J. Meteor. Res.  2019, Vol. 33 Issue (1): 46-65 PDF
http://dx.doi.org/10.1007/s13351-019-8101-6
The Chinese Meteorological Society
0

#### Article Information

LU, Bo, and Hong-Li REN, 2019.
ENSO Features, Dynamics, and Teleconnections to East Asian Climate as Simulated in CAMS-CSM. 2019.
J. Meteor. Res., 33(1): 46-65
http://dx.doi.org/10.1007/s13351-019-8101-6

### Article History

in final form November 19, 2018
ENSO Features, Dynamics, and Teleconnections to East Asian Climate as Simulated in CAMS-CSM
Bo LU1,2, Hong-Li REN1,3
1. Laboratory for Climate Studies & China Meteorological Administration–Nanjing University Joint Laboratory for Climate Prediction Studies, National Climate Center, China Meteorological Administration, Beijing 100081;
2. Xinjiang Climate Center, Urumqi 830002;
3. Department of Atmospheric Science, School of Environmental Studies, China University of Geoscience, Wuhan 430074
ABSTRACT: This study evaluates the performance of CAMS-CSM (the climate system model of the Chinese Academy of Meteorological Sciences) in simulating the features, dynamics, and teleconnections to East Asian climate of the El Niño–Southern Oscillation (ENSO). In general, fundamental features of ENSO, such as its dominant patterns and phase-locking features, are reproduced well. The two types of El Niño are also represented, in terms of their spatial distributions and mutual independency. However, the skewed feature is missed in the model and the simulation of ENSO is extremely strong, which is found—based on Bjerknes index assessment—to be caused by underestimation of the shortwave damping effect. Besides, the modeled ENSO exhibits a regular oscillation with a period shorter than observed. By utilizing the Wyrtki index, it is suggested that this periodicity bias results from an overly quick phase transition induced by feedback from the thermocline and zonal advection. In addition to internal dynamics of ENSO, its external precursors—such as the North Pacific Oscillation with its accompanying seasonal footprinting mechanism, and the Indian Ocean Dipole with its 1-yr lead correlation with ENSO—are reproduced well by the model. Furthermore, with respect to the impacts of ENSO on the East Asian summer monsoon, although the anomalous Philippine anticyclone is reproduced in the post-El Niño summer, it exhibits an eastward shift compared with observation; and as a consequence, the observed flooding of the Yangtze River basin is poorly represented, with unrealistic air–sea interaction over the South China Sea being the likely physical origin of this bias. The response of wintertime lower-tropospheric circulation to ENSO is simulated well, in spite of an underestimation of temperature anomalies in central China. This study highlights the dynamic processes that are key for the simulation of ENSO, which could shed some light on improving this model in the future.
Key words: model evaluation     ENSO     dynamics     teleconnection     CAMS-CSM
1 Introduction

El Niño–Southern Oscillation (ENSO) is the most dominant climatic fluctuation on the interannual timescale. It is characterized by large-scale sea surface temperature (SST) anomalies over the central to eastern Pacific, associated with changes of thermal condition in the upper ocean and convective anomalies in the atmosphere. Since Bjerknes (1969), great progress has been made in understanding the dynamics of ENSO (e.g., Wyrtki, 1985; Suarez and Schopf, 1988; Battisti and Hirst, 1989; Neelin, 1991; Li, 1997). ENSO is well-recognized as resulting from ocean–atmosphere interactions over the tropical Pacific, and theories have been proposed to explain its formation and development. For example, Jin (1997a, b) proposed a recharge oscillator framework, which highlights the importance of changes of upper-ocean heat content in the phase transition of ENSO.

In spite of the advanced theoretical understanding of ENSO, accurately simulating ENSO with coupled global circulation models (CGCMs) remains challenging. For instance, Guilyardi (2006) showed that the simulated ENSO often exhibits large spread in amplitude among those CGCMs that participated in the Coupled Model Intercomparison Project phase 3 (CMIP3). Although this diversity in amplitude has been reduced in CMIP5, the biases in ENSO periodicity are still large (Bellenger et al., 2014; Lu et al., 2018). The modeled oscillation is often too regular (AchutaRao and Sperber, 2006), as compared to the broad spectrum ranging from 2 to 7 yr found in observation. Besides, the asymmetry (Bellenger et al., 2014) and seasonal phase-locking feature (Guilyardi et al., 2009) of ENSO are often poorly simulated by CGCMs.

The biases in ENSO behaviors originate from the poor representation of key physical processes in ENSO dynamics. For example, Bellenger et al. (2014) showed that current CGCMs generally underestimate ENSO-induced surface wind, which results in a weakened thermocline and zonal advective feedbacks. In addition, due to biases in the modeled relationship between shortwave radiation and SST, the heat-flux-induced SST damping is also underestimated, which has been demonstrated in models for both CMIP3 (Lloyd et al., 2009) and CMIP5 (Bellenger et al., 2014) models. The errors in the modeled long-term mean state in the tropical Pacific Ocean can also be a source of bias in the modeled behaviors of ENSO (van Oldenborgh et al., 2005; Guilyardi, 2006). Since the physics of ENSO is complex, its modeled behaviors might be good due to error compensations (Kim S. T. et al., 2014b). Thus, it is necessary to evaluate the modeled physics of ENSO in detail.

Based on the linear framework of the recharge/discharge oscillator (Jin, 1997a), Jin et al. (2006) proposed the Bjerknes stability index (BJ-index), which measures the growth rate of ENSO and captures the effects of thermocline feedback, zonal advective feedback, Ekman feedback, thermal damping, and dynamical damping comprehensively. Many studies have shown that the BJ-index provides a basis for a better understanding of the biases in the modeled amplitude of ENSO (e.g., Kim and Jin, 2011; Kim S. T. et al., 2014a). Recently, Lu et al. (2018) formulated the Wyrtki index to capture the periodicity of ENSO, which describes the thermocline and zonal advective feedbacks multiplied by the efficiency of the discharge–recharge process of the equatorial oceanic heat content driven by ENSO wind stress anomalies. The diversity of the simulated periodicity of ENSO in CMIP5 models can be explained well by this index (Lu et al., 2018). In this paper, the simulation of ENSO by a coupled climate model, CAMS-CSM (the climate system model of the Chinese Academy of Meteorological Sciences), is evaluated. In particular, the dynamical origin of the biases in ENSO behaviors is investigated by adopting the BJ-index and Wyrtki Index.

As the strongest climate phenomenon on the interannual timescale, ENSO can induce remote climate impacts by stimulating teleconnections. A number of studies have shown that El Niño might weaken the East Asian winter monsoon (EAWM) by suppressing convection over the northwestern Pacific (Li, 1990; Zhang et al., 1996). In addition, the relationship between ENSO and the East Asian summer monsoon (EASM) relationship has been extensively studied. In the decaying summer of two super-strong El Niño events, in 1982/83 and 1997/98, destructive flooding occurred in the Yangtze River basin. A number of hypotheses have been proposed to explain this well-known ENSO–EASM relationship (Wang et al., 2000; Xie et al., 2009, 2016; Stuecker et al., 2015; Zhang et al., 2016). Models with a better representation of ENSO-induced teleconnections are capable of better predicting the East Asian climate (Li et al., 2016; Lu et al., 2017; Liu et al., 2018). Thus, a realistic simulation of ENSO itself is insufficient for East Asian climate simulation, although successful ENSO simulation is often necessary for a reasonable simulation of ENSO teleconnections (Gong et al., 2014, 2015, 2018). In this paper, the modeled impacts of ENSO on East Asian climate in CAMS-CSM are also assessed.

The remainder of the paper is organized as follows. The observational dataset and a description of CAMS-CSM are given in Section 2. The simulated mean state of the tropical Pacific and the modeled behaviors of ENSO are presented in Section 3. ENSO’s dynamics and the origin of the simulated ENSO biases are investigated in Section 4. Section 5 documents the simulated influence of ENSO teleconnections on East Asian climate. Finally, a summary and discussion are provided in Section 6.

2 Data and methodology

This study utilizes Hadley Centre Global Sea Surface Temperature dataset (HadISST; Rayner et al., 2006) for the observed monthly mean SST, Climate Prediction Center Merged Analysis of Precipitation (CMAP; Xie and Arkin, 1997) for the observed precipitation, Hadley Centre/Climatic Research Unit version 4.6 (HadCRUT- 4.6; Morice et al., 2012) for the observed surface air temperature, and Wave and Anemometer-based Sea-surface Wind dataset (WASWind ; Tokinaga and Xie, 2011) for the observed surface wind over ocean. Their horizontal grid resolutions are 1° × 1°, 2.5° × 2.5°, 5° × 5°, and 4° × 4°, respectively. A common period after 1979 is analyzed for the observational data. Reanalysis-1 (NCEP–NCAR reanalysis data; Kalnay et al., 1996) is used to retrieve the horizontal wind, geopotential height, and surface heat flux. The horizontal grid resolution of the atmospheric reanalysis data is 2.5° × 2.5°, and a common period of 1979–2016 is analyzed to be consistent with the observational data. Global Ocean Data Assimilation System (GODAS) data (Behringer et al., 1998) are used to represent the objective estimations of the monthly 3-D current and temperature reanalysis data in ocean, as well as the momentum at sea surface since 1980. GODAS reanalysis data contain 40 vertical levels, with a horizontal resolution of 1° × 1/3°.

CAMS-CSM is composed of an atmospheric model (ECHAM5), oceanic model [Modular Ocean Model, version 4 (MOM4); Griffies et al., 2004], sea-ice model (SIS), and land-surface model (CoLM). The resolution of CAMS-ECHAM is T106, with 31 vertical levels from the surface to 10 hPa. The oceanic component, MOM4, uses a tripolar grid (Murray, 1996), with a zonal resolution of 1° globally. The meridional resolution is 1/3° within 10°S–10°N, which changes to 1° from 30°S (N) to the poles. There are 50 vertical layers in CAMS-MOM. For more detailed information about the model design, please refer to Rong et al. (2018).

In this paper, the historical simulation of CAMS-CSM is evaluated. CAMS-CSM is initialized at the quasi-equilibrium state derived from the control experiment. Then, the historical simulation is performed under the historical forcings of ozone, solar forcing, greenhouse gases, and anthropogenic aerosol from 1900–2013. The atmospheric output is available during the whole simulation period, while the three-dimensional oceanic output is only available from 1980–2013.

3 Modeled features of ENSO

It has been reported that the behaviors of ENSO are sensitive to the mean state of the tropical Pacific, such as the mean thermocline depth (Zebiak and Cane, 1987; Timmermann et al., 1999; Fedorov and Philander, 2001), mean trade wind (Fedorov and Philander, 2001; Zheng et al., 2008), and zonal SST gradient (Knutson et al., 1997; Fedorov and Philander, 2001). Thus, the simulated mean state in CAMS-CSM is firstly assessed. Figures 1a and 1b compare the annual mean SST and surface wind patterns between observation and simulation. Generally, CAMS-CSM captures the fundamental features of the mean state over the tropical Pacific, including the prevailing easterly trade wind and the “warm pool (WP)–cold tongue (CT)” distribution of SST. However, some biases still exist. For example, the trade wind over the southeastern Pacific is underestimated by the model. As a result, the SST in the southeastern Pacific is warmer than observed. Meanwhile, the WP in the Southern Hemisphere exhibits an eastward extension. Associated with this SST mean-state bias, excessive precipitation is evident in the South Pacific Convergence Zone (Figs. 1c, d). This so-called “double ITCZ (Intertropical Convergence Zone) problem” is suffered by most state-of-the-art CGCMs (Lin, 2007; Hwang and Frierson, 2013). In addition, compared with observation, an excessive westward extension of the modeled equatorial Pacific CT is evident. This so-called CT bias is another common error in many CGCMs (Misra et al., 2008; Li and Xie, 2014), which can have impacts on the simulation of ENSO (Vannière et al., 2013).

 Figure 1 Climatological annual mean (a, b) SST (color shaded; °C) overlapped with surface wind (vectors; m s–1) and (c, d) precipitation (mm day–1) from (a, c) observation and (b, d) CAMS-CSM.

Mean thermocline depth is an important indicator of the strength of thermocline feedback (Fedorov and Philander, 2001). Figure 2 demonstrates the mean thermocline depth (indicated by the 20°C isotherm) and equatorial zonal wind stress along 120°E–80°W. The modeled climatological thermocline is slightly shallower in the central to western Pacific, which is consistent with the westward extension of the simulated CT. It implies that the modeled thermocline is less tilted than observed. The simulated easterly wind stress is slightly weaker in the central to eastern Pacific.

 Figure 2 The mean (a) thermocline depth (indicated by the 20°C isotherm) and (b) equatorial (5°S–5°N average) zonal wind stress along 120°E–80°W in observation (black curve) and the CAMS-CSM simulation (red curve).

To obtain the principal modes of SST variability over the tropical Pacific, empirical orthogonal function (EOF) analysis is performed. As shown in Fig. 3a, the leading EOF mode indicates the CT El Niño, which is characterized by a strong anomalous warming along the equatorial central to eastern Pacific. The simulated leading EOF mode is similar to that observed in terms of the SST spatial distribution, except the overly strong anomalous warming (Fig. 3b). As shown in Fig. 3c, the observed second EOF mode indicates the WP El Niño, which is characterized by an anomalous warming in the central tropical Pacific and cooling in the eastern tropical Pacific (Ashok et al., 2007; Weng et al., 2007). Such WP El Niño events have occurred more frequently since 2000, which can be attributed to the La Niña-like interdecadal SST mean state (Chung and Li, 2013). Associated with this distinct SST pattern, the climatic impact of WP El Niño is very different to that of conventional (CT) El Niño (Weng et al., 2009). Here, it is demonstrated that CAMS-CSM generally captures the observed features of WP El Niño (Fig. 3d). Anomalous warming extending from the subtropical northeastern Pacific to the central Pacific is simulated well (Yu et al., 2010), with anomalous cooling in the eastern Pacific. It should be noted, however, that the modeled equatorial warming shows a westward extension, with an unrealistic warming to the west of 150°E. In addition, the explained variance of the modeled second EOF mode is 7%, which is much smaller than the observed.

 Figure 3 Spatial patterns of the (a, b) first and (c, d) second EOF modes of SST anomalies in the tropical Pacific in (a, c) observation and (b, d) the CAMS-CSM simulation.

Ham and Kug (2012) showed that the Niño3 and Niño4 indices are significantly correlated with each other in most CMIP3 models, implying the existence of one single type of El Niño rather than two types in CMIP3 simulations. Such a performance is slightly improved in CMIP5 models, in that the two types of El Niño are simulated with greater independence compared with CMIP3 (Kug et al., 2012). Here, the dependence between the two types of El Niño is examined for CAM-CSM. Following Ham and Kug (2012), the modified Niño3 (SST averaged over 5°S–5°N, 170°–110°W) and Niño4 (SST averaged over 5°S–5°N, 140°E–170°W) indices are adopted to indicate the CT El Niño and WP El Niño, respectively. As demonstrated in Fig. 4a, the magnitude of the modified Niño3 index is not linearly proportional to the modified Niño4 index, implying the potential existence of two types of El Niño in observation. In the CAMS-CSM simulations (Fig. 4b), the CT El Niño events and WP Niño events depart from the one-to-one line. In other words, the independence between CT El Niño and WP El Niño can be simulated to some degree. Unlike the El Niño cases, CT La Niña events and WP La Niña events gather along the one-to-one line (Fig. 4c), which implies that the flavors of La Niña events are not so clear as those for El Niño. In the model simulation, the observed relations between CT La Niña and WP La Niña events are reproduced but with greater dependency (Fig. 4d). Kug et al. (2009, 2011) suggested that La Niña events are harder to separate into two types because of the similarity in SST patterns. Here, we show that this asymmetric characteristic between El Niño and La Niña events is captured well by CAM-CSM. However, it is noted that, although the model captures the spatial patterns and mutual independence of the two types of El Niño events, more evaluations should be performed regarding their distinct dynamics, which is beyond the scope of this study.

 Figure 4 Scatterplots between the normalized modified Niño3 and Niño4 indices in boreal winter (December–February) during (a, b) El Niño events and (c, d) La Niña events from (a, c) observation and (b, d) the CAMS-CSM simulation. Note that the red (blue) dots denote the CT (WP) El Niño and La Niña. The correlation coefficients are also given. The definition of the modified Niño3 index is the seasonal mean SST anomalies averaged over 5°S–5°N, 170°–110°W, while that of Niño4 is over 5°S–5°N, 140°E–170°W.

The evolutions of El Niño and La Niña events are assessed in Fig. 5. The seasonal phase-locking feature is successfully captured by CAMS-CSM. Similar to observed, the modeled El Niño and La Niña events develop during boreal summer, and mature in boreal winter. This may due to the realistic seasonally dependent coupled instability (Li, 1997) in CAMS-CSM. In observation (Fig. 5b), most El Niño events turn to a La Niña event in the following winter (Chen M. C. et al., 2016), and few El Niño events persist for more than 2 yr (Chen and Li, 2017), while La Niña is often re-intensified in the second year (Fig. 5d). This evolutionary asymmetry is caused by the asymmetric wind response in the western Pacific and the asymmetric cloud–radiation–SST and evaporation–SST feedbacks in different phases of ENSO (Chen M. C. et al., 2016). Although CAMS-CSM simulates the phase transition for El Niño (Fig. 5a), the re-intensification of La Niña is poorly simulated. As demonstrated in Fig. 5c, most modeled La Niña events turn into El Niño events in the following winter, indicating a failure to simulate the observed evolutionary asymmetry. It is also clear that the simulated El Niño and La Niña events are generally stronger than those observed. Figure 6a demonstrates the standard deviation of Niño3.4 SST anomalies. It is clear that the overestimation of ENSO amplitude is pronounced in each calendar month. In addition, this bias in amplitude exists for both types of El Niño, since the standard deviation of the modified Niño3 and Niño4 anomalies are both overestimated (Figs. 6b, c). Observationally, El Niño events are usually stronger than La Niña events. The skewness of the observed Niño3.4 SST anomaly is 0.4. However, this asymmetry is poorly represented in the simulation. The modeled El Niño and La Niña events have comparable amplitude, and the skewness of the modeled Niño3.4 SST anomaly is –0.1. ENSO’s periodicity is further evaluated in Fig. 7. The observed spectrum exhibits a broad band from 2–7 yr. The modeled spectrum, however, shows a strong peak every 2.8 yr, implying ENSO in CAMS-CSM tends to be more frequent and regular in terms of its oscillation pattern, with a shorter period compared to observed.

 Figure 5 Evolution of Niño3.4 anomalies during (a, b) El Niño and (c, d) La Niña events from (a, c) the CAMS-CSM simulation and (b, d) observation. The composites are indicated by the thick red curves. Here, the events with an absolute peak value exceeding 0.5°C are selected.
 Figure 6 Standard deviation of SST anomalies (°C) averaged within the (a) Niño3.4 region (5°S–5°N, 170°–120°W), (b) modified Niño3 region (5°S–5°N, 170°–110°W), and (c) modified Niño4 region (5°S–5°N, 140°E–170°W), in each calendar month, in observation (blue bars) and the CAMS-CSM simulation (yellow bars).
 Figure 7 Spectra of Niño3.4 anomalies in observation (black curve) and the CAMS-CSM simulation (red curve).
4 ENSO dynamics in CAMS-CSM 4.1 Amplitude bias and its physical origin

To improve the model, it is necessary to understand the physical origins of the errors in simulating the behaviors of ENSO. It has been shown that CAMS-CSM overestimates the amplitude of ENSO (Figs. 5, 6). The BJ-index, proposed by Jin et al. (2006), provides a powerful dynamical estimation of ENSO’s growth rate (Kim and Jin, 2011; Kim S. T. et al., 2014a), and thus it is adopted here to diagnose the origin of this amplitude bias in CAMS-CSM. Based on the recharge–discharge framework, the BJ-index can be formulated as follows:

 $\hspace{-170pt} {I_{\rm BJ}} = \dfrac{{R - \varepsilon }}{2} \approx \dfrac{R}{2},$ (1)
 $\begin{array}{l} R \!=\! {\mu _a}{\beta _u} < \!-\! \dfrac{{\partial \overline T }}{{\partial x}} > \!+\! {\mu _a}{\beta _h}{a_h} < \dfrac{{H(\bar w)\bar w}}{{{H_{\rm m}}}} > \!+\! {\mu _a}{\beta _w} < H(\bar w)\dfrac{{\partial \overline T }}{{\partial z}} > \\ \;\;\;\;\;\;\;\; - ({a_1}\dfrac{{ < \Delta \overline u > }}{{{L_x}}} + {a_2}\dfrac{{ < \Delta \overline v > }}{{{L_y}}}) - {\alpha _{\rm s}}, \end{array}$ (2)

where T denotes ocean temperature averaged over the mixed layer; u, v, and w represent the zonal, meridional, and vertical ocean current, respectively; Lx and Ly are the zonal and meridional extents of the Pacific; Hm is the mixed-layer depth, which is fixed at 50 m for simplicity; a1 (a2) is derived by regressing the SST anomalies at zonal (meridional) boundaries against the area-averaged SST anomalies; H(x) is a step function, which only considers regions with upward vertical advection; the sign “< >” denotes the average of quantities over the eastern Pacific (5°S–5°N, 170°–80°W); and the overbars represent the climatological mean state. For more details about the formulation of the BJ-index, readers are referred to Jin et al. (2006).

To obtain the BJ-index in Eq. (2), several balance relationships are applied. The parameter ${\mu _a}$ indicates the strength of the wind stress response to ENSO SST forcing; ${\beta _u}$ measures the wind-driven zonal current anomalies; ${\beta _h}$ represents the response of the thermocline slope to the surface wind stress anomalies; ${\beta _w}$ measures the wind-driven upwelling anomalies; ${\alpha _s}$ indicates the surface heat flux changes in response to ENSO SST forcing; and ah denotes the relationship between the subsurface temperature anomaly and thermocline depth anomaly. As demonstrated in Fig. 8, these balance relationships hold relatively well for both the GODAS reanalysis data and the CAMS-CSM simulation, which ensures the robustness of the BJ-index in representing ENSO’s growth rate.

 Figure 8 Scatterplots of (a, g) $[{\tau _x}] = {\mu ^*}_a < T >$ , (b, h) $< {T_{\rm sub}} > = {a_h} < h >$ , (c, i) $< w > = - {\beta _w}[{\tau _x}]$ , (d, j) $< Q > = - {\alpha _s} < T >$ , (e, k) $< h > - [h] = {\beta _h}[{\tau _x}]$ , and (f, l) $< u > - {\beta _{uh}}[h] = {\beta _u}[\tau ]$ , in (a–f) observation and (g–l) the CAMS-CSM simulation. The linear fitting lines are indicated by the red straight lines. Please refer to Jin et al. (2006) for more details about these balance equations.

In Eq. (2), the BJ-index consists of three positive feedbacks (zonal advective feedback: ${\mu _a}{\beta _u} < - \dfrac{{\partial \overline T }}{{\partial x}} >$ ; thermocline feedback: ${\mu _a}{\beta _h}{a_h} < \dfrac{{H(\bar w)\bar w}}{{{H_{\rm m}}}} >$ ; and Ekman feedback: ${\mu _a}{\beta _w} < H(\bar w)\dfrac{{\partial \overline T }}{{\partial z}} >$ ) and two negative feedbacks (mean advection damping: $- ({a_1}\dfrac{{ < \Delta \overline u > }}{{{L_x}}} + {a_2}\dfrac{{ < \Delta \overline v > }}{{{L_y}}})$ ; and thermal damping: $- {\alpha _{\rm s}}$ ). Figure 9 compares each contributing processes of the BJ-index in the GODAS reanalysis data and the CAMS-CSM simulation. Clearly, the thermocline feedback is the dominant term among all three feedbacks. CAMS-CSM tends to produce a weaker thermocline feedback than observed (Fig. 9). This underestimation of thermocline feedback can be attributed to three physical processes. First, the wind stress response to ENSO SST forcing ( ${\mu _a}$ ) is too weak in the CAMS-CSM simulation. Figure 10 demonstrates the regression patterns of precipitation and surface wind onto Niño3.4 anomalies. It can be seen that the ENSO-induced atmospheric convection is fairly weak in the CAMS-CSM simulation and, as a consequence, the westerly wind response over the central to western Pacific is underestimated. Second, the thermocline response to surface wind stress is also weak in CAMS-CSM. As shown in Fig. 11a, the westerly wind stress tends to lift the thermocline in the western Pacific and deepen the thermocline in the eastern Pacific. However, the simulated thermocline changes are weaker than observed (Fig. 11b), which gives rise to parameter ${\beta _h}$ being smaller. Third, the modeled mean upwelling in the central to eastern Pacific (4.2 × 10–6 m s–1) is also slightly weaker than observed (4.6 × 10–6 m s–1). Above all, in the simulation by CAMS-CSM, the ENSO-SST-induced weak westerly wind stress can further induce weak subsurface temperature changes, which are then brought to the mixed layer by the weak mean upwelling. As a result, the modeled thermocline feedback is weaker than observed.

 Figure 9 The BJ-index (BJ) and its five contributing terms (MA: mean advective damping, ZA: zonal advective feedback, TH: thermocline feedback, EK: Ekman feedback, and TD: thermal damping) in observation (blue bars) and the CAMS-CSM simulation (red bars). The calculation of the BJ-index is given in Eqs. (1) and (2).
 Figure 10 Regression patterns of the anomalous precipitation (color shaded; mm day–1 °C–1) and surface wind (vectors; m s–1 °C–1) against Niño3.4 anomalies in (a) observation and (b) the CAMS-CSM simulation.
 Figure 11 Regression patterns of anomalous thermocline depth (m Pa–1) against the surface wind stress anomalies along the equator (5°S–5°N, 140°–80°W) in (a) observation and (b) the CAMS-CSM simulation.

 Figure 12 Regression of anomalous shortwave radiation (SW), longwave radiation (LW), sensible heat flux (SH), and latent heat flux (LH) against the SST anomaly in the Niño3.4 region in observation (blue bars) and the CAMS-CSM simulation (red bars).
 Figure 13 Regression patterns of anomalous shortwave radiation flux (W m–2 °C–1) against the Niño3.4 index in (a) observation and (b) the CAMS-CSM simulation.

As shown in Fig. 9, the gross effect, as indicated by the BJ-index, is negative in observation, which is consistent with the theory that ENSO is a damped oscillation triggered by stochastic forcing (e.g. Kleeman and Moore, 1997; Kirtman and Schopf, 1998). In the simulation by CAMS-CSM, the BJ-index is also negative but the absolute value is smaller. This implies that the modeled ENSO is a less damped mode, which explains the overestimation of ENSO’s amplitude (Fig. 6). Here, we show that the BJ-index is useful in diagnosing the physical origin of the bias in ENSO’s amplitude, especially when error compensation occurs. For CAMS-CSM, the weak negative shortwave radiation feedback overwhelms the weak thermocline feedback, giving rise to a stronger modeled ENSO.

4.2 Periodicity bias and its physical origin

Realistically simulating ENSO’s periodicity remains a challenging issue in current CGCMs (Bellenger et al., 2014; Lu and Ren, 2016). For CAMS-CSM, the modeled ENSO oscillates regularly with a period of 2.8 yr, which is shorter than observed. Here, the so-called Wyrtki index, proposed by Lu et al. (2018), is adopted to estimate the physical origin of this periodicity bias. Similar to the BJ-index, the Wyrtki index is also derived from the linear recharge–discharge framework:

 $\dfrac{{{\rm d} < \!T \! > }}{{{\rm d}t}} = R < \!T\! > + F [h],$ (3)
 $\!\!\!\!\!\!\!\! \dfrac{{{\rm d}[h]}}{{{\rm d}t}} = - \varepsilon [h] - B < \!T \! >,$ (4)

where T denotes SST; h represents thermocline depth; the sign “< >” denotes the average of quantities over the eastern Pacific (5°S–5°N, 170°–80°W); and the sign “[ ]” denotes the average of quantities over the entire equatorial Pacific domain (5°S–5°N, 140°E–80°W);R indicates the damping rate of eastern Pacific SST, and is often referred to as the Bjerknes instability index (Jin et al., 2006); $\varepsilon$ denotes the damping rate of thermocline depth due to the energy leak at the western boundary and mixing; F represents the impact of the discharged or recharged state of the ocean heat content on SST; and B indicates the efficiency of the discharging or recharging of the equatorial heat content driven by the equatorial wind stress curl anomalies induced by the anomalies of ENSO SST. The noise forcing and nonlinear terms are omitted here for brevity. This simple linear framework is capable of representing the linear low-frequency dynamics of ENSO’s growth rate (Jin et al., 2006) and frequency (Lu et al., 2018). The Wyrtki index can be formulated as:

 ${I_{\rm Wyrtki}} = \dfrac{{4\pi }}{{\sqrt {4B \cdot F - {{(R + \varepsilon)}^2}} }} \approx \dfrac{{2\pi }}{{\sqrt {B \cdot F} }}.$ (5)

The simulated parameter B is 3.9 × 10–7 m (s K)–1, which is quite close to the observed value of 4.3 × 10–7 m (s K)–1. However, the modeled parameter F is 1.3 × 10–8 K (s m)–1, which is much larger than the observed value of 8.3 × 10–9 K (s m)–1. As a result, the modeled Wyrtki index is 2.8 yr, which is slightly shorter than the observed Wyrtki index of 3.4 yr. In other words, the phase transition of ENSO induced by the thermocline and zonal advective feedbacks is quicker than observed, which contributes to the shorter period of ENSO in CAMS-CSM.

4.3 Triggers for ENSO in CAMS-CSM

In addition to the internal dynamics of ENSO as discussed above, assessing the external forcings for ENSO is also very important, given the fact that ENSO is a damped oscillation mode (Fig. 14) and can be triggered by several precursors, such as North Pacific Oscillation (NPO; Vimont et al., 2001; Alexander et al., 2010), the Indian Ocean Dipole (IOD; Izumo et al., 2010), Arctic Oscillation in spring (Chen S. F. et al., 2014), an anomalous EAWM (Li, 1990, 1996), thermocline changes in the western Pacific (Chen et al., 2016a), and the consecutive occurrence of westerly wind bursts (Chen et al., 2017). We begin by evaluating the precursor of NPO during the previous winter.

 Figure 14 Seasonal composites of negative-minus-positive SLP index cases for SST (color shaded; °C) and 10-m winds (arrows; m s–1) in observation (left-hand panels) and the CAMS-CSM simulation (right-hand panels).

As shown in Fig. 14a, the sea level pressure (SLP) changes (negative minus positive) in North Pacific might reduce the trade winds, which in turn may induce footprint-like SST anomalies in the subtropics via latent heat flux changes. This footprint signal persists throughout the spring and potentially affects the tropical Pacific throughout the following year (Figs. 14c, e, f; Vimont et al., 2001). This so-called “seasonal footprinting mechanism” is reproduced well by CAMS-CSM, in terms of the seasonal evolution of SST and surface wind (Fig. 14). In fact, the SLP anomalies in North Pacific (10°–25°N, 175°–145°W) during the previous winter are significantly correlated with ENSO (Fig. 15a). CAMS-CSM tends to simulate a similar NPO–ENSO relationship as observed (Fig. 15b), implying a realistic seasonal footprinting mechanism in this model.

 Figure 15 Scatterplots between the normalized Niño3.4 index averaged during November (0) to January (1) and the normalized SLP index averaged during November (–1) to March (0) from (a) observation and (b) the CAMS-CSM simulation. The correlation coefficients are given in parentheses on top of each panel.

Izumo et al. (2010) pointed out that the IOD is another precursor of El Niño, at a lead time of 14 months. Figure 16 compares the lead–lag correlations between the IOD [averaged from September to October of Year (0)] and the Niño3.4 index in observation [1980–2016, following Izumo et al. (2010)] and in the CAMS-CSM simulation, separately. A significant positive correlation is evident in Year (0). The El Niño condition is often accompanied by a positive IOD phase, with a strengthening of the easterly winds off Sumatra in summer, which may in turn aid the development of El Niño (Annamalai et al., 2005; Luo et al., 2010). This covariability of ENSO and the IOD is simulated well by the CAMS-CSM, with a positive correlation (0.6) in Year (0) (Fig. 16). In addition, a significant negative correlation is clear one year after the IOD peak. In other words, a positive (negative) IOD phase leads the ENSO peak by around 14 months. This relationship has been applied in an ENSO prediction model with reliable skill (Izumo et al., 2010). As shown in Fig. 16b, such a relation of the prior IOD signal with ENSO is also captured by CAMS-CSM.

 Figure 16 Lag-correlation between the IOD [averaged from September to November of Year (0)] and the Niño3.4 index (three-month running average applied) in (a) observation and (b) the CAMS-CSM simulation. The dashed lines indicate the 95% confidence level. The vertical dashed lines denote the September–October–November in Year (0) for a co-occurring IOD.
5 Impact of ENSO on East Asian climate

ENSO teleconnections are of great interest from the perspective of East Asian climate. The EASM (Chen et al., 1992; Zhang et al., 1996; Wang et al., 2000; Li et al., 2017) and EAWM (Li, 1990) are both influenced by ENSO. Realistically simulating ENSO teleconnections has the potential to enable skillful seasonal predictions in East Asia (Li et al., 2016; Lu et al., 2017; Liu et al., 2018). Figure 17 demonstrates the anomalous patterns of precipitation, 850-hPa horizontal wind, and SST during the boreal summer following strong El Niño events. The Indo-western Pacific capacitor effect is evident in observation (Xie et al., 2016). The SST warming and easterly wind anomalies are pronounced over the Indo-western Pacific, which stimulates an anomalous anticyclone (AAC) over the northwestern Pacific. As a result of the increased water vapor transport, summer rainfall is enhanced over the Yangtze River valley. As shown in Fig. 17d, the SST warming over the Indo-western Pacific is simulated by CAMS-CSM during post-El Niño summer. However, the AAC exhibits an eastward displacement compared with observation (Fig. 17b) and, as a consequence, the rainfall band shows a southeastward shift following the position of the AAC. In other words, the observed flooding in the Yangtze River valley in post-El Niño summer is poorly presented by CAMS-CSM.

 Figure 17 Composite patterns of the (a, b) anomalous precipitation (color shaded; mm day–1) overlapped with horizontal wind at 850 hPa (vectors; m s–1) and (c) SST (°C) during boreal summer in the decaying year of strong El Niño events from (a, c) observation (1982/83, 1997/98, and 2015/16) and (b, d) the CAMS-CSM simulation. A modeled strong El Niño event is defined as occurring when the value of the November–December–January mean Niño3.4 index exceeds 1.5 times its standard deviation.

The cause of the eastward-displaced AAC in the model is key for understanding the unrealistic ENSO–EASM relationship in CAMS-CSM. In observation, the convection activity is weakened from the tropical northwestern Pacific to the South China Sea (SCS), which is associated with the location of the observed AAC. However, the simulated convection is enhanced over the SCS, which induces the eastward displacement of the modeled AAC. Note that the underlying SST is warming in both observation and the CAMS-CSM simulation. Thus, the observed precipitation over the SCS is uncoupled with local SST warming. In fact, the correlation between the observed SST and local precipitation over the SCS is –0.47 during boreal summer (Fig. 18a), while the correlation between SCS SST and surface shortwave radiation is 0.37 (Fig. 18c). This implies that SCS SST is passively modulated by the atmospheric heat flux, which is consistent with previous studies (e.g., Wang et al., 2006). However, in CAMS-CSM, the air–sea interaction is unrealistically strong over the SCS. As shown in Fig. 18b, the modeled precipitation over the SCS is positively correlated with local SST (Fig. 18b), which explains the simulated biases in SCS precipitation and the shift in the location of the AAC in post-El Niño summer.

 Figure 18 Scatterplots of (a, b) local precipitation against the local SST anomaly and (c, d) the local SST anomaly against local shortwave radiation, over the South China Sea (8°–22°N, 110°–120°E), during boreal summer in (a, c) observation and (b, d) the CAMS-CSM simulation. Linear fitting lines are indicated by the straight lines. Correlation coefficients are given in parentheses on top of each panel.

Although the EAWM is mainly influenced by mid–high-latitude climatic modes (Gong et al., 2001; Wu et al., 2006), ENSO can also modulate the strength of the EAWM (Li, 1990; Zhang et al., 1996). Recent studies suggest that ENSO can influence the tropical component of the EAWM, while its extratropical component is affected by midlatitude factors (Liu et al., 2012; Chen Z. et al., 2014). A combined effect of Arctic Oscillation and ENSO on the EAWM has also been discovered (Chen W. et al., 2013). Figure 19 demonstrates the running correlation of the Niño3.4 index with two EAWM indices, designed by Li and Yang (2010) and Shi (1996), respectively. The former EAWM index is defined by the zonal wind at 200hPa, while the latter is defined by the SLP. Negative correlation is evident in both observation and in the CAMS-CSM simulation, which exhibits pronounced interdecadal variation. Extensive studies have shown that this interdecadal modulation is caused by Pacific Decadal Oscillation (Wang et al., 2008; Kim J. W. et al., 2014; Wang and Lu, 2016), Atlantic Multidecadal Oscillation (Li and Bates, 2007; Jiang et al., 2014), and their combination (He and Wang, 2013). As shown in Fig. 19a, the ENSO-induced change in the EAWM in the upper troposphere is pronounced after 1995, which is overestimated by CAMS-CSM, with significant correlation during a large portion of the simulation period. For the low-level EAWM (Fig. 19b), its correlation with ENSO exhibits interdecadal changes, which is captured well by CAMS-CSM. Figure 20 shows the differences in winter surface air temperature and horizontal wind at 850 hPa between El Niño and La Niña years. Observationally, a southwesterly wind anomaly is evident over southeastern China, which is consistent with previous studies (Chen et al., 2000). This southwesterly wind anomaly brings warm air from the SCS and prevents the southward movement of cold air from the mid to high latitudes. As a result, the surface air temperature over China tends to be warmer in El Niño years than La Niña years. In the CAMS-CSM simulation, the southwesterly wind anomaly is stronger than observed, which is consistent with the overestimation of the ENSO–EAWM relationship (Fig. 19). However, this modeled southwesterly wind anomaly is located along coastal regions of China, implying an eastward shift of the modeled AAC over the northwestern Pacific. Thus, an eastward-located western North Pacific anticyclone is a common bias in both summer (Fig. 17) and winter (Fig. 20). Consequently, the simulated warming is remarkable only in southern and eastern China, with an underestimation of observed warming in central China. It should be noted that the EAWM can also affect ENSO (Li, 1990, 1996) and its teleconnections in the extratropics (Ma et al., 2018), which needs to be evaluated for CAMS-CSM in the future.

 Figure 19 The 11-yr moving correlations between wintertime Niño3.4 anomalies and the EAWM indices designed by (a) Li and Yang (2010) and (b) Shi (1996) in observation (black curves) and the CAMS-CSM simulation (red curves). The horizontal dashed line denotes the 95% confidence level.
 Figure 20 Differences in wintertime anomalous surface air temperature (color shaded; °C) and horizontal wind at 850 hPa (vectors; m s–1) between El Niño years and La Niña years in (a) observation and (b) the CAMS-CSM simulation.
6 Summary and discussion

This study evaluates the performance of ENSO in the historical simulation of CAMS-CSM. The mean state of the tropical Pacific, ENSO’s properties and dynamics, as well as its impact on East Asian climate, are assessed in detail. Furthermore, the physical origins of model biases are discussed. The main conclusions can be summarized as follows:

(1) In general, the mean state of the tropical Pacific is simulated well by CAMS-CSM, including the prevailing easterly trade winds, CT and WP. However, the well-known double ITCZ problem and CT bias still exist, which are common errors in most CGCMs (Li and Xie, 2014). CAMS-CSM captures the two types of El Niño well, with realistic spatial patterns and independence. The annual cycle and phase-locking features of ENSO are also reproduced; however, ENSO’s amplitude is overestimated in each calendar month. Observationally, ENSO is skewed to a positive phase. However, El Niño and La Niña events tend to have similar amplitude in CAMS-CSM. The observed ENSO spectrum is broad (2–7 yr), while the modeled ENSO spectrum exhibits a sharp peak at around 2.8 yr.

(2) The dynamics of ENSO in CAMS-CSM are examined in terms of the BJ-index and Wyrtki index. It is found that biases of positive thermocline feedback and negative heat flux feedback are dominant in CAMS-CSM. The thermocline feedback is underestimated by CAMS-CSM, which is contributed by an underestimation of ENSO-induced westerly wind and the thermocline changes in response to surface wind stress. This weakened thermocline feedback is overwhelmed by the biases of heat flux damping. The modeled ENSO-induced shortwave reduction is one third of that observed. As a result, the modeled ENSO is more easily stimulated, as indicated by the gross BJ-index. This explains the overestimation of ENSO’s amplitude in CAMS-CSM. In addition, the simulated periodicity bias is estimated by the Wyrtki index. It is found that the phase transition induced by the thermocline and zonal advective feedbacks is too quick, which explains the relatively shorter period in CAMS-CSM. The triggers of ENSO, such as the seasonal footprinting mechanism and IOD, are captured well by CAM-CSM.

(3) The relationship between ENSO and the East Asian monsoon is further assessed. During post-El Niño summer, the northwestern Pacific AAC is reproduced by CAMS-CSM. However, the modeled AAC exhibits eastward displacement, such that water vapor cannot be transported to the Yangtze River valley and the observed flooding in this region in post-El Niño summer is poorly simulated. It is further demonstrated that the eastward displacement of the modeled AAC is caused by an unrealistic enhancement of convection over the SCS. Observationally, the SCS SST is a passive response to surface heat flux, and is negatively correlated with local precipitation. However, the modeled air–sea coupling is too strong over the SCS, which leads to an unrealistic positive correlation between precipitation and local SST. For the EAWM, CAMS-CSM reproduces the observed negative correlation between wintertime lower-tropospheric circulation and ENSO. However, the response of surface air temperature in the central China is weaker than observed.

This study shows that a stronger thermocline feedback and stronger heat flux damping is needed to give a better representation of ENSO dynamics. Improving the convective scheme could be a way to develop the model in the future. For example, modifying the cumulus entrainment may provide a more realistic response of convection to ENSO (Watanabe et al., 2011), which may in turn give a better representation of ENSO’s properties (Lu et al., 2017). In addition, the triggering conditions of convection over the SCS need to be improved, such that a better relationship between ENSO and the EASM can be obtained.

Acknowledgments. We would like to thank the editors and reviewers for their valuable comments.

References