaly), implying that the SSTA tends to move eastward. On the other hand, under the strong background SST gradient, the maximum SST tendency appears in phase with or slightly to the west of the SSTA; as a result, the SSTA does not move toward the EP and is confined to the CP.

Fig. 12.

(a) As in Fig. 10(right), but for a specified CP SSTA (red contour) in (top) N0.5 and (bottom) N2.0 runs. (b) As in Fig. 11a, but the surface wind forcing was derived from the CP SSTA case. Solid and dashed lines in (b) denote N0.5 and N2.0 runs, respectively. The unit of the vertical axis is °C month−1.

Fig. 12.

(a) As in Fig. 10(right), but for a specified CP SSTA (red contour) in (top) N0.5 and (bottom) N2.0 runs. (b) As in Fig. 11a, but the surface wind forcing was derived from the CP SSTA case. Solid and dashed lines in (b) denote N0.5 and N2.0 runs, respectively. The unit of the vertical axis is °C month−1.

The idealized atmospheric and ocean model experiments above imply that the interdecadal change of the mean SST gradient may have a dynamic impact on the El Niño’s behavior. During the weak SST gradient ID1 period, due to warmer SST over the eastern Pacific and weaker trade wind, EP-like El Niño occurred more frequently. During the strong SST gradient ID2 period, the mean state favors the growth of the SSTA in the central Pacific, regardless of the initial SSTA location.

7. Conclusions and discussion

The relationship between the interdecadal changes of the mean state and El Niño’s behavior was examined based on the analysis of observational data from 1980 to 2010. A linear trend analysis indicates that the tropical Pacific SST undergoes a trend toward a La Niña–like pattern, with a significant cooling trend over the tropical eastern Pacific (east of 160°W) and a significant warming trend in the tropical western Pacific (west of 160°E). Associated with the SST trend are suppressed (enhanced) convection over the central–eastern equatorial Pacific (Maritime Continent), strengthened trades across the central Pacific, and the increase (decrease) of SSH in the western (eastern) Pacific. The trend patterns are well represented by the differences between two interdecadal periods (ID1 and ID2) with a changing point in 1999. The interdecadal change in the tropical Pacific is consistent with the PDO phase transition from a warm to a cold phase.

The interannual SST variability differs between the two interdecadal periods. In ID1 (ID2), the maximum interannual SST variability is located in the eastern (central) Pacific. Such a feature is consistent with the fact that ID1 (ID2) is dominated by EP- (CP-) type El Niño.

The Bjerknes dynamic air–sea feedback strength for the two types of El Niño was examined. It is found that, for given a unit SSTA forcing, atmospheric precipitation and wind responses are greater in CP El Niño than in EP El Niño. The difference is primarily attributed to a higher background mean SST between the central and eastern Pacific. Given a unit wind forcing, however, the ocean subsurface temperature response is greater in EP El Niño than in CP El Niño. This is attributed to a shallower (deeper) mean thermocline in the eastern (central) Pacific, which leads to a larger (smaller) subsurface temperature variation given the same wind or thermocline depth change. The diagnosis of the overall Bjerknes air–sea feedback strength during the developing phase shows that the EP El Niño possesses a greater growth rate than the CP El Niño.

A mixed layer heat budget analysis indicates that for both types of El Niño, the ocean dynamics contribute to the SSTA development while the surface heat fluxes tend to damp the SSTA. In EP El Niño, the thermocline feedback is a dominant process for the El Niño growth, and the surface latent heat flux is a major damping process. On the other hand, in CP El Niño, the zonal advection feedback plays a crucial role. The result is consistent with Kug et al. (2009) and Yu et al. (2010).

One question related to the heat budget is how the decadal mean state change affects each of the budget terms. Since the zonal advection is important for the CP El Niño, the La Niña–like mean state change is expected to strengthen the zonal advection term. However, our calculation shows that such a change is small. This implies that the direct impact of the mean state on the heat budget terms is weak. The mean state effect is primarily through the change of interannual anomalies, through an atmospheric bridge.

The role of the background zonal SST gradient on El Niño behavior was investigated through idealized atmospheric and oceanic model experiments. Under a weak SST gradient mean state, the maximum precipitation and wind responses to a specified EP SSTA are confined to the area east of 150°W. With the increase of the mean SST gradient, the anomaly centers shift westward. Forcing a CZ-type ocean model with the above wind anomalies, we derived a longitude-dependent characteristic for the SSTA tendency: that is, so-derived maximum SSTA tendency appears in the central (eastern) Pacific under the strong (weak) background SST gradient condition. Similar westward shift is found when a CP SSTA is prescribed. The maximum SSTA tendency appears to the east of the SSTA center under a weak background SST gradient, which favors eastward propagation. Under a strong background SST gradient, the maximum SST tendency appears in phase with or slightly to the west of the SSTA center; as a result, the SSTA remains in the CP. The numerical model results imply that a change toward a La Niña–like (stronger SST gradient) mean state since 1999 may be responsible for more frequent occurrence of CP El Niño; an El Niño–like (weaker SST gradient) mean state favors a more frequent occurrence of EP El Niño.

It is argued that the stronger background zonal SST gradient may cause the westward shift of the SSTA tendency through the following three processes. First, the enhanced mean SST cooling over the EP may prohibit the development of convective activity in situ (Ham and Kug 2011), leading to the westward shift of the anomalous precipitation center. Second, the enhanced trade wind may advect the anomalous low-level convergence center westward. Third, the change of the mean wind may induce a convergence (divergence) in the western (eastern) Pacific, which may further affect anomalous convection through the change of the background specific humidity.

The argument above implies that both local SST over the EP and the strengthened zonal SST gradient may play a role in modulating El Niño behavior. To further understand the local SST effect, we conducted an additional AGCM experiment, in which we kept the same background SST gradient (N1.0) but added a 3°C uniform SST warming across the equatorial Pacific. The sensitivity model result shows that both the convection and wind anomalies shift eastward toward the SSTA center in the EP because of a warmer background SST there (figures not shown). This suggests that the local background SST indeed plays a role in affecting the anomalous response.

The interaction between the background mean state and the interannual perturbations may involve a two-way interaction. On one hand, the mean state may control the location of El Niño development. On the other hand, more frequent occurrence of CP- or EP-type El Niño events can modulate the mean state (e.g., Rodgers et al. 2004; Sun and Yu 2009; Lee and McPhaden 2010; McPhaden et al. 2011). A recent study by B. Xiang et al. (2012, unpublished manuscript) showed that the decadal mean state change patterns remain little changed even when extreme El Niño (La Niña) episodes were removed from ID1 (ID2). This indicates that the La Niña–like mean state change is robust. To the first order, the mean state change is attributed to the decadal variability itself, not to the nonlinear rectification of higher-frequency interannual variability.

The occurrence of a couple of CP El Niño episodes during ID1 does not contradict the conclusion derived from this study. Our main point is that, because of the strengthening of the background zonal SST gradient and colder local SST in ID2, EP-type El Niño becomes less favorable, and the CP-type El Niño becomes more frequent because of the westward shift of the background low-level convergence. The relatively weak mean SST gradient in ID1 favors more frequent EP El Niño but does not prevent the occurrence of the CP El Niño.

The background SST gradient–El Niño relationship illustrated in this study is consistent with the model diagnosis of Choi et al. (2011) but different from Ashok et al. (2007) and Yeh et al. (2009), who suggested an opposite relationship between the background zonal SST gradient and the El Niño behavior. Ashok et al. (2007) and Yeh et al. (2009) emphasized the effect of decadal thermocline change, but it is not clear whether or not a small decadal thermocline change in the CP can efficiently affect El Niño behavior. A more detailed diagnosis of the above relationship with a longer record that covers both the changing points (in 1978 and 1999) is needed. It is likely that both anthropogenic forcing and natural variability may affect El Niño behavior. Addressing this problem requires further observational and modeling studies.

Acknowledgments

We thank three anonymous reviewers for their constructive and valuable comments. P. H. C. was supported by NSC 101-2111-M-133-001. T. L. was supported by NSF Grant AGS-1106536; ONR Grant N00014-0810256; and by the International Pacific Research Center, which is sponsored by the Japan Agency for Marine-Earth Science and Technology (JAMSTEC), NASA (NNX07AG53G), and NOAA (NA17RJ1230).

REFERENCES

REFERENCES
Ashok
,
K.
,
S. K.
Behera
,
S. A.
Rao
,
H.
Weng
, and
T.
Yamagata
,
2007
:
El Niño Modoki and its possible teleconnection
.
J. Geophys. Res.
,
112
,
C11007
,
doi:10.1029/2006JC003798
.
Battisti
,
D. S.
,
1988
:
Dynamics and thermodynamics of a warming event in a coupled tropical atmosphere–ocean model
.
J. Atmos. Sci.
,
45
,
2889
2919
.
Battisti
,
D. S.
, and
A. C.
Hirst
,
1989
:
Interannual variability in the tropical atmosphere/ocean system: Influence of the basic state and ocean geometry
.
J. Atmos. Sci.
,
46
,
1687
1712
.
Cane
,
M. A.
,
1979a
: The response of an equatorial ocean to simple wind-stress patterns:. I. Model formulation and analytic results.
J. Mar. Res.
,
37
,
233
252
.
Cane
,
M. A.
,
1979b
: The response of an equatorial ocean to simple wind-stress patterns: II. Numerical results.
J. Mar. Res.
,
37
,
253
299
.
Carton
,
J. A.
,
B. S.
Giese
, and
S. A.
Grodsky
,
2005
:
Sea level rise and the warming of the oceans in the Simple Ocean Data Assimilation (SODA) ocean reanalysis
.
J. Geophys. Res.
,
110
,
C09006
,
doi:10.1029/2004JC002817
.
Choi
,
J.
,
S.-I.
An
,
J.-S.
Kug
, and
S.-W.
Yeh
,
2011
:
The role of mean state on changes in El Niño’s flavor
.
Climate Dyn.
, 37, 1205–1215, doi:10.1007/s00382-010-0912-1.
Feng
,
M.
,
M. J.
McPhaden
, and
T.
Lee
,
2010
:
Decadal variability of the Pacific subtropical cells and their influence on the southeast Indian Ocean
.
Geophys. Res. Lett.
,
37
,
L09606
,
doi:10.1029/2010GL042796
.
Gill
,
A. E.
,
1980
:
Some simple solutions for heat-induced tropical circulation
.
Quart. J. Roy. Meteor. Soc.
,
106
,
447
462
.
Ham
,
Y.-G.
, and
J.-S.
Kug
,
2011
:
How well do current climate models simulate two types of El Niño?
Climate Dyn.
, 38, 383–398,
doi:10.1007/s00382-011-1157-3
.
Jin
,
F.-F.
,
1997
:
An equatorial ocean recharge paradigm for ENSO. Part I: Conceptual model
.
J. Atmos. Sci.
,
54
,
811
829
.
Kanamitsu
,
M.
,
W.
Ebisuzaki
,
J.
Woollen
,
S. K.
Yang
,
J. J.
Hnilo
,
M.
Fiorino
, and
G. L.
Potter
,
2002
:
NCEP–DOE AMIP-II Reanalysis (R-2)
.
Bull. Amer. Meteor. Soc.
,
83
,
1631
1643
.
Kao
,
H.-Y.
, and
J.-Y.
Yu
,
2009
:
Contrasting eastern-Pacific and central-Pacific types of ENSO
.
J. Climate
,
22
,
615
632
.
Kim
,
H.-M.
,
P. J.
Webster
, and
J. A.
Curry
,
2009
:
Impact of shifting patterns of Pacific Ocean warming on North Atlantic tropical cyclones
.
Science
,
325
,
77
80
,
doi:10.1126/science.1174062
.
Kug
,
J.-S.
,
F.-F.
Jin
, and
S.-I.
An
,
2009
:
Two types of El Niño events: Cold tongue El Niño and warm pool El Niño
.
J. Climate
,
22
,
1499
1515
.
Kug
,
J.-S.
,
J.
Choi
,
S.-I.
An
,
F.-F.
Jin
, and
A. T.
Wittenberg
,
2010
:
Warm pool and cold tongue El Niño events as simulated by the GFDL 2.1 coupled GCM
.
J. Climate
,
23
,
1226
1239
.
Larkin
,
N. K.
, and
D. E.
Harrison
,
2005
:
On the definition of El Niño and associated seasonal average U.S. weather anomalies
.
Geophys. Res. Lett.
,
32
,
L13705
,
doi:10.1029/2005GL022738
.
Lau
,
N.-C.
,
S. G. H.
Philander
, and
M. J.
Nath
,
1992
:
Simulation of ENSO-like phenomena with a low-resolution coupled GCM of the global ocean and atmosphere
.
J. Climate
,
5
,
284
307
.
Lee
,
T.
, and
M. J.
McPhaden
,
2010
:
Increasing intensity of El Niño in the central-equatorial Pacific
.
Geophys. Res. Lett.
,
37
,
L14603
,
doi:10.1029/2010GL044007
.
Li
,
T.
,
1997
:
Phase transition of the El Niño–Southern Oscillation: A stationary SST mode
.
J. Atmos. Sci.
,
54
,
2872
2887
.
Li
,
T.
, and
B.
Wang
,
1994
:
A thermodynamic equilibrium climate model for monthly mean surface winds and precipitation over the tropical Pacific
.
J. Atmos. Sci.
,
51
,
1372
1385
.
Liu
,
L.
,
W.
Yu
, and
T.
Li
,
2011
:
Dynamic and thermodynamic air–sea coupling associated with the Indian Ocean dipole diagnosed from 23 WCRP CMIP3 models
.
J. Climate
,
24
,
4941
4958
.
McPhaden
,
M. J.
,
T.
Lee
, and
D.
McClurg
,
2011
:
El Niño and its relationship to changing background conditions in the tropical Pacific Ocean
.
Geophys. Res. Lett.
,
38
,
L15709
, doi:10.1029/2011GL048275.
Philander
,
S. G.
,
1990
: El Niño, La Niña, and the Southern Oscillations. Academic Press, 293 pp.
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
,
354
384
.
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
,
doi:10.1029/2002JD002670
.
Rodgers
,
K. B.
,
P.
Friederichs
, and
M.
Latif
,
2004
:
Tropical Pacific decadal variability and its relation to decadal modulations of ENSO
.
J. Climate
,
17
,
3761
3774
.
Roeckner
,
E.
, and
Coauthors
,
2003
: The atmospheric general circulation model ECHAM5, Part I: Model description. Max Planck Institute for Meteorology Rep. 349, 127 pp.
Schopf
,
P. S.
, and
M. J.
Suarez
,
1988
:
Vacillations in a coupled ocean–atmosphere model
.
J. Atmos. Sci.
,
45
,
549
566
.
Sun
,
F.
, and
J.-Y.
Yu
,
2009
:
A 10–15-yr modulation cycle of ENSO intensity
.
J. Climate
,
22
,
1718
1735
.
Weng
,
H.
,
S. K.
Behera
, and
T.
Yamagata
,
2009
:
Anomalous winter climate conditions in the Pacific Rim during recent El Niño Modoki and El Niño events
.
Climate Dyn.
,
32
,
663
674
.
Xie
,
P. 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
.
Yeh
,
S.-W.
,
J.-S.
Kug
,
B.
Dewitte
,
M.-H.
Kwon
,
B.
Kirtman
, and
F.-F.
Jin
,
2009
:
El Niño in a changing climate
.
Nature
,
461
,
511
515
.
Yu
,
J.-Y.
, and
H.-Y.
Kao
,
2007
:
Decadal changes of ENSO persistence barrier in SST and ocean heat content indices: 1958–2001
.
J. Geophys. Res.
,
112
,
D13106
,
doi:10.1029/2006JD007654
.
Yu
,
J.-Y.
, and
S. T.
Kim
,
2011
:
Relationships between extratropical sea level pressure variations and the central Pacific and eastern Pacific types of ENSO
.
J. Climate
,
24
,
708
720
.
Yu
,
J.-Y.
,
H.-Y.
Kao
, and
T.
Lee
,
2010
:
Subtropics-related interannual sea surface temperature variability in the central equatorial Pacific
.
J. Climate
,
23
,
2869
2884
.
Yu
,
L.
,
X.
Jin
, and
R. A.
Weller
,
2008
: Multidecade global flux datasets from the objectively analyzed air-sea fluxes (OAFlux) project: Latent and sensible heat fluxes, ocean evaporation, and related surface meteorological variables. Woods Hole Oceanographic Institution OAFlux Project Tech. Rep. OA-2008-01, 64 pp.
Zebiak
,
S. E.
, and
M. A.
Cane
,
1987
:
A model El Niño-Southern Oscillation
.
Mon. Wea. Rev.
,
115
,
2262
2278
.

Footnotes

*

School of Ocean and Earth Science and Technology Contribution Number 8704 and International Pacific Resource Center Contribution Number 898.