Asymmetric responses of the Philippine Sea anomalous anticyclone/cyclone to two types of El Niño–Southern Oscillation during the boreal winter

The Philippine Sea anomalous anticyclone/cyclone (abbreviated to PSAC/PSCC) induced by El Niño/La Niña is a key atmospheric circulation system that connects El Niño/La Niña to the East Asian climate. The asymmetry between PSAC and PSCC is analyzed after El Niño–Southern Oscillation (ENSO) is divided into eastern Pacific (EP) ENSO and central Pacific (CP) ENSO. For EP ENSO, evident asymmetry exists between EP El Niño‐induced PSAC and EP La Niña‐induced PSCC, with the former being much stronger than the latter, caused by the nonlinearity of the convective anomalies over the western Pacific. Much stronger sea surface temperature (SST) anomalies for EP El Niño than for EP La Niña is responsible for this convective heating asymmetry.


| INTRODUCTION
The variability of the atmospheric circulation around the Philippine Sea in winter is mainly affected by ENSO. An El Niño (La Niña) event tends to force an anomalous low-level anticyclone (cyclone) over the western North Pacific in winter, with a center around the Philippine Sea, which is referred to as the Philippine Sea anomalous anticyclone (cyclone; hereafter abbreviated to PSAC [PSCC]) (e.g., Zhang et al., 1996;Wang et al., 2000;Wu et al., 2003). The anomalous diabatic cooling (heating) over the western Pacific, produced by El Niño (La Niña) through modifying the Walker circulation, is the reason for the formation of the PSAC (PSCC) via the Gill-Matsuno response (Matsuno, 1966;Gill, 1980;Wang et al., 2000). It is the PSAC (PSCC) by which the impacts of ENSO on the East Asian winter climate are predicted (Tomita and Yasunari, 1996;Zhang et al., 1996;Chen et al., 2000;Chen, 2002;Wang et al., 2008). Therefore, a deep and thorough understanding of the features of the ENSO-induced PSAC (PSCC) in winter is vitally important to achieve.
In recent years, a consensus has been reached that there are two types of El Niño, with distinct physical air-sea interaction and different climate impacts (e.g., Ashok et al., 2007;Kao and Yu, 2009;Yeh et al., 2009). One type is referred to as the central Pacific (CP) El Niño (but also El Niño Modoki, Dateline El Niño, or warm pool El Niño) (Larkin and Harrison, 2005;Ashok et al., 2007;Kao and Yu, 2009;Kug et al., 2009), in which the anomalous sea surface temperature (SST) warming emerges over the tropical central Pacific in winter. The other is correspondingly referred to as eastern Pacific (EP) El Niño (but also conventional El Niño, or cold tongue El Niño) and is characterized by the anomalous SST warming being situated in the tropical eastern Pacific in winter. Therefore, owing to the existence of these different types of ENSO, the relationship between ENSO and its related PSAC/PSCC becomes more complicated (Ashok et al., 2007;Kao and Yu, 2009;Yeh et al., 2009;Yuan et al., 2012). In addition, both of EP ENSO and CP ENSO reach their peak phase in winter, and correspondingly they exert the most remarkable influences on the atmospheric circulation in winter. Using linear correlation and regression, Feng et al. (2010) demonstrated that a much stronger and larger PSAC/PSCC is forced by EP ENSO during its mature winter than by CP ENSO, due to the stronger anomalous diabatic cooling over the western Pacific. Thus, severe climate anomalies associated with EP ENSO strike the southern part of China and Southeast Asia in winter, which cannot be observed with CP ENSO (Feng et al., 2010).
ENSO is known to exert strong asymmetric impacts on atmospheric circulation in winter and the related climate (e.g., Hannachi, 2001;Lin and Derome, 2004;Zhang et al., 2014;Zhang et al., 2015). More importantly, previous works have documented that the asymmetric impacts exerted by the two types of ENSO are significantly different (Cai et al., 2010;Karori et al., 2013;Frauen et al., 2014;Feng et al., 2017). For example, compared with CP ENSO, EP ENSO has strong asymmetric impacts on extratropical teleconnections in winter (Feng et al., 2017). PSAC and PSCC have been reported to manifest strong nonlinear responses to ENSO (Wu et al., 2010), wherein the PSAC center shifts eastward for El Niño relative to its counterpart for La Niña. However, ENSO was measured by the Niño3.4 index in their study, which indicates EP and CP ENSO are mixed together (Wu et al., 2010). Considering the importance of PSAC/PSCC to local climate anomalies, it is necessary to clarify the different asymmetric impacts exerted on them by the two types of ENSO, which is the target of this study. Understanding this issue may to some extent help us to improve the prediction skill with respect to East Asian climate.
The paper is structured as follows: Following this introduction, we describe the data and methods used in the study in section 2. Section 3 focuses on analyzing the asymmetric features of the PSAC/PSCC associated with the two types of ENSO. Subsequently, a possible physical mechanism responsible for this asymmetry is examined in section 4. Finally, a summary and some further discussion are provided in section 5.

| DATA, METHODS AND MODEL DESCRIPTION
The reanalysis data used in this study are from NCEP-NCAR. The horizontal resolution of the dataset is 2.5 (latitude) × 2.5 (longitude). In the vertical direction, there are 17 levels, from 1,000 to 10 hPa (Kalnay et al., 1996). The SST data are from the Hadley Center Global Sea Ice and Sea Surface Temperature (HadISST), spanning the period from 1870 to the present day, with a horizontal resolution of 1 (latitude) × 1 (longitude) (Rayner et al., 2003). In this study, the data from 1950 to 2011 are used, having been detrended to avoid long-term trend influences.
EP and CP ENSO are measured by the winter mean Niño3 index and El Niño Modoki index (EMI), respectively (Ashok et al., 2007). The EMI is constructed according to Ashok et al. (2007) by the area-averaged SST anomalies over the tropical central Pacific (10 S-10 N, 165 E-140 W) minus those over the western (10 S-20 N, 125 -145 E) and eastern (15 S-15 N, 110 -70 W) Pacific (Ashok et al., 2007). The Niño3 index is defined as the averaged SST anomalies over the region (5 S-5 N, 150 -90 W). According to the criterion used in Feng et al. (2017), the same EP and CP El Niño (La Niña) years are selected (shown in Table 1). The validity of this criterion is thoroughly discussed in their study, which for brevity is omitted here. In addition, we visually checked the spatial distributions of the SST anomalies for each EP and CP ENSO event to ensure the accuracy of the selection. In addition, the composite method is mainly employed in this study. A dry baroclinic model is used in this study in order to verify the observational results, which is built based on the dynamic core from the Geophysical Fluid Dynamics Laboratory (Wang et al., 2003;Jiang and Li, 2005;Li, 2006). This model has a horizontal resolution of T42 and five σ levels in the vertical. The basic state is set as the NCEP/NCAR winter mean reanalysis data. The external forcing is imposed based  1957, 1965, 1972, 1976, 1982, 1986, 1991, 19971968, 1977,1994, 2002, 2004La Niña 1955, 1967, 1970, 1975, 1984, 20071973, 1983, 1988, 1998, 1999 on the observational diabatic heating associated with EP and CP ENSO. Figure 1 shows the composite anomalous 850-hPa winds and streamfunction during the EP El Niño and EP La Niña, separately. The EP El Niño mature winter is accompanied by a large PSAC, with its center situated at about (10 N, 130 E) ( Figure 1a). By contrast, the EP La Niña-induced PSCC is much weaker and located more westward, with a center at about (8 N, 120 E) ( Figure 1b). The asymmetric component shows an obvious anticyclone (Figure 1c), implying the amplitude of PSAC is much larger than that of PSCC. The result acquired in Figure 1 is similar to that in the study by Wu et al. (2010), who investigated the asymmetry between El Niño-induced PSAC and La Niña-induced PSCC without separating El Niño and La Niña into two . (d-f) As in (a-c) but for CP El Niño and CP La Niña conditions. Streamfunction anomalies above the 95% confidence level are dotted. Wind anomalies less than 0.5 m/s are omitted types. It indicates that, if EP and CP ENSO are mixed together, the atmospheric circulation response over the western North Pacific tends to show the EP ENSO-forced results rather than CP ENSO-forced results.

| ASYMMETRIC PHILIPPINE SEA ANOMALOUS ANTICYCLONE/CYCLONE
In comparison to the asymmetry between PSAC and PSCC induced by EP ENSO, CP El Niño and CP La Niña cause distinct asymmetric features between PSAC and PSCC. A northeast-southwest-oriented PSAC appears over the western North Pacific (Figure 1d). A similarly oriented PSCC is seen for CP La Niña (Figure 1e), indicating an analogous spatial pattern between PSAC and PSCC for CP ENSO. Furthermore, the intensity between PSAC and PSCC induced by CP ENSO is roughly comparable. Therefore, the asymmetric component of the CP ENSO-related atmospheric circulation anomalies over the western North Pacific is much weaker than that associated with EP ENSO (Figure 1c,f). It is found that to the west of the Philippines the asymmetric signal is weak, whereas to the east of the Philippines there is obvious asymmetric signal ( Figure 1f). Nevertheless, we should note that this strong asymmetric signal is caused not by the CP El Niño-related PSAC and CP La Niña-related PSCC, while it is caused by the atmospheric circulation anomalies over the central Pacific that are forced by the convective anomalies there. Therefore, this asymmetric signal is out of the focus of this study.
To further demonstrate the asymmetry of the two types of ENSO-induced PSAC/PSCC, the intensity of PSAC/ PSCC is calculated via the area-averaged 850-hPa streamfunction anomalies over the region (100 -140 E, 0 -20 N) for the EP ENSO cases, and the region (100 -130 E, 0 -20 N) for the CP ENSO cases, as shown in Figure 2. The area-averaged region is selected by the domain of PSAC/ PSCC shown in Figure 1. The signs for the EP/CP La Niña cases are reversed for convenience of comparison. For the EP El Niño cases (Figure 2a), the intensity of PSAC in five out of eight EP El Niño cases is above 1.0 × 10 6 m 2 /s, whereas the intensity of PSCC is below that value for all the EP La Niña cases except one. Therefore, the averaged PSAC intensity is much higher than the averaged PSCC intensity, showing strong asymmetric responses of atmospheric circulation to EP ENSO. In contrast, the situation is different for CP ENSO (Figure 2b). In most of CP La Niña cases, the PSCC intensity varies around the value of 1.0 × 10 6 m 2 /s, only except the CP La Nina case of 1983. In four out of six CP El Niño cases, the PSAC intensity is fluctuated around the value of 1.0 × 10 6 m 2 /s. Therefore, the averaged intensity of the PSAC induced by CP El Niño is equivalent to that induced by CP La Niña, indicative of the insignificant asymmetric responses between PSAC and PSCC.

| PHYSICAL MECHANISMS
In this part, we focus on the question as to why EP ENSO, and not CP ENSO, produces the large asymmetric impacts. Figures 3 and 4 show the composite SST anomalies and zonal circulation anomalies for the EP and CP ENSO cases, respectively. For the EP El Niño cases, a prominent dipole pattern of the tropical SST anomalies exists, with the positive pole over the eastern Pacific and the negative pole over the tropical western Pacific (Figure 3a). Comparatively, for the EP La Niña cases, this dipole SSTA pattern disappears (Figure 4b). Instead, a monopole pattern dominates the tropical Pacific, that is, negative SST anomalies cover the central-eastern Pacific (Figure 3b). One asymmetric feature of the SST anomalies between EP El Niño and EP La Niña is positive SST anomalies over the central-eastern Pacific (Figure 3c), which indicates EP El Niño warming is much stronger than EP La Niña cooling (Figure 3a,b). This asymmetry results in much stronger anomalous upward motion over the central-eastern Pacific generated by EP El Niño warming (Figure 4a). Meanwhile, because of the mass balance law, anomalous sinking motion associated with EP El Niño appears over the western Pacific, and thus anomalous Walker circulation is produced by EP El Niño (Figure 4a). This anomalous zonal circulation is much stronger than its counterpart associated with EP La Niña (Figure 4b,c). Therefore, convective activity is severely suppressed over the western Pacific for EP El Niño (Figure 3a,c), by which a strengthened PSAC is forced through the Gill-Matsuno response. Comparatively, for EP La Niña, convective activity is weakly enhanced, thereby producing a weak PSCC (Figure 3b). Hence, asymmetric intensity between PSAC and PSCC is formed.
Note that significant negative SST anomalies appear over the western Pacific for EP El Niño, but not for EP La Niña (Figure 3a,b). Therefore, a strong asymmetric signal of SST anomalies appears over the western Pacific in addition to that over the central-eastern Pacific (Figure 3c). Such strong negative SST anomalies for EP El Niño favor to suppress local convective activity (Wu et al., 2010). Therefore, the stronger convective anomalies over the western Pacific for EP El Niño are a result of the combined effects of stronger EP El Niño SST warming over the central-eastern Pacific, and cooling over the western Pacific.
In contrast, the CP El Niño (La Niña)-related anomalous SST warming (cooling) shifts westward relative to that for EP El Niño (La Niña) (Figure 3d,e). Specifically, CP El Niño is characterized by positive SST anomalies over the central Pacific and negative ones over the western Pacific (Figure 3d), while CP La Niña shows a similar SST anomaly distribution but with the opposite poles. Therefore, the SST anomalies between CP El Niño and CP La Niña show a similar spatial pattern and equivalent magnitude (Figure 3d,e), showing weak anomalous SST asymmetry (Figure 3f). The anomalous zonal circulation, with the ascending flow over the central Pacific and the descending flow over the western Pacific, is induced by CP El Niño (Figure 4d), while similar anomalous zonal circulation with the reversed direction is generated by CP La Niña (Figure 4e). Due to the weak asymmetry of the anomalous SST anomalies between CP El Niño and CP La Niña, the anomalous zonal circulation respectively induced by CP El Niño and CP La Niña is nearly symmetric (Figure 4d-f). Correspondingly, the convective anomalies over the central Pacific and western Pacific are comparable between CP El Niño and CP La Niña (Figure 3d,e), and thereby a weak asymmetry between PSAC and PSCC is produced.   (a, b). The right-hand column is the same as the left, but for CP ENSO. The contour interval is 1 × 10 −2 Pa/s. Dotted areas indicate SST anomalies above the 95% confidence level, based on the two-tailed Student's t-test Note that although the SST anomalies in the central Pacific between CP El Niño and CP La Niña show nearly symmetric distribution, the center of the convective anomalies in the central Pacific is located slightly more westward during CP La Niña than during CP El Niño due to the asymmetry of zonal SST anomalies. This westward convective center in the central Pacific during CP La Niña is in favor of pushing the convective center over the western Pacific more westward (Figure 3d,e), which further produces the PSCC associated with CP La Niña to be located more westward compared with the PSAC associated with CP El Niño (Figure 1d,e).
In order to verify the reliability of the observational results that the asymmetry in the convective activity over the western Pacific plays a crucial role in the asymmetry between PSAC and PSCC, a dry baroclinic model is used. The prescribed diabatic heating has a horizontal distribution, which is estimated by the observed precipitation anomalies associated with EP and CP ENSO. To highlight the roles of heating over the western Pacific, the precipitation anomalies over the region (20 S-20 N, 90 -160 E) are retained, and further normalized by the maximum precipitation anomalies (4.8 mm/day) among EP and CP ENSO composited fields. This estimated method of the horizontal heating distribution is followed by the previous studies (Chen and Tam, 2010;Feng et al., 2010). The heating profile in the vertical peaks at 500 hPa (σ = 0.5), with the magnitude of 0.6 K/day, and decreases toward upward and downward. The column mean of this vertical profile is 0.25 K/day, equivalent to the latent heating released by 1 mm/day precipitation. The simulated results on day 30 are shown here after the steady state is reached on Day 10.
Four experiments are designed according to the EP (CP) El Niño and EP (CP) La Niña precipitation fields. Figure 5 shows the low-level (850 hPa) wind responses to the prescribed diabatic heating in EP and CP ENSO experiments. For EP El Niño, the atmospheric response shows a strong anticyclone over the Philippines (Figure 5a). For EP La Niña, because of the weak diabatic heating over the WNP, the atmospheric response is quite weak relative to that for EP El Niño (Figure 5b). Therefore, an asymmetry in the intensity between PSAC and PSCC is forced by different  (a, b). The right-hand column is the same as the left, but for CP ENSO. Contours denote the vertical p-velocity anomalies exceeding 90% confidence level, based on the two-tailed Student's t-test diabatic heating intensity, which is in good agreement with the observational results. In contrast, since the diabatic heating is comparable between CP El Niño and CP La Niña, the forced PSAC and PSCC have a similar feature and show a weak asymmetry (Figure 5c,d). Consequently, the simulated result further confirms that the asymmetry in the anomalous convective activity over the western Pacific is the main reason for the asymmetry between PSAC and PSCC.

| CONCLUSIONS AND DISCUSSION
The asymmetric impacts of ENSO on PSAC and PSCC are analyzed after ENSO is divided into its EP and CP types. There is strong asymmetry between EP El Niño-induced PSAC and EP La Niña-induced PSCC, with the amplitude of the former being greater than the latter. However, CP El Niño-induced PSAC and CP La Niña-induced PSCC bear a resemblance in their intensity and spatial pattern, implying a much weaker asymmetry compared with that for EP ENSO.
To explore the possible mechanism behind why EP ENSO causes more significant asymmetry between PSAC and PSCC than CP ENSO, the tropical SST anomalies and convective anomalies are analyzed. EP El Niño and EP La Niña demonstrate remarkable asymmetry over the western and eastern Pacific, in which the SST anomalies are much stronger for EP El Niño than EP La Niña. Correspondingly, the anomalous Walker circulation and the resultant convective anomalies are amplified more by EP El Niño, producing a much stronger PSAC. As such, the asymmetry between PSAC and PSCC is formed. However, for CP ENSO, the situation is completely different. The distribution and intensity of the SST anomalies between CP El Niño and CP La Niña are similar, demonstrating a weak asymmetry in SST anomalies. Moreover, CP ENSO preferentially occurs over the tropical central Pacific, where the zonal climatological SST shows weak asymmetry. Therefore, the anomalous zonal circulation associated with CP ENSO shows a nearly symmetric feature, as do the convective anomalies over the western Pacific. Such results lead to a weak asymmetry in CP El Niño-induced PSAC and CP La Niña-induced PSCC.
Previous works have also shown that EP ENSO generates teleconnections with much stronger asymmetric features than those under CP ENSO (Frauen et al., 2014;Feng et al., 2017;Zhang et al., 2018). In comparison to CP ENSO, the strong asymmetric impacts caused by EP ENSO are inherently unavoidable, because the preference location that EP ENSO often occurs is over the eastern Pacific, where the zonal climatological SST is characterized by strong asymmetry. Therefore, even if the strength and spatial pattern are nearly symmetric for EP ENSO, the convective anomalies over the eastern Pacific can show asymmetric features owing to this strong climatological SST asymmetry, and thus emanate the asymmetric teleconnections.