The Asymmetric Effects of El Niño and La Niña on the East Asian Winter Monsoon and Their Simulation by CMIP5 Atmospheric Models
  J. Meteor. Res.  2017, Vol. 31 Issue (1): 82-93   PDF    
http://dx.doi.org/10.1007/s13351-017-6095-5
The Chinese Meteorological Society
0

Article Information

GUO Zhun, ZHOU Tianjun, WU Bo . 2017.
The Asymmetric Effects of El Niño and La Niña on the East Asian Winter Monsoon and Their Simulation by CMIP5 Atmospheric Models. 2017.
J. Meteor. Res., 31(1): 82-93
http://dx.doi.org/10.1007/s13351-017-6095-5

Article History

Received June 13, 2016
in final form September 18, 2016
The Asymmetric Effects of El Niño and La Niña on the East Asian Winter Monsoon and Their Simulation by CMIP5 Atmospheric Models
Zhun GUO1,2, Tianjun ZHOU1, Bo WU1     
1. State Key Laboratory of Numerical Modeling for Atmospheric Sciences and Geophysical Fluid Dynamics,Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing 100029;
2. Climate Change Research Center,Chinese Academy of Sciences,Beijing 100029
ABSTRACT: El Niño–Southern Oscillation (ENSO) events significantly affect the year-by-year variations of the East Asian winter monsoon (EAWM). However, the effect of La Niña events on the EAWM is not a mirror image of that of El Niño events. Although the EAWM becomes generally weaker during El Niño events and stronger during La Niña winters, the enhanced precipitation over the southeastern China and warmer surface air temperature along the East Asian coastline during El Niño years are more significant. These asymmetric effects are caused by the asymmetric longitudinal positions of the western North Pacific (WNP) anticyclone during El Niño events and the WNP cyclone during La Niña events; specifically, the center of the WNP cyclone during La Niña events is westward-shifted relative to its El Niño counterpart. This central-position shift results from the longitudinal shift of remote El Niño and La Niña anomalous heating, and asymmetry in the amplitude of local sea surface temperature anomalies over the WNP. However, such asymmetric effects of ENSO on the EAWM are barely reproduced by the atmospheric models of Phase 5 of the Coupled Model Intercomparison Project (CMIP5), although the spatial patterns of anomalous circulations are reasonably reproduced. The major limitation of the CMIP5 models is an overestimation of the anomalous WNP anticyclone/cyclone, which leads to stronger EAWM rainfall responses. The overestimated latent heat flux anomalies near the South China Sea and the northern WNP might be a key factor behind the overestimated anomalous circulations.
Key words: AMIP (Atmospheric Model Intercomparison Project)     East Asian winter monsoon     ENSO     asymmetric effects     cloud     precipitation    
1 Introduction

The East Asian winter monsoon (EAWM) is an essential member of the global monsoon system. The EAWM can exert a large influence on the economy and society of eastern China, Korea, Japan, and surrounding regions, by affecting precipitation and temperature variations (Chang et al., 2006). Predictions of the interannual variability of the EAWM rely heavily on the El Niño–Southern Oscillation (ENSO) phenomenon (Webster et al., 1998;Wang et al., 2008). Boreal winter is the mature phase of El Niño events. At this time, as a response to the equatorial eastern Pacific warming, a weak EAWM thus establishes under the effects of the anomalous anticyclone over the western North Pacific (WNP) (Zhang et al., 1996;Wang et al., 2000). Therefore, following a weaker EAWM during El Niño winters, southern China, eastern central China, and southern Japan are warmer and wetter, while northeastern China becomes drier (Li, 1990;Zhang et al., 1999;Chen et al., 2000;Wu et al., 2003). In La Niña years, the EAWM becomes stronger and the associated climate anomalies over East Asia are traditionally regarded as a mirror image of El Niño years (Tomita and Yasunari, 1996;Zhang et al., 1996;Ji et al., 1997;Wang et al., 2000).

In fact, many previous studies have demonstrated that El Niño is asymmetric to La Niña (e.g.,Burgers and Stephenson, 1999;Jin et al., 2003;An and Jin, 2004;An et al., 2005;Zhang et al., 2015). Moreover, sensitivity experiments confirm that such asymmetry is highly related to nonlinear responses in the atmosphere to the underlying sea surface temperature (SST) anomalies (SSTAs) (Hoerling et al., 1997;Kang and Kug, 2002). The stronger convection over the equatorial central Pacific during La Niña appears to the west of the suppressed convection during El Niño (Hoerling et al., 1997). Thus, the cooling of the equatorial Pacific associated with La Niña events may not lead to an opposite effect to the equatorial Pacific warming associated with El Niño events. This asymmetric feature of tropical atmospheric circulation has been confirmed by both data diagnosis and numerical modeling (Wu et al., 2010).

The nonlinear responses in the atmosphere to the underlying SSTAs are highly related to the physical schemes of atmospheric models, even when they are forced by identical historical SST. It is therefore expected that the atmospheric responses of the EAWM will also vary greatly among different models. However, whether or not models are similar in their simulation of the asymmetric effects of ENSO on EAWM interannual variability needs to be investigated. Therefore, it is necessary to understand these similarities. The Atmospheric Model Intercomparison Project (AMIP) experiments from Phase 5 of the Coupled Model Intercomparison Project (CMIP5) provide multiple samples and simplify these questions in atmospheric models that the underlying SST anomalies are fixed. Thus, our objectives are to: (1) evaluate the performance of CMIP5 models in simulating the asymmetric effects of ENSO on the EAWM, and (2) identify the possible sources of any model biases uncovered.

Following this introduction, Section 2 describes the models, datasets, and analysis methods. The asymmetric effects in both observations and models are discussed in Section 3. Section 4 provides a summary of the study's key findings.

2 Data and methods

The datasets used in the present study are as follows.

(1) The wind fields from the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis data, for the period 1948–2010 (Kalnay et al., 1996).

(2) The precipitation and total cloud fraction from 560 meteorological stations, for the period 1950–2005, provided by the China Meteorological Administration.

(3) Precipitation data from the Global Precipitation Climatology Project (GPCP) and the Climate Prediction Center Merged Analysis of Precipitation (CMAP;Xie and Arkin, 1997;Adler et al., 2003).

(4) The AMIP experiments from 27 CMIP5 models. The details of the CMIP5 models are listed in Table 1.

Table 1 Details of the 27 CMIP5 models used in this study
Model Institute/Country Atmospheric resolution (lat. × lon., level)
ACCESS1.0 CSIRO–BoM/Australia 145 × 192, L38
BCC_CSM1.1 BCC/China 64 × 128, L26
BCC_CSM1.1(m) BCC/China 160 × 320, L26
BNU-ESM Beijing Normal University/China 64 × 128, L26
CanESM2 CCCma/Canada 64 × 128, L35
CCSM4 NCAR/USA 192 × 288, L27
CESM1(CAM5) NSF–DOE–NCAR/USA 192 × 288, L27
CMCC-CM Canadian Centre for Climate Modelling and Analysis/Italy 240 × 480, L27
CNRM-CM5 Centre National de Recherches Météorologiques–Centre Européen de Recherche et de Formation Avancée en Calcul Scientifique/France 128 × 256, L31
CSIRO Mk3.6.0 CSIRO–Queensland Climate Change Centre of Excellence/Australia 96 × 192, L18
EC-EARTH Irish Center of High-End Computing/Belgium 160 × 320, L16
FGOALS-g2 IAP–Tsinghua University/China 60 × 128, L26
FGOALS-s2 IAP-LASG/China 64 × 128, L26
GFDL-HIRAM-C180 NOAA-GFDL/USA 90 × 144, L24
GFDL-HIRAM-C360 NOAA-GFDL/USA 90 × 144, L24
GISS-E2-R NASA-GISS/USA 89 × 144, L40
HadGEM2-A Korean Meteorological Administration-National Institute of Meteorological Research/South Korea 144 × 192, L38
INM-CM4.0 Institute of Numerical Mathematics/Russia 120 × 180, L21
IPSL-CM5A-LR IPSL/France 96 × 96, L39
IPSL-CM5B-LR IPSL/France 96 × 96, L39
IPSL-CM5B-LR IPSL/France 96 × 96, L39
MIROC5 Atmosphere and Ocean Research Institute (AORI), NIES, Japan Agency for Marine–Earth Science and Technology (JAMSTEC)/Japan 128 × 256, L40
MPI-ESM-LR MPI-M/Germany 96 × 192, L47
MPI-ESM-MR MPI-M/Germany 96 × 192, L96
MRI-AGCM3.2H Meteorological Research Institute (MRI)/Japan 160 × 320, L48
MRI-AGCM3.2S MRI/Japan 160 × 320, L48
NorESM1-M Norwegian Climate Centre/Norway 96 × 144, L26

Eight-year filtered data are employed through a Lanczos filter (Duchon, 1979), to focus on the interannual variability. The asymmetric component of El Niño and La Niña events is defined as their sum (Hoerling et al., 1997). A composite analysis method is applied. We choose six El Niño events and six La Niña events (Table 2) in the period 1979–2005 to evaluate the models, based on the threshold of one standard deviation of the winter mean (December–February; DJF) Niño3.4 index. The Niño3.4 index is defined as the averaged SSTA over the central eastern Pacific (5°N–5°S, 120°–170°W), and the data source is the Climate Prediction Center, NOAA.

Table 2 El Niño and La Niña events during the period 1979–2005
Event Year
El Niño 1982, 1986, 1991, 1994, 1997, 2002
La Niña 1984, 1988, 1995, 1998, 1999, 2000

As in Song and Zhou (2014), we also quantify the model performance in the simulation of interannual precipitation patterns over southeastern China (20°–35°N, 100°–130°E), by applying a skill score (S):

$S = \frac{{{{\left( {1 + R} \right)}^2}}}{{{{\left( {{\rm {SDR}} + \dfrac{1}{\rm{SDR}}} \right)}^2}}}, $

where R and SDR are the pattern correlation and the ratio of spatial standard deviation of the model against the observation, respectively.

3 Results

The interannual variability of the EAWM is significantly affected by ENSO activities. Composite DJF mean precipitation and cloudiness anomalies for both El Niño and La Niña events are shown in Fig. 1, as well as their asymmetric components. The asymmetric effects of El Niño and La Niña events are confirmed. Southeastern China has more precipitation in the winters of El Niño years, with a central value of 9.0 mm day–1, which is about 50% of the climatology. The corresponding change of cloudiness exhibits an increase of 8%, which is about 12% of the climatology. On the contrary, the precipitation anomalies for La Niña years are not evident (Fig. 1b), and thus the responses of precipitation to El Niño and La Niña events are highly asymmetric (Fig. 1c). The cloudiness in southeastern China is less than normal in La Niña years, but the amplitude is far weaker than that of El Niño years (Figs. 1d and 1e), with –2% versus 8%, thus also showing a highly asymmetric response (Fig. 1f).

Moreover, the correlation coefficient between area-averaged precipitation/cloudiness anomalies and the Niño3.4 index is 0.84/0.71 for years with a positive Niño3.4 index, which is statistically significant at the 1%/5% level. However, corresponding statistics for years with a negative Niño3.4 index are not evident. The high (low) correlation confirms the significant (insignificant) effects of El Niño (La Niña) events on winter precipitation (cloudiness) over southeastern China. All of this evidence shows that the effect of La Niña on the EAWM is not simply a mirror image of El Niño years.

Fig. 1 Composite December–February (a–c) mean precipitation (pr) anomalies (mm day–1) and (d–f) total cloud (cld; %) for (a, d) eight El Niño (El) events, (b, e) seven La Niña (La) events, and (c, f) the asymmetric components between El Niño and La Niña events. The data are derived from observations at 560 meteorological stations in China for the period 1958–2001.
Fig. 2 Composite December–February mean 850-hPa wind (vectors; m s–1) and surface air temperature (color-shaded; K) anomalies for (a) nine El Niño (El) events and (b) nine La Niña (La) events. (c) Asymmetric component estimated by the sum of (a) and (b). The data from NCEP are from 1958 to 2005.

Previous diagnostic analysis has revealed that the anomalous lower-tropospheric anticyclone/cyclone in the tropical WNP connects the warm/cold events in the eastern Pacific with a weak/strong EAWM (Wang et al., 2000). To examine the lower-tropospheric circulation changes,Fig. 2 compares the composite DJF mean 850-hPa wind and surface air temperature anomalies. To extend the local precipitation changes shown in Fig. 1 to a broader domain and present a larger picture, the surface air temperature anomalies derived from the NCEP–NCAR data are also shown in Fig. 2, since the change in surface air temperature is an essential indicator of the southern component of the EAWM (Wang et al., 2000). In the wintertime of El Niño events, East Asia has a strong anomalous anticyclone in the WNP (hereafter, WNPAC), and a weak anomalous cyclone in southeastern China. Anomalous southwesterly flow prevails along the East Asian coastline. Deficient precipitation is thus generated over the western Pacific with the dominance of the strong anticyclone, while excessive precipitation is seen in southeastern China due to the control of the anomalous cyclone. In El Niño winters, continental East Asia experiences a warmer climate associated with the dominance of the WNPAC (Fig. 2a), and thereby a weakening cold and dry monsoonal circulation.

The pattern of La Niña events is not simply a mirror image of El Niño events, as shown in Fig. 2b; the central position of the WNP cyclone (WNPC) is located in the South China Sea along 110°–120°E, which shifts westward and becomes much weaker relative to the WNPAC (120°–130°E) during El Niño events. The spatial phase shift causes the anomalous northeasterly wind to be weak in southeastern China, and therefore cannot effectively influence the moisture transport and associated changes of precipitation over that region (Fig. 2b). In La Niña winters, however, the enhanced winter monsoon associated with the WNPC in the South China Sea results in a colder climate, especially west of 110°E, along the Indochina Peninsula. Due to the westward shift of the WNPC relative to the WNPAC, the cooling signal is most robust around (20°N, 105°E). The asymmetric responses of surface circulations are more evident in the sums of El Niño and La Niña (Fig. 2c). The amplitudes of precipitation anomalies extending from the South China Sea to the WNP along 10°–14°N in La Niña years are stronger than those of El Niño years. The asymmetric response of surface air temperature is most robust over continental East Asia, with an amplitude of near 1.0°C (Fig. 2c). The resemblance of Fig. 2c to Fig. 2a again confirms that the effects related to El Niño events on the East Asian climate are stronger than those of La Niña events.

The above analyses demonstrate that El Niño and La Niña events have robust asymmetric effects on the EAWM. Although El Niño years generally lead to a weaker winter monsoon and thereby a warmer climate, while La Niña years witness a stronger winter monsoon and thereby a colder climate, due to the westward shift of the WNPC relative to the WNPAC, the effect of El Niño on precipitation (temperature) over southeastern China (the East Asian coastline) is more significant. In addition, the intensity of the WNPC is much weaker than that of the WNPAC, such that the WNPC cannot influence the EAWM as efficiently as the WNPAC does in El Niño years. Therefore, the asymmetry of the WNPAC and WNPC plays crucial roles in dominating the asymmetric climate anomalies.

Fig. 3 Composite December–February mean precipitation anomalies (mm day–1) for selected El Niño events. (a) Station data, (b) CMAP, (c) GPCP, (d) multi-model ensemble mean (MME), and (e) MME of high-skill models (BE5). The dotted areas indicate that the precipitation anomalies are statistically significant at the 1% level derived by Student's t-test.
Fig. 4 As in Fig. 3, but for La Niña events.

The surface cooling/warming and the subsidence over the WNP lead to suppressed/enhanced convective heating that further induces a Rossby-wave response and thereby the WNPAC/WNPC. Many factors contribute to the asymmetric responses of the WNPAC and WNPC. First, anomalous heating shows a significant spatial asymmetry between El Niño and La Niña events (Wu et al., 2010). In La Niña events, the negative precipitation anomalous belt appears in the west of its positive counterpart in El Niño events. Therefore, the WNPC during La Niña is located west of the WNPAC during El Niño. Second, the SSTAs over the WNP also show an asymmetry in their amplitudes; specifically, the cold SSTA during El Niño is stronger than the warm SSTA during La Niña. These mechanisms have been demonstrated by numerical model experiments (Wu et al., 2010).

To investigate the simulation of the interannual EAWM patterns,Figs. 3 and 4 show composite DJF-averaged anomalous precipitation for El Niño and La Niña derived from observations and the multi-model ensemble mean (MME) of all 27 atmospheric general circulation models from CMIP5, respectively. To better evaluate the performance of the models, precipitation anomalies derived from the GPCP and CMAP data are also compared. The patterns from CMAP and GPCP show almost the same results as the station data, with pattern correlation coefficients of 0.98 and 0.96 for El Niño events. Among the CMIP5 models, 16 out of 27 show similar positive EAWM precipitation responses to the El Niño events as observed. However, the simulation skills in most models are less than 0.6 (Table 3), partly because of the westward shift of the simulated precipitation centers (figure omitted). The MME of the CMIP5 models is able to reproduce the positive precipitation anomalies over southeastern China, with a skill score of 0.42, but its magnitude is smaller than observed.

Table 3 Skill scores for precipitation in the CMIP5 models for El Niño, La Niña, and their asymmetric components. The skill is relative to the station data over the region (20°–30°N, 105°–125°E)
Model Skill score of El Niño events Skill score of La Niña events Asymmetric components
ACCESS1.0 0.73884 0.28505 0.21061
BCC_CSM1.1 0.35859 0.64383 0.23087
BCC_CSM1.1(m) 0.25639 0.40646 0.10421
BNU-ESM 0.41938 0.36966 0.15503
CanESM2 0.49987 0.18572 0.09283
CCSM4 0.23585 0.28705 0.06770
CESM1(CAM5) 0.54187 0.36636 0.19852
CMCC-CM 0.34965 0.30505 0.10666
CNRM-CM5 0.51815 0.17773 0.09209
CSIRO Mk3.6.0 0.32501 0.18913 0.06147
EC-EARTH 0.50936 0.22347 0.11382
FGOALS-g2 0.54640 0.23661 0.12928
FGOALS-s2 0.29582 0.13798 0.04082
GFDL-HIRAM-C180 0.31212 0.24428 0.07624
GFDL-HIRAM-C360 0.57343 0.40800 0.23396
GISS-E2-R 0.54718 0.28728 0.15719
HadGEM2-A 0.41755 0.36083 0.15066
INM-CM4.0 0.55964 0.24957 0.13967
IPSL-CM5A-LR 0.13558 0.23946 0.03247
IPSL-CM5B-LR 0.30372 0.50449 0.15322
IPSL-CM5B-LR 0.19347 0.24458 0.04732
MIROC5 0.42487 0.50769 0.21570
MPI-ESM-LR 0.62299 0.45562 0.28384
MPI-ESM-MR 0.46322 0.22243 0.10303
MRI-AGCM3.2H 0.54276 0.32700 0.17748
MRI-AGCM3.2S 0.65373 0.25368 0.16584
NorESM1-M 0.06700 0.15094 0.01011

The MME and most of the CMIP5 models fail to reproduce the asymmetric responses of precipitation to El Niño and La Niña events, as they reproduce incorrect negative precipitation anomalies when La Niña events appear (Fig. 4). Accordingly, the skill scores of precipitation patterns are also poor (Table 3). Similar biases are also evident in the responses of cloud fraction, in that the cloud anomalies are also highly symmetric (Fig. 5).

Fig. 5 Composite December–February cloud faction anomalies for the asymmetric component. (a) Station data, (b) International Satellite Cloud Climatology Project (ISCCP), (c) multi-model ensemble mean (MME), and (e) high-skill models (BE5, see Table 3). The dotted areas indicate that the precipitation anomalies are statistically significant at the 1% level by Student's t-test.

Among the CMIP5 models, ACCESS1.0 performs best in terms of the precipitation pattern during El Niño events, with the highest skill score for precipitation of 0.74, followed by MRI-AGCM3.2S (0.65), MPI-ESM-LR (0.62), GFDL-HIRAM-C360 (0.58), and INM-CM4.0 (0.56). To further investigate the model performance, these five models, based on their precipitation skill score during El Niño events, are selected to represent the high-skill models ensemble (hereafter, BE5). The positive precipitation anomalies over southeastern China are better captured in BE5, with a skill score of 0.63. However, BE5 still overestimates the negative anomalies in precipitation and cloud fraction in La Niña winters (Fig. 4).

In the observation, the zonally asymmetric positions of the WNPAC and WNPC are regarded as one of the most influential factors in dominating the asymmetric EAWM precipitation anomalies.Figure 6 compares the composite DJF mean 850-hPa wind anomalies and surface air temperature for El Niño, La Niña, and their asymmetric components. Generally, the CMIP5 models perform reasonably in simulating the spatial patterns and locations of the WNPAC and WNPC, as evidenced by the MME skill scores of 0.64 in El Niño events and 0.88 in La Niña events. In addition, the asymmetric component of the lower-atmospheric circulation is also reasonably reproduced (Figs. 6gi). However, the CMIP5 models overestimate the magnitudes of the WNPAC/WNPC, and thereby the anomalies of precipitation, especially in those models that perform well with respect to El Niño events. For example, in La Niña year winters, the simulated WNPC is stronger than observed. Accordingly, the anomalous northeasterly wind is much stronger than observed over southeastern China, and effectively influences the moisture transport and results in deficient local precipitation. Similar biases are also evident in El Niño events. The magnitudes of 850-hPa winds averaged over the northwestern Pacific (NWP; 0°–30°N, 100°–150°E), which are defined as $\sqrt {{u^2} + {v^2}} $ , are significantly proportional to the absolute value of the precipitation anomalies averaged over southeastern China (Fig. 7a), affirming the impact of the lower-atmospheric circulation on the simulation of EAWM interannual variability.

Fig. 6 Composite December–February mean surface air temperature anomalies (K) and 850-hPa winds (m s–1) for (a–c) El Niño events, (d–f) La Niña events, and (g–i) the asymmetric component of ENSO. The dotted areas indicate that the precipitation anomalies are statistically significant at the 1% level.
Fig. 7 Scatter plot of (a) the absolute value (ABS) of anomalous precipitation averaged over southeastern China (20°–30°N, 105°–125°E), and (b) the absolute value of anomalous latent heat flux averaged over the NWP (0°–30°N, 100°–150°E), versus the magnitudes of 850-hPa anomalous winds averaged over the NWP. Red dots denote El Niño events and blue dots denote La Niña events.

Since the CMIP5 models are forced by historical SST, surface air temperature anomalies do not show significant disagreements between simulations and observations over ocean. Over land, however, the anomalies of surface air temperature anomalies, which are the responses to anomalous circulation, show large inconsistencies with observations. Owing to the overestimation of the WNPAC in El Niño winters, the MME and BE5 tend to overestimate the positive anomalies of surface temperature over the Indochina Peninsula. This is also true for La Niña events, which has a strong cold bias (Fig. 6).

As the spatial patterns of the WNPAC and WNPC are simulated well by the CMIP5 models, the overestimations of circulation and precipitation may be related to the lack of air-sea coupling that suppresses the local negative feedback between surface wind/latent heating flux and the underlying SST, as the SST cannot respond to the surface heating fluxes (Wu and Kirtman, 2005). When SST is specified, such biases in latent heating flux and circulation are further amplified by the positive wind-evaporation feedback over the WNP, which is essential for maintaining the local SSTA and anomalous lower-tropospheric circulation (Wang et al., 2000).Figure 8 compares composite DJF mean surface latent heat flux anomalies for El Niño and La Niña derived from OAFlux (objectively analyzed air–sea fluxes) data and the models. In El Niño winters, a negative latent heating anomaly extends from the East/South China Sea and the northern Philippine Sea to the northern WNP, while the WNP witnesses a positive latent heating anomaly. The La Niña winter shows almost opposite spatial patterns, although the positive anomalous latent heat flux center is located farther to the northwest of the composite negative latent heat flux anomaly center during El Niño (Fig. 8d). These dipole patterns in anomalous latent heat flux can be reproduced well by the CMIP5, MME, and BE5. However, the simulated latent anomalies are twice as those observed in both El Niño and La Niña winters, especially in the South China Sea and northern Philippine Sea. Thus, the WNPAC in El Niño events and the WNPC in La Niña events are overestimated.

Fig. 8 As in Fig. 6, but for latent heat fluxes (W m–2).

To show the positive feedback more clearly, the magnitudes of 850-hPa anomalous winds and absolute values of latent heat flux averaged over the NWP are shown in Fig. 7b. For both El Niño and La Niña events, there is a robust positive linear relationship between the magnitudes of lower-atmospheric circulation and latent heat fluxes. It indicates that the simulation of abnormal lower-atmospheric circulation is highly related to that of latent heat flux, as evidenced by the statistically significant correlation coefficient of 0.83 and 0.69 for El Niño and La Niña years, respectively. Thus, the overestimated latent heat flux anomalies tend to enhance the Rossby-wave response, and thereby a low skill score of interannual EAWM variability.

The anomalous lower-tropospheric circulations in turn enhance the anomalies of latent heat flux by influencing the surface moisture advection. Following the bulk aerodynamic scheme, the latent heat flux is defined as:

$E = - \rho {L_{\rm{e}}}{C_{\rm{e}}}\left( {{q_{\rm{a}}} - {q_{\rm{s}}}} \right)U, $

where E is latent heat flux,ρ is atmospheric surface density,Le is latent heat of vaporization,U is the horizontal component of the near-surface mean wind magnitude, and Ce represents the turbulent exchange coefficients of moisture. As the surface saturation specific humidity,qs, is a function of SST,Fig. 8 only shows the composite DJF mean specific humidity (qa) and wind anomalies (U) at the bottom of the atmosphere for El Niño and La Niña, derived from observations and models, separately.

In the observation, there are dipole specific humidity anomalous patterns, with a negative/positive southern lobe over the NWP and a positive/negative northern lobe over the South China Sea and northern Philippine Sea in El Niño/La Niña winters, which resemble the patterns of latent heat flux anomalies. Associated with surface air temperature and circulation patterns, the dipole patterns of specific humidity in La Niña winter are also located farther to the northwest of the composite negative specific humidity anomaly in El Niño winter.

Fig. 9 As in Fig. 6, but for the specific humidity (g kg–1) and winds (m s–1) at the surface.

The CMIP5 MME shows similar patterns to the observation, except for a stronger positive/negative northern lobe of specific humidity. In El Niño events, the positive northern lobe over the South China Sea and north of the Philippines in the CMIP5, MME, and BE5 are much stronger than observed (Fig. 9). Accordingly, the local negative latent heat flux anomalies are stronger. In La Niña events, the simulated dipole specific humidity anomalous patterns are also stronger than observed, with a stronger negative northern lobe over the coastal areas of China. This favors a stronger positive anomalous latent heat flux over the northern WNP, and thereby a stronger WNPC.

In El Niño winters, the stronger local anomalous circulation further amplifies the bias of the positive specific humidity anomalies by enhancing northward moisture transport over the South China Sea and the northern WNP (Figs. 9b and 9c). Conversely, the anomalous northward winds further amplify northward moisture transport and thus enhance the negative specific humidity anomalies in the marginal seas of East Asia, in La Niña winters.

4 Conclusion

The interannual variability of the EAWM is significantly influenced by the variation of ENSO. In this study, based on data diagnosis, the authors investigate the asymmetric effects of El Niño and La Niña events on the EAWM. Although El Niño years generally show weaker winter monsoons and La Niña years show stronger winter monsoons, the enhanced precipitation over southeastern China and warmer air temperature along the East Asian coastline in El Niño years are more significant. In El Niño winters, southeastern China experiences a significant increase in precipitation and total cloud. However, the corresponding changes in La Niña winters are weak. The asymmetric effects are dominated by the asymmetric longitudinal positions and amplitude magnitudes of the WNPAC in El Niño years and the WNPC in La Niña years; the center of the weak WNPC in La Niña winters appears 10° west of the center of the strong WNPAC in El Niño winters, as a result of the longitudinal shift between El Niño and La Niña anomalous heating, and the amplitude asymmetry of the SSTA in the WNP.

Twenty-seven CMIP5 atmospheric models are evaluated in this study, in terms of their ability to simulate the asymmetric effects of El Niño and La Niña events on EAWM interannual variability. Such asymmetric effects are barely reproduced. First, a westward shift and underestimation of the precipitation centers are seen in most of the models, and thereby the MME. Second, the models commonly overestimate the negative precipitation anomalies over southeastern China during La Niña events. They thus overestimate the symmetric component of ENSO effects on EAWM interannual variability. Similar bias is also evident with respect to cloud fraction over southeastern China.

In both El Niño and La Niña winters, the spatial patterns of lower-atmospheric circulations are well reproduced by the CMIP5 models. Moreover, the zonally asymmetric positions of the WNPAC and WNPC are also reasonably reproduced. However, the magnitudes of the simulated WNPAC in El Niño years and the WNPC in La Niña years are stronger than observed, which lead to stronger EAWM rainfall responses over southeastern China.

Further analysis confirms that the overestimation of the WNPAC and WNPC occurs because of the overestimated latent heat flux anomalies, which amplify the Rossby-wave response over the WNP to the suppressed/enhanced central Pacific convective heating. This is evidenced by the statistically significant correlation coefficient of 0.83 and 0.69 for El Niño and La Niña years, respectively. The stronger dipole pattern of latent heating flux anomalies results in stronger anomalous lower-atmospheric circulation. Moreover, the enhanced WNPAC and WNPC in turn further enhance the anomalies of surface specific humidity, and thereby latent heating fluxes, by enhancing/ blocking southward water vapor transport over the marginal seas of East Asia and the northern WNP.

Note that the present analysis focuses on the composite conditions of El Niño and La Niña years. While the results are robust for most events, one exception is the winter of 1986, whose El Niño event was related to an anomalous cyclone in the WNP rather than an anomalous anticyclone. Analysis shows that the equatorial eastern Pacific SSTA during the winter of 1986 was not as warm as a normal El Niño event (figure omitted), and thus the warming of the equatorial eastern Pacific was not strong enough to suppress the convective heating in the western Pacific, and thereby failed in creating the anticyclone. The detailed processes of this special event warrant further study. Moreover, the intraseasonal oscillation is also a key factor for the asymmetric responses in anomalous circulation to ENSO (Zhang et al., 2015). Whether models can reasonably capture this mechanism needs to be evaluated in future work.

References
F. Adler, J. Huffman R., Chang G., et al ,2003: The version-2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present). J. Hydrometeor. , 4 , 1147–1167.
S.-I. An, F.-F. Jin ,2004: Nonlinearity and asymmetry of ENSO. J. Climate , 17 , 2399–2412.
S.-I. An, Y. G. Ham, J. S. Kug, et al ,2005: El Niño–La Niña asymmetry in the coupled model intercomparison project simulations. J. Climate , 18 , 2617–2627. DOI:10.1175/JCLI3433.1
G. Burgers, D. B. Stephenson ,1999: The “normality” of El Niño. Geophys. Res. Lett. , 26 , 1027–1030.
Chang, C.-P., Z. Wang, and H. Hendon, 2006: The Asian winter monsoon.The Asian Monsoon. Wang, B., ed. Berlin Heidelberg, Springer, 89–127.
W. Chen, H.-F. Graf, R. H. Huang ,2000: The interannual variability of East Asian winter monsoon and its relation to the summer monsoon. Adv. Atmos. Sci. , 17 , 46–60.
C. E. Duchon ,1979: Lanczos filtering in one and two dimensions. J. Appl. Meteor. , 18 , 1016–1022.
M. P. Hoerling, A. Kumar, M. Zhong ,1997: El Niño, La Niña, and the nonlinearity of their teleconnections. J. Climate , 10 , 1769–1786.
L. R. Ji, S. Q. Sun, K. Arpe, et al ,1997: Model study on the interannual variability of Asian winter monsoon and its influence. Adv. Atmos. Sci. , 14 , 1–22.
An Jin, Timmermann F.-F, et al ,2003: Strong El Niño events and nonlinear dynamical heating. Geophys. Res. Lett. , 30 . DOI:10.1029/2002GL016356
E. Kalnay, M. Kanamitsu, R. Kistler, et al ,1996: The NCEP/NCAR 40-year reanalysis project. Bull. Am. Meteor. Soc. , 77 , 437–472.
I.-S. Kang, J.-S. Kug ,2002: El Niño and La Niña sea surface temperature anomalies: Asymmetry characteristics associated with their wind stress anomalies. J. Geophys. Res. , 107 , ACL 1-1–ACL 1-10. DOI:10.1029/2001JD000393
C. Y. Li ,1990: Interaction between anomalous winter monsoon in East Asia and El Niño events. Adv. Atmos. Sci. , 7 , 36–46.
F. F. Song, T. J. Zhou ,2014: The climatology and interannual variability of East Asian summer monsoon in CMIP5 coupled models: Does air–sea coupling improve the simulations?. J. Climate , 27 , 8761–8777.
T. Tomita, T. Yasunari ,1996: Role of the northeast winter monsoon on the biennial oscillation of the ENSO/monsoon system. J. Meteor. Soc. Japan , 74 , 399–413.
B. Wang, R. G. Wu, X. H. Fu ,2000: Pacific–East Asian teleconnection: How does ENSO affect East Asian climate?. J. Climate , 13 , 1517–1536.
L. Wang, W. Chen, R. H. Huang ,2008: Interdecadal modulation of PDO on the impact of ENSO on the East Asian winter monsoon. Geophys. Res. Lett. , 35 , L20702. DOI:10.1029/2008-GL035287
J. Webster, O. Magaña P., N. Palmer V., et al ,1998: Monsoons: Processes, predictability, and the prospects for prediction. J. Geophys. Res. , 103 , 14451–14510.
B. Wu, T. Li, T. J. Zhou ,2010: Asymmetry of atmospheric circulation anomalies over the western North Pacific between El Niño and La Niña. J. Climate , 23 , 4807–4822. DOI:10.1175/2010JCLI3222.1
R. G. Wu, B. P. Kirtman ,2005: Roles of Indian and Pacific Ocean air–sea coupling in tropical atmospheric variability. Climate Dyn. , 25 , 155–170.
R. G. Wu, Z.-Z. Hu, B. P. Kirtman ,2003: Evolution of ENSO-related rainfall anomalies in East Asia. J. Climate , 16 , 3742–3758.
P. P. Xie, P. A. Arkin ,1997: Global precipitation: A 17-year monthly analysis based on gauge observations, satellite estimates, and numerical model outputs. Bull. Amer. Meteor. Soc. , 78 , 2539–2558.
R. H. Zhang, A. Sumi, M. Kimoto ,1996: Impact of El Niño on the East Asian monsoon: A diagnostic study of the '86/87 and '91/92 events. J. Meteor. Soc. Japan , 74 , 49–62.
R. H. Zhang, A. Sumi, M. Kimoto ,1999: A diagnostic study of the impact of El Niño on the precipitation in China. Adv. Atmos. Sci. , 16 , 229–241.
R. H. Zhang, T. R. Li, M. Wen, et al ,2015: Role of intraseasonal oscillation in asymmetric impacts of El Niño and La Niña on the rainfall over southern China in boreal winter. Climate Dyn. , 45 , 559–567.