Abstract

The sea surface temperature gradient (SSTG) between the southwest Pacific (SWP; 40°–20°S, 160°E–170°W) and the western Pacific warm pool (WWP; 0°–16°N, 125°–165°E) in boreal spring was revealed to contribute significantly to the interannual variability of tropical cyclone (TC) genesis frequency over the western North Pacific (WNP) since 1980. The present study extends the analysis back to 1951. Results show that the interannual relationship between the SSTG and WNP TC genesis frequency is statistically significant only after the mid-1970s while insignificant prior to 1974. Further analyses show that the SSTG is strongly negatively correlated with the SST anomaly (SSTA) in the central Pacific (CP) in the recent period while with the SSTA in the equatorial eastern Pacific (EP) in the prior period. In the prior period, the spring positive (negative) EP SSTA induces an anticyclonic (cyclonic) circulation over the WNP. This partly offsets the impact of the SSTG anomaly on the atmospheric circulation over the WNP. In addition, the SSTG anomaly in spring shows poor persistence through summer. As a result, the relationship between the spring SSTG anomaly and WNP TC genesis frequency is largely weakened in the prior period. In the recent period, the response of the atmospheric circulation over the WNP to the CP SSTAs enhances the atmospheric circulation response to the spring SSTG anomaly, thus strengthening the relationship between the SSTG and WNP TC genesis frequency. These results are further confirmed by the atmospheric general circulation model (AGCM) simulations.

1. Introduction

Tropical cyclone (TC) activity over the western North Pacific (WNP) is subject to strong interannual variability. To understand the physical mechanisms that control the interannual variability and their potential changes is critical to skillful seasonal prediction of TC activity. There is an increasing need for accurate seasonal prediction of TC activity for early preparedness to reduce destructive influence of TCs. Therefore, it is of great importance to understand the interannual variability of TC activity and investigate its controlling factors with the ultimate goal to improve seasonal prediction skills of TC activity.

El Niño–Southern Oscillation (ENSO), a prominent interannual signal in the tropics, has been considered as the most crucial factor modulating the interannual variability of TC activity over the WNP (Chan 1985, 2000; Wu and Lau 1992; Wang and Chan 2002; Camargo and Sobel 2005; Chen et al. 2006). Other factors such as the quasi-biennial oscillation (QBO; Camargo and Sobel 2010), the eastern Indian Ocean (EIO) sea surface temperature (SST) anomaly (Du et al. 2011; Zhan et al. 2011a,b), and the spring SST anomaly east of Australia (Zhou and Cui 2010) have all been reported to contribute to the interannual variability of TC activity over the WNP. More recently, Zhan et al. (2013) found that the cross-equatorial SST gradient (SSTG) between the southwest Pacific (SWP; 40°–20°S, 160°E–170°W) east of Australia and the western Pacific warm pool (WWP; 0°–16°N, 125°–165°E) in boreal spring contributes largely to the interannual variability of TC genesis frequency over the WNP in typhoon season (June–October). They showed that the negative SSTG anomaly in boreal spring can induce an anomalous cross-equatorial pressure gradient and tropical westerly anomalies, leading to a strengthened monsoon trough and increased TC genesis frequency over the WNP.

In addition to the interdecadal change of TC activity itself (Zhang et al. 2013; Choi et al. 2015; He et al. 2015), some of the relationships on the interannual time scale as mentioned above are found to have experienced interdecadal or decadal changes as well. For instance, the relationship between the QBO and WNP TC genesis frequency was statistically significant from the 1950s to the mid-1980s but became insignificant thereafter (Camargo and Sobel 2010). Recently, Zhan et al. (2014) revealed that the impact of the EIO SSTA on TC genesis frequency over the WNP documented in Zhan et al. (2011a,b) is statistically significant only after the late 1970s, suggesting the existence of an interdecadal shift. They found that this interdecadal shift of the interannual relationship between the EIO SSTA and WNP TC genesis frequency results mainly from the different covariability of the EIO SSTA with SSTAs over the central and eastern equatorial Pacific and the change in the area coverage of the EIO SSTA before and after 1980. In addition, the interannual relationships of WNP TC activity with the North Atlantic Oscillation (Zhou and Cui 2014), ENSO (Zhao and Wang 2016), and the Arctic Oscillation (Cao et al. 2015) have been revealed to experience significant interdecadal shifts.

It is unknown whether the interannual relationship between WNP TC genesis frequency and the boreal spring SSTG between the SWP and the WWP mentioned above experienced any interdecadal changes. Zhan et al. (2013) demonstrated that the SSTG affects WNP TC genesis frequency initially through inducing atmospheric circulation anomalies over the tropical WNP and equatorial central Pacific in spring and the effect is then maintained and intensified by the equatorial air–sea interaction in the central-western Pacific. Therefore, any changes in the equatorial central Pacific on a decadal time scale might cause changes in the interannual relationship between the SSTG and WNP TC genesis frequency.

The equatorial central Pacific is a key region associated with ENSO. Previous studies (Ashok et al. 2007; Ashok and Yamagata 2009) have shown that there are two types of El Niño—that is, the eastern Pacific (EP) El Niño (or the canonical El Niño) and the central Pacific (CP) El Niño (or Modoki). It is evident that the CP El Niño has become more dominant than the EP El Niño in the last decade or so. It is also shown that the two types of El Niño can lead to different spatial and temporal characteristics of the atmospheric circulation response, affecting WNP TC activity in different ways (Chan 2008; Kao and Yu 2009; Kim et al. 2009; Chen 2011; Lin et al. 2015). For example, the CP El Niño (EP El Niño) often induces a large-scale anomalous cyclonic (anticyclonic) circulation in the main TC genesis region over the WNP (Chen and Tam 2010; Yuan and Yang 2012; Lin et al. 2015). Therefore, the atmospheric circulation response over the WNP to the SSTG anomaly may be enhanced by the CP El Niño or suppressed by the EP El Niño. It is still an issue whether the increasing frequency of the CP El Niño in the last decade or so has contributed to the significant interannual relationship between the SSTG and WNP TC genesis frequency as documented in Zhan et al. (2013).

The main objectives of this study are to document the decadal change of the interannual relationship between the SSTG and TC genesis frequency over the WNP and to investigate the physical mechanisms that are responsible for such a change. We will show that this interannual relationship experienced a decadal shift around the mid-1970s from insignificant prior to the mid-1970s to significant after the mid-1970s. This change is mainly attributed to the increase in the CP-type El Niño in the recent period and the increasing persistence of the SSTG anomaly from boreal spring through summer. The rest of the paper is organized as follows. Section 2 presents the datasets and methodology used in this study. The decadal shift of the interannual relationship between WNP TC genesis frequency and the SSTG is statistically analyzed based on observational and reanalysis data in section 3. The possible physical mechanisms revealed based on both data analysis and numerical model simulations are discussed in section 4. Finally, main conclusions are given in section 5.

2. Data and methodology

The annual TC numbers over the WNP from 1951 to 2013 are calculated based on the 6-hourly best-track TC data obtained from the Shanghai Typhoon Institute of the China Meteorological Administration (CMA; Ying et al. 2014) and the Joint Typhoon Warning Center (JTWC). In the present study, only those TCs reaching the maximum sustained surface wind speed equal to or larger than 17 m s−1 are considered. Similar to Zhan et al. (2013), we focus only on TCs in the typhoon season (1 June–31 October of every year) over the WNP, which extends from 105°E to the date line and thus includes the South China Sea (SCS). It should be mentioned that there is an issue on the reliability of the best-track TC data before the mid-1960s because no meteorological satellite was available to monitor TCs over the open oceans. To confirm the robustness of our results, we repeated all of our analyses discussed in this paper but based on the best-track TC data for the period of 1961–2013 and got very similar results. Therefore, our following discussions will be based on results from the analyses using a relatively longer dataset back to 1951. This allows us to better examine the statistical significance.

The 1951–2013 monthly mean SST data with 2.0° horizontal resolution from National Oceanic and Atmospheric Administration (NOAA) (ERSST.v3; Smith and Reynolds 2004) are used to analyze the influence of SST on the atmospheric circulation over the WNP. The SST data on the 2.0° latitude by 2.0° longitude grids from the Hadley Centre Sea Ice and Sea Surface Temperature dataset (HadISST; Rayner et al. 2003) are used to confirm the robustness of the results shown in Fig. 1. The monthly mean atmospheric reanalysis data, including sea level pressure (SLP), vertical pressure velocity at 500 hPa, and wind fields at 200 and 850 hPa from 1951–2013, are taken from the National Centers for Environmental Prediction–National Center for Atmosphere Research (NCEP–NCAR) reanalysis with a spatial resolution of 2.5° longitude × 2.5° latitude (Kalnay et al. 1996). The monthly mean 1958–2001 ERA-40 data from the European Centre for Medium-Range Weather Forecasts (ECMWF; Simmons and Gibson 2000) with 2.5° horizontal resolution are used to verify the results obtained based on the NCEP–NCAR reanalysis data.

Fig. 1.

The 20-yr running correlations of the spring SSTG derived from the ERSST and HadISST data and the WNP TC frequency in the typhoon season (June–October) derived from the CMA and JTWC best-track data during 1951–2013 based on (a) original and (b) filtered time series using the LHPF. The horizontal gray line shows the 95% significance level. The vertical dashed gray lines show the years 1974 and 1979.

Fig. 1.

The 20-yr running correlations of the spring SSTG derived from the ERSST and HadISST data and the WNP TC frequency in the typhoon season (June–October) derived from the CMA and JTWC best-track data during 1951–2013 based on (a) original and (b) filtered time series using the LHPF. The horizontal gray line shows the 95% significance level. The vertical dashed gray lines show the years 1974 and 1979.

Following Zhan et al. (2013), the SSTG is defined as the difference in SSTAs between the SWP (40°–20°S, 160°E–170°W) and the WWP (0°–16°N, 125°–165°E) in boreal spring. The negative SSTG anomaly means that the difference of SWP and WWP SST (SWP minus WWP) is lower than the climatological SSTG, and vice versa. The EP SSTA index in spring is defined as the SSTA averaged over the EP (10°S–10°N, 160°–80°W), and the CP SSTA index in the typhoon season is defined as the SSTA averaged over the CP (10°S–10°N, 180°–120°W) according to the regression pattern (see Figs. 6a,d, which will be discussed later). In addition, we also used the Niño-3 index as the EP SSTA index and the ENSO Modoki index (Ashok et al. 2007) as the CP SSTA index and got similar results (not shown).

To isolate the effect of the SSTG from that of the ENSO signal on the large-scale atmospheric circulation, the method developed by Ashok et al. (2003) is employed:

 
formula

where Iorg stands for the original time series of ISSTG, Iυ stands for the EP or CP SSTA index, denotes the normalized Iυ, denotes the standard deviation of the original SSTG index, and r(Iorg, Iυ) denotes the correlation coefficient between the Iorg and Iυ. So the ISSTG is the remainder of Iorg with the signal Iυ removed.

The model employed to conduct sensitivity numerical experiments is the ECHAM4.8 atmospheric general circulation model (AGCM) developed at the Max Planck Institute (MPI; Roeckner et al. 1996). The AGCM is run at spectral T42 horizontal resolution with 19 layers in the vertical. The basic prognostic variables of the model are represented in the horizontal by truncated series of spherical harmonics and in the finite differencing in the vertical in the hybrid sigma–pressure coordinate. A semi-implicit time-differencing scheme is used for model time integration. The model physics include the surface turbulent fluxes calculated based on the Monin–Obukhov similarity theory, the horizontal diffusion in a hyper-Laplacian form, the ECMWF longwave and shortwave radiation schemes, and the convective parameterization developed by Tiedtke (1989) and later modified by Nordeng (1994).

3. Statistical analysis of the decadal shift

To examine the possible decadal change of the interannual relationship between TC genesis frequency in the typhoon season (1 June–31 October) over the WNP and the boreal spring SSTG (March–May), we first calculated the 20-yr running correlation between them during 1951–2013. To avoid the interference of the low-frequency variability, the Lanczos high-pass filter (LHPF; period ≥ 10 yr retained) is used to keep the interannual variability only in the time series for a comparison. As we can see from Fig. 1, the correlations from both the original (Fig. 1a) and the filtered (Fig. 1b) time series show a significant decadal shift around 1974. Prior to 1974, the correlation is statistically insignificant, while it becomes significant over the 95% confidence level after 1975. This indicates that the relationship between the SSTG and WNP TC genesis frequency is statistically significant only after 1975. Considering the possible dependence of the running correlation on the length of the sliding window, we also calculated the correlations using the 19-, 21-, and 23-yr sliding windows and got results similar to those using the 20-yr sliding window. Note that sliding windows (longer than 25 yr) are not relevant because we focus on possible decadal change not multidecadal change. Similar correlation calculations are repeated using the two best-track TC datasets (CMA and JTWC) and two different SST datasets (ERSST and HadISST); the results remain consistent, indicating that the decadal shift is robust. Since the results based on the two SST datasets are similar, only results based on the ERSST dataset are discussed below.

To understand the decadal shift of the interannual relationship between WNP TC genesis frequency and the spring SSTG, we divided the whole time series into two periods: 1951–74 (the prior period) and 1979–2013 (the recent period). The prior period contains 24 years while the recent period contains 35 years. Note that we chose 1979 as the starting year for the second period and considered the four years of 1975–78 as the transition year. In addition, we objectively examined the transition year using the standard normal homogeneity test (SNHT; Alexandersson 1986). A shift occurring in the late 1970s can be detected by the SNHT method as well (not shown), suggesting that the division of the two periods is reasonable.

To investigate different characteristics of the environmental conditions that determine the different SSTG influences on WNP TC genesis in the two periods identified above, we first compare the anomalies of the TC genesis potential index (GPI) in the different boreal spring SSTG anomalies in the two periods based on regression analysis, as shown in Fig. 2. Here GPI is calculated according to Emanuel and Nolan (2004). The GPI anomalies regressed onto the spring SSTG are quite different in the two periods. Overall, the GPI anomalies in response to the spring SSTG are insignificant in the prior period, but significant positive values appear north of 20°N in the WNP and south of 15°N east of the Philippines in the recent period. This indicates that the significant correlation between the spring SSTG and TC genesis frequency over the WNP is expected in the recent period but not in the prior period.

Fig. 2.

Regressions of the TC GPI in the typhoon season with respect to the spring SSTG during (a) 1951–74 and (b) 1979–2013. The shading in blue (red) indicates areas where the negative (positive) differences are statistically significant at the 95% confidence level by the Student’s t test.

Fig. 2.

Regressions of the TC GPI in the typhoon season with respect to the spring SSTG during (a) 1951–74 and (b) 1979–2013. The shading in blue (red) indicates areas where the negative (positive) differences are statistically significant at the 95% confidence level by the Student’s t test.

We further examine the differences in the atmospheric circulation regressed onto the boreal spring SSTG between the prior and recent periods. Figure 3 displays the mean SLP, 850-hPa winds, and 500-hPa vertical pressure velocity fields in the typhoon season (June–October) regressed upon the boreal spring SSTG index in the two periods. In the prior period (Fig. 3a), the SLP field shows a northwest–southeast-oriented large negative anomaly distribution elongated from the SWP to the EIO and very weak (insignificant) positive anomalies in the WNP main TC genesis region during a positive spring SSTG anomaly. In the recent period (Fig. 3b), the regressed SLP field shows a much stronger correlation with the spring SSTG anomaly, with significant positive anomalies over the WNP and with negative anomalies in the SWP slightly reduced, leading to a prominent north–south SLP gradient across the equator in the central-western Pacific. The low-level easterly anomalies related to the positive SSTG anomaly are much weaker and restricted in the lower latitudes in the prior period (Fig. 3c) than in the recent period (Fig. 3d). In particular, a well-organized low-level anomalous anticyclonic circulation related to the positive SSTG anomaly appears in the main TC genesis region over the WNP in the recent period (Fig. 3d), consistent with the positive SLP anomalies (Fig. 3b). Corresponding to the changes in the SLP and low-level wind fields, the vertical motion in the midtroposphere shows a region with overall subsidence anomalies in the main TC genesis region over the WNP in the recent period (Fig. 3f) while the change in vertical motion is insignificant in the prior period (Fig. 3e).

Fig. 3.

Regression fields of the seasonal mean (a),(b) SLP (hPa); (c),(d) 850-hPa winds (m s−1); and (e),(f) 500-hPa vertical velocity (ω; Pa s−1) in the typhoon season based on the NCEP–NCAR reanalysis data with respect to the spring SSTG for (a),(c),(e) 1951–74 and (b),(d),(f) 1979–2013. Areas where the positive (negative) difference is statistically significant at the 95% confidence level based on the Student’s t test are shaded in red (blue). In (a) and (b), the solid boxes indicate the SLP gradient, and in (c) and (d), the solid (dashed) box indicates the area for the cross-equatorial flow index (the equatorial westerly index). These refer to regions that will be discussed in Fig. 5.

Fig. 3.

Regression fields of the seasonal mean (a),(b) SLP (hPa); (c),(d) 850-hPa winds (m s−1); and (e),(f) 500-hPa vertical velocity (ω; Pa s−1) in the typhoon season based on the NCEP–NCAR reanalysis data with respect to the spring SSTG for (a),(c),(e) 1951–74 and (b),(d),(f) 1979–2013. Areas where the positive (negative) difference is statistically significant at the 95% confidence level based on the Student’s t test are shaded in red (blue). In (a) and (b), the solid boxes indicate the SLP gradient, and in (c) and (d), the solid (dashed) box indicates the area for the cross-equatorial flow index (the equatorial westerly index). These refer to regions that will be discussed in Fig. 5.

The above results indicate that the atmospheric circulation anomaly over the WNP in response to the positive (negative) boreal spring SSTG anomaly is unfavorable (favorable) for TC genesis over the WNP and the effect is much stronger in the recent period than in the prior period. The results for the recent period are consistent with those documented in Zhan et al. (2013) based on data in the period of 1980–2011. To examine whether the regressed results discussed above are independent of the data used, we repeated the analysis using the ERA-40 data. Although the ERA-40 data have a shorter record for both periods than the NCEP–NCAR reanalysis data, the regressed fields (Fig. 4) are quite similar to those based on the NCEP–NCAR reanalysis data (Fig. 3). In addition, we also analyzed other reanalysis datasets including Japanese 55-year Reanalysis Project (JRA-55) and ECMWF twentieth-century reanalysis (ERA-20C). Results from all these datasets are very consistent with each other (not shown). This confirms that results from the regression analysis discussed above are independent of the dataset used.

Fig. 4.

As in Fig. 3, but based on the ERA-40 data for (a),(c),(e) 1958–74 and (b),(d),(f) 1979–2001.

Fig. 4.

As in Fig. 3, but based on the ERA-40 data for (a),(c),(e) 1958–74 and (b),(d),(f) 1979–2001.

To examine how the environmental parameters evolve related to the spring SSTG, following Zhan et al. (2013) we calculated the lagged correlations of three parameters (from April to October) with the boreal spring SSTG based on a brief inspection of Figs. 3 and 4. Figure 5 shows the lagged correlations between the spring SSTG and monthly mean cross-equatorial SLP gradient, defined as the difference between SLP averaged over 10°–25°N, 105°–165°E and that over 25°–10°S, 105°–165°E, low-level equatorial westerly anomalies averaged over 5°S–5°N, 140°E–180°, and low-level cross-equatorial flow averaged over 5°S–0°, 100°–130°E. Overall the SSTG anomaly in spring shows poor persistence through summer in the prior period (Fig. 5a) and good persistence in the recent period with the self-correlation coefficient by one month significant over the 95% confidence level (Fig. 5e). The SLP gradient is positively correlated with the spring SSTG from April to October (Figs. 5b,f). The correlation is statistically significant over the 95% confidence level throughout 7 months in the recent period (Fig. 5f), while it is weak and unstable in the prior period (Fig. 5b). The negative correlation between the low-level equatorial zonal wind and the spring SSTG is significant from May through October in the recent period while only significant in August and October in the prior period. In sharp contrast, the low-level cross-equatorial flow is negatively correlated with the spring SSTG in both the prior and recent periods except for in April in the prior period.

Fig. 5.

Lagged correlations between the spring SSTG and monthly mean (a),(e) SSTG; (b),(f) SLP gradient (SLPG) between 10°–25°N, 105°–165°E and 25°–10°S, 105°–165°E; (c),(g) equatorial westerly (EW) averaged over 5°S–5°N, 140°E–180°; and (d),(h) cross-equatorial flow (CEF) averaged over 5°S–0°, 100°–130°E in (a)–(d) the prior period and (e)–(h) the recent period. The dashed transverse line indicates correlations significant at the 95% confidence level.

Fig. 5.

Lagged correlations between the spring SSTG and monthly mean (a),(e) SSTG; (b),(f) SLP gradient (SLPG) between 10°–25°N, 105°–165°E and 25°–10°S, 105°–165°E; (c),(g) equatorial westerly (EW) averaged over 5°S–5°N, 140°E–180°; and (d),(h) cross-equatorial flow (CEF) averaged over 5°S–0°, 100°–130°E in (a)–(d) the prior period and (e)–(h) the recent period. The dashed transverse line indicates correlations significant at the 95% confidence level.

The above lagged correlation analysis suggests that the immediate response to the positive spring SSTG anomaly is the anomalous low-level northerly cross-equatorial flow over the western equatorial Pacific (between 100° and 130°E). This response is similar in the prior and recent periods. Distinct difference in the anomalies of the low-level equatorial zonal wind related to the spring SSTG anomaly occurs between the two periods. Persistent low-level anomalous equatorial easterly remains from May through October related to the positive spring SSTG anomaly in the recent period, while this anomaly is only statistically significant in both August and October in the prior period. Since the equatorial zonal wind in the tropical western Pacific affects TC genesis over the WNP significantly (Zhan et al. 2011a, 2013, 2014), it is not surprising that the spring SSTG has a prolonged and significant effect on TC genesis frequency over the WNP in the recent period. Both the low-level anomalous equatorial zonal wind and the anomalous cross-equatorial SLP gradient in response to the positive SSTG anomaly maintain the anomalous anticyclonic circulation over the main TC genesis region over the WNP in the recent period. The weak anomaly in the low-level equatorial zonal wind related to the spring SSTG in the prior period seems to explain the insignificant correlation between the SSTG and WNP TC genesis frequency in the prior period. The issue of why the low-level equatorial wind anomaly related to the spring SSTG is so different will be discussed in the next section.

Zhan et al. (2013) suggested that in spring, the positive SSTG anomaly induces a cross-equatorial SLP gradient and equatorial easterly anomalies. The latter trigger the equatorial upwelling and cold SSTA in the central Pacific, increasing the zonal SST gradient along the equator and thus strengthening the easterly anomalies in the typhoon season. This is a positive feedback via air–sea interaction (see their schematic diagram in Fig. 5). It is thus a natural extension of the above regression analysis to examine how the global SST is correlated with the spring SSTG. Figure 6 shows the regressed SSTA fields in spring and the typhoon season, onto the spring SSTG. The spring SSTG is highly correlated with SSTA over the SWP and the equatorial EP in spring in the prior period (Fig. 6a), while its correlation with the spring EP SST disappears in the recent period (Fig. 6b). These results suggest that the development of the positive SSTG anomaly and the equatorial EP cooling occur in spring simultaneously in the prior period while they are almost independent in the recent period. In the typhoon season, the correlation between the spring SSTG and SSTA over the SWP remains high in the recent period, while it considerably weakens in the prior period. This suggests that the SWP SSTA in spring can persist through the typhoon season in the recent period while its persistence is quite weak in the prior period. The correlation between the spring SSTG and the EP SSTA in the typhoon season is significantly negative in the prior period (Fig. 6c). In the recent period, a significant negative correlation region appears in the CP (Fig. 6d). This negative correlation is explained partially as a result of the equatorial upwelling in the CP in response to the persistent equatorial easterly wind anomaly in the western Pacific (Fig. 5e) by Zhan et al. (2013).

Fig. 6.

Regressions of global SSTAs (°C) in (a),(b) spring and (c),(d) the typhoon season with respect to the spring SSTG during (a),(c) 1951–74 and (b),(d) 1979–2013. Shading in blue (red) indicates areas where the negative (positive) differences are statistically significant at the 95% confidence level by the Student’s t test. In (a), the solid (dashed) box indicates the area of the SSTG (EP), and in (d), the dashed box indicates the CP. These refer to regions with imposed SSTA in numerical experiments that are discussed in section 4b.

Fig. 6.

Regressions of global SSTAs (°C) in (a),(b) spring and (c),(d) the typhoon season with respect to the spring SSTG during (a),(c) 1951–74 and (b),(d) 1979–2013. Shading in blue (red) indicates areas where the negative (positive) differences are statistically significant at the 95% confidence level by the Student’s t test. In (a), the solid (dashed) box indicates the area of the SSTG (EP), and in (d), the dashed box indicates the CP. These refer to regions with imposed SSTA in numerical experiments that are discussed in section 4b.

The above analyses demonstrate that the decadal shift of the interannual relationship between WNP TC genesis frequency and the spring SSTG is closely related to different responses of the large-scale air–sea conditions to the spring SSTG anomaly in the two periods. In the prior period, the positive spring SSTG anomaly is accompanied by cooling in the EP in spring and strengthening through the typhoon season. The cooling in the EP largely offsets the anomaly in the atmospheric circulation related to the SSTG anomaly in spring, which otherwise works as a trigger for the atmospheric circulation anomaly over the WNP in the typhoon season (Wang and Chan 2002; Chen and Tam 2010). The strengthening of EP cooling through the typhoon season may play a role in suppressing the persistence of the SSTG through summer. In the recent period, in sharp contrast, the positive spring SSTG anomaly is followed by cooling in the CP in the typhoon season, a result of equatorial air–sea interaction, as previously revealed by Zhan et al. (2013). Since the CP El Niño becomes more frequent after the late 1970s, this suggests that the decadal shift of the interannual relationship between the spring SSTG and WNP TC genesis frequency may result from the change of the ENSO pattern from EP El Niño to CP El Niño. We thus hypothesize that the concurrent EP cooling in spring and its subsequent strengthening in the typhoon season suppressed the negative effect of the positive spring SSTG anomaly and its persistence in the prior period, while the effect of the positive spring SSTG anomaly on WNP TC genesis is largely enhanced by cooling in the CP in the recent period. This hypothesis will be further analyzed in detail in the next section through both data analysis and AGCM simulations.

4. Possible physical mechanisms

In the previous section, we have revealed the existence of an interdecadal change of the interannual relationship between the boreal spring SSTG and WNP TC genesis frequency around 1974 and examined the different anomalies in TC GPI and the large-scale atmospheric circulation over the WNP related to the boreal spring SSTG in the prior and recent periods. To understand the physical mechanisms responsible for the above interdecadal change, results from various composite analyses and regression analysis with the signal removal algorithm described in section 2 and results from AGCM sensitivity numerical experiments will be discussed in this section.

a. Composite and regression analyses

Previous studies (Wolter and Timlin 1993; Barnston and Chelliah 1997; Ashok et al. 2007) have defined various indices for ENSO (El Niño and La Niña) events (e.g., Niño-3, Niño-4, Niño-3.4, the multivariate ENSO index, and ENSO Modoki index). Here we simply define the EP events based on the SSTAs averaged over 10°S–10°N, 160°–80°W and the CP events based on the SSTAs averaged over 10°S–10°N, 180°–120°W, respectively, according to the regression pattern shown in Fig. 6. As shown in section 3, the SSTG index is well correlated with the EP SSTA in the prior period and the CP SSTA in the recent period in the typhoon season. We first focus on the issue whether the EP (CP) SSTAs defined in the present study have any distinct impacts on the anomalies in the large-scale atmospheric circulation over the WNP related to the spring SSTG. To facilitate composite analysis, we define an EP (CP) event if the absolute SSTA in the EP (CP) is larger than 80% of its standard deviation. Note that to avoid any coexistence of the strong SSTA anomalies in both the EP and CP associated with strong events, those with both EP and CP events (1982, 1987, 1997, and 2009) are excluded in our composite analysis below. Based on these criteria, we found six EP El Niño events (1951, 1957, 1963, 1965, 1969, and 1972) in the prior period and four CP El Niño events (1991, 1994, 2002, and 2004) in the recent period. In the following discussion, we compose the low-level wind fields in all El Niño events, respectively, in the prior and recent periods to reveal the differences in the atmospheric circulation over the WNP in response to different distributions in SSTAs in the two periods.

Consistent with previous studies (Ashok et al. 2007; Chen and Tam 2010), in EP El Niño years in the prior period, positive SSTAs occupy the whole tropical central-eastern Pacific with the maximum in the EP (Fig. 7a). Westerly anomalies at 850 hPa occur in the equatorial western-central Pacific, and a cyclonic–anticyclonic circulation couplet appears over the WNP, with the anticyclonic circulation centered at around 30°N, 145°E and the cyclonic circulation located east of Philippines over the western WNP (Fig. 7c). This pattern is consistent with that in the GPI anomalies with an increase in the southeast WNP and a decrease in the northwest WNP. As indicated in previous studies (e.g., Wang and Chan 2002), the cyclonic circulation east of the Philippines to the date line favors the eastward shift of the TC genesis region in the WNP. In the recent period, the center of the positive SSTAs shifted from the equatorial EP to the equatorial CP (Fig. 7b), and the corresponding wind anomalies over the equatorial Pacific and the WNP show considerable differences from the EP events. The CP El Niño corresponds to low-level equatorial westerly anomalies, a large-scale cyclonic circulation anomaly, and significant positive anomalies in the GPI over the WNP in the recent period (Fig. 7d), favorable for TC genesis over the WNP. This enhances the effect of the spring SSTG on WNP TC genesis frequency as discussed earlier (Fig. 3).

Fig. 7.

Composite (a),(b) SSTAs (contours; °C) and (c),(d) 850-hPa wind anomalies (vectors; m s−1) in the typhoon season for (a),(c) EP El Niño events in the prior period and (b),(d) CP El Niño events in the recent period. The red (blue) shading indicates areas where the positive (negative) values are statistically significant at the 95% confidence level by the Student’s t test.

Fig. 7.

Composite (a),(b) SSTAs (contours; °C) and (c),(d) 850-hPa wind anomalies (vectors; m s−1) in the typhoon season for (a),(c) EP El Niño events in the prior period and (b),(d) CP El Niño events in the recent period. The red (blue) shading indicates areas where the positive (negative) values are statistically significant at the 95% confidence level by the Student’s t test.

We also show in section 3 that in spring high correlation exists only between the SSTG index and the EP SSTA in the prior period (Fig. 6a), while the anomaly in the atmospheric circulation over the WNP related to the spring SSTG is weak. To examine whether the EP SSTA in spring suppresses the response of the atmospheric circulation over the WNP to the SSTG, we compare in Fig. 8 the regressed SLP and 850-hPa wind fields in spring onto the spring SSTG (Figs. 8a,c) and the corresponding regression with the EP SST signal removed using the algorithm described in section 2 (Figs. 8b,d) in the prior period. In spring, the anomalies in SLP and 850-hPa winds in response to the SSTG anomaly are quite weak over the WNP. However, with the EP SSTA signal removed, the response is significantly enhanced with a positive SLP anomaly and low-level anomalous anticyclonic circulation over the WNP and stronger anomalous equatorial zonal flow in the western Pacific. This suggests that the effect of the spring SSTG on the large-scale atmospheric circulation over the WNP is partly suppressed by the EP SSTA in spring in the prior period. This explains the weak anomalies in the equatorial zonal wind and the cross-equatorial SLP gradient in response to the spring SSTG anomaly in the prior period as shown in Figs. 5b,c. Since the EP SSTA strengthens through the typhoon season, the effect of the spring SSTG is further suppressed by the EP SSTA in the following typhoon season. This explains why the interannual correlation between the spring SSTG and WNP TC genesis frequency is insignificant in the prior period.

Fig. 8.

Regressed (a) SLP (hPa) and (c) 850-hPa winds (m s−1) in spring with respect to the spring SSTG during 1951–74. (b),(d) As in (a),(c) but with the EP SSTA signal removed. The red (blue) shading indicates areas where the positive (negative) differences are statistically significant at the 95% confidence level by the Student’s t test.

Fig. 8.

Regressed (a) SLP (hPa) and (c) 850-hPa winds (m s−1) in spring with respect to the spring SSTG during 1951–74. (b),(d) As in (a),(c) but with the EP SSTA signal removed. The red (blue) shading indicates areas where the positive (negative) differences are statistically significant at the 95% confidence level by the Student’s t test.

As demonstrated by Zhan et al. (2013), the CP SSTA in the typhoon season, initiated by the equatorial air–sea interaction in spring, plays an important role in enhancing the effect of the SSTG on WNP TC genesis in the typhoon season. This can be clearly seen from Fig. 9, which compares the regressed SLP and 850-hPa wind fields onto the SSTG in the typhoon season in the recent period (Figs. 9a,c) and the corresponding results with the CP SSTA signal removed (Figs. 9b,d). With the removal of the CP SST signal the anomaly in the large-scale circulation over the WNP is still significant but substantially weaker than that with the CP SST signal included. This demonstrates that the impact of the SSTG on the atmospheric circulation and thus WNP TC genesis is substantially enhanced by the CP SSTA in the typhoon season, which in turn is a result of the equatorial air–sea interaction triggered by the SSTG in spring.

Fig. 9.

Regressed (a) SLP (hPa) and (c) 850-hPa winds (m s−1) in the typhoon season with respect to the spring SSTG during 1979–2013. (b),(d) As in (a),(c) but with the CP SSTA signal removed. The red (blue) shading indicates areas where the positive (negative) differences are statistically significant at the 95% confidence level by the Student’s t test.

Fig. 9.

Regressed (a) SLP (hPa) and (c) 850-hPa winds (m s−1) in the typhoon season with respect to the spring SSTG during 1979–2013. (b),(d) As in (a),(c) but with the CP SSTA signal removed. The red (blue) shading indicates areas where the positive (negative) differences are statistically significant at the 95% confidence level by the Student’s t test.

Since the EP and CP SSTAs play significant but different roles in modulating the impact of the spring SSTG on the large-scale atmospheric circulation and TC genesis over the WNP, it is interesting to examine how the monthly mean SSTA regressed onto the spring SSTG evolves in the prior and recent periods (note that the spring and the typhoon season means have already been given in Fig. 6). The left column in Fig. 10 shows the regressed monthly SSTA onto the spring SSTG in the prior period. The SSTA in the equatorial EP develops and strengthens from March to May after the prominent positive SSTG anomaly occurs in March. From May to September, although the SSTG anomaly weakens rapidly (because of the weakening of SSTA over the SWP), the EP SSTA continues intensifying. This suggests that the SSTG anomaly established in early spring might contribute to the development of the EP El Niño–La Niña events in the prior period, and then the impact of the SSTG on the atmospheric circulation and TC genesis over the WNP is greatly suppressed by the EP SSTA in the typhoon season. This is a topic for a future study. In the recent period, the SSTG anomaly in March is dominated by the SSTA over the SWP with no evident SSTA signal over the central Pacific. The CP SSTA develops after May (Fig. 10, right). Since the CP SSTA plays a role in enhancing the impact of the SSTG, the development and strengthening through the typhoon season in the CP SSTA explains the significant impact of the boreal spring SSTG on the atmospheric circulation and TC genesis over the WNP in the recent period, as documented in Zhan et al. (2013). These results demonstrate that the different responses of the large-scale atmospheric circulation over the WNP to the EP and CP SSTAs explain well the interdecadal shift of the interannual relationship between WNP TC genesis frequency and the spring SSTG.

Fig. 10.

Regressed monthly mean SSTAs (°C) with respect to the spring SSTG during (left) 1951–74 and (right) 1979–2013. Here, only regression fields in March, May, July, and September are displayed because of limited space. The red (blue) shading indicates areas where the positive (negative) differences are statistically significant at the 95% confidence level by the Student’s t test.

Fig. 10.

Regressed monthly mean SSTAs (°C) with respect to the spring SSTG during (left) 1951–74 and (right) 1979–2013. Here, only regression fields in March, May, July, and September are displayed because of limited space. The red (blue) shading indicates areas where the positive (negative) differences are statistically significant at the 95% confidence level by the Student’s t test.

Another remarkable feature in Fig. 10 is the difference in the persistence of the SWP SSTA in the two periods. Consistent with Figs. 5a,e, in the recent period significant correlations between the spring SSTG and the SWP SSTA are kept through the whole typhoon season. However, in the prior period the correlation of the spring SSTG with May SSTA in the SWP is slightly weaker than that in the recent period and becomes much weaker in July. The poor persistence of the SWP SSTA might also contribute to the insignificant impact of the spring SSTG on TC genesis frequency over the WNP in the prior period, while the strong persistence contributes positively to the impact in the recent period. This point will be further confirmed with the results from AGCM simulations in the next subsection. The enhanced persistence of the SWP SSTA for the recent period may be due to 1) the intensified thermocline feedback and/or 2) the strengthened large-scale connection in the tropical Indo-Pacific regions as shown in Figs. 6c,d. The former could be inferred from a deepening trend in thermocline averaged in the SWP (not shown). An in-depth analysis of the reason for the enhanced persistence of the SWP SSTA is beyond the scope of this work and reserved for a future study.

b. Results from AGCM simulations

To further confirm the findings from data analysis discussed above, we performed a series of numerical experiments using the ECHAM4.8 AGCM as briefly described in section 2. Two sets of experiments (Table 1) were performed with the observed climatological monthly mean SST in the prior and recent periods, respectively, and with imposed SSTA in key regions (over either EP, CP, WWP, or SWP). In the first group, the CTR_1 run was performed with the observed climatological monthly mean SST in the prior period (1951–74). The sensitivity experiment SSTG+_1 (SSTG−_1) is identical to the CTR_1 run except that 1°C was added to (subtracted from) SST in the SWP region (40°–20°S, 160°E–170°W) and 0.35°C was subtracted from (added to) SST in the WWP (0°–16°N, 125°–165°E) from March to May according to the regression pattern in Fig. 6a. The difference between SSTG+_1 and SSTG−_1 can help evaluate the response of the atmospheric circulation to the spring SSTG anomaly. Experiment EP− (EP+) is identical to CTR_1 but with 2°C subtracted from (added to) SST in the EP region (10°S–10°N, 160°–80°W) to isolate the response of the atmospheric circulation over the WNP to the spring EP SSTA. Experiment SSTG+_EP− (SSTG−_EP+) is identical to the SSTG+_1 (SSTG−_1) except that 2°C was subtracted from (added to) SST in the EP region from March to May. Another experiment SWP+_1 (SWP−_1) is identical to the CTR_1 run but with 1°C added to (subtracted from) SST in the SWP region from June to October. This allows us to address whether the SWP SSTA in the typhoon season could continuously affect the atmospheric circulation over the WNP without the effect of the EP SSTA. In the second group of experiments, the CTR_2 run was performed with the observed climatological monthly mean SST in the recent period (1979–2013). Experiment SSTG+_2 (SSTG−_2) is similar to experiment SSTG+_1 (SSTG−_1) but uses the climatological mean SST in the recent period. The experiment CP− (CP+) is identical to CTR_2 but with 2°C subtracted from (added to) SST in the CP region (10°S–10°N, 180°–120°W) from June to October to isolate the response of the atmospheric circulation over the WNP to the CP SSTA. Experiment SWP+_2 (SWP−_2) is identical to CTR_2 but with 1°C added to (subtracted from) SST in the SWP region from June to October to confirm the effect of the persistence of the spring SSTG (via SWP SSTA in the typhoon season) on the atmospheric circulation over the WNP in the recent period. Note that the SWP SSTA in the typhoon season is a vital signal to maintain the effect of the spring SSTG on the subsequent atmospheric circulation over the WNP while the WWP SSTA in the typhoon season might result from TC activity and could not impose a significant impact on local TC activity. Therefore, the SWP–CP SSTA relationship will be examined in the last experiment to discuss the role of the CP SSTA during the typhoon season in enhancing the effect of the spring SSTG on the atmospheric circulation over the WNP in the typhoon season. The last experiment SWP+_CP− (SWP−_CP+) is identical to experiment SWP+_2 (SWP−_2) except that 2°C was subtracted from (added to) SST in the CP region from June to October. The model was integrated for 30 years for each experiment. The results of the first five years are considered as the model spinup and are excluded from the following discussions. It should be mentioned that both spatial pattern and magnitude of the SST anomalies in the sensitivity experiments are selected based on observed SST anomalies. For example, since the standard deviation of EP and CP SSTA is about twice as large as that of SWP SSTA, we subtracted (or added) 2°C in the EP and CP region in our relevant sensitivity experiments. We also performed the sensitivity experiments with 1°C for the EP and CP region. Although the atmospheric circulation response becomes weaker and also changes in some details, the overall circulation pattern response is still similar (not shown).

Table 1.

Experimental design of the AGCM simulations.

Experimental design of the AGCM simulations.
Experimental design of the AGCM simulations.

Figure 11 shows the results from the first group of experiments. The difference between experiments with the positive and negative spring SSTG anomalies shows a coherent positive SLP anomaly (Fig. 11a) and a low-level anomalous anticyclonic circulation (Fig. 11b) over the WNP in spring, very similar to those based on the regression analysis shown in Figs. 8b,d. This indicates that the model can reproduce the observed response of the atmospheric circulation over the WNP to the spring SSTG anomaly. The cold EP SSTA, as inferred from the difference between experiments EP− and EP+, induces a large-scale negative SLP anomaly (Fig. 11c) and a low-level anomalous cyclonic circulation (Fig. 11d) over the WNP. Note that the circulation response over the WNP to the EP SSTA in spring is slightly different from that in the typhoon season (Fig. 7c). Comparing Figs. 11c,d with Figs. 11a,b, we can see that the EP SSTA plays a role in suppressing the response of the atmospheric circulation over the WNP to the SSTG anomaly. This can be seen clearly from the difference fields between experiments SSTG+_EP− and SSTG−_EP+ shown in Figs. 11e,f. Now the original positive SLP anomaly induced by the SSTG anomaly is replaced by the negative SLP anomaly, and the low-level anomalous anticyclonic circulation is replaced by an anomalous cyclonic circulation, very similar to those induced by the negative EP SSTA shown in Figs. 11c,d. The imposed positive SSTA over the SWP in the typhoon season in experiment SWP maintains and enhances the SSTG anomaly, followed by a similar atmospheric circulation response over the WNP to the spring SSTG, with positive SLP anomaly and an anomalous anticyclonic circulation over the WNP (Figs. 11g,h). These results strongly suggest that the poor persistence of the SWP SSTA also contributes to the insignificant impact of the spring SSTG on TC genesis frequency over the WNP in the prior period.

Fig. 11.

The simulated differences in (left) SLP (hPa) and (right) 850-hPa winds (m s−1) in the prior period between (a),(b) the SSTG+_1 and SSTG−_1 runs; (c),(d) the EP− and EP+ runs; (e),(f) the SSTG+_EP− and SSTG−_EP+ runs; and (g),(h) the SWP+_1 and SWP−_1 runs. The differences in (a)–(f) are averaged in spring and those in (g),(h) are averaged in the typhoon season.

Fig. 11.

The simulated differences in (left) SLP (hPa) and (right) 850-hPa winds (m s−1) in the prior period between (a),(b) the SSTG+_1 and SSTG−_1 runs; (c),(d) the EP− and EP+ runs; (e),(f) the SSTG+_EP− and SSTG−_EP+ runs; and (g),(h) the SWP+_1 and SWP−_1 runs. The differences in (a)–(f) are averaged in spring and those in (g),(h) are averaged in the typhoon season.

In the recent period, the spring SSTG anomaly induces positive SLP anomalies over the WNP, equatorial easterly anomalies over the tropical western Pacific and SCS in spring (Figs. 12a,b), similar to those experiments for the prior period (Figs. 11a,b). The negative CP SSTA in the typhoon season, as inferred from the difference between experiments CP− and CP+, induces strong equatorial easterly anomalies over the central-western Pacific, positive SLP anomalies over the WNP and negative SLP anomalies over the Maritime Continent (Figs. 12c,d). More interestingly, the negative CP SSTA in the typhoon season induces an anomalous anticyclonic circulation over the WNP, instead of an anomalous cyclonic circulation induced by the negative EP SSTA in the prior period (Figs. 11c,d). As a result, the CP SSTA in the typhoon season plays a significant role in enhancing the atmospheric circulation anomaly over the WNP induced by the spring SSTG in the recent period. This is further confirmed by results from the difference fields between experiments SWP+_CP− and SWP−_CP+ shown in Figs. 12g,h. With either the negative CP SSTA or the positive SWP SSTA (Figs. 12e,f), the anomalous positive SLP and cyclonic circulation in the typhoon season are relatively weak. However, with both the SWP and CP SSTAs, the anomalous positive SLP and equatorial easterlies are remarkably enhanced. Note that although the model produces the overall features that can help isolate the effects of the SWP and CP SSTAs on the WNP circulation in the typhoon season, the simulations show a westward shift of the anticyclonic circulation anomaly over the WNP. Nevertheless, these numerical results confirm that the CP SSTA in the typhoon season, triggered by the spring SSTG and amplified by the equatorial air–sea interaction (Zhan et al. 2013), enhances the impact of the spring SSTG on WNP TC genesis frequency in the recent period. In addition, results from the SWP experiments in the recent period also demonstrate that the persistence of the SWP SSTA contributes positively to the intensified impact of the spring SSTG on the atmospheric circulation and thus TC genesis over the WNP in the recent period.

Fig. 12.

The simulated differences in (left) SLP (hPa) and (right) 850 hPa winds (m s−1) in the recent period between (a),(b) the SSTG+_1 and SSTG−_1 runs; (c),(d) the CP− and CP+ runs; (e),(f) the SWP+_1 and SWP−_1 runs; and (g),(h) the SWP+_CP− and SWP−_CP+ runs. The differences in (a),(b) are averaged in spring and those in (c)–(h) are averaged in the typhoon season.

Fig. 12.

The simulated differences in (left) SLP (hPa) and (right) 850 hPa winds (m s−1) in the recent period between (a),(b) the SSTG+_1 and SSTG−_1 runs; (c),(d) the CP− and CP+ runs; (e),(f) the SWP+_1 and SWP−_1 runs; and (g),(h) the SWP+_CP− and SWP−_CP+ runs. The differences in (a),(b) are averaged in spring and those in (c)–(h) are averaged in the typhoon season.

5. Conclusions and discussion

This study has documented the decadal shift of the interannual relationship between TC genesis frequency over the WNP and the spring SSTG between the SWP and the WWP and revealed the possible physical mechanisms based on data analyses and AGCM numerical experiments. Results from data analyses showed that the previously revealed interannual relationship between the spring SSTG and WNP TC genesis frequency (Zhan et al. 2013) is statically significant only after 1974 while insignificant prior to 1974. It is found that this decadal shift results mainly from the change of the SSTA pattern over the CP and EP and the change in the persistence of the SWP SSTA from spring through the typhoon season.

In the prior period, the response of the atmospheric circulation to the spring SSTG anomaly is weak, with the spring SSTG being well correlated with the EP SSTA from spring through the typhoon season. In the recent period, however, the SSTA pattern in the equatorial Pacific in association with the spring SSTG anomaly changes from the EP to the CP. Results from regression analysis with the EP SST signal removed in the prior period demonstrated that the spring EP SSTA suppresses the atmospheric circulation response to the spring SSTG anomaly and thus TC genesis frequency over the WNP. In addition, the SSTG anomaly weakens considerably and shows poor persistence through summer. As a result, the SSTG displays insignificant correlation with WNP TC genesis frequency in the prior period. In contrast to the EP SSTA, the CP SSTA in the typhoon season enhances the response of the atmospheric circulation anomalies over the WNP to the spring SSTG anomaly and thus amplifies the relationship between the spring SSTG and WNP TC genesis frequency in the recent period. Therefore, the change in the SSTA pattern from the EP to the CP in the equatorial Pacific is most likely responsible for the decadal shift of the interannual relationship between the spring SSTG and WNP TC genesis frequency. A series of numerical experiments were performed using an AGCM to examine the different responses of the atmospheric circulation to various SSTAs, including the spring SSTG anomaly, SSTAs in the EP, CP, and SWP, in the prior and recent periods. Results from the numerical experiments generally confirm the findings from data analyses.

The results from both data analysis and AGCM numerical experiments can be schematically depicted in Fig. 13. The negative (positive) spring SSTG anomaly induces equatorial easterly (westerly) anomalies over the central-western Pacific and a low-level anomalous anticyclonic (cyclonic) circulation over the WNP, suppressing (enhancing) WNP TC genesis, as revealed by Zhan et al. (2013). The negative (positive) EP SSTA in both spring and the typhoon season induces an anomalous cyclonic (anticyclonic) circulation over the WNP that largely offsets the anomalous anticyclonic (cyclonic) circulation in response to the positive (negative) spring SSTG anomaly (Fig. 13a). As a result, the EP SSTA pattern in the prior period offsets the impact of the spring SSTG anomaly on WNP TC genesis, leading to the statistically insignificant correlation between the spring SSTG and TC genesis frequency over the WNP in the prior period. The poor persistence of the SSTG anomaly from spring through the typhoon season in the prior period also contributes to the observed insignificant correlation.

Fig. 13.

A schematic diagram showing the physical mechanism for (a) the suppressed impact of the spring SSTG on the WNP TC genesis in the prior period and (b) the intensified impact in the recent period. See text for details.

Fig. 13.

A schematic diagram showing the physical mechanism for (a) the suppressed impact of the spring SSTG on the WNP TC genesis in the prior period and (b) the intensified impact in the recent period. See text for details.

In the recent period, the spring SSTG anomaly is well correlated with the CP SSTA. The positive (negative) spring SSTG anomaly is accompanied by the development of a negative (positive) SSTA over the CP in the typhoon season. The negative (positive) CP SSTA in the typhoon season induces the equatorial easterly (westerly) anomalies over the central-western Pacific and an anomalous anticyclonic (cyclonic) circulation over the WNP, enhancing the response of the atmospheric circulation over the WNP to the spring positive (negative) SSTG anomaly (Fig. 13b). As a result, the relationship between the spring SSTG anomaly and the WNP atmospheric circulation is substantially enhanced by the CP SSTA in the following typhoon season in the recent period, leading to a significant interannual relationship between the spring SSTG and TC genesis frequency over the WNP.

We noticed that the EIO SSTA is not correlated with the spring SSTG in the prior period, while it shows an increasing trend to correlate positively with the spring SSTG in the recent period (Figs. 6 and 10). Since the positive SSTA over the EIO also suppresses WNP TC genesis (Zhan et al. 2011a,b), the EIO SSTA may also play some roles in enhancing the impact of the spring SSTG in the recent period. In addition, the relatively more rapid warming of the EIO than other tropical oceans could be another candidate that may have contributed to the amplified impact of the spring SSTG on TC genesis over the WNP in recent period. We performed three numerical experiments using the ECHAM model as described in section 2. The control (CTR) experiment was run with the observed climatological monthly mean SST in the recent period (1979–2013). Experiment SSTG+ is identical to the CTR run except that 0.7°C was added to SST in the SWP region and 0.3°C was subtracted from SST in the WWP from March to May and 0.5°C was added to SST in the SWP region from June to October. The experiment SSTG+_EIO+ is identical to SSTG+ but with 0.5°C added to SST in the EIO region (10°S–22.5°N, 75°–100°E) from March to October. The results show that the anticyclonic circulation over the WNP associated with the SSTG was intensified when 0.5°C was added to SST in the EIO region (not shown). An important issue is whether the change in the relationship documented in this study may be due to the consistent variations of SST in the EIO and the SWP. To address this issue, we compared the regressed 850-hPa wind fields in the typhoon season onto the spring SSTG and the corresponding regression with the EIO SST signal removed using the algorithm described in section 2 in the recent period. With the removal of the EIO SST signal, the response of the large-scale circulation over the WNP to the spring SSTG is still very significant and similar to that with the EIO SST signal included except that the magnitude in the former is slightly weaker than that in the latter. This suggests that the EIO SSTA plays a role in enhancing the relationship between the spring SSTG and WNP TC genesis frequency in the recent period, but such a role is secondary. The possible impact of the EIO warming on the WNP TC activity is currently under investigation, and the results will be reported in due course.

It should be mentioned that only AGCM was used to verify the physical mechanisms proposed in this study. Some uncertainties could exist regarding how the large-scale atmospheric circulation responds to the SST forcing without air–sea interaction involved. However, it should be feasible to use AGCM since we mainly verify the prompt responses of the atmospheric circulation to the SSTAs in different key regions. In future studies, coupled climate models or TC-permitting coupled climate models will be used to better address relevant issues.

Previous studies have also suggested that the Pacific decadal oscillation (PDO) has an impact on TC activity over the WNP with fewer TCs in its positive phase and more TCs in its negative phase (Leung et al. 2005; Goh and Chan 2010). We calculated the correlation between TC genesis frequency and the SSTG in different phases of the PDO. According to the time series of the PDO index, the PDO shows two cold phases, 1951–78 (period I) and 1998–2013 (period III), and a warm phase, 1979–97 (period II) in the studied period. The correlations are significant in the recent two periods but insignificant in the first period. We further composited the differences of SSTAs over the Pacific in the typhoon season among the three phases. The results show that the difference in SSTA in the SWP is insignificant between the first two periods but significant between the recent two periods. This suggests that the SWP SSTA seems to be associated with the PDO in period III. The significant warming in the SWP may contribute to the enhanced persistence of the SWP SSTA through the typhoon season and thus the intensified relationship between the spring SSTG anomaly and WNP TC genesis.

Finally, in this study, we have not examined why the SSTA pattern changed around 1974. McGregor et al. (2014) revealed that the North Atlantic warming in the recent period led to the strengthening of the Walker circulation and the cooling in the EP, which might be responsible for the westward shift of the ENSO-related SSTA from the EP to the CP. Verifying this possibility is beyond the scope of this paper and will be a topic for a future study.

Acknowledgments

This study has been supported by the National Basic Research Program 973 (Grant 2012CB956003) and the China NSFC Grants 41375093 and 41375098. Yuqing Wang acknowledges the financial support by USGS Grant G12AC20501 to University of Hawai‘i at Mānoa.

REFERENCES

REFERENCES
Alexandersson
,
H.
,
1986
:
A homogeneity test applied to precipitation data
.
J. Climatol.
,
6
,
661
675
, doi:.
Ashok
,
K.
, and
T.
Yamagata
,
2009
:
Climate change: The El Niño with a difference
.
Nature
,
461
,
481
484
, doi:.
Ashok
,
K.
,
Z.
Guan
, and
T. A.
Yamagata
,
2003
:
A look at the relationship between the ENSO and the Indian Ocean dipole
.
J. Meteor. Soc. Japan
,
81
,
41
56
, doi:.
Ashok
,
K.
,
K.
Behera
,
S. A.
Rao
,
H.
Weng
, and
T.
Yamagata
,
2007
:
El Niño Modoki and its possible teleconnection
.
J. Geophys. Res.
,
112
,
C11007
, doi:.
Barnston
,
A. G.
, and
M.
Chelliah
,
1997
:
Documentation of a highly ENSO-related SST region in the equatorial Pacific
.
Atmos.–Ocean
,
35
,
367
383
, doi:.
Camargo
,
S. J.
, and
A. H.
Sobel
,
2005
:
Western North Pacific tropical cyclone intensity and ENSO
.
J. Climate
,
18
,
2996
3006
, doi:.
Camargo
,
S. J.
, and
A. H.
Sobel
,
2010
:
Revisiting the influence of the quasi-biennial oscillation on tropical cyclone activity
.
J. Climate
,
23
,
5810
5825
, doi:.
Cao
,
X.
,
S. F.
Chen
,
G. H.
Chen
,
W.
Chen
, and
R. G.
Wu
,
2015
:
On the weakened relationship between spring Arctic Oscillation and following summer tropical cyclone frequency over the western North Pacific: A comparison between 1968–1986 and 1989–2007
.
Adv. Atmos. Sci.
,
32
,
1319
1328
, doi:.
Chan
,
J. C. L.
,
1985
:
Tropical cyclone activity in the northwest Pacific in relation to the El Niño/Southern Oscillation phenomenon
.
Mon. Wea. Rev.
,
113
,
599
606
, doi:.
Chan
,
J. C. L.
,
2000
:
Tropical cyclone activity over the western North Pacific associated with El Niño and La Niña events
.
J. Climate
,
13
,
2960
2972
, doi:.
Chan
,
J. C. L.
,
2008
:
Decadal variations of intense typhoon occurrence in the western North Pacific
.
Philos. Trans. Roy. Soc. London
,
464A
,
249
272
, doi:.
Chen
,
G. H.
,
2011
:
How does shifting Pacific Ocean warming modulate on tropical cyclone frequency over the South China Sea?
J. Climate
,
24
,
4695
4700
, doi:.
Chen
,
G. H.
, and
C. Y.
Tam
,
2010
:
Different impacts of two kinds of Pacific Ocean warming on tropical cyclone frequency over the western North Pacific
.
Geophys. Res. Lett.
,
37
,
L01803
, doi:.
Chen
,
T. C.
,
S. Y.
Wang
, and
M. C.
Yen
,
2006
:
Interannual variation of the tropical cyclone activity over the western North Pacific
.
J. Climate
,
19
,
5709
5720
, doi:.
Choi
,
Y.
,
K. J.
Ha
,
C. H.
Ho
, and
C. E.
Chung
,
2015
:
Interdecadal change in typhoon genesis condition over the western North Pacific
.
Climate Dyn.
,
45
,
3243
3255
, doi:.
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
, doi:.
Emanuel
,
K. A.
, and
D. S.
Nolan
,
2004
: Tropical cyclone activity and global climate. Preprints, 26th Conf. on Hurricanes and Tropical Meteorology, Miami, FL, Amer. Meteor. Soc.,
240
241
.
Goh
,
A. Z. C.
, and
J. C. L.
Chan
,
2010
:
Interannual and interdecadal variations of tropical cyclone activity in the South China Sea
.
Int. J. Climatol.
,
30
,
827
843
, doi:.
He
,
H.
,
J.
Yang
,
D.
Gong
,
R.
Mao
,
Y.
Wang
, and
M.
Gao
,
2015
:
Decadal changes in tropical cyclone activity over the western North Pacific in the late 1990s
.
Climate Dyn.
,
45
,
3317
3329
, doi:.
Kalnay
,
E.
, and Coauthors
,
1996
:
The NCEP/NCAR 40-Year Reanalysis Project
.
Bull. Amer. Meteor. Soc.
,
77
,
437
471
, doi:.
Kao
,
H. Y.
, and
J. Y.
Yu
,
2009
:
Contrasting eastern‐Pacific and central‐Pacific types of El Niño
.
J. Climate
,
22
,
615
632
, doi:.
Kim
,
H. M.
,
P. J.
Webster
, and
J. A.
Curry
,
2009
:
Impact of shifting patterns of Pacific Ocean warming on North Atlantic tropical cyclones
.
Science
,
325
,
77
80
, doi:.
Leung
,
Y.-K.
,
M.-C.
Wu
, and
W.-L.
Chang
,
2005
: Variations of tropical cyclone activity in the South China Sea. Hong Kong Observatory Reprint 675, 12 pp.
Lin
,
C.-C.
,
Y.-J.
Liou
, and
S.-J.
Huang
,
2015
:
Impacts of two-type ENSO on rainfall over Taiwan
.
Adv. Meteor.
,
2015
, 658347, doi:.
McGregor
,
S.
,
A.
Timmerman
,
M. F.
Stuecker
,
M. H.
England
,
M.
Merrifield
, and
F.-F.
Jin
,
2014
:
Recent Walker circulation strengthening and Pacific cooling amplified by Atlantic warming
.
Nat. Climate Change
,
4
,
888
892
, doi:.
Nordeng
,
T. E.
,
1994
: Extended versions of the convective parameterization scheme at ECMWF and their impact on the mean and transient activity of the model in the tropics. ECMWF Research Department Tech. Memo., 41 pp.
Rayner
,
N. A.
,
D. E.
Parker
,
E. B.
Horton
,
C. K.
Folland
,
L. V.
Alexander
,
D. P.
Rowell
,
E. C.
Kent
, and
A.
Kaplan
,
2003
:
Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century
.
J. Geophys. Res.
,
108
,
4407
, doi:.
Roeckner
,
E. K.
, and Coauthors
,
1996
: The atmospheric general circulation model ECHAM-4: Model description and simulation of present-day climate. Max Planck Institute Rep. 218, 90 pp.
Simmons
,
A. J.
, and
J. K.
Gibson
,
2000
: The ERA-40 project plan. ECMWF ERA-40 Project Rep. Series 1, 63 pp.
Smith
,
T. M.
, and
R. W.
Reynolds
,
2004
:
Improved extended reconstruction of SST (1854–1997)
.
J. Climate
,
17
,
2466
2477
, doi:.
Tiedtke
,
M.
,
1989
:
A comprehensive mass flux scheme for cumulus parameterization in large-scale models
.
Mon. Wea. Rev.
,
117
,
1779
1800
, doi:.
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
, doi:.
Wolter
,
K.
, and
M. S.
Timlin
,
1993
: Monitoring ENSO in COADS with a seasonally adjusted principal component index. Proc. 17th Climate Diagnostics Workshop, Norman, OK, NOAA/National Meteorological Center/Climate Analyses Center,
52
57
.
Wu
,
G.
, and
N. C.
Lau
,
1992
:
A GCM simulation of the relationship between tropical-storm formation and ENSO
.
Mon. Wea. Rev.
,
120
,
958
977
, doi:.
Ying
,
M.
,
W.
Zhang
,
H.
Yu
,
X.
Lu
,
J.
Feng
,
Y.
Fan
,
Y.
Zhu
, and
D.
Chen
,
2014
:
An overview of the China Meteorological Administration tropical cyclone database
.
J. Atmos. Oceanic Technol.
,
31
,
287
301
, doi:.
Yuan
,
Y.
, and
S.
Yang
,
2012
:
Impacts of different types of El Niño on the East Asian climate: Focus on ENSO cycles
.
J. Climate
,
25
,
7702
7722
, doi:.
Zhan
,
R.
,
Y.
Wang
, and
X.
Lei
,
2011a
:
Contributions of ENSO and East Indian Ocean SSTA to the interannual variability of northwest Pacific tropical cyclone frequency
.
J. Climate
,
24
,
509
521
, doi:.
Zhan
,
R.
,
Y.
Wang
, and
C.-C.
Wu
,
2011b
:
Impact of SSTA in the East Indian Ocean on the frequency of northwest Pacific tropical cyclones: A regional atmospheric model study
.
J. Climate
,
24
,
6227
6242
, doi:.
Zhan
,
R.
,
Y.
Wang
, and
W.
Min
,
2013
:
The SST gradient between the southwestern Pacific and the western Pacific warm pool: A new factor controlling the northwestern Pacific tropical cyclone genesis frequency
.
J. Climate
,
26
,
2408
2415
, doi:.
Zhan
,
R.
,
Y.
Wang
, and
L.
Tao
,
2014
:
Intensified impact of East Indian Ocean SST anomaly on tropical cyclone genesis frequency over the western North Pacific
.
J. Climate
,
27
,
8724
8739
, doi:.
Zhang
,
W.
,
Y.
Leung
, and
J. Z.
Min
,
2013
:
North Pacific gyre oscillation and the occurrence of western North Pacific tropical cyclones
.
Geophys. Res. Lett.
,
40
,
5205
5211
, doi:.
Zhao
,
H.-K.
, and
C.-Z.
Wang
,
2016
:
Interdecadal modulation on the relationship between ENSO and typhoon activity during the late season in the western North Pacific
.
Climate Dyn.
, doi:, in press.
Zhou
,
B.-T.
, and
X.-P.
Cui
,
2010
:
Sea surface temperature east of Australia: A predictor of tropical cyclone frequency over the western North Pacific?
Chin. Sci. Bull.
,
55
,
196
201
, doi:.
Zhou
,
B.-T.
, and
X.-P.
Cui
,
2014
:
Interdecadal change of the linkage between the North Atlantic Oscillation and the tropical cyclone frequency over the western North Pacific
.
Sci. China
,
57D
,
2148
2155
, doi:.