This study focuses on statistical analysis of anomalous tropical cyclone (TC) activities and the physical mechanisms behind these anomalies. Different patterns of decaying of the warm sea surface temperature anomaly (SSTA) over the equatorial central-eastern Pacific are categorized into three types: eastern Pacific warming decaying to La Niña (EPWDL), eastern Pacific warming decaying to a neutral phase (EPWDN), and a central Pacific warming decaying year (CPWD). Differences in TC activity over the western North Pacific (WNP) corresponding to the above three types are discussed, and possible mechanisms are proposed. For EPWDL, TC genesis shows a significant positive (negative) anomaly over the northwestern (southeastern) WNP and more TCs move westward and make landfall over the southern East Asian coast. This is attributed primarily to the combined modulation of La Niña and the warm equatorial east Indian Ocean SSTA. For EPWDN, enhanced TC genesis is observed over the northeastern WNP, and suppressed TC activity is located mainly in the zonal region extending from the Philippine Sea to the eastern WNP, close to 160°E. Most of the TCs formed over the eastern WNP experience early recurvature east of 140°E, then move northeastward; hence, fewer TCs move northwestward to make landfall over the East Asian coast. For CPWD, the enhanced TC activity appears over the western WNP. This is due to the weak anomalous cyclonic circulation over the Philippines, primarily caused by the weaker, more westward-shifting warm SSTA compared to that in the previous warming year over the central Pacific.
El Niño–Southern Oscillation (ENSO) is regarded as one of the most important factors to modulate the interannual variability of tropical cyclone (TC) activity over the western North Pacific (WNP). Although total genesis frequency over the entire WNP has little change relative to the climatological mean, TC activity over the southeastern (northwestern) WNP shows a higher (lower) genesis frequency, stronger (weaker) intensity, and a longer (shorter) life span during El Niño. In contrast, the opposite features of TC activity appear during La Niña (Chan 1985, 2000; Chia and Ropelewski 2002; Wang and Chan 2002; Camargo and Sobel 2005; Wu et al. 2005; Chen et al. 2006; Camargo et al. 2007; Zhan et al. 2011; Zhao et al. 2011; Ha et al. 2013a). During the El Niño (La Niña) years, the positive (negative) anomaly of TC genesis in the southeastern WNP is generally attributed to increased (decreased) relative vorticity, caused by enhanced equatorial westerly (easterly) anomalies. Meanwhile, the negative (positive) anomaly of TC genesis in the northwestern WNP is closely related to the anomalous descending (ascending) motion associated with the strengthening (weakening) of the WNP subtropical high (WPSH). The warm (cold) equatorial central-eastern Pacific sea surface temperature anomaly (SSTA) plays a dominant role in these processes (Walsh and Ryan 2000; Chia and Ropelewski 2002; Wang and Chan 2002). It should be mentioned that the influences of ENSO on the WNP TC activity not only exist during the El Niño/La Niña years but continue to linger after the extreme phases. Chan (2000) investigates the anomaly of the annual TC genesis frequency before and after El Niño/La Niña years. He proposes that TC activity following strong warm (cold) years is significantly below (above) normal, indicating that the anomalous atmospheric circulation, induced by ENSO, can modulate WNP TC activity until after the mature phases of ENSO. In recent years, great efforts have been made to study the impact that the warm SSTA over the tropical Indian Ocean has on the East Asian climate in spring/summer, when El Niño is decaying. The “capacitor effect” mechanism was established by Xie et al. (2009) to explain how El Niño affects the remote equatorial east Indian Ocean (EEIO) SSTA, which subsequently causes an anomalous atmospheric circulation over East Asia as well as the WNP TC activity anomaly in the summer (Klein et al. 1999; Yoo et al. 2006; Yang et al. 2007; Wu et al. 2010; Xie et al. 2009; Kim et al. 2010; Du et al. 2011; Tao et al. 2012). The EEIO warming, induced by El Niño teleconnection, is similar to a capacitor charged by the battery of ENSO. It persists through the following summer after the warm SSTA over the equatorial eastern Pacific has dissipated, exerting its climatic effect through the equatorial baroclinic Kelvin wave in the lower troposphere (Wu et al. 2010). This baroclinic Kelvin wave serves as a discharging capacitor to strengthen and maintain the WNP anomalous anticyclonic circulation. Consequently, suppressed TC activity is observed in the southeastern WNP due to this anomalous anticyclone, which has low-level divergence and increasing zonal vertical shear. Meanwhile, increased TC frequency, which is attributed to reduced vertical shear and enhanced convection, is found in the western WNP and the South China Sea (SCS; Du et al. 2011). In general, a significant negative correlation exists between the WNP TC genesis frequency and the EEIO SSTA (Zhan et al. 2011; Tao et al. 2012).
In recent studies, the equatorial central-eastern Pacific warming has been separated into two regimes based on the spatial distribution of the maximum SSTA (Larkin and Harrison 2005; Ashok et al. 2007; Weng et al. 2007, 2009; Yu and Kao 2007; Kao and Yu 2009; Kug et al. 2009; Yeh et al. 2009; Yu and Kim 2010). An El Niño event, linked with a maximum warm anomaly in the cold tongue region of the eastern Pacific (EP), is defined as EP warming/El Niño, or canonical El Niño. The other type of El Niño, referred to as central Pacific (CP) El Niño, or El Niño Modoki (Ashok et al. 2007; Weng et al. 2007), is associated with CP warming. This is a phenomenon with the maximum SST anomalies appearing over the equatorial central Pacific (around 160°E–140°W), with colder SSTAs on its western and eastern sides. Yeh et al. (2009) argues that a CP El Niño event can occur more frequently under the projected global warming scenario. Based on satellite data, Lee and McPhaden (2010) find that the CP El Niño events nearly doubled in the past three decades. They conclude that the trend of CP warming is primarily a result of intense El Niño events; therefore, they excluded the possibility that the CP warming is related to a general increasing of the background SST associated with global warming. Previous studies (Larkin and Harrison 2005; Ashok et al. 2007; Ashok and Yamagata 2009; Yeh et al. 2009; Cai and Cowan 2009; D.-W. Kim et al. 2011) reveal the distinct climatic and synoptic variability (induced by latitudinal shifting of the warm tropical Pacific SSTA) and the physical mechanisms behind it. In particular, impacts of different types of Pacific warming on global TC activity have been widely investigated in recent years (Kim et al. 2009; Kim et al. 2010; Lee et al. 2010; H. M. Kim et al. 2011; Wang et al. 2013). Kim et al. (2009) reveal that TC activity over the North Atlantic shows increasing frequency, increasing the potential for landfall along the Gulf of Mexico coast and Central America during CP warming episodes. This is attributed primarily to decreased vertical wind shear in the main region of TC genesis, forced by the anomalous teleconnection pattern closely related to CP heating. Lee et al. (2010) argues, however, that longer time series data are needed to support the linkage between CP warming and increased TC frequency over the North Atlantic. H. M. Kim et al. (2011) compares TC activity in different Pacific SSTA regimes over the North Pacific, finding evident suppression of TC activity over the eastern Pacific in the CP warming years. Meanwhile, the TC genesis location shifts westward due to the anomalous westerly wind associated with CP warming and an anomalous low-level cyclonic circulation over the western WNP, which is a favorable condition for TC activity due to enhanced convection (Chen and Tam 2010; Chen 2011; Hong et al. 2011; Wang et al. 2013). CP warming also promotes cyclogenesis in the central southwest Pacific region (Chand et al. 2013). Zhang et al. (2012) examines TC landfall in the WNP, modulated by CP warming, concluding that TCs are more likely to make landfall over East Asia (especially over Japan and Korea in summer) due to a strong easterly flow anomaly induced by the westward shift of the WPSH, as well as the northward shifting of the TC genesis locations.
It should be noted that a robust asymmetry of the duration between EP El Niño and La Niña has been revealed in recent studies (Larkin and Harrison 2002; McPhaden and Zhang 2009; Ohba and Ueda 2009; Okumura and Deser 2010). Different from the slower transition from La Niña to EP El Niño, most EP El Niño events reach their peak phase near the end of the calendar year, then terminate rapidly and develop into La Niña events in the subsequent summer/fall. This phenomenon can be explained by the nonlinear response of surface winds to the tropical Indo-Pacific SSTA (Ohba and Ueda 2009; Okumura and Deser 2010). Moreover, most studies on the WNP TC activity in a decaying EP El Niño focus on the transitional period into a La Niña phase (Du et al. 2011; Tao et al. 2012). Figure 1 shows a composite of SSTAs averaged from July to October (JASO) for different decaying types of equatorial Pacific warming. The WNP TC activity during the decaying El Niño is not only modulated by the warm EEIO SSTA (Fig. 1a), as proposed in previous studies (Du et al. 2011; Zhan et al. 2011; Tao et al. 2012), but also by EP cooling (Ha and Zhong 2013); however, this combined effect of La Niña and EEIO warming on the WNP TC activity has not previously been well discussed. It is noteworthy that some EP El Niño events decay to a neutral ENSO in the following summer, instead of decaying to a La Niña phase (Fig. 1b), causing a potential difference in the East Asian summer climate and TC activity between these two decaying EP El Niño regimes. It is therefore necessary to regroup the EP warming decaying year into the La Niña (EPWDL) and neutral ENSO (EPWDN) phases and to further examine whether any significant differences exist in the TC activity under the two regimes. In addition, until now, relatively little research has been conducted concerning TC activity in the CP warming decaying (CPWD) year; therefore, this remains an open problem.
The objective of this study is to investigate the characteristics of TC activity in the three types of Pacific warming decaying phases mentioned above. We first analyze the anomalies of the TC genesis location, frequency, track, intensity, and life span over the WNP. We then investigate the combined effects of modulation factors on WNP TC activity by detecting the corresponding variation of large-scale circulation and in terms of the tropical synoptic-scale barotropic energy conversion. This paper is organized as follows: datasets and methodology are described in section 2. Statistical analyses, features of TC activity, and effects of environmental conditions during the EPWDL, EPWDN, and CPWD events are presented in section 3. The possible mechanisms of TC anomalies are discussed in section 4. Finally, section 5 provides conclusions.
2. Data and methodology
TC data used in this study are from the best-track dataset provided by the Regional Specialized Meteorological Center (RSMC) of the Japan Meteorological Agency (JMA). This dataset covers the WNP and SCS at a 6-h interval. It includes satellite data after the 1970s and records the maximum sustained wind speed [units: knots (kt);1 kt = 0.51 m s−1] after 1976. We employ this dataset to determine the anomaly of TC frequency during the TC peak season over JASO from 1960 to 2010. Only those TCs that reach tropical storm intensity (10-min average maximum sustained wind speed ≥17 m s−1 based on the JMA scale) are selected for this study. To calculate the anomalies of TC frequency, each TC position is binned into its corresponding 5° × 5° grid box. The monthly Niño-3 index and the extended reconstructed SST (ERSST; Smith et al. 2008) are obtained from the National Oceanic and Atmospheric Administration (NOAA). Wind and relative humidity fields are extracted from the National Centers for Environmental Prediction–National Center for Atmospheric Research Reanalysis Project (NNRP) dataset (Kalnay et al. 1996).
In this study, the Niño-3 index and the El Niño Modoki index (EMI; Ashok et al. 2007) are employed to distinguish EP and CP events, respectively. The EMI is used to measure the SSTA over the central Pacific and is defined as , where , , and are the SSTA averaged over the central (10°S–10°N, 165°E–140°W), eastern (15°S–5°N, 110°W–70°W), and western (10°S–20°N, 125°–145°E) Pacific, respectively. The EEIO SSTA index is also used to evaluate the climatic effect from the tropical Indian Ocean, with a mean SST over 15°S–15°N, 80°–110°E, averaged from March to October. Note that the EEIO SSTA index is used here because it is closely leading and simultaneously correlated with the WNP TC genesis frequency (Zhan et al. 2011, their Table 1). Linear trends are removed from all above time series of indices. Figure 2 shows the interannual variability of the normalized Niño-3 index and the EMI over JASO from 1960 to 2010. The extreme ENSO years are determined based on the threshold of one standard deviation from the Niño-3 index/EMI. It should be pointed out that, although 1998 is categorized as the year of an EPWDN event, strong CP cooling and EEIO warming are found simultaneously with more than 2 times the standard deviation of the indices. Thus, this extreme year might affect the statistical significance for the composite results of the EPWDN event. For this reason we remove the year 1998 from the EPWDN category. Based on the Niño-3 index of the EP warming decaying year, four EPWDL years (1964, 1973, 1988, and 2010) and three EPWDN years (1966, 1977, and 1984) are determined (Fig. 2a). For the CP condition, only five CPWD years are selected among the seven CP warming events because the consecutive 1966/67 and 1990/91 are strong CP years and are thus excluded (Fig. 2b). The CPWD years are therefore determined to be 1968, 1992, 1995, 2003, and 2005. Although using only three samples of the EPWDN event seems insufficient, the threshold criteria for one standard deviation makes the selected samples typical and we do not feel it is necessary to extend the study period for the purpose of including more samples, primarily due to the poor quality of the TC best-track dataset before the 1960s.
The composite analysis is applied in this study, and a two-tailed Student's t test is used to examine the statistical significance of the mean value. We use the genesis potential index (GPI) to evaluate the combined effect of several factors for climatological TC genesis. The factors used for this empirical index calculation is derived from the reanalysis and model data (Emanuel and Nolan 2004; Camargo et al. 2007; Walsh et al. 2007). In section 3b, the GPI proposed by Emanuel and Nolan (2004) is calculated over the WNP to examine the contribution of environmental fields to TC genesis. The index is defined as , where is the absolute vorticity at 850 hPa, is the relative humidity at 600 hPa, is the potential intensity, and is the magnitude of the vertical wind shear between 200 and 850 hPa. It is found that the seasonal variation and spatial distribution of the calculated GPI are generally consistent with those of Camargo et al. (2007, not shown); thus, GPI data is considered as the acceptable results for analysis in this study.
3. Features of TC activity in EPWDL, EPWDN, and CPWD
a. Statistical analysis of TC genesis
Table 1 lists TC genesis frequency (TCGF) in the WNP from July to October. It shows that the number of cyclogenesis west (east) of 140°E in EPWDL is significantly higher (lower) than the climatological mean. This is typical during La Niña events, as is revealed by many previous studies (Chan 2000; Chia and Ropelewski 2002; Wang and Chan 2002; Camargo and Sobel 2005; Zhan et al. 2011). TC genesis in EPWDN, however, does not exhibit the east–west oriented characteristics, as is shown in the EPWDL years. In particular, the EPWDN TCGFs over the entire WNP and east of 140°E are both higher than the climatological mean, but not statistically as significant as those in EPWDL. In CPWD, the number of TC genesis is very close to the climatological mean. Table 2 lists long-lived TC genesis frequency and the mean life span of each TC in the WNP. One primary feature is the significant decrease of long-lived TC frequency in EPWDL, relative to the climatological mean. There are annually only about five long-lived TCs with longer than a 10-day life span. This is significantly less than the climatological mean of 7.2. The frequency of long-lived TCs with longer than a 7-day life span also shows the similar decreasing feature. In response to the decreased long-lived TC frequency in EPWDL, the mean TC life span significantly shortens to 7.7 days relative to the climatological mean of 9.2 days. For the EPWDN years, the frequency of long-lived TCs with longer than a 10-day life span is slightly less (by 0.9) than the climatological mean, and the mean TC life span is significantly shorter (by 1.0 day) than the climatological mean. Different from the EPWDN and EPWDL scenarios, the long-lived TC numbers in the CPWD years are higher than the annual mean in both the 7- and 10-day spans, and the mean TC life span is close to that of normal years. We also examine the genesis frequency of strong TCs (with the averaged maximum sustained wind speed ≥41.5 m s−1). These results are listed in Table 3, which shows that the strong TC number in EPWDL is significantly lower than that in EPWDN and CPWD and is also significantly lower than the climatological mean. This is because in EPWDL more TCs form in the western WNP and move west-northwestward quickly, leaving insufficient time for these TCs to stay in a favorable environmental condition to reach their full development. The shift of the TC genesis location in EPWDL explains why the strong TC number is lower than that in EPWDN and CPWD. In contrast to the scenario in EPWDL, there is no obvious change in strong TC frequency during EPWDN and CPWD, compared to the climatological mean. In general, the TCGF exhibits distinct features between the EPWDL and EPWDN years, although they are both in the EP El Niño decaying phase; however, TCGF in the CPWD years does not show any obvious anomalies in the statistical analysis.
b. Spatial distribution of TC genesis
Figure 3 shows the spatial distribution of anomalies of TC genesis, 850-hPa wind, and GPI in the composite EPWDL, EPWDN, and CPWD events. For EPWDL, TC genesis exhibits a dipole pattern, similar to that in the typical La Niña years, with a significant positive anomaly in the northwestern and a negative anomaly in the southeastern WNP (Fig. 3a). The large-scale circulation pattern shows that an anomalous anticyclone is between 10° and 30°N over the WNP in the lower troposphere. This is induced by the intense equatorial easterly anomaly (Fig. 3b) and is considered to be a crucial circulation system in the EP warming decaying years that are associated with the EEIO heating. Moreover, the anomalous anticyclone strongly suppresses TC activity in the WNP (Du et al. 2011; Zhan et al. 2011). The positive (negative) GPI anomaly, which is generally consistent with the distribution of TCGF anomaly, is located over the northwestern (southeastern) WNP (Fig. 3b). The weaker-than-normal WNP monsoon trough, with a negative relative vorticity anomaly, is observed east of 130°E, and the suppressed convection and anomalous divergence in the monsoon trough suppress TC genesis in the eastern WNP. Meanwhile, a positive relative vorticity anomaly in the SCS and enhanced convection over the western WNP are favorable for the local TC activity. The large-scale circulation and environmental variables over the WNP in EPWDL are therefore favorable for the formation of the west–east-oriented pattern of TC genesis. For EPWDN, enhanced TC activity is observed over the northeastern WNP, and a significant negative anomaly of TCGF is located mainly in a zonal region extending from the Philippine Sea to the eastern WNP, close to 160°E (Fig. 3c). The GPI anomaly shows a similar pattern to that of TCGF (Fig. 3d). Different from the anomalous anticyclonic circulation observed in EPWDL, an anomalous cyclone with an evident positive anomaly of relative vorticity dominates over the WNP (centered around 27°N, 145°E), promoting TC genesis in the northeastern WNP. In contrast, the negative TCGF and GPI anomalies are located along the southwest flank of the anomalous cyclone due to the negative anomaly of relative vorticity over the Philippine Sea. For CPWD, the significant positive anomaly of TCGF appears over the western WNP near the Philippines. This phenomenon is accompanied by a weak anomalous cyclone over the Philippine Sea (centered around 15°N, 135°E), in addition to a positive anomaly of GPI that also appears over this region. There are, however, no evident anomalies in other regions of WNP (Fig. 3e).
Significant interannual variability of TC genesis is explained by the shift of the East Asian monsoon trough. Synoptic-scale disturbances closely related to the change of the monsoon trough are considered to be an important forcing mechanism for cyclogenesis in the ENSO cycle (Chen and Weng 1998; Wu et al. 2012; Ha et al. 2013b). In addition, barotropic energy conversion (KmKe) is an essential energy source for the development of tropical depressions in the lower troposphere (Shapiro 1978; Lau and Lau 1992; Maloney and Hartmann 2001). In this section, considering the obvious differences in the large-scale circulations and equatorial flows over the WNP in the three types of events discussed above, we try to explore the contribution of the barotropic energy source to the WNP TC genesis by calculating synoptic-scale disturbances described by eddy kinetic energy (EKE) and its KmKe at 850 hPa. Following Zhan et al. (2011), we assume that KmKe provides an extra energy source for EKE, which later contributes to TC genesis under favorable conditions. Following the approach of Seiki and Takayabu (2007), an environmental variable is decomposed into a basic state (defined as an 11-day running mean) and an eddy component (defined as the residual), and the EKE can be calculated by the eddy winds
Four dynamical process terms, the zonal convergence and the meridional shear of zonal wind (, ) and the zonal shear and the meridional convergence of meridional wind (, ), contribute to the EKE tendency caused by KmKe. The EKE tendency can be expressed in the Cartesian coordinates as
where is EKE, and are the mean zonal and meridional winds, and are the eddy zonal and meridional winds, is the three-dimensional wind vector, is the horizontal wind vector, and the over bar represents a time average over JASO. Figure 4 shows the 850-hPa EKE and KmKe anomalies in the WNP over JASO. In the EPWDL years, the negative anomaly of EKE is located over most of the WNP, except the SCS. The suppressed synoptic-scale disturbances are unfavorable for TC activity in the eastern WNP. In contrast, a nearly opposite pattern of an EKE anomaly appears in the EPWDN years. The maximum positive EKE anomaly extends farther eastward from 140°E, and the negative anomaly resides primarily around 18°N, 125°E (Fig. 4b). The enhanced (suppressed) synoptic-scale disturbances correspond to the increased (decreased) TC activity over the eastern (western) WNP (Fig. 3c), indicating that EKE change is possibly an important mechanism for TC genesis in the EP warming decaying years; however, the anomaly of EKE is insignificant in the CPWD years (Fig. 4c). To further investigate how the variation of the large-scale mean flow modulates the synoptic-scale disturbances (as well as the TC activity) over the WNP, we analyze the impact of KmKe on EKE tendency. In the EPWDL years, the significant positive anomaly of KmKe is located over the western WNP, with anomalous centers near 15°N, 120°E and 25°N, 130°E. The zonal tongue of a significant negative anomaly covers about 10°–20°N, 130°–150°N in the monsoon trough region (Fig. 4d), suggesting that the local eddy kinetic energy decreases through the barotropic eddy-mean flow interaction. In contrast, during the EPWDN years, the positive anomaly of KmKe is significant around 20°N, 145°N, where the anomalous cyclonic circulation is located. Meanwhile, the significant negative anomaly is found east of the Philippines (Fig. 4e). For CPWD, the significant positive anomaly of KmKe is found over the Philippine Sea, near 10°N, 135°E (Fig. 4f). These results clearly demonstrate that patterns of anomalous KmKe are corresponding well to those of EKE. This suggests that KmKe makes prominent contributions to the growth of the synoptic-scale disturbances over the WNP. Apparently, the change in mean flow modulates the conversion of barotropic energy to the synoptic-scale disturbances and affects the WNP TC genesis, subsequently.
c. TC track
Figure 5 shows TC tracks and anomalies of TC occurrence frequency (TCOF) from July to October. TC tracks are largely dominated by the steering flow in the midtroposphere (Harr and Elsberry 1991, 1995) and beta drift (Wang and Holland 1996), and those over the WNP generally tend to move along the western flank of the WPSH. In this section, the 5870-gpm contour, which represents the center of the WPSH, is marked to further explain the cause of the TC track anomaly. For EPWDL, most TCs tend to move westward after their formation in the Philippine Sea and the western WNP (Fig. 5a), making landfall over the South China coast and Indo-China peninsula. Thus, the most significant positive anomaly of TC occurrence is found over the SCS, while the negative anomaly appears in most areas of the WNP, with a significant anomaly over the eastern WNP (Fig. 5b). The WPSH center covers a larger area and extends farther westward, in comparison with the climatological mean (Fig. 5b), while a strong easterly anomaly prevails over the western WNP due to the intense and westward-shifting WPSH. Consequently, more TCs are steered by the sustained easterly to make landfall over the southern East Asian coast in EPWDL, while reduced TC recurvature shortens the mean life span (Table 2). In EPWDN, most TCs that form over the eastern WNP first experience early recurvature east of 140°E and then move northeastward (Fig. 5c), which can explain their relatively short mean life spans. Meanwhile, fewer TCs move northwestward and make landfall over the East Asian coast (Fig. 5c); hence, a significant positive (negative) anomaly of TC occurrence located east (west) of 140°E (Fig. 5d). This is consistent with the trajectories of TCs in EPWDN. This pattern is closely related to the much weaker WPSH with a remarkably eastward retreat (Fig. 5d), as well as strong westerly anomalies that prevail over the western WNP. The features of large-scale circulation in EPWDN are therefore prone to cause early recurvature of TCs over the central WNP and to inhibit their landfall over Japan, Korea, and East Asia. In the CPWD years, the frequency of westward-moving TCs is slightly more than the climatological mean. This is a result of the mean flow steering by the weak easterly anomaly and the somewhat westward shifting of WPSH (Fig. 5f).
4. Possible mechanisms
As shown in Fig. 1, a strong cold SSTA over the equatorial central-eastern Pacific in EPWDL indicates that strong La Niña events develop rapidly during the EP warming decaying years (Fig. 1a). As a direct Rossby wave response to cooling over the equatorial central-eastern Pacific, an anomalous anticyclonic circulation dominates over the WNP with the intense equatorial easterly anomaly in the lower troposphere (Wang et al. 2000). At the same time, an extensive warming appears over the EEIO in spring/summer of the EPWDL years (Fig. 1a). The warm SSTA is very robust since the standard deviations of the normalized EEIO indices in all EPWDL years are higher than 0.8 (Fig. 6). The warm EEIO SSTA forces an equatorial baroclinic Kelvin wave with an anomalous easterly, and divergence of the atmospheric boundary layer over the tropical Pacific. This further reinforces the low-level anomalous anticyclone over the WNP (Xie et al. 2009; Wu et al. 2009). Under the combined influence of the cold SSTA over the equatorial central-eastern Pacific and the warm EEIO, the East Asian summer monsoon trough becomes much weaker, and significant negative vorticity anomalies are present in the zonal region from the SCS to the date line, close to the equator, in the EPWDL years (not shown). These anomalous environmental conditions lead to a sharp decrease of TC frequency, east of 140°E. To further illustrate the consistent impact of the two factors on TC activity, we calculate correlations of TC genesis/occurrence frequency with the EEIO and inversed signed Niño-3 indices from 1960 to 2010 (Fig. 7). The significant negative (positive) correlation between TCGF and the negative Niño-3 index exists over the southeastern (northwestern) WNP, where the correlation reaches a maximum around the area of 10°–17°N, 150°E–180°, with the correlation coefficient above 0.5 (Fig. 7a). Meanwhile, regarding the relationship between the WNP TC genesis frequency and the EEIO SSTA, the significant negative correlation is located mainly over the central-eastern WNP (Fig. 7b). Apparently the spatial distribution of correlation between the WNP TC genesis frequency and the EEIO SSTA exhibits a similar pattern to that of the Niño-3 index (especially over the eastern WNP) but with a slightly weaker relationship. This indicates that the impact of La Niña on WNP TC activity is more significant than that of the warm EEIO SSTA. In addition, the correlations between TC occurrence and the indices also exhibit a similar relationship (Figs. 7c,d). These correlation results suggest that the modulation of La Niña events on WNP TC activity resembles the effect of the warm EEIO SSTA in the EPWDL years. The WNP TC activity is therefore influenced by the combined effect of equatorial central-eastern Pacific cooling and the warm EEIO SSTA. The two modulation factors exert synergistic impacts and jointly contribute to the anomaly of TC activity over the WNP in the EPWDL years.
TC activity and large-scale circulation over the WNP present distinct features between the EPWDL and EPWDN events, despite the fact that both events appear in the EP warming decaying phase, indicating that they have different modulation factors. For EPWDN, a positive SSTA is located over the CP region, while a negative SSTA appears over the equatorial eastern Pacific and EEIO (Fig. 1b); that is, the SSTA pattern for EPWDN over the CP region and EEIO is nearly opposite to that in the EPWDL years. Note that the SSTA pattern over the EEIO/EP region in EPWDN resembles that of typical CP warming but with weaker amplitude (Fig. 1d). Actually, two years (1966 and 1977) out of the three EPWDN years exhibit a warm SSTA over the CP region, with the normalized EMI indices higher than 0.8 (Fig. 1b). This suggests that the TC activity and large-scale circulation in EPWDN are similar to those in the CP warming years. To better compare the features of TC activity between EPWDN and the CP event, the anomalies of TC activity and related environmental variables in the composite CP warming year are shown in Fig. 8. Since the SSTA in the CP warming years is stronger than that in EPWDN (Figs. 5b,d), the cyclonic circulation anomaly is more southwestward and stronger than that in EPWDN (Fig. 8b). Such a difference in large-scale circulations can explain the northeastward shifting of TCGF anomalies in EPWDN, relative to those in the CP warming years. In addition, the GPI anomalies show much larger amplitude than those in EPWDN (Fig. 8b), and positive anomalies of EKE and KmKe also exhibit larger amplitudes in CP warming years, relative to those in EPWDN (Figs. 8c,d). For the TC track, it is found that more TC recurvature happens in the CP warming summer (Fig. 8e), which agrees with recent studies (H. M. Kim et al. 2011; Zhang et al. 2012). The positive TC anomalies appear over most of WNP, with a significant anomaly located mainly near Korea and the southern Japanese coast. In contrast, an area of suppressed TC activity is concentrated east of the Philippines and the SCS, between 10° and 20°N (Fig. 8f). It is interesting to note that the location of TC recurvature in EPWDN shifts eastward, compared with that in the typical CP warming years, and hence TCs are more likely to experience early recurvature east of 140°E, instead of making landfall in Japan and Korea. The WPSH in the CP warming years is weaker than the climatological mean, and it covers a much smaller area over the WNP in EPWDN when compared with that in the CP warming years. These features can be detected by the latitudinal positions of the WPSH's east flank in EPWDN and CP warming years (Figs. 4d, 8f). The differences in the large-scale circulation discussed above explain why the TC occurrence anomaly is different between the two events. In general, the WNP TC activities in the EPWDN years resemble those during the CP warming years, given the similar tropical SSTA pattern over the CP region. The features of TC activity in EPWDN can generally be seen as a duplicate of those in the CP warming years. As a result of the relatively weaker CP warming in EPWDN, however, the locations of positive and negative TC anomalies (including TC genesis, occurrence, and recurvature) shift northeastward as a whole, compared to those in the CP warming years. Additionally, it is found that a cold SSTA appears over the EEIO in EPWDN. Based on the correlation relationship (Figs. 7b,d), the cold EEIO SSTA may also have contributions to promote the WNP TC activity, to some extent.
During CPWD years, enhanced TC activity is found near the Philippines, but no other significant features appear over any other areas of the WNP. When comparing the transition process from EP warming to the subsequent EPWDL, two prominent characteristics are found in the CP warming turnabout. On one hand, CP warming does not induce the robust EEIO warming in the following years (Fig. 9a), thus having a limited remote impact on the East Asian TC activity. This phenomenon is attributed mainly to the relative weaker intensity of CP warming, compared to that in the EP. Consequently, the EEIO warming induced by CP warming is weak (Wang and Zhang 2002; Yuan et al. 2012). On the other hand, different from the sharp interannual variation of the EP ENSO cycle, the evolution of the CP SSTA presents a gradual transition from CP warming events to the following CPWD years (Fig. 1), demonstrating an evident interdecadal variability of CP SSTAs (Matsuura et al. 2003; Kim et al. 2010). We calculate the interannual variability of normalized Niño-3 and EMI indices from 1960 to 2010. The result reveals that the variability of Niño-3 is almost double that of the EMI (1.5 versus 0.8), and their difference is statistically significant at the 99% confidence level by the Student's t test. In particular, CP warming gradually decays from its peak phase but still maintains the warm SSTA with the maximum anomaly near 170°E in the CPWD summer/fall (Figs. 1c, 9a). Meanwhile, no significant anomalies exist over the EEIO and equatorial Pacific (Fig. 9a), indicating that the summer climate and TC activity over East Asia are mainly modulated by the weak CP warming in CPWD. Figure 9b shows the composite time–longitude cross section of 850-hPa equatorial wind and relative vorticity over the WNP. In CPWD, the positive anomaly of relative vorticity, which is induced by the westerly anomaly concentrated east of 150°E (due to CP heating), is observed over the Philippine Sea. This environmental condition is favorable for TC activity near the Philippines. Furthermore, since the maximum warm SSTA in CPWD is located more westward than that in the CP warming years, and the amplitude of the SSTA is also much weaker, enhanced TC activity shifts westward, corresponding to the anomalous cyclonic circulation over the western WNP, with weaker intensity compared to that in the previous CP warming year.
5. Summary and conclusions
There are distinct spatial–temporal features of interannual variability of TC activities at different phases of ENSO episodes. TC activities are modulated by the tropical Pacific and the EEIO SSTA. A better understanding of TC characteristics in the following years after warm events will be beneficial for the improvement of the seasonal TC prediction. Because different relationships exist between ENSO and the EEIO SSTA at various phases of the ENSO cycle, their impact on WNP TC activity in the EP/CP warming decaying years becomes an interesting issue to explore. In this study, three types of El Niño decaying (EPWDL, EPWDN, and CPWD) are defined based on the evolution of the tropical equatorial Pacific SSTA. Anomalies of TC activity corresponding to the above three types are analyzed to investigate the impact of various decaying patterns of Pacific Ocean warming on TC activity, and the underneath mechanisms are explored. The main conclusions are as follows.
In the EPWDL years, TC genesis frequency shows a significant positive anomaly in the northwestern WNP and a negative anomaly in the southeastern WNP, exhibiting a dipole pattern similar to that in the typical La Niña years. For TC tracks, the significant positive anomaly of TC occurrence frequency is located over the SCS, while the negative anomaly appears over the eastern WNP. It is found that the WPSH covers a larger area and extends farther westward in EPWDL, and a strong easterly anomaly prevails over the western WNP due to the intense and westward-shifting WPSH. Thus, after their formation in the Philippine Sea and the western WNP, more TCs are steered by the sustained easterly to move westward and finally make landfall over the southern East Asian coast and the Indo-China peninsula. Under the combined influence of the cold SSTA over the equatorial central-eastern Pacific and the EEIO warming, the anomalous anticyclonic circulation is intensified over the WNP, and the East Asian summer monsoon trough becomes much weaker with significant negative vorticity anomalies from the SCS to the date line. These anomalous environmental conditions lead to a sharp suppression of TC activity east of 140°E. Meanwhile, the correlation results suggest that the impact of La Niña is more significant than the EEIO heating.
In the EPWDN years, enhanced TC activity observed over the northeastern WNP is attributed primarily to the anomalous cyclone in the northeastern WNP. In contrast, a significant negative anomaly of TC genesis is located mainly in a zonal region extending from the Philippine Sea to the eastern WNP, along the southwest flank of the anomalous cyclone. The enhanced (suppressed) synoptic-scale disturbances are responsible for the increased (decreased) TC genesis over the eastern (western) WNP. It is noted that the SSTA pattern over the EEIO to the EP region in EPWDN resembles that of typical CP warming years but with weaker amplitude. Thus, the WNP cyclonic circulation anomaly in EPWDN is weaker, and resides more northeastward, than that in the CP warming years. For TC tracks, most TCs that form over the eastern WNP experience early recurvature east of 140°E and then move northeastward, while fewer TCs move northwestward and make landfall over Japan, Korea, and the East Asian coast. The features of large-scale circulation in EPWDN are prone to cause early recurvature of TCs over the central WNP, therefore inhibiting their landfall over East Asia. In general, the WNP TC activities in the EPWDN years resemble those in the CP warming years, as they have a similar tropical SSTA pattern over the CP region for EPWDN and CP warming years, although the SSTA intensity is different between the two events. The features of TC activity in EPWDN can therefore generally be seen as duplicates of those in the CP warming years, except that the locations of TC activity anomalies (including TC genesis, occurrence, and recurvature) shift northeastward, as a whole, in comparison to those in the CP warming events.
In the CPWD years, the significant positive anomaly of TC genesis frequency appears over the western WNP near the Philippines, accompanied by a weak anomalous cyclone and positive anomaly of GPI/EKE/KmKe over the Philippine Sea. This suggests that KmKe provides great contributions to the development of the synoptic-scale disturbances over the western WNP, and hence modulates the TC genesis there. There is no distinct anomaly in other regions of the WNP, however. For TC tracks, the frequency of westward-moving TCs is slightly higher than the climatological mean due to the steering of the weak easterly anomaly, partly as a result of the westward shift of WPSH. Compared with the transition process from EP warming events to the EPWDL, the CP warming years turnabout shows two prominent features. First, the EEIO warming induced by the CP event is neither clear nor consistent and thus has limited impact on the WNP TC activity in the CPWD years. Second, the CP SSTA transition from CP warming years to the following CPWD is a slow and gradual process, which is significantly different from the sharp change that is often seen during the EP ENSO cycle. The CP warming gradually decays but still maintains the warm phase in the CPWD summer/fall. Meanwhile, the neutral SSTAs in the EEIO and equatorial eastern Pacific indicate that the summer climate and TC activity over East Asia are mainly modulated by the weak CP warming in CPWD. Since the maximum warm SSTA in CPWD is weaker, and located more westward than that in the previous CP warming years, enhanced TC activity mainly appears near the Philippines, corresponding to the anomalous cyclonic circulation over the western WNP.
This study focuses on statistical analysis of TC activities. The different patterns of decaying of Pacific Ocean warming are categorized into three types, and their impacts on TC activities are discussed. The current study, however, is solely based on observational data analysis, while our knowledge of the underneath physical mechanism between SSTA and TC anomalies is far less than adequate. Our future work will combine observations and numerical model experiments to further investigate the TC activity anomalies and their associated physical mechanisms.
This work is jointly sponsored by National Key Basic Research Program of China (2013CB956203), the R&D Special Fund for Public Welfare Industry (Meteorology) (GYHY201306025), and National Natural Science Foundation of China (41175090).