Advances in Research of ENSO Changes and the Associated Impacts on Asian-Pacific Climate

Transcription

1 Asia-Pac. J. Atmos. Sci., 50(4), , 2014 pissn / eissn DOI: /s REVIEW Advances in Research of ENSO Changes and the Associated Impacts on Asian-Pacific Climate Tianjun Zhou 1,2, Bo Wu 1, and Lu Dong 1,3 1 LASG, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China 2 Climate Change Research Center, Chinese Academy of Sciences, Beijing, China 3 Graduate University of Chinese Academy of Sciences, Beijing, China (Manuscript received 18 February 2014; accepted 25 June 2014) The Korean Meteorological Society and Springer 2014 Abstract: This review provides a summary on the recent major advances in research of ENSO changes and the associated impacts on Asian-Pacific climate. Achievements in the following topics are summarized: 1) the asymmetry between El Niño and La Niña; 2) the different features of central Pacific (CP) El Niño and eastern Pacific (EP) El Niño; 3) the change of ENSO in a warming world, including analysis of pre-industrial control simulation, historical simulation and climate projections of coupled climate system model; 4) Impact of EP ENSO on warm-pool air-sea interaction and East Asianwestern North Pacific summer monsoon; 5) Impacts of CP ENSO on Asian-Pacific climate, with focus on East Asian seasonal precipitation and tropical cyclones in the western Pacific. Research results published in the recent 5 years are the major sources for this review. Based on the review of the current progresses, some challenging issues needed to be investigated in the future are highlighted. Key words: ENSO, monsoon, central Pacific El Niño, eastern Pacific El Niño, global warming 1. Introduction The El Niño-Southern Oscillation (ENSO) is a coupled atmosphere-ocean phenomenon with preferred time scales of two to about seven years. A canonical El Niño event features a basin-wide warming of the tropical Pacific Ocean east of the dateline and usually peaks at the end of the calendar year. The atmospheric counterpart of El Niño is the Southern Oscillation, which is a fluctuation of tropical surface pressure between the western Pacific and eastern Pacific. The opposite phase to El Niño is La Niña, which features a sea surface temperature (SST) cooling in the eastern equatorial Pacific Ocean. Basin scale sea surface temperature fluctuations associated with ENSO events induce tropical circulation and thereby heating source changes, which further affects climate conditions globally through atmospheric tele-connections (Trenberth et al., 1998; McPhaen et al., 2006; See Wang et al., 2012 for a recent review). ENSO is the most dominant phenomenon of earth s climate variability on interannual time scale and has significant Corresponding Author: Tianjun Zhou, LASG, Institute of Atmospheric Physics, Chinese Academy of Sciences, P.O. Box 9804, Beijing , China. zhoutj@lasg.iap.ac.cn impacts on the water resources as well as agriculture around many parts of the world. Understanding changes of ENSO flavors in a changing climate is of crucial scientific and socioeconomic needs. The classical feature of the El Niño exhibits a SST warming center in the eastern equatorial Pacific attached to the coast of South America. In recent decades, El Niño warming center tends to occur more frequently in the central Pacific and is sandwiched by anomalously cooling to its east and west (Latif et al., 1997; Ashok et al., 2007; Kug et al., 2009). Diverse names are defined for these new types of El Niño, ranging from central Pacific El Niño, dateline El Niño, El Niño- Modoki, to warm pool El Niño (IPCC, 2013; Christensen et al., 2013a). To facilitate the discussion, we use the terminology of central Pacific El Niño (hereafter CP) instead of various names here. Correspondingly, we use eastern Pacific (EP) El Niño to term the conventional El Niño. Understanding the mechanism of more frequent occurrence of CP and projecting its future evolution have been a focus of ENSO study community. Asia supports the largest population on the earth, being home to 3.6 billion people. The economy and society across the region are critically influenced by the evolution and variability of the monsoon. The interannual variability of the monsoon is impacted by ENSO activities. Predictions of monsoon anomalies across the region rely heavily on the prediction of ENSO events. Great efforts have been devoted to the studies of monsoon-enso connection in the past decades (See reviews by Li and Wang, 2005; Lau and Wang, 2006). How the change of ENSO flavors would impact the monsoon variability is a new question for monsoon study community. The primary focus of this review is the change of ENSO flavors and its impact on Asian-Pacific climate. After presenting an overview of ENSO flavor changes, we focus on the impacts of ENSO on Asian-Pacific climate instead of a review on our understanding of ENSO physics as Wang et al. (2012). In section 2, we summarize the progress in the studies of asymmetry between El Niño and La Niña. The characteristics of central Pacific El Niño and eastern Pacific El Niño are compared in section 3. Recent progresses in the simulation and projections of ENSO are reviewed in section 4. Impact of EP ENSO on warm-pool air-sea interaction and East Asian-

2 406 ASIA-PACIFIC JOURNAL OF ATMOSPHERIC SCIENCES western Pacific monsoon are reviewed in section 5. Impacts of CP ENSO on Asian-Pacific climate are summarized in section 6. Section 7 presents a summary. 2. Asymmetry between El Niño and La Niña El Niño and La Niña are two opposite phases of ENSO, but they are not symmetric in many aspects, such as amplitude (e.g., An and Jin, 2004), duration (e.g., Ohba and Ueda, 2009) and zonal phase propagation (e.g., McPhaden and Zhang, 2009). In the section, we review the primary recent progresses in understanding the asymmetries of ENSO amplitude and duration. a. Amplitude ENSO amplitude shows a remarkable asymmetric feature, that is, amplitude of El Niño-related positive SST anomaly (SSTA) in the equatorial eastern Pacific is significantly stronger than that of La Niña-related negative SSTA (Burgers and Stephenson, 1999; Jin et al., 2003; An and Jin, 2004; Duan et al., 2008). An (2009) has given a detailed review for the topic. In the section, we just focus on the new progresses. The asymmetry can be statistically measured by normalized third statistical moment, referred to as skewness (Burgers and Stephenson, 1999). The significant positive skewness is strongest in the far equatorial eastern Pacific. It reaches maximum in boreal winter, consistent with the mature phase of ENSO (Su et al., 2010). Two types of mechanisms have been proposed to understand the asymmetry of ENSO amplitude. (1) Nonlinear ocean processes An and Jin (2004) proposed that the asymmetry of the amplitude between El Niño and La Niña results from nonlinear vertical advection in the far equatorial eastern Pacific based on mixed-layer heat budget analysis. The anomalous mixed layer temperature tendency equation is written as T = ( v T) w T -----, (1) t z Q ρc p H + R where T represent mixed layer temperature, v and w represent horizontal and vertical currents, respectively, Q represent net surface heat flux, ρ, C p and H, represent seawater density, specific heat of seawater and mixed-layer depth. R is the residual term, including sub-grid scale contributions. ( )' denotes anomalous variables. The horizontal and vertical advection terms can be further decomposed into linear and nonlinear components. We take the vertical advection as an example, w T (2) z = w T w T w T z z z An and Jin (2004) showed that during the developing phase of El Niño, the far equatorial eastern Pacific is dominated by anomalous upwelling (w' > 0) and subsurface warming caused by deepening of the thermocline ( T'/ z < 0). Thus nonlinear vertical advection warms the SST. During La Niña, both the vertical motion and subsurface temperature change signs. Thus nonlinear vertical advection term is also positive, which reduces the SST cooling. Based on the similar analysis method, Su et al. (2010) argued that the asymmetric amplitude is caused by the horizontal temperature advection instead of the vertical advection. The different conclusions are derived from the different reanalysis data. Su et al. (2010) found that though the far equatorial eastern Pacific is covered by easterly (westerly) surface wind stress anomalies during El Niño (La Niña) developing phase, the zonal current anomalies are primarily eastward (westward) geostrophic currents driven by thermocline depth variation. To the east of the maximum (minimum) SSTA [ T'/ x >0 ( T'/ x < 0)], both the eastward geostrophic currents (u' > 0) associated El Niño and westward geostrophic currents (u' <0) associated with La Niña causes warm nonlinear advection, which enhances El Niño but reduces La Niña amplitude. Meanwhile, the vertical motion in the far eastern Pacific is primarily driven by the convergence (divergence) of the geostrophic current instead of the Ekman current driven by the wind stress anomalies. Therefore, they obtain an opposite conclusion with that obtained by An and Jin (2004). Certainly, the opposite conclusion may be also caused by their differences in how to make the temporal and spatial averaging. In addition to the nonlinear temperature advection, An (2008) found that the activity of tropical instability waves (TIWs) also has a contribution to the asymmetric amplitude of ENSO. The activity of TIWs are intensified during La Niña, but suppressed during El Niño. Though TIWs suppress both El Niño and La Niña amplitude, the suppression caused by the active TIWs during La Niña is clearly more effective, which amplifies the asymmetry of ENSO amplitude. (2) Nonlinear atmospheric processes Nonlinear dynamics of tropical atmosphere variability are also alternative types of mechanisms to explain the asymmetric amplitude of ENSO. The types of mechanisms include nonlinear atmospheric response to underlying SST forcing and nonlinear interactions between ENSO and shorter time scale variability ranging from synoptical to intra-seasonal time scales. The easterly wind stress anomalies over the equatorial central Pacific associated with La Niña tends to be shifted westward by about degrees relative to westerly wind stress anomalies associated with El Niño, which is consistent with the zonal shift of ENSO-induced convection anomalies (Hoerling et al., 1997; Kang and Kug, 2002). Kang and Kug (2002) demonstrated that the asymmetric amplitude of ENSO can be generated by the longitudinal shift of zonal wind stress anomalies. Frauen and Dommenget (2010) further used a hybrid coupled model to verify the mechanism. The atmospheric component of the model is a state-of-the-art atmospheric general circulation

3 31 August 2014 Tianjun Zhou et al. 407 model (AGCM), while ocean component is a simple 2- Dimension linear model. Thus the nonlinear characteristics can only be rooted in atmospheric processes in the model. The model can simulate the skewness of the SSTA in the equatorial eastern Pacific, indicating that atmospheric process has a contribution to the asymmetry of amplitude between El Niño and La Niña (Choi et al., 2013). High-frequency atmospheric variability with much shorter time scale than ENSO can be seen as a multiplicative (ENSOstate-dependent) stochastic forcing on ENSO coupled system and can change ENSO properties (Eisenman et al., 2005; Perze et al., 2005; Jin et al., 2007). Because the multiplicative stochastic forcing is modulated by low-frequency variability and thus ENSO-state-dependent, it represents an interaction between high-frequency and low-frequency variability. Based on intermediate coupled model, Perze et al. (2005) demonstrated that the warm ENSO skewness is amplified when introducing nonlinear multiplicative stochastic forcing. They made an assumption that during El Niño regime, the amplitude of high-frequency variability linearly increases with the Nino- 3 index, while that during La Niña is set to zero, that is, stochastic forcing only reinforces El Niño. The asymmetry of stochastic forcing is the primarily reason for the amplification of the warm ENSO skewness under the framework. Except for the asymmetry of the stochastic forcing amplitude, another nonlinear mechanism about how the upscale feedback of high-frequency variability influences ENSO skewness are revealed by Rong et al. (2011). They stressed the role of nonlinearly rectified low-frequency wind stress by the highfrequency wind. The rectification is caused by nonlinear quadratic relationship between the wind stress and wind. The high-frequency wind tends to enhance both westerly wind stress anomalies during El Niño and easterly wind stress anomalies during La Niña. But because the high-frequency wind is multiplicative and is enhanced (suppressed) during El Niño (La Niña), the westerly wind stress anomalies during El Niño are more reinforced. The asymmetric rectified wind stress anomalies contribute to the warm ENSO skewness through modulating oceanic advection. In addition, the relationship between the ENSO amplitude and location of zonal wind patch also exhibited inter-decadal variations (An and Wang, 2000). Decadal variation of ENSO is another important topic of ENSO dynamics, but not covered in this review. b. Decaying rate and duration The temporal evolution of El Niño and La Niña are asymmetric. After their peak phase (boreal winter), El Niño generally decays rapidly and tends to evolve to the opposite phase in the next winter, whereas La Niña generally decays much slower and tends to maintain in the next winter (Fig. 1, Kessler, 2002; Larkin and Harrison, 2002; Ohba and Ueda, 2009; Ohba et al., 2010; Okumura and Deser, 2010; Wu et al., 2010a). The asymmetric decaying rate or duration between El Fig. 1. Temporal evolution of monthly mean Nino-3.4 (5 o S-5 o N, o E) SST anomalies (units: K) from the January of ENSO developing year to the December of following year for (a) El Niño events (1957, 1965, 1972, 1982, 1986, 1991, 1994, 1997 and 2002) and (b) La Niña events (1949, 1955, 1970, 1973, 1975, 1984, 1988, 1998 and 1999, thin lines). Composite Nino-3.4 SST anomalies (solid thick line) and the tendency of the solid thick line (dashed thick line) are also shown. Nino-3.4 indices are filtered by 5-point smoother. Y(0) and Y(1) indicate ENSO developing year and decaying year, respectively. (After Wu et al., 2010a). Niño and La Niña is another important nonlinear property of ENSO. The asymmetry of duration between El Niño and La Niña is associated with the asymmetry of zonal wind anomalies over the equatorial far western Pacific. During El Niño mature winter, the western North Pacific (WNP) is dominated by an anomalous anticyclone, referred to as WNPAC, with the equatorial far western Pacific covered by easterly anomalies to the southern flank of WNPAC (Fig. 2a). In contrast, during La Niña mature winter, though an anomalous cyclone is seen in the WNP (named WNPC), it is far weaker and shifted further westward than WNPAC (Figs. 2b, c). Meanwhile, the equatorial western Pacific is dominated by the easterly anomalies stimulated by the La Niña-induced negative precipitation anomalies over the equatorial central Pacific (Ohba and Ueda, 2009; Wu et al., 2010a). The asymmetric zonal wind anomalies over the western Pacific contribute to the asymmetric duration of ENSO. The strong easterly anomalies associated with El Niño tend to generate strong oceanic upwelling Kelvin waves propagating eastward and accelerating decay of El Niño. In contrast, because westerly anomalies do not establish during La Niña winter, there are no corresponding downwelling Kelvin waves generated. Therefore, El Niño generally decays more rapidly than La Niña (Ohba and Ueda, 2009;

4 408 ASIA-PACIFIC JOURNAL OF ATMOSPHERIC SCIENCES Fig. 2. Composite DJF mean 850 hpa streamfunction anomalies for (a) nine El Niño events and (b) nine La Niña events, (listed in Figure caption of Fig. 1, Units: 10 6 m 2 s 1 ). (c) Asymmetric component estimated by the sum of (a) and (b). The shading represent 10% significance level. (d) Symmetric component estimated by the difference of (a) and (b). The solid lines denote poistive values and the dashed lines denote negative ones. The contour interval is 0.5. (After Wu et al., 2010a). Ohba et al., 2010; Okumura and Deser, 2010; Wu et al., 2010a; Wu and Zhou, 2013). Three different mechanisms are proposed to explain the asymmetric low-level atmospheric circulation anomalies over the equatorial far western Pacific. (1) Nonlinear atmospheric response to the SSTA in the equatorial central-eastern Pacific As noted in Section 2.1, suppressed convection over the equatorial central Pacific during La Niña winter and associated easterly anomalies to its west are shifted westward relative to their counterparts during El Niño winter, owing to nonlinear atmospheric response to underling SSTA (Hoerling et al., 1997; Kang and Kug, 2002). The easterly anomalies associated with Fig. 3. (a) Climatological JJA mean 1000 hpa wind. The box represents the target region of the WNP (2-15 o N, o E). (b) Scatter diagram of the JJA mean 1000 hpa zonal wind anomalies against the 1000 hpa wind speed anomalies for each grid in the target region of WNP in 18 ENSO events (listed in figure caption in Fig. 1). The thick line is calculated based on a theoretical model in which that mean wind is set to be 0.7 m s 1. (After Wu et al., 2010a). La Niña extend into the western Pacific, causing the westerly anomalies (or WNPC) have no space to develop (Ohba and Ueda, 2009; Wu et al., 2010a; Okumura et al., 2011). (2) Asymmetry of the SSTA in the WNP During El Niño mature winter, WNPAC is coupled with underlying cold SSTA through positive wind-evaporation-sst feedback (Wang et al., 2000, also seen in Sect. 6). The amplitude of cold SSTA in the WNP is significantly stronger than that of warm SSTA associated with La Niña. The asymmetric SSTA in the WNP has a contribution to the asymmetry of the convection anomalies in situ and thus the asymmetry between WNPAC (westerly anomalies) and WNPC (easterly anomalies) (Wu et al., 2010a). The asymmetry of the SSTA in the WNP is originated from the nonlinear air-sea interaction processes during preceding summer (ENSO developing summer). Although SSTA, precipitation and surface wind anomalies over the WNP are approximately symmetric during that time, the latent heat flux

5 31 August 2014 Tianjun Zhou et al. 409 Table 1. Lists of years for Eastern Pacific and Central Pacific El Niño events. A: Types based on EOF by (Ashok et al., 2007); B: Types based on the relative ration between NINO3 and NINO4 SST index by (Yeh et al., 2009); C: similar to B but including Mixed type, M by Kug et al. (2009); D: Types based on subsurface temperature by Yu et al. (2011), E: Types based on sea surface salinity for by Singh et al. (2011). (Adapted and modified from Christensen et al., 2013b). A B C D E 50/51, 57/58, 63/64, 65/66, 69/70, 72/73, EP 76/77, 79/80, 82/83, 86/87, 87/88, 91/92, 97/98, 03/04, 06/07 72/73, 76/77, 82/83, 97/98 97/98 82/83, 91/92, 97/98 CP 63/64, 68/69, 77/78, 79/80, 86/87, 90/91, 91/92, 92/93, 94/95, 02/03, 04/05, 09/10 68/69, 77/78, 90/91, 92/93, 94/95, 01/02, 02/03, 04/05, 09/10 77/78, 90/91, 94/95, 02/03, 04/05, 09/10 63/64, 68/69, 69/70, 77/78, 77/78, 86/87, 90/91, 92/93, 94/95, 02/03, 04/05, 06/07 anomalies are highly asymmetric over the key WNP region, where the climatological surface wind speed is small (Fig. 3a). Both the westerly anomalies associated with El Niño and the easterly anomalies associated with La Niña in the region enhance local total surface wind speed (Fig. 3b) and evaporation, and thus strengthen (weaken) the cold (warm) SSTA in situ during El Niño (La Niña) (Wu et al., 2010a). (3) Indian Ocean basin mode During El Niño (La Niña) mature winter and following spring, remote forcing from the SSTA in the equatorial centraleastern Pacific can generate an Indian Ocean basin-wide warming (cooling) through an atmospheric bridge, referred to as IOBW (IOBC) (Klein et al., 1999; Alexander et al., 2002; An, 2003, 2004). Hong et al. (2010) found that the amplitude of IOBW be significantly stronger than that of IOBC owing to their asymmetric mixed layer depths. Obha and Watanabe (2012) proposed that the asymmetry of amplitude between IOBW and IOBC has a contribution to the asymmetric zonal wind anomalies over the equatorial western Pacific. The argument is based on the mechanism that IOBW has impacts on the easterly anomalies during El Niño mature winter through driving atmospheric Kelvin waves (Watanabe and Jin, 2002; Annamalai et al., 2005; Kug and Kang, 2006). Dry atmospheric model driven by idealized diabatic heating (e.g., Obha et al., 2012) or state-of-the-art AGCM driven by IOBW-related regional SSTA (e.g., Kug and Kang, 2006) were used to testify the mechanism. However, Wu et al. (2009) pointed out that, in the observation, the convection over the eastern tropical Indian Ocean is suppressed by El Niño remote forcing during El Niño mature winter. Therefore they argued that the SSTA in the tropical Indian Ocean is a slavery of the atmospheric forcing and cannot force atmosphere freely during the season. Wu et al. (2012) further demonstrated that when using diabatic heating in the tropical Indian Ocean with dipole pattern (positive west and negative east) to drive an dry atmospheric model, the westerly instead of easterly anomalies are generated in the equatorial western Pacific, suggesting that IOBW may have no positive contributions to WNPAC or easterly anomalies over the equatorial western Pacific during El Niño mature winter. 3. Central Pacific El Niño and eastern Pacific El Niño The standard El Niño events, i.e., EP events, feature the maximum warming over the equatorial eastern Pacific. It is also called the cold tongue (CT) El Niño, which is regarded as the conventional El Niño (Kug et al., 2009). But it has long been recognized that the SST anomaly pattern associated with El Niño varies from event to event, some events exhibit the warming center in the central Pacific rather than the eastern Pacific (Fu et al., 1986). Since the late 1990s, the maximum SST warming during El Niño has been frequently observed in the central Pacific (Ashok et al., 2007; Kao and Yu, 2009; Kug et al., 2009; Yeh et al., 2009). Such kinds of El Niño events with the maximum warming center in the central Pacific are usually termed as central Pacific (CP) El Niño, which is also called the warm pool (WP) El Niño (Kug et al., 2009). The years for EP and CP El Niño events defined by different indexes are listed in Table 1. The structure, evolution, and teleconnections of two ENSO types are compared by Kao and Yu (2009). Major differences are found in SST warming center location (Fig. 4), spatial patterns of associated atmospheric and oceanic variables (Fig. 5), teleconnection with the Indian Ocean, dominant period, as well as transition mechanisms. The thermocline feedback plays a crucial role in the development of SST anomaly associated with the EP El Niño, while zonal advective feedback (i.e., zonal advection of mean SST by anomalous zonal currents) is a key process during the CP El Niño Compared to the CMIP3 models, the pre-industrial simulations of the CMIP5 models are found to better simulate the observed spatial patterns of the two types of ENSO and have a significantly smaller inter-model diversity in ENSO intensities (Kim and Yu, 2012). 4. ENSO in a warming world: simulation and projections a. Simulations Recent studies show that the EP El Niño has become less frequent and that CP El Niño has become more common during the late twentieth century (Fig. 6; Yeh et al., 2009, 2011)

6 410 ASIA-PACIFIC JOURNAL OF ATMOSPHERIC SCIENCES Fig. 4. Leading EOF patterns obtained from a combined EOF-regression analysis for (a) the eastern-pacic type of ENSO and (b) the central-pacic type of ENSO. Contour intervals are 0.1. The numbers indicated on the top of the panels are the percentage of residual SST variance explained by each EOF mode. (After Kao and Yu, 2009). due to the La Niña-like pattern in the tropical Pacific (Chung and Li, 2013; Xiang et al., 2013; Yeh et al., 2014). Coupled climate models are used to show many weaknesses in the simulation of ENSO. The too large diversity in CMIP3 ENSO amplitude is however reduced by a factor of two in CMIP5 and the ENSO life cycle (location of surface temperature anomalies, seasonal phase locking) is modestly improved (Bellenger et al., 2013; IPCC, 2013; Christensen et al., 2013a). This adds fidelity to the diagnosis based on long-term simulation of coupled models. Numerical simulation results support the notion that a shift to the La Niña-like inter-decadal mean state is responsible for more frequent occurrence of CP El Niño (Choi et al., 2011; Chung and Li, 2013). By using a longterm CGCM simulation, focusing on the role of climate state in the regime change between more and fewer CP El Niño events, it is found that the higher occurrence regime of CP El Niño is associated with a strong zonal gradient of mean sea surface temperature in the equatorial Pacific. This changed climate states obviously intensify zonal advective feedback, which promotes increased generation of the CP El Niño (Choi et al., 2011). Numerical experiments with the Cane and Zebiak model demonstrate that the decadal changes in the surface winds qualitatively reproduce the observed coherent changes in El Niño properties (Wang and An, 2002). The fundamental factor that altered the model's El Niño is the decadal changes of the background equatorial winds and associated upwelling (Wang and An, 2002). On the other hand, the El Niño-La Niña asymmetries also provide a possible mechanism for ENSO to exert a nonzero residual effect, and the accumulated residuals of these asymmetric anomalies could lead to slow changes in the Pacific mean state (Sun and Yu, 2009; Choi et al., 2012). Liang et al. (2012) further elucidated the role of ENSO events in shaping Fig. 5. The distribution of correlation coefcients between the principal components of the EP- and CP-EOFs and surface wind, SST, and precipitation anomalies. (a), (c) The correlations with the EP-EOF; (b), (d) the correlations with the CP-EOF. In (a) and (b) the correlations with the surface wind anomalies are shown as vectors and contours show the correlations with the SST anomalies; correlations with the precipitation anomalies are shown in (c) and (d). Contour intervals are 0.3 for all panels; dashed lines denote negative values. Highlighted vectors indicate the correlation coefcients pass the 95% signicance level estimated by a two-tailed t test. (After Kao and Yu, 2009). the tropical mean climate state and suggested that the decadal warming in the recent decades in the eastern tropical Pacific may be more a consequence than a cause of the elevated ENSO activity. It is demonstrated by a CGCM that the tropical Pacific decadal oscillation can be driven by an interaction between the ENSO and the slowly varying mean background climate state (Choi et al., 2009). Hu et al. (2012) has assessed the prediction skills of two types of ENSO in the Climate Forecast System of National Centers for Environmental Prediction. The hindcast skill for EP ENSO is higher than that of CP ENSO. The preceding westerly wind signal and air-sea interaction differences are responsible for the predication skill difference. They also argued that the frequent increase of the CP ENSO in the recent decades may not be directly connected with the linear trend of the tropical climate. In addition to the changes of ENSO types, the ENSO also

7 31 August 2014 Tianjun Zhou et al. 411 Fig. 6. The time series of the principal component of the (a) eastern- Pacic type of ENSO and (b) central-pacic type of ENSO. Dashed lines denote the 95% signicant level and the values below the curves are uncertain. (After Kao and Yu, 2009). has different flavors. Data diagnosis have identified two modes associated with two flavors of El Niño in terms of the three dimensional structure of atmospheric temperature. The first is a deep-warm mode, which features a coherent zonal mean warming throughout the troposphere from 30 o N to 30 o S with cooling aloft. The second is a shallow-warm mode, which features strong wave signatures in the troposphere with warmth (coolness) over the central Pacific (western Pacific) (Trenberth and Smith, 2006, 2009). These two dominant modes are useful metrics in gauging model performances. Zhou and Zhang (2011) found that the atmospheric temperature anomaly is generally well captured by AMIP II models, demonstrating that the observational changes documented here are driven by SST changes during the El Niño events and the variety of vertical temperature structures associated with two flavors of El Niño are highly reproducible (Fig. 7). The model skill for the first mode is slightly higher than the second mode (Zhou and Zhang, 2011). The model s ability in reproducing the dominant modes is sensitive to the convection schemes employed in the model (Zhang et al., 2010). b. Projections Based on the results of projected global warming scenarios from CMIP3, the projections of anthropogenic climate change are associated with an increased frequency of the CP El Niño Fig. 7. Cross sections of the zonal mean temperature anomaly correlations with the EOF time series: (a) EOF1 and (b) EOF3 from ERA-40; (c) EOF1 and (d) EOF2 from the MME for 11 AMIP AGCMs (after Zhou and Zhang, 2011, Fig. 2).

8 412 ASIA-PACIFIC JOURNAL OF ATMOSPHERIC SCIENCES Fig. 8. Observed composite patterns of monthly mean precipitation anomalies (left panel, units: mm day 1 ) and 850 hpa wind anomalies (right panel, units: m s 1 ) from June to August during the El Niño decaying summer (1982, 1991, 1994, 1997 and 2004 events). The contour values for the precipitation anomaly are ± 1, ± 2.5, ± 4. The shading denotes the 5% significant level for the precipitation (left panel) and the zonal wind (right panel). (After Wu et al., 2010b). compared to the EP El Niño, which is related to a flattening of the thermocline in the equatorial Pacific (Yeh et al., 2009). Ensemble means of the CMIP5 models indicate that the intensity of the CP ENSO increases steadily from the preindustrial to the historical and the RCP4.5 simulations, but the intensity of the EP ENSO increases from the pre-industrial to the historical simulations and then decreases in the RCP4.5 projections (Kim and Yu, 2012). Different from the multi-model ensemble, the overall future prospect on which type of El Niño will occur more frequently in each model is rather uncertain (Yeh et al., 2014). However, a control simulation of the Kiel Climate Model, which was run for 4200 years with the present values of greenhouse gases, exhibit large variations of the occurrence frequency of the CP El Niño versus the EP El Niño. Thus, we cannot exclude the possibility that an increasing of occurrence frequency of CP El Niño during recent decades in the observation could be a part of natural variability in the tropical climate system (Yeh et al., 2011). In addition, McPhaden et al. (2011) found that the changes in background conditions during are opposite to those expected from greenhouse gas forcing in climate models and opposite to what is expected if changes in the background state are mediating more frequent occurrences of CP El Niño. A possible interpretation of these results is that the character of El Niño over has varied naturally and that these variations projected onto changes in the background state because of the asymmetric spatial structures of CP and EP El Niños (McPhaden et al., 2011). Thus, whether the mean climate state change leads to more frequent emergence of CP El Niño or the other way around is not yet known (IPCC,

9 31 August 2014 Tianjun Zhou et al. 413 Fig. 10. Schematic diagram illustrating how the tropical Indian Ocean heating influences the WNP monsoon during the El Niño decaying summer. Ps, Pe and Pt denote the pressure at the surface, the top of the planetary boundary layer and the top of the troposphere, respectively. In response to the enhanced convection over the tropical Indian Ocean, easterly wind anomalies (solid black arrowheads on Pe) are established as an atmospheric Kelvin wave response to the anomalous heating. The anomalous anticyclonic shear vorticity induces the anomalous subsidence (thick black arrowheads) due to the Ekman Pumping effect and thus the boundary layer divergence (solid black arrowheads between Pe and Ps). The anomalous boundary layer divergence further leads to suppressed convection (dashed curve and hollow arrowhead) and low-level anomalous anticyclone (dashed circle) over the WNP. (After Wu et al., 2009). Fig. 9. Observed composite patterns of monthly mean SST anomalies (units: o C) from June to August during the El Niño decaying summer (1982, 1991, 1994, 1997 and 2004 events). The solid lines represent the 5% significant level. (After Wu et al., 2010b). 2013; Christensen et al., 2013a). 5. Impact of EP ENSO on warm-pool air-sea interaction and East Asian - western North Pacific summer monsoon Canonical El Niño events reach mature phase in boreal winter, and then decay rapidly after that. For the following boreal summer, warm SSTA in the equatorial central-eastern Pacific generally disappears or even evolves to an opposite phase (Rasmusson and Carpenter, 1982). However, East Asian (EA)-western North Pacific (WNP) monsoon has strong variations during El Niño decaying summer (See reviews by Li and Wang, 2005; Lau and Wang, 2006; Zhou et al., 2011; Hsu et al., 2014). It is found that WNPAC, maintaining from El Niño mature winter to decaying summer, plays an essential role in linking El Niño and the EA-WNP summer monsoon (Fig. 8, see Zhang et al., 1996; Chang et al., 2000; Wang et al., 2000; Chou, 2003 for previous studies). The WNPAC anomaly corresponds to the westward extension of the western Pacific subtropical high and can enhance the Meiyu-Baiu-Changma precipitation in boreal summer (Fig. 8, see Chang et al., 2000 for earlier study). To understand how El Niño influences the EA-WNP summer monsoon during its decaying phase, we must firstly explore what mechanisms are responsible for the maintenance of WNPAC. Until now, it is still a controversial issue. Four different mechanisms have been proposed to explain it. a. Indian Ocean basin-wide warming IOBW can persist throughout El Niño decaying summer (Fig. 9). It enhances local convection and stimulates atmospheric easterly Kelvin waves to the east and thus modulates the EA-WNP summer monsoon (Yang et al., 2007; Li et al., 2008; Huang et al., 2010). For atmospheric easterly Kelvin waves, the strongest easterlies are located in the equator, which gradually decrease polward. The anticyclonic shear induces a boundary layer divergence over the off-equatorial WNP through Ekman pumping processes. The boundary divergence suppresses convections associated with the WNP summer monsoon and further drives atmospheric Rossby-waves-like anticyclone anomalies to its west in terms of the Gill model (Fig. 10, see Wu et al., 2009; Xie et al., 2009 for details). Wu et al. (2010b) further demonstrated that the Kelvin wave

10 414 ASIA-PACIFIC JOURNAL OF ATMOSPHERIC SCIENCES forcing mechanism relies to the mean state of the WNP summer monsoon. Because the WNP monsoon trough does not fully establish until late summer, the impacts of IOBW on WNPAC primarily occur in late summer. This mechanism has following two potential problems. Firstly, WNPAC does not always concur with IOBW (Fig. S2 of Wang et al., 2013). Secondly, when AGCMs is forced by IOBW-related regional SSTA, the simulated precipitation anomalies in the tropical Indian Ocean are much stronger than those in the observation. Thus these model results may overestimate the contribution of the Indian Ocean remote forcing to WNPAC. Interannual variability of East Asian summer monsoon is partly dominated by the WNPAC. A comparison of interannual variability of East Asian summer monsoon simulated by CMIP3 and CMIP5 AGCMs found that there is a skill dependent on Indian Ocean-western Pacific anticyclone teleconnection, highlighting the importance of IOBM forcing to East Asian climate (Song and Zhou, 2014). Similar idea has been used to assess the performances of CMIP5 coupled models (Zhou et al., 2014). b. Cold SSTA in the WNP remaining from preceding winter and spring During El Niño mature winter, WNPAC is coupled with underlying cold SSTA through a positive wind-evaporation- SST feedback. The northeasterly anomalies to the southeastern flank of WNPAC enhance northeasterly trade wind and thus induce upward latent heat flux, which cools the SST in situ. The cold SSTA suppresses local convection and further reinforces WNPAC to its west in terms of the Gill model (Wang et al., 2000). The work of the positive feedback relies on the mean northeasterly trade wind in the background. For the El Niño decaying summer, the positive feedback changes to a negative feedback owing to the onset of the WNP monsoon southwesterly. Thus the cold SSTA decays under the combining effects of reduced upward latent heat flux and enhanced downward shortwave radiation flux (Chou et al., 2009; Wu et al., 2010b). However, the decay of the cold SSTA from June to August is a gradual process (Fig. 9). Before its complete disappearance, the cold SSTA still suppresses local convection and thus maintains WNPAC in the early summer (Wu et al., 2010b). Xiang et al. (2013) argued that during El Niño decaying summer, though the cold SSTA in the WNP is very weak, it still plays an essential role in maintaining WNPAC and its contribution may be underestimated in Wu et al. (2010b). c. Positive feedback between WNPAC and Indo-western Pacific warm pool WNPAC covers the Indian monsoon region, WNP monsoon region and extends eastward to the trade wind region. On one hand, the northeasterly anomalies to the southeastern flank of WNPAC enhance the mean trade wind, and thus cause positive upward latent heat flux anomalies, which maintain the cold SSTA in the WNP. On the other hand, the easterly anomalies to the southwestern flank of WNPAC weaken the Indian monsoon, and cause negative latent heat flux anomalies, which maintain the warm SSTA in the northern Indian Ocean. The remote warmer SSTA can suppress local convection and reinforce WNPAC through atmospheric Kelvin waves (Wang et al., 2013; Xiang et al., 2013). It is clear that this viewpoint is conflict to the second mechanism discussed above. The key question is whether WNPAC is coupled with underlying cold SSTA through a positive feedback or a negative feedback. Even though a positive wind-evaporation-ssta feedback does exist, whether it is strong enough to suppress the opposite effects of the negative cloud-shortwave feedback remains unknown. d. Tropical-extratropical interactions During El Niño mature winter, atmospheric transient activities can generate cold SSTA in the mid-latitude North Pacific through modulating surface heat fluxes. The cold SSTA gradually moves southwestward to the subtropics in the following spring and summer. The process is called seasonal footprinting mechanism (Vimont et al., 2001). Chen et al. (2013) proposed that the mechanism has a contribution to the cold SSTA in the WNP during El Niño decaying summer and thus WNPAC. But as noted in the second mechanism, the cold SSTA in the WNP may be also generated by the local wind-evaporation- SST feedback in the preceding winter and spring. In the observation, it is difficult to distinguish their relative contributions. The controversy about which one or several mechanisms are responsible for the maintenance of WNPAC during El Niño decaying summer will be resolved through more detailed observational diagnosis and well-designed numerical experiments. The importance of local air-sea interaction in the simulation of western Pacific climate anomalies has been demonstrated by a set of regional ocean-atmosphere coupled model simulations (Zou and Zhou, 2011, 2012, 2013a, b). ENSO also influences the EA-WNP summer monsoon during its developing phase (Chou et al., 2003; Wang et al., 2003; Wu et al., 2005; Wu et al., 2009). During El Niño developing summer, the WNP monsoon is enhanced by the anomalous cyclone stimulated by enhanced convection over the equatorial central-eastern Pacific (Chou et al., 2003; Wang et al., 2003). On the other hand, weakened Indian summer monsoon precipitation, which is associated with El Niño remote forcing, reduces precipitation over northern China through an anomalous atmospheric Rossby wave propagating along o N (Wu et al., 2005). Analyses of climate models also improve our understanding of the forcing mechanism. The interannual variability of the western North Pacific subtropical high is characterized by two distinct modes associated with different SSTA patterns (e.g., He et al., 2013; He and Zhou, 2014). The positive phases of

11 31 August 2014 Tianjun Zhou et al. 415 both two modes feature an anomalous anticyclone over the western North Pacific (WNP), but with easterly (westerly) wind anomalies in equatorial Pacific associated with the first (second) mode. The first mode is associated with warm SSTA over the TIO and the phase transition from El Niño into La Niña over the Tropical Pacific, whereas the second mode is associated with cold SSTA over the WNP. The first mode is reasonably reproduced by the CMIP5 AGCMs, suggesting it is an SST-forced mode. In contrast, the second mode is not well reproduced by the AMIP models due to the models failure in simulating the local SST-WNPAC relationship over the WNP, suggesting the second mode is not an SST-forced phenomenon, and local air-sea interaction may be essential (He and Zhou, 2014). The existences of two modes with similar anticyclone in the WNP adds difficulty in the understanding of the forcing mechanism of interannual variability of WNP subtropical high (Wu and Zhou, 2008). Another interesting feature deserving further study is the nonlinearity of ENSO impacts on the EA-WNP summer monsoon. Firstly, the impacts are intensity-dependent. In moderate ENSO years, there is excessive rainfall along the Yellow- Huaihe River valleys but deficient rainfall along the Yangtze River valley (Xue and Liu, 2008). This pattern is far different from the classical ENSO-related rainfall anomalies (e.g., Zhang et al., 1996). Secondly, the response of the EA summer monsoon leading mode to El Niño and La Niña is asymmetric (Liu et al., 2008). 6. Impacts of CP ENSO on Asian-Pacific climate a. Impacts on tropical Pacific climate and beyond The two types of ENSO events significantly influence the temperature and precipitation over many parts of the globe. EP ENSO is associated with signicant wind stress anomalies covering a large part of the tropical Pacic (Fig. 5a), while surface wind anomalies associated with the CP ENSO are limited to the equatorial central-to-western Pacic (130 o E- 160 o W) and off Australia (Fig. 5b). For EP ENSO, the precipitation anomalies extend from the equatorial eastern to central Pacic, which reects zonal shifts of the Pacic Walker circulation (Fig. 5c), while for CP ENSO, the precipitation anomalies are characterized by a dipole pattern within the tropical Pacic with largest anomalies located mainly in the far western and eastern Pacic (Fig. 5d) (Kug et al., 2009; Yeh et al., 2009). In addition to the tropics, the eastern Pacic, North America and North Atlantic are the major regions having robust climate differences between the EP and CP El Niños. Nevertheless, the climate contrasts seem not robust over the Eurasian continent (Hu et al., 2012). Depending on the season, the impacts of CP El Niño over regions such as the Far East including Japan, New Zealand, western coast of United States, etc., are opposite to those of the EP ENSO (Ashok et al., 2007). Weng et al. (2011) investigated the possible influences of Fig. 11. Partial correlation patterns of zonal wind component at 850 hpa for (a) Nino3, (b) El Niño Modoki index, and (c) IOD. Instead of using coefcients, the signicant levels are given for reference. The red (blue) shading shows the westerly (easterly) anomalies. The difference between the area-averaged zonal wind anomaly in the southern and northern black boxes are used to dene the WNPSM index (Wang and Fan, 1999) (After Weng et al., 2011). three coupled ocean-atmosphere phenomena in the Indo- Pacific Oceans, EP El Niño, CP El Niño Modoki and the Indian Ocean Dipole (IOD), on summer climate in China based on partial correlation/regression analysis. Among the three phenomena, CP El Niño has the strongest relationship with the WNP summer monsoon (Fig. 11). When two or three phenomena coexist with either positive or negative phase, the influences exerted by one phenomenon on summer climate in different regions of China may be enhanced or weakened by other phenomena. In 1994 when both CP El Niño and IOD are prominent without El Niño, a strong WNP summer monsoon is associated with severe flooding in southern China and severe drought in the Yangtze River Valley (YRV). The 500

12 416 ASIA-PACIFIC JOURNAL OF ATMOSPHERIC SCIENCES hpa high systems over China are responsible for heat waves in most parts of China. In 1983, when a strong negative phase of CP El Niño is accompanied by moderate El Niño and IOD, a weak WNP summer monsoon is associated with severe flooding in the YRV and severe drought in southern China. The 500 hpa low systems over China are responsible for the cold summer in the YRV and northeastern China. For rainfall, the influence path seems largely through the low-level tropospheric circulations including the WNP summer monsoon. For temperature, the influence path seems largely through the midlevel tropospheric circulations over East Eurasia/western North Pacific Ocean. The 2004 event, among all the CP El Niño cases, is distinct from the point that it is a pure event; viz. there is no accompanying IOD event in the tropical Indian Ocean (Ashok et al., 2009). Several parts of the globe experienced anomalous climate conditions during the boreal summer of 2004, such as severe drought in Japan, Maritime Continent, and southern Mexico and Ecuador, surplus rainfall in Philippine Islands and central tropical Pacific (Ashok et al., 2009). Ashok et al. (2007) attribute these anomalous summer conditions in 2004 to a CP El Niño event. Using observed data and an AGCM, it is showed that during boreal summer anomalous twin Walker circulation cells associated with the CP El Niño SSTA give rise to the rainfall anomalies in the tropics (Ashok et al., 2009). b. Impacts of CP ENSO on seasonal precipitation in Asia Fig. 12. Schematic diagrams showing the circulation anomlaoes associated with warm events of (a) CP ENSO and (b) EP ENSO. WPSH: western Pacific subtropical high. Dashed lines represent the climatological mean. Solid lines indicate anomalous circulation, and heavy arrows represent nomalous wind directions. C and AC indicate cyclonice and antoicyclonice circulation anomalies, respectively (After Feng and Li, 2011, Fig. 11). The impact of CP ENSO on precipitation in China is seasonal-dependent and also different to EP ENSO. Feng and Li (2011) investigated the relationship between CP ENSO and spring rainfall over SC (SC) by using observed data sets from 1979 to They found that while warm events of CP ENSO are accompanied by a significant reduction in rainfall over SC, there is enhanced rainfall associated with warm events of EP ENSO. The different responses of SC precipitation are dominated by the different responses of atmospheric circulations associated with two types of ENSO (Fig. 12). The western Pacific subtropical high (WPSH) shifts southeastward associated with the conventional El Niño / EP ENSO, but extends northwestward associated with warm phase of CP ENSO (Fig. 12). Moreover, there exists a strong asymmetry in the relationship between SC spring rainfall, EP ENSO and CP ENSO events. These relationships are only statistically significant for positive events; the cold events of two types ENSO on SC precipitation in spring are not significant. The summer precipitation anomalies over eastern China associated with the two types of ENSO are compared by Karori et al. (2013). It is found that the two types of ENSO have asymmetric features with respect to the impact of their positive and negative phases on boreal summer rainfall over the YRV and SC. Following a warm phase of CP ENSO (called El Niño Modoki in their paper), the precipitation anomalies over eastern China features a tripolar pattern, with excessive precipitation along the YRC but deficient precipitation over the southern China and the northern China. This pattern is different to the precipitation anomalies associated with canonical or EP ENSO (Fig. 13). The precipitation anomalies are dominated by an asymmetric atmospheric response to the asymmetric sea surface temperature anomalies (SSTAs) in the lower troposphere (Fig. 14). The location of the center of the anomalous circulations differs during the opposite phases of the two types of ENSO. Numerical simulations with AGCM show that the asymmetric response of the lower atmosphere is driven mainly by differing SSTA patterns in the equatorial Pacific Ocean (Karori et al. 2013). The boreal autumn precipitations over eastern China associated with types of ENSO are different (Zhang et al., 2011). The conventional warm EP ENSO year autumns usually see excessive precipitation in the southern China, but deficient precipitation is witnessed in warm CP ENSO year autumns. They also suggested that the zonal shift of warmer SSTA associated with two types of ENSO explains the drying trend observed in SC since the 1990s. A similar mechanism is used to explain the drought tendency occurred in the southwestern part of the United States of America (Zhang et al., 2012). During boreal winter, the precipitation anomalies associated with two types of ENSO are different (Feng et al., 2010). In the case of warm EP ENSO events, wet conditions are observed over SC, and dry conditions are seen over the Philippines, Borneo, Celebes, and Sulawesi. In contrast, for warm CP ENSO events, the negative rainfall anomalies around the Philippines are weaker and are located more northward compared to the EP ENSO counterpart. The different precipitation

13 31 August 2014 Tianjun Zhou et al. 417 Fig. 13. Normalized regression/correlation anomaly pattern of JJA rainfall over China for (a) WP El Niño events, (b) WP La Niña events, (c) CT El Niño events, and (d) CT La Niña events. Hatched regions indicate regression coefficients significant above the 90% confidence level (After Karori et al., 2013, Fig. 3). anomalies are attributed to the different anomalous Walker circulation and low-level anticyclone around the Philippines. Both the Philippine anticyclone and the descending branch center of the Walker circulation over the WNP occupy a smaller domain and are located more northward during CP ENSO warm events than during EP ENSO warm phases. Numerical experiments suggest that above different low-level atmospheric responses are driven by different diabatic cooling over the western North Pacific associated with two types of ENSO warm phases (Feng et al., 2010). c. Impacts of CP ENSO on tropical cyclone in the western Pacific The different impacts of two types of El Niños on tropical cyclone (TC) tracks over the WNP are examined by Hong et al. (2011) based on observational data. The difference of TC tracks between CP El Niño and EP El Niño is not evident in boreal summer (JJA) but is great in boreal autumn (SON). During CP El Niño the TCs re-curve northward at a further westward location near the coastline of East Asia, thus more TCs land in Taiwan island and SC continent. The westward shift of the subtropical high and associated steering flow during CP El Niño is a key factor that causes the difference in the TC tracks in autumn. Numerical experiments demonstrate the dominance of different local SST anomalies in the WNP between CP El Niño and EP El Niño. Chen et al. (2011) investigated the modulation of two types of ENSO on TC frequency over the South China Sea (SCS) during boreal summer and fall for An abovenormal TC frequency over the SCS is found during June- August (JJA) for the CP El Niño years, whereas the belownormal TC frequency is seen during September-November (SON) for the EP El Niño years. The broad-scale convection anomaly over the WNP during JJA for CP El Niño drives a zonally elongated cyclonic anomaly over the WNP and SCS, leading to enhanced TC activity. In contrast, during SON for the EP El Niño, the cooling source centered in the Maritime

14 418 ASIA-PACIFIC JOURNAL OF ATMOSPHERIC SCIENCES Fig. 14. Schematic representation of the atmospheric circulation anomalies at 850-hPa level associated with (a) WP El Niño, (b) WP La Niña, (c) CT El Niño, and (d) CT La Niña. The areas shaded in red (blue) represent positive (negative) SSTAs. The STH is for subtropical high, gray lines show the climatological mean, and black lines show anomalous circulations. The letters C and A represent cyclonic and anticyclonic circulation, respectively (After Karori et al., 2013, Fig. 8). Continent drives an anticyclonic anomaly over the SCS, resulting in a suppressed TC activity. 7. Summary and concluding remarks ENSO is the most important driver of global climate variability on interannual time scale. More and more observational evidences have revealed changing flavors of ENSO in the past decades. Understanding the changes of ENSO including its flavors and exploring the underlying mechanisms are of crucial scientific needs. Based on our understanding of past changes, how to make reliable prediction/projection of climate anomalies associated with changes of ENSO types and flavors are also of great socioeconomic importance. In this article, progresses in the studies of the changes of ENSO and their impacts on Asian-Pacific climate are reviewed. Major progresses are summarized below: 1) The asymmetric behaviors between El Niño and La Niña are one important aspect of ENSO s nonlinear characteristics. The most striking asymmetry is their asymmetric amplitude, that is, amplitude of El Niño is significantly stronger than that of La Niña. The asymmetric amplitude is closely associated with the nonlinear advections of ocean temperature. But whether the horizontal advection works or the vertical advection does is still a controversial issue. In addition to nonlinear ocean dynamics, the other mechanisms, such as the nonlinear atmospheric response to underlying SST forcing and nonlinear interactions between ENSO and shorter time scale variability ranging from synoptic to intra-seasonal time scales may also play some roles in the asymmetric amplitude of ENSO. Another interesting asymmetric behavior is La Niña tends to persist much longer than El Niño. The asymmetric duration is associated with the asymmetric low-level circulation anomalies over the tropical western Pacific.