1. Introduction
Over the past few decades, many studies have focused on the relationship between western North Pacific (WNP) tropical cyclones (TCs) and the El Niño–Southern Oscillation (ENSO) (Chia and Ropelewski 2002; Wang and Chan 2002; Camargo and Sobel 2005; Chan 2007; Huang and Xu 2010; Kim et al. 2011). For example, Wang and Chan found that a southeastward (northwestward) shift in the positions of TC genesis in the WNP occurs during El Niño (La Niña) years, which in turn may favor (suppress) the development of intense TCs. Camargo and Sobel (2005) discovered that the accumulated cyclone energy (ACE) (Bell et al. 2000) is positively correlated with ENSO indices, suggesting that stronger (weaker) and longer-lasting (shorter lived) TCs tend to form during an El Niño (La Niña) event. Following this, Camargo et al. (2007b,a) investigated the impacts of ENSO on the tracks and genesis locations of TCs. Making use of a genesis potential index (Emanuel and Nolan 2004), they suggested that vorticity and relative humidity play an important role in the eastward shift in the mean genesis location of TCs in the WNP. Chan (2007) also pointed out that interannual variations in intense typhoons (TYs) in the WNP are unlikely to be determined by local sea surface temperature (SST) but are related to changes in planetary-scale atmospheric circulation (vorticity and wind shear) and thermodynamic structure [moist static energy (MSE)] associated with the El Niño phenomenon. More recently, Huang and Xu (2010) attributed the increase in the number of super typhoons (STYs) in El Niño years to changes in SST, the monsoon trough, and vertical wind shear (VWS).
These studies focused mainly on the effect of ENSO on intense TCs (Chan 2007; Huang and Xu 2010) or considered all TCs as a whole regardless of their intensity (Chia and Ropelewski 2002; Wang and Chan 2002; Camargo et al. 2007a; Kim et al. 2011). Thus, the effect of ENSO on weaker typhoons or tropical depressions is unclear, and the connection between ENSO and TCs with different intensities remains uncertain. In addition, less effort has been focused on studying the TC–ENSO relationship during the ENSO transition phase. Frank and Young (2007) recently suggested that the variance in factors that control the formation of TCs can ultimately result in storms with different degrees of intensity. Thus, the impact of ENSO on TCs with different intensities is also expected to vary.
In contrast to most ENSO–TC studies that use anomalies in SST in the Niño-3 or Niño-3.4 regions as a method of diagnosing the impact of ENSO, the present study uses upper-ocean heat content (OHC) as a proxy for ENSO since previous studies have shown that this factor also serves as an important indicator of ENSO (Kessler 1990; Kinter et al. 2002; Zhou and Chan 2007). Wang et al. (1999) indicated that the strongest ENSO signal occurs in the subsurface as a result of wind stress driven by ENSO. However, few previous investigations have used OHC to study the impact of ENSO on TCs. A recent study by Wada and Chan (2008) suggested that a decrease in OHC in the WNP is related to the passage of TCs, though they did not delve deeply into the relationship between OHC, ENSO, and TCs. Therefore, this study uses OHC to examine the impact of ENSO on TCs with different intensities.
Section 2 describes the datasets used and the classification of different groups of TCs. Sections 3 and 4 investigate the interannual variability of different groups of TCs and their relationship with ENSO. Section 5 examines the connection of TC frequency with large-scale environmental parameters. Finally, a discussion and summary are given in section 6.
2. Datasets and definitions
The TC datasets of the WNP were obtained from the Joint Typhoon Warning Center (http://www.usno.navy.mil/JTWC/), which records the location and intensity of all TCs at 6-h intervals. July–November (JASON) is taken to be the TC season, during which about 80% of the total number of TCs in the WNP (0°–30°N, 120°E–180°) occur. The study period from 1965 to 2006 was chosen based on the availability of routine satellite observations. Tropical cyclones formed during this period are categorized into one of the three groups according to their maximum attainable intensity. The super typhoon (STY) group includes TCs that reach at least 114 knots (kt, 1 kt = 0.51 m s−1; Category 4 and 5 on the Saffir–Simpson Hurricane Wind Scale), accounting for about 26% (223/850) of the total number of TCs and representing the most intense TCs. The typhoon (TY) group comprises moderately strong TCs with a maximum sustained wind speed between 64 and 114 kt (Category 1–3 on the Saffir–Simpson Hurricane Wind Scale) and accounts for a further 37% of the total frequency (316/850). The remaining 37%, with a maximum sustained wind speed of less than 64 kt (311/850), is classified as the tropical storm and tropical depression (TSTD) group and corresponds to the weakest type of TC.
3. Interannual variability of TCs in association with OHC
a. Variability of TCs with different intensities
During 1965–2006 an average of 5.31 STYs, 7.52 TYs, and 7.40 TSTDs formed in the WNP, with a standard deviation of 2.35, 2.70, and 2.78, respectively. Figure 1 shows the corresponding standardized time series of the STY, TY, and TSTD frequencies in the WNP during the TC season (JASON). Interannual as well as interdecadal variations are noted for all three TC groups. However, it should be pointed out that the interrelationship between these three groups of TCs is weak, such that no significant correlation can be revealed (Table 1). In other words, a higher occurrence of a certain TC group does not necessarily imply a higher frequency of the others. For example, in the period 2002–06, a below-average number of TSTDs and TYs were recorded in the WNP (Figs. 1b and 1c), while more STYs were actually noted during this period (Fig. 1a).
Standardized time series of JASON (a) STY, (b) TY, and (c) TSTD frequencies in the period 1965–2006. For each TC group, the standardized time series is calculated by first removing the mean and then dividing the resultant deviations from the mean by the standard deviation. The means (standard deviations) are 5.31 (2.35), 7.53 (2.70), and 7.40 (2.78) for STY, TY, and TSTD, respectively. The dotted line denotes the 10-yr running mean of the standardized time series.
Citation: Journal of Climate 25, 17; 10.1175/JCLI-D-11-00430.1
Correlation coefficients between frequencies of different TC intensity groups during JASON obtained by calculating the correlations between the 32-yr time series of the frequency of each group of TCs with that of the other groups.
To clearly identify the dominant modes of variability, the real-valued Mexican hat wavelet, the second derivation of a Gaussian (DOG; derivative m = 2), was also applied to the three frequency time series (Fig. 2). This method has also been employed in previous TC studies (Chan 2008; Chan and Xu 2009) for mode identification, and further details can be found in Torrence and Compo (1998). Consistent with the aforementioned weak correlations between the three TC groups, the wavelet power spectra of the three groups of TCs demonstrate different features. For instance, the 2–7-yr signal is more prominent in the late 1990s for STYs (Fig. 2a) in a way similar to that in Chan (2008), whereas it appears before the 1990s in the TY spectrum (Fig. 2b). For the weaker TSTDs, the 2–7-yr band is much smaller and the 16–32-yr band dominates in the 1990s (Fig. 2c). The results again suggest that variations exist among different groups of TCs.
The normalized wavelet power spectrum of JASON (a) STY, (b) TY, and (c) TSTD frequency over the period 1965–2006. The cross-hatched region represents the cone of influence, and the thick solid line denotes values that are over 95% confidence.
Citation: Journal of Climate 25, 17; 10.1175/JCLI-D-11-00430.1
b. The impact of OHC on TC intensity
The aforementioned wavelet spectra of different groups of TCs show differences in power in the 2–7-yr band. As the 2–7-yr band is often linked to ENSO, the impact of ENSO on different groups of TCs is also expected to vary.
As discussed in the introduction, the OHC, rather than SST anomalies, was used as a proxy for ENSO. Therefore, so as to obtain a general picture of how OHC is related to ENSO, a correlation map between the Niño-3.4 SST with basinwide OHC was first computed (Fig. 3). The correlation pattern exhibits a distinct east–west dipole in the Pacific with a significant positive (negative) correlation in the eastern (western) Pacific, which is similar to the pattern found by Zhou and Chan (2007). Such a pattern is consistent with the accumulation of warm water in the eastern Pacific associated with El Niño events. The results here support the previous view that OHC is a good indicator of ENSO (Kessler 1990; Kinter et al. 2002; Zhou and Chan 2007) and is thus suitable to use in the present study.
Correlation between Niño-3.4 SST and OHC in JASON for the period 1965–2003. The contour represents values over 95% confidence.
Citation: Journal of Climate 25, 17; 10.1175/JCLI-D-11-00430.1
Correlation between OHC and STY frequency: (a) OHC leads STY by one year, (b) simultaneous correlation, and (c) OHC lags STY by one year during JASON for the period 1965–2003. The contour represents values over 95% confidence.
Citation: Journal of Climate 25, 17; 10.1175/JCLI-D-11-00430.1
Simultaneous partial correlation between OHC and STY frequency during JASON for the period 1965–2003 after removal of the ENSO effect. The contour represents values over 95% confidence.
Citation: Journal of Climate 25, 17; 10.1175/JCLI-D-11-00430.1
In contrast to STYs, the correlation between OHC and TYs shows different patterns (Fig. 6). Instead, in year 0, the correlation with an ENSO-like dipole pattern appears in both year −1 and year 1, which suggests that the frequency of TYs tends to increase (decrease) during the transition from La Niña to El Niño (El Niño to La Niña) events. These results imply that different phases of ENSO affect STYs and TYs in different ways. Modulation of STYs tends to occur during the mature phase of ENSO, while that of TYs appears to be related to the transition phase. Finally, the simultaneous correlation for the TSTD group (Fig. 7) appears to be similar to that of the STY group, although the sign in year 0 is reversed and the correlations in year −1 and year 1 are weak. All of these findings imply that the timing with which ENSO impacts STYs, TYs, and TSTDs differs and that the corresponding changes in their frequencies appear to follow the ENSO cycle closely.
As in Fig. 4, but for the correlation between OHC and TY frequency.
Citation: Journal of Climate 25, 17; 10.1175/JCLI-D-11-00430.1
As in Fig. 4, but for the correlation between OHC and TSTD frequency.
Citation: Journal of Climate 25, 17; 10.1175/JCLI-D-11-00430.1
4. The TC–ENSO relationship
To confirm the aforementioned correlation results between the frequency of different TC groups and OHC, composite analysis is also carried out to illustrate the impact of ENSO on different groups of TCs. The OHC anomaly composites are based on the more/fewer TC years (denoted by year 0), when the JASON standardized TC frequency is more/less than 1/−1 (refer to Table 2 for the selected years). In addition, the evolution of OHC anomalies two years before and after year 0 (denoted by year −2 and year 2, respectively) are also shown. Figure 8a shows the longitude–time sections of OHC anomaly composites along the equator for the fewer-STY case. In year −2, positive (negative) OHC anomalies appear in the eastern (western) Pacific, which is typical of the El Niño pattern. Accordingly, a generally higher-than-normal STY frequency can be found during this period, which agrees well with the significant positive correlation of STYs and OHC as revealed in section 3b. The negative OHC anomalies in the western Pacific begin to propagate eastward in year −1. The variation in number of STYs during this period is small, while a generally below-average total TC number is recorded. The negative OHC anomalies finally reach the eastern Pacific in early year 0, which signifies the mature stage of the La Niña event, during which the occurrence of STYs is significantly reduced. At the same time, it is also interesting to note that the number of total TCs during this period is generally above normal, which again suggests that the increase in total TC formation does not necessarily support greater STY development. Finally, in year 1, the positive anomalies in the western Pacific start to move eastward and the whole cycle is repeated. The reverse is generally true for the years with a greater number of STYs, as shown in Fig. 8b, although the OHC anomalies and the reduction in STY number in year −2 are less prominent. The aforementioned results further confirm that the frequency of STYs follows the ENSO cycle, with greater (reduced) frequency during the mature phase of El Niño (La Niña), while the variation in frequency is relatively smaller during the transition phase. These results conform well to previous studies (Camargo and Sobel 2005; Chan 2007), which found that El Niño favors the development of intense TCs.
More (fewer) TC and STY years selected when the JASON standardized anomalies are greater (less) than or equal to 1 (−1).
Longitude–time section of OHC anomaly composites (108 J m−2) averaged over 5°S–5°N for (a) fewer STY years and (b) more STY years, where year 0 refers to the less and more STY years, respectively, and year −n (n) refers to n years before (after) year 0. The red (black) line on the right represents the standardized frequencies of STYs (total TCs) during the composite years.
Citation: Journal of Climate 25, 17; 10.1175/JCLI-D-11-00430.1
A similar analysis is also applied to TYs (Fig. 9a). The OHC anomalies depict an El Niño–like dipole pattern in year −1, indicating the mature stage of an El Niño event, which differs from the case of STYs (Fig. 8a). Year 0 is the transition phase when the negative OHC anomalies in the western Pacific begin to migrate eastward. During this period, the TY frequency shows a significant decrease, consistent with the previous correlation result that fewer TYs will generally be found during an El Niño to La Niña transition. Moreover, an above-normal TY number can be observed in year −2, which is the transition phase from a La Niña to an El Niño event. The reverse is also found for years with a greater frequency of TYs (figure not shown), which indicates that ENSO impacts TYs mainly during transition phases.
As in Fig. 8, but for (a) fewer TY years and (b) fewer TSTD years. The red line on the right represents the standardized frequencies of TYs and TSTDs, respectively, during the composite years, while the black line denotes the total TC frequency.
Citation: Journal of Climate 25, 17; 10.1175/JCLI-D-11-00430.1
Finally, the pattern of OHC anomalies for the fewer TSTDs (Fig. 9b) is almost the reverse of that for STYs (Fig. 8a), with significant reduction (increase) in year 0 (−2) corresponding to the mature phase of an El Niño (La Niña) event, though the OHC anomalies appear to be weaker and less organized. While for years with a greater TSTD frequency, the OHC pattern is generally the reverse of that of the fewer-TSTD case (figure not shown). This indicates that the weaker TSTDs generally exhibit activity opposite to that of intense STYs during the mature phase of an ENSO event.
To summarize, TCs of different intensities are modulated during different phases of an ENSO cycle. In particular, the increase (decrease) in the frequency of intense STYs is usually found to be associated with a mature El Niño (La Niña) event, while the converse is found for the weak TSTD group. This is in agreement with Camargo and Sobel (2005) and Camargo et al. (2007b). In contrast, modulation of the number of TYs occurs mainly during the transition period of an ENSO cycle. The results here suggest that weak and intense TCs could respond differently to external forcing induced by an ENSO event, which will be discussed in detail in section 5.
As an extension, we also investigate other TC parameters, including genesis position, TC lifetime, and accumulated cyclone energy (ACE) so as to achieve a more comprehensive understanding of the interannual variability of TCs in the WNP. Differences in these parameters between the high and low occurrence years are considered to be significant when exceeding 95% confidence based on the Student’s t test. Figure 10 shows the genesis positions of STYs in the WNP during the high and low occurrence years. The average number of STYs formed in the high occurrence years is 8.75, which is about 5.5 times greater than that in the low occurrence years (1.6), while the total TC number remains approximately the same (22.1 versus 21). In addition to the number of STYs generated, another noticeable feature is the difference in the genesis positions, whereby the TCs formed in the years with a higher occurrence of STYs exhibit a clear shift to the southeast (0°–15°N, 140°E–180°). In fact, such a shift in genesis positions is closely related to changes in atmospheric circulation influenced by ENSO (refer to section 5). Consistent with previous studies (Chia and Ropelewski 2002; Wang and Chan 2002; Camargo and Sobel 2005; Camargo et al. 2007b), the increase in the number of STYs, together with the southeastward shift in genesis locations, results in TCs with a much longer life span and contributes to a significant increase in ACE in high occurrence years (Table 3).
Composite distribution of TCs in (a) more STY and (b) fewer STY years. The black dots represent STY genesis positions, while the triangles represent genesis positions other than STY. The numbers within parentheses (top-left corner) indicate the number of total TCs and STYs, respectively, formed during the composite years.
Citation: Journal of Climate 25, 17; 10.1175/JCLI-D-11-00430.1
Accumulated cyclone energy (ACE) and average life span of TCs in different years. Bold values indicate the differences between the more and fewer TC years are significant at 95% confidence based on a Student’s t test.
With regard to TYs, the average number recorded in the high occurrence years is 11.25, compared with 3.2 in the years with a lower occurrence, which also contributes to a significant difference in total TC number during the two periods (22.75 versus 15.8). In contrast to STYs, no prominent shift in the genesis position (Fig. 11) or changes in life span (Table 3) can be observed. This means that the significant increase in ACE is mainly associated with an increase in the number of TYs during high occurrence years. Finally, for the TSTD group (Fig. 12), the significant change in their number (11.9 versus 3.25) in the high and low occurrence years does not lead to a prominent difference in ACE. This is because ACE is mainly dominated by intense TCs, while weak TSTDs make only a minor contribution (Bell et al. 2000). Indeed, the average TC life span in the low occurrence years is longer than that in the high occurrence years, which contributes to the overall larger ACE. Nevertheless, the results here indicate that, in addition to the genesis numbers, the shifts in genesis position and the changes in mean life span, at least in the case of intense STYs, may also be associated with ENSO. This is consistent with previous studies (Chia and Ropelewski 2002; Wang and Chan 2002; Camargo and Sobel 2005; Camargo et al. 2007b) in which the southeastward shift in genesis positions favored the development of more intense and longer-lived TCs in El Niño years.
As in Fig. 10, but for (a) more TY and (b) fewer TY years.
Citation: Journal of Climate 25, 17; 10.1175/JCLI-D-11-00430.1
As in Fig. 10, but for (a) more TSTD and (b) fewer TSTD years.
Citation: Journal of Climate 25, 17; 10.1175/JCLI-D-11-00430.1
5. Impact of large-scale environmental parameters on TC activity
In the previous section, the change in frequency of different TC intensity groups is found to be related to different phases of ENSO events. Therefore, ENSO must exert its impact on TC activity through differences in the modulation of large-scale environmental parameters. In this section, the six TC-related parameters suggested by Gray (1979) are investigated, including four dynamic parameters [850-hPa relative vorticity, 200–850-hPa total vertical wind shear (VWS), 200-hPa divergence, and 500-hPa omega] and two thermodynamic parameters [600-hPa relative humidity and 1000–500-hPa average moist static energy (MSE)]. Following the method of Chan and Liu (2004), empirical orthogonal function (EOF) analysis is first applied to the standardized time series of the different parameters during the TC season. Correlations are then computed for the principal component (PC) time series of each mode with the Pacific dipole index (PDI) and the TC frequency. The correlation results are summarized in Tables 4 and 5, respectively. Similar to Zhou and Chan (2007), the PDI is defined as the difference in OHC averaged over the eastern region (6°S–4°N, 110°–150°W) minus the western region (0°–10°N, 130°–170°E) to reflect the dipole pattern during an ENSO event. Thus, a large positive (negative) value of the PDI corresponds to the warm (cold) phase of ENSO. This was shown previously (Zhou and Chan 2007) to be a good index to represent ENSO, with a correlation of 0.91 (over 99% confidence) with the Niño-3.4 index in this case. Finally, stepwise regressions are performed to determine the relative contribution of each environmental parameter to the prediction of TC frequency.
Correlation coefficients between PC time series of different environmental parameters with the frequencies of different TC groups. Bold values indicate the correlation is significant at least at the 90% confidence level.
Correlation coefficients between PC time series of different environmental parameters with the PDI at different leads and lags. PC lags (leads) refers to the correlation when the PC time series lags (leads) the PDI by one year. Only values over the 90% confidence level are shown. Single (double) asterisks denote the correlation between the PDI with the time series of PC1 (PC2) of the parameter.
a. Dynamic factors
EOF1 of the low-level relative vorticity reveals a north–south dipole pattern (Fig. 13), with a maximum positive center situated in the southeastern part of the western Pacific. Such a pattern is consistent with that found by Chan and Liu (2004) and explains 22.4% of the total variance. This represents the interannual variability of the strength of the monsoon trough in the WNP. The corresponding PC time series reveals a significant positive (negative) correlation of 0.34 (−0.36) at 95% confidence with the STY (TSTD), while it correlates only weakly with the TY frequency (0.076). At the same time, the PC time series shows a significant positive correlation (0.85 at 99% confidence) with the PDI, indicating that it is closely linked to ENSO. The importance of a monsoon trough and relative vorticity associated with the eastward shift in the mean genesis location has also been mentioned in previous studies (Wu et al. 2004; Camargo et al. 2007b,a; Kim et al. 2011). Consistently the EOF1 pattern here suggests that an increase in low-level relative vorticity, with a maximum in the southeast region associated with the strengthening of the monsoon trough, favors (suppresses) formation of intense STYs (weaker TSTDs) during an El Niño event.
(top) EOF1 and (bottom) EOF2 of the JASON mean 850-hPa relative vorticity. The sign of the EOF is chosen such that the simultaneous correlation is positive between the PC time series with the PDI. The value above the top-right corner of each panel indicates the percentage of the total variance explained by each mode.
Citation: Journal of Climate 25, 17; 10.1175/JCLI-D-11-00430.1
In contrast, the PC time series of EOF2 depicts a much more significant correlation with TY frequency (0.33 at 95% confidence) compared with that of STY (0.24) and TSTD (0.12). The positive center migrates northward to about 20°N, 155°E. This pattern was not discussed by Chan and Liu. Instead of showing significant simultaneous correlations, the time series of PC2 reveals a significant lag-1 correlation (−0.30 at 90% confidence) as well as a lead-1 correlation (0.26 at 90% confidence) with the PDI. This suggests that a positive (negative) EOF2 pattern usually occurs during La Niña to El Niño (El Niño to La Niña) transitions, which in turn affects the formation of TYs. Therefore, a change in the number of TYs during the transition phase, as revealed in the previous section, can be attributed partly to a concomitant change in low-level vorticity induced by ENSO.
The first EOF of the total VWS shows an alternating pattern with a positive–negative–positive orientation (Fig. 14). This pattern is found to be significantly correlated only with STY frequency (0.31 at 95% confidence), while at the same time being closely linked to the PDI (0.64 at 99% confidence). This implies that a weaker wind shear in the southeast (5°–15°N, 165°E–180°) during an El Niño event is particularly favorable for STY development, even though a higher-than-normal value appears in the western region. The pattern is consistent with that in Camargo et al. (2007b) and Kim et al. (2011), who have similarly found a reduction in shear near the date line during El Niño years.
As in Fig. 13, but of the total vertical wind shear.
Citation: Journal of Climate 25, 17; 10.1175/JCLI-D-11-00430.1
Although the time series of PC2 is found to be significantly correlated with the frequency of TYs (−0.46 at 99% confidence), it reveals no significant relationship with the PDI. This suggests that the significant impact of EOF2 on TY frequency is related to factors other than ENSO. This may be due to a weaker ENSO forcing during the transition phase compared with the mature phase, such that other types of climate variability may also play a role in affecting the number of TYs. However, determination of the exact nature of this climate variability is beyond the scope of this study and requires future investigation.
With regard to midlevel vertical motion (Fig. 15), only the EOF2 mode with a southeast–northwest dipole pattern reveals a significant correlation with the frequency of STYs, while EOF1 does not show any significant relationship with any of the TC groups. In addition, the time series of PC2 also correlates positively with the simultaneous PDI (0.71 at 99% confidence). This suggests that an anomalous rising motion (denoted by the negative center) in the southeastern region during an El Niño event favors the formation of STYs. Such a change in vertical motion is also consistent with the strengthening of the monsoon trough, which results in a stronger cyclonic inflow at low levels during an El Niño event, as shown in Fig. 13.
As in Fig. 13, but of the 500-hPa omega.
Citation: Journal of Climate 25, 17; 10.1175/JCLI-D-11-00430.1
The first EOF of 200-hPa divergence reveals a north–south dipole pattern (Fig. 16). Although the time series of PC1 correlates positively (0.71 at 99% confidence) with the PDI, it shows no significant relationship with the frequency of different groups of TCs. This means that the upper-level divergence induced by ENSO plays a relatively minor role compared with other dynamic factors. In contrast, the time series of PC2 shows a significant correlation with TY frequency but reveals no significant relationship with the PDI. Similar to the EOF2 of the VWS, the results here suggest that the change in TY frequency through modulation of the 200-hPa divergence is related to factors other than ENSO.
As in Fig. 13, but of the 200-hPa divergence.
Citation: Journal of Climate 25, 17; 10.1175/JCLI-D-11-00430.1
b. Thermodynamic factors
EOF1 of the 600-hPa relative humidity again shows a southeast–northwest dipole pattern (Fig. 17). The correlations with both STY frequency (0.58 at 99% confidence) and the PDI (0.52 at 99% confidence) are significantly positive. This means that an increase in midlevel relative humidity in the southeast region during an El Niño event is beneficial for the development of STYs. Our results here agree with previous studies (Camargo et al. 2007a; Kim et al. 2011), which reveal a similar increase in humidity near the date line during an El Niño event. On the contrary, EOF2 shows no significant relationship with different TC parameters, though PC2 reveals a significant lag-1 correlation (−0.31 at 90% confidence) with the PDI.
As in Fig. 13, but of the 600-hPa relative humidity.
Citation: Journal of Climate 25, 17; 10.1175/JCLI-D-11-00430.1
Finally, for the MSE (Fig. 18), only EOF2 is found to be related to the number of both STYs (0.55 at 99% confidence) and TSTDs (−0.26 at 90% confidence), while EOF1 is not. Again, EOF2 reveals a southeast–northwest dipole pattern with a significant positive correlation of 0.72 with the PDI, which implies that an increase in MSE in the southeastern region would result in an increase (decrease) in the number of STYs (TSTDs) during an El Niño event. Such a pattern also resembles the EOF2 pattern of MSE obtained by Chan and Liu (2004), which further supports our results.
As in Fig. 13, but of the 1000–500-hPa average moist static energy.
Citation: Journal of Climate 25, 17; 10.1175/JCLI-D-11-00430.1
c. Stepwise regression
6. Summary and discussion
This study examines the interannual variability of three groups of TCs—intense STYs, moderately strong TYs, and weak TSTDs—and their relationship with ENSO. In contrast to many previous studies, we use upper-ocean heat content (OHC) instead of SST as a proxy for the ENSO signal to determine the subsurface oceanic response to both ENSO and TCs.
Both wavelet analysis and correlation studies reveal significant differences among the three groups of TCs. Specifically, the simultaneous correlation of OHC with STYs and TSTDs depicts a significant ENSO-like dipole pattern, while that with TYs does not. In contrast, the lead-1 and lag-1 correlation of OHC with TYs reveals a similar ENSO-like dipole pattern. The results suggest that the timing with which ENSO impacts STYs, TYs, and TSTDs differs and the corresponding changes in their frequencies closely follows the evolution of an ENSO cycle. This is further supported by the composite analysis. The increase (decrease) in STY frequency in more (fewer) STY years is usually associated with the mature phase of El Niño (La Niña) events, while the opposite is found for TSTDs. In contrast, the frequency of TYs shows an increase (decrease) during the transition period from La Niña to El Niño (El Niño to La Niña). All of these findings provide new insight into the ENSO–TC relationship and extend the results of previous studies (Chia and Ropelewski, 2002; Wang and Chan 2002; Kim et al. 2011).
To examine how the frequency of different groups of TCs is related to various types of environmental forcing, EOF analysis is also performed. The EOF patterns of the environmental factors associated with ENSO, including relative vorticity, omega, relative humidity, and MSE, reveal a prominent southeast–northwest dipole pattern. The enhanced low-level cyclonic vorticity, midlevel vertical motion, relative humidity, and MSE as well as weaker VWS in the southeast part of the WNP provide favorable conditions for STY development during El Niño years, while unfavorable conditions in the northwest suppress the formation of TSTDs. These results are consistent with previous studies (Chia and Ropelewski 2002; Wang and Chan 2002; Chan and Liu 2004; Wu et al. 2004; Camargo et al. 2007b) that showed that the strengthening of the monsoon trough during an El Niño year causes a southeastward shift in the TC genesis location, which in turn favors the development of intense TCs. Stepwise regression emphasizes the importance of VWS and MSE (both are correlated significantly with the PDI) in the prediction of the frequency of STYs, while the frequency of TSTDs is found to be related largely to low-level relative vorticity.
On the other hand, TY frequency is found to be related to low-level relative vorticity, vertical wind shear, and upper-level divergence. However, only the change in relative vorticity is found to be related to the transition phase of ENSO, while the change in VWS and divergence might be related to other factors. This may be due to weaker ENSO forcing during the transition phase, which implies that other climate variability might also contribute to the variability of TYs. Using stepwise regression, only relative vorticity and VWS are identified as dominant factors that control TY frequency, highlighting the importance of these two factors in forecasting TYs.
In a recent study, Du et al. (2011) found that VWS increases in the summer following strong El Niño events through the development of a warm Kelvin wave from the tropical Indian Ocean that then suppresses TC formation. Thus, the EOF2 pattern of VWS might be induced by forcing from the Indian Ocean, which requires further investigation. In addition, Zhan et al. (2011) also suggested that the SST anomaly in the east Indian Ocean (EIO) (10°S–22.5°N, 75°–100°E) affects the basinwide genesis of TCs in the WNP [which also included the South China Sea (SCS)]. As a comparison, we also determined whether the frequency of different TC groups is related to the EIO OHC. As shown by the correlation maps of OHC with different TC groups (Figs. 4, 6, and 7), the correlation in the EIO is not prominent in most cases, which indicates that the impact of EIO OHC on TCs is generally weak. The only exception is the simultaneous TY–OHC correlation, with a significant negative value of −0.35 at 95% confidence, suggesting that a warming in the EIO will lead to a decrease in the number of TYs in the WNP. This is consistent with the results of Zhan et al. (2011), although the correlation coefficient in the present study is much smaller [−0.69 in the case of Zhan et al. (2011)]. This discrepancy may be related to the chosen study region. Zhan et al. took into account all TCs that formed within the WNP as well as the SCS, whereas we consider only those formed within the WNP. This suggests that the impact of anomalous warming in the EIO on TC genesis in the WNP and SCS may be different. Thus, the interaction between EIO warming and ENSO on cyclogenesis in the WNP and SCS, especially during the transition phase of ENSO, still requires future investigation. Nevertheless, the results from this study clearly demonstrate that TCs with different intensities are modulated differently by dynamic and thermodynamic factors during the various phases of ENSO.
Acknowledgments
The research described in this paper was supported by 973 Basic Research Program Grant 2009CB421400, National Nature Science foundation of China Grant 41175079, and Hong Kong Croucher Foundation Grant 9220055. The authors are grateful to the two reviewers and to the editor for their constructive comments and suggestions.
REFERENCES
Bell, G. D., and Coauthors, 2000: Climate assessment for 1999. Bull. Amer. Meteor. Soc., 81, S1–S50.
Camargo, S. J., and A. H. Sobel, 2005: Western North Pacific tropical cyclone intensity and ENSO. J. Climate, 18, 2996–3006.
Camargo, S. J., K. A. Emanuel, and A. H. Sobel, 2007a: Use of a genesis potential index to diagnose ENSO effects on tropical cyclone genesis. J. Climate, 20, 4819–4834.
Camargo, S. J., A. W. Robertson, S. J. Gaffney, P. Smyth, and M. Ghil, 2007b: Cluster analysis of typhoon tracks. Part II: Large-scale circulation and ENSO. J. Climate, 20, 3654–3676.
Chan, J. C. L., 2007: Interannual variations of intense typhoon activity. Tellus, 59, 455–460.
Chan, J. C. L., 2008: Decadal variations of intense typhoon occurrence in the western North Pacific. Proc. Roy. Soc. London, 464A, 249–272.
Chan, J. C. L., and K. S. Liu, 2004: Global warming and western North Pacific typhoon activity from an observational perspective. J. Climate, 17, 4590–4602.
Chan, J. C. L., and M. Xu, 2009: Inter-annual and inter-decadal variations of landfalling tropical cyclones in East Asia. Part I: Time-series analysis. Int. J. Climatol., 29, 1285–1293.
Chen, G. H., and R. Huang, 2008: Influence of monsoon over the warm pool on interannual variation on tropical cyclone activity over the western North Pacific. Adv. Atmos. Sci., 25, 319–328.
Chia, H. H., and C. F. Ropelewski, 2002: The interannual variability in the genesis location of tropical cyclones in the northwest Pacific. J. Climate, 15, 2934–2944.
Du, Y., L. Yang, and S.-P. Xie, 2011: Tropical Indian Ocean influence on northwest Pacific tropical cyclones in summer following strong El Niño. J. Climate, 24, 315–322.
Emanuel, K. A., and D. Nolan, 2004: Tropical cyclone activity and the global climate system. Preprints, 26th Conf. on Hurricanes and Tropical Meteorology, Miami, FL, Amer. Meteor. Soc., 10A.2. [Available online at http://ams.confex.com/ams/26HURR/techprogram/paper_75463.htm.]
Frank, W. M., and G. S. Young, 2007: The interannual variability of tropical cyclones. Mon. Wea. Rev., 135, 3587–3598.
Gray, W. M., 1979: Hurricanes: Their formation, structure and likely role in the tropical circulation. Meteorology over the Tropical Oceans, D. B. Shaw, Ed., Royal Meteorological Society, 155–218.
Huang, F., and S. Xu, 2010: Super typhoon activity over the western North Pacific and its relationship with ENSO. J. Ocean Univ. China, 9, 123–128.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437–471.
Kessler, W. S., 1990: Observations of long Rossby waves in the northern tropical Pacific. J. Geophys. Res., 95 (C4), 5183–5217.
Kim, H.-M., P. J. Webster, and J. A. Curry, 2011: Modulation of North Pacific tropical cyclone activity by three phases of ENSO. J. Climate, 24, 1839–1849.
Kinter, J. L., K. Miyakoda, and S. Yang, 2002: Recent change in the connection from the Asian monsoon to ENSO. J. Climate, 15, 1203–1215.
Smith, T. M., and R. W. Reynolds, 2004: Improved extended reconstruction of SST (1854–1997). J. Climate, 17, 2466–2477.
Torrence, C., and G. P. Compo, 1998: A practical guide to wavelet analysis. Bull. Amer. Meteor. Soc., 79, 61–78.
Wada, A., and J. C. L. Chan, 2008: Relationship between typhoon activity and upper ocean heat content. Geophys. Res. Lett., 35, L17603, doi:10.1029/2008GL035129.
Wang, B., and J. C. L. Chan, 2002: How strong ENSO events affect tropical storm activity over the western North Pacific. J. Climate, 15, 1643–1658.
Wang, B., R. Wu, and R. Lukas, 1999: Roles of the western North Pacific wind variation in thermocline adjustment and ENSO phase transition. J. Meteor. Soc. Japan, 77, 1–16.
Wu, M. C., W. L. Chang, and W. M. Leung, 2004: Impacts of El Niño–Southern Oscillation events on tropical cyclone landfalling activity in the western North Pacific. J. Climate, 17, 1419–1428.
Zhan, R., Y. Wang, and X. Lei, 2011: Contributions of ENSO and east Indian Ocean SSTA to the interannual variability of northwest Pacific tropical cyclone frequency. J. Climate, 24, 509–521.
Zhou, W., and J. C. L. Chan, 2007: ENSO and South China Sea summer monsoon onset. Int. J. Climatol., 27, 157–167.
