Abstract

The poleward migration of the annual mean location of tropical cyclone (TC) lifetime maximum intensity (LMI) has been identified in the major TC basins of the globe over the past 30 years, which is particularly robust over the western North Pacific (WNP). This study has revealed that this poleward migration consists mainly of weak TCs (with maximum sustained surface wind speed less than 33 m s−1) over the WNP. Results show that the location of LMI of weak TCs has migrated about 1° latitude poleward per decade since 1980, while such a trend is considerably smaller for intense TCs. This is found to be linked to a significant decreasing trend of TC genesis in the southern WNP and a significant increasing trend in the northwestern WNP over the past 30 years. It is shown that the greater sea surface temperature (SST) warming at higher latitudes associated with global warming and its associated changes in the large-scale circulation favor more TCs to form in the northern WNP and fewer but stronger TCs to form in the southern WNP.

1. Introduction

The western North Pacific (WNP) is the most active ocean basin for tropical cyclones (TCs), where about one-third of TCs worldwide form and often cause serious property damage and loss of life. Supertyphoon Haiyan in 2013 was one such example, which devastated the Philippines with more than one million houses destroyed and 6300 people killed (NDRRMC 2014). Therefore, understanding the long-term changes of TC activity over the WNP is particularly important.

Many previous studies have shown that the total number of annual TC geneses over the WNP has experienced a downward trend in recent decades, which is closely linked to the significant decrease of TC genesis frequency over the southeastern WNP (Liu and Chan 2013; He et al. 2015; Choi et al. 2016). It is also evident that TCs in category 4 and 5 of the Saffir–Simpson scale over the WNP have almost doubled in number (Webster et al. 2005), and the strong TCs are getting stronger (Elsner et al. 2008; Kang and Elsner 2016). Using the power dissipation index (PDI) as a metric of TC activity, Emanuel (2005) found that the PDI over the Atlantic and the WNP has increased remarkably since the mid-1970s, which is attributed to increasing trends in both duration and peak intensity of TCs. Besides TC frequency and intensity, the prevailing TC tracks over the WNP have shown a westward shift, imposing an increasing influence over subtropical East Asia and decreasing influence over the South China Sea (Wu et al. 2005; He et al. 2015; Mei and Xie 2016). However, whether the trends in TC activity in the past decades are significant is still under debate due to the uncertainties in TC data quality (Kossin et al. 2013) and the short data record (Klotzbach 2006). For example, Klotzbach and Landsea (2015) found that by adding an additional 10 years of data to the analysis of Webster et al. (2005) the upward trend in category 4 and 5 TC numbers was flattened.

More recently, Kossin et al. (2014) proposed a new TC metric: the annual mean latitudinal location where TCs reach their lifetime maximum intensity (LMI), which is comparatively insensitive to some of the past data uncertainties. They revealed a significant poleward migration of the mean latitudinal location of the LMI in most of ocean basins over the past 30 years. This poleward migration is attributed to a poleward shift of large-scale vertical wind shear (VWS) and maximum potential intensity (MPI), which in turn are plausibly associated with tropical expansion (Lucas et al. 2014). In a follow-up study, Kossin et al. (2016) have further demonstrated that TCs over the WNP have the largest contribution to this global trend. The poleward migration of the mean location of LMI over the WNP is consistent with changes in TC exposure (track and frequency of occurrence) and is most likely to continue in the twenty-first century. Previous studies have also shown that the activities of weak and intense TCs exhibit very different trends (Yoshimura et al. 2006; Oouchi et al. 2006). This has been considered to be closely linked to different mechanisms affecting TC genesis and intensity under global climate change (Kang and Elsner 2016).

It is our interest in this study to examine contributions to the poleward shift in the mean location of LMI by weak and intense TCs over the WNP. Here the category of “intense TCs” defines those TCs with maximum near-surface wind speeds of 33 m s−1 or greater, which includes typhoons, severe typhoons, and supertyphoons in the basin. Correspondingly, the category of “weak TCs” defines those TCs with maximum near-surface winds less than 33 m s−1, which includes tropical storms and severe tropical storms over the WNP. We will show that the poleward migration of the mean latitudinal location of LMI over the WNP is largely contributed by weak TCs. The rest of the paper is organized as follows. Section 2 describes the data and methodology. Contributions by weak and intense TCs to the poleward migration of the mean location of TC LMI over the WNP are examined in section 3. The relevant changes in the large-scale dynamical and thermodynamic environments are discussed in section 4. Finally, the main conclusions together with a discussion are given in section 5.

2. Data and methodology

The TC best-track data used in this study were obtained from the Joint Typhoon Warning Center (JTWC). Considering the fact that TC intensity might differ in the datasets from different agencies, the best-track datasets from the Japan Meteorological Agency (JMA) and the Shanghai Typhoon Institute of the China Meteorological Administration (CMA; Ying et al. 2014) were used to confirm the robustness of the results. Since different TC intensity definitions are used among these best-track datasets, the reported maximum sustained near-surface wind speeds were first converted to meters per second in 10-min average, similar to that done in Park et al. (2011). TCs in this study are systems with at least tropical storm (TS) intensity (with maximum sustained 10-m wind speed Vmax ≥ 17 m s−1) between 100°E and 180° over the WNP. Note that this is slightly different from the definition used in Kossin et al. (2016) in which TCs with LMI ≥ 35 kt (1 kt ≈ 0.51 m s−1) were used. We will show in section 3 that the results from this study are in general consistent with those reported in Kossin et al. (2016). As mentioned in section 1, we classified TCs into weak TCs with 17 ≤ Vmax < 33 m s−1 and intense TCs with Vmax ≥ 33 m s−1. There are two reasons for choosing 33 m s−1 as the split between weak and intense classification of TCs: 1) 33 m s−1 is very close to the 50th percentile (36 m s−1) of all TC intensity, and 2) a TC case exceeding 33 m s−1 is generally classified as a typhoon in the TC best-track data from the JMA, which includes typhoon, severe typhoon, and supertyphoon. The sample sizes of weak and intense TCs are comparable (Table 1), with the sample size ratio of weak TCs to intense TCs ranging from 0.7 to 0.8 in the three TC best-track datasets, suggesting that the classification of weak and intense TCs is meaningful for statistical analysis. We focus our study on the period of 1980–2016. Because the TC best-track data for 2016 are not available yet, the TC data of 2016 were taken from the real-time operational analyses in our following analyses.

Table 1.

Linear trends (° lat decade−1) and p values in the annual mean ϕLMI for all TCs, weak TCs (WTCs), and intense TCs (ITCs) over the WNP from three TC best-track datasets and their ensemble mean (ENS) during 1980–2013 (T1) and 1980–2016 (T2). Values with an asterisk (in boldface) denote trends that are statistically significant at the 95% (90%) confidence level using the F test. Numbers in the parentheses indicate the sample size.

Linear trends (° lat decade−1) and p values in the annual mean ϕLMI for all TCs, weak TCs (WTCs), and intense TCs (ITCs) over the WNP from three TC best-track datasets and their ensemble mean (ENS) during 1980–2013 (T1) and 1980–2016 (T2). Values with an asterisk (in boldface) denote trends that are statistically significant at the 95% (90%) confidence level using the F test. Numbers in the parentheses indicate the sample size.
Linear trends (° lat decade−1) and p values in the annual mean ϕLMI for all TCs, weak TCs (WTCs), and intense TCs (ITCs) over the WNP from three TC best-track datasets and their ensemble mean (ENS) during 1980–2013 (T1) and 1980–2016 (T2). Values with an asterisk (in boldface) denote trends that are statistically significant at the 95% (90%) confidence level using the F test. Numbers in the parentheses indicate the sample size.

The monthly sea surface temperature (SST) data were obtained from the Met Office Hadley Centre (Rayner et al. 2003). The monthly mean atmospheric data were obtained from the Interim European Centre for Medium-Range Weather Forecasts (ECMWF) interim reanalysis (ERA-Interim; Dee et al. 2011), which has a horizontal resolution of 2.5° latitude × 2.5° longitude. Following Kossin et al. (2016), seasonal averages of the atmospheric and oceanic fields were taken over the typhoon season of the WNP, namely July–November (JASON).

3. Statistical analysis

Figure 1 shows trends in the annual mean latitudinal location of TC LMI ϕLMI over the WNP by quantile from 0.2 to 0.8 using the JTWC best-track data. The solid (dashed) curve is the trend based on TCs with intensity below (above) the corresponding percentile. For example, the 50th percentile in the solid curve indicates TCs with intensity from 17 to 36 m s−1 (the 50th percentile) and that in the dashed curve indicates TCs with intensity greater than 36 m s−1. For the solid curve, trends at all quantiles are statistically significant at the 95% confidence level but smaller, while still positive, at the higher quantiles, with the largest trend noted for the lowest quantile (the 20th percentile). This suggests that the poleward trends in the annual mean ϕLMI are getting smaller when the stronger TCs are included in the statistics.

Fig. 1.

Trends in the annual mean ϕLMI over the WNP by quantile from 0.2 to 0.8 using JTWC best-track data. The solid (dashed) line is the trend based on TCs with intensity below (above) the corresponding percentile. The solid dots indicate the trends are significant at the 95% confidence level based on the F test.

Fig. 1.

Trends in the annual mean ϕLMI over the WNP by quantile from 0.2 to 0.8 using JTWC best-track data. The solid (dashed) line is the trend based on TCs with intensity below (above) the corresponding percentile. The solid dots indicate the trends are significant at the 95% confidence level based on the F test.

For the dashed curve, the poleward trends at all quantiles are also seen but are considerably smaller, with the lowest trend at the 50th percentile. It seems that weak TCs contribute predominantly to the poleward migration of the annual mean ϕLMI over the WNP as previously documented in Kossin et al. (2014, 2016). It is thus interesting to examine the poleward trends of the annual mean ϕLMI for weak and intense TCs (see section 2), separately.

Figure 2 presents time series of the annual mean ϕLMI for, all TCs, weak TCs, and intense TCs over the WNP based on the JTWC best-track dataset during 1980–2016. The corresponding time series from the CMA and JMA best-track datasets, and an ensemble mean of the three datasets, are also shown in Fig. 2 for a comparison. The corresponding trends during 1980–2013 were also calculated and listed in Table 1 for a direct comparison with the results of Kossin et al. (2016). All time series for all TCs based on three individual datasets and their ensemble mean show an upward trend (Fig. 2a). During 1980–2013, the trends ranged from 0.47° to 0.68° latitude per decade (lat decade−1), all statistically significant at the 95% confidence level, with the p values varying between 0.009 and 0.040, less than 0.05 (Table 1). Such a significant poleward trend is consistent with the results reported in Kossin et al. (2014, 2016) except for some slight differences due to the different TC definitions as mentioned in section 2. We also calculated the trends using the same TC definition (with sustained near-surface maximum wind speeds ≥35 kt) as used in Kossin et al. (2016) and got almost the same trend values as given in Table 1 (not shown).

Fig. 2.

Time series of the annual mean ϕLMI (° lat) for (a) all TCs, (b) WTCs, and (c) ITCs over the WNP from three TC best-track datasets and their ensemble mean during 1980–2016. The thin black line is the trend based on the ensemble mean.

Fig. 2.

Time series of the annual mean ϕLMI (° lat) for (a) all TCs, (b) WTCs, and (c) ITCs over the WNP from three TC best-track datasets and their ensemble mean during 1980–2016. The thin black line is the trend based on the ensemble mean.

Similar to that during 1980–2013, the annual mean ϕLMI of all TCs during 1980–2016 shows an increasing trend but with a relatively smaller value (Table 1), which might be related to the prominent influence of the strong El Niño event during 2014/15. Nevertheless, the positive trends ranged from 0.34° to 0.58° lat decade−1, which are all statistically significant at the 95% confidence level except for those based on the JMA best-track dataset (Table 1). As implied by Fig. 1, the poleward trends in the annual mean ϕLMI are more evident for weak TCs but less pronounced for intense TCs (Fig. 2 and Table 1). All datasets present nearly uniform and robust positive trends in the annual mean latitudinal location of LMI for weak TCs. The trends are all statistically significant at the 90% confidence level during both 1980–2013 and 1980–2016, including those in the JMA best-track dataset. The upward trends for weak TCs ranged from 0.65° to 0.94° lat decade−1 during 1980–2016, which are much larger than those for all TCs. However, the annual mean latitudinal location of LMI for intense TCs shows generally much smaller upward trends than those for either all TCs or weak TCs. Changes in the p values are similar to those in trends. Since the three TC best-track datasets show consistent results, we will focus on our analysis using the JTWC TC best-track dataset in the next section.

4. Possible mechanisms

Results from the statistical analyses discussed in section 3 suggest that weak TCs contributed predominantly to the poleward migration of the annual mean location of TC LMI over the WNP. To understand possible mechanisms responsible for the significant poleward migration in the annual mean ϕLMI of weak TCs, we will examine changes in TC genesis location and track that could substantially affect the annual mean ϕLMI and the associated changes in the large-scale environment over the WNP in this section.

a. Changes in TC genesis location and track

Figure 3 shows time series of the annual mean genesis latitude ϕgene, the difference between ϕLMI and ϕgene, and the genesis frequency for all TCs, weak TCs, and intense TCs over the WNP during 1980–2016 based on the JTWC best-track dataset. Similar to the annual mean ϕLMI in Fig. 2, the annual mean ϕgene shows a poleward migration trend for all TCs, weak TCs, and intense TCs. The trends are statistically significant at the 95% confidence level for all TCs and weak TCs, but do not reach the 90% confidence level for intense TCs (Table 2). The poleward trend for weak TCs is larger than that for all TCs, suggesting that the trend for all TCs is dominated by the trend for weak TCs. More interestingly, the poleward trends in the annual mean ϕgene for all TCs and weak TCs (Table 2) are very close to those in the annual mean ϕLMI (Table 1). This indicates that the poleward migration in the annual mean ϕgene is a major factor controlling the poleward trend in the annual mean ϕLMI. Consistently, the difference between ϕLMI and ϕgene (Fig. 3b) shows a nearly null trend (Table 2), suggesting that trends in the annual mean ϕgene explain well the trends in the annual mean ϕLMI. Another feature is the pronounced decreasing trend in the genesis frequency for both all TCs (number of TCs with intensity not less than TS) and weak TCs (number of TCs with intensity of TS and severe TS).

Fig. 3.

Time series of the annual mean (a) ϕgene (° lat) and (b) difference between ϕLMI and ϕgene (° lat). (c) Time series of genesis frequency for all TCs (left ordinate) and WTCs and ITCs (right ordinate) over the WNP based on the JTWC best-track dataset during 1980–2016.

Fig. 3.

Time series of the annual mean (a) ϕgene (° lat) and (b) difference between ϕLMI and ϕgene (° lat). (c) Time series of genesis frequency for all TCs (left ordinate) and WTCs and ITCs (right ordinate) over the WNP based on the JTWC best-track dataset during 1980–2016.

Table 2.

Linear trends in the annual mean ϕgene (° lat decade−1), difference between ϕLMI and ϕgene (° lat decade−1), and genesis frequencies over the entire WNP (Fre; decade −1), the southern WNP south of 20°N (FreS20; decade−1), and the northern WNP north of 20°N (FreN20; decade −1), for all TCs, WTCs, and ITCs over the WNP based on the JTWC best-track datasets during 1980–2016. Values with an asterisk (in boldface) denote trends that are statistically significant at the 95% (90%) confidence level using the F test.

Linear trends in the annual mean ϕgene (° lat decade−1), difference between ϕLMI and ϕgene (° lat decade−1), and genesis frequencies over the entire WNP (Fre; decade −1), the southern WNP south of 20°N (FreS20; decade−1), and the northern WNP north of 20°N (FreN20; decade −1), for all TCs, WTCs, and ITCs over the WNP based on the JTWC best-track datasets during 1980–2016. Values with an asterisk (in boldface) denote trends that are statistically significant at the 95% (90%) confidence level using the F test.
Linear trends in the annual mean ϕgene (° lat decade−1), difference between ϕLMI and ϕgene (° lat decade−1), and genesis frequencies over the entire WNP (Fre; decade −1), the southern WNP south of 20°N (FreS20; decade−1), and the northern WNP north of 20°N (FreN20; decade −1), for all TCs, WTCs, and ITCs over the WNP based on the JTWC best-track datasets during 1980–2016. Values with an asterisk (in boldface) denote trends that are statistically significant at the 95% (90%) confidence level using the F test.

To further examine changes in TC genesis and track, we show in Fig. 4 the spatial distribution of trends in track density for all TCs, weak TCs, and intense TCs over the WNP during 1980–2016 based on the JTWC best-track dataset. The track density, which is determined primarily by both TC genesis location and track, is defined as the annual number of TCs passing through each 5° latitude × 5° longitude grid box. It can be seen from Fig. 4a that the TC exposure (track density) shows significant decreasing trends south of 20°N over the southern WNP, including the South China Sea and the Philippine Sea, but increasing trends over the northern WNP, especially around Taiwan and Japan. These are generally consistent with the results of Kossin et al. (2016, their Fig. 4). A similar trend pattern is found for TC genesis frequency (figure not shown). This suggests that the poleward migration of the annual mean ϕLMI for all TCs is largely attributed to the large reduction of TC genesis over the southern WNP and the increase over the northern WNP. The southern reduction contributes much more than the northern increase since the basinwide TC frequency over the WNP shows a large decreasing trend (Fig. 3c).

Fig. 4.

Trends in track density (decade−1 for each 5° × 5° latitude–longitude grid box) for all (a) TCs, (b) WTCs, and (c) ITCs over the WNP based on the JTWC best-track dataset during 1980–2016. Orange (blue) color indicates areas where the positive (negative) trend is statistically significant at the 90% confidence level by the F test.

Fig. 4.

Trends in track density (decade−1 for each 5° × 5° latitude–longitude grid box) for all (a) TCs, (b) WTCs, and (c) ITCs over the WNP based on the JTWC best-track dataset during 1980–2016. Orange (blue) color indicates areas where the positive (negative) trend is statistically significant at the 90% confidence level by the F test.

The trends in track densities for both weak and intense TCs exhibit similar spatial distributions, with a large decrease over the southern WNP and an increase over the northern WNP. However, the increase in weak TCs occurs over most of the northern WNP, whereas the increase in intense TCs is only located in a small region over the northwestern WNP. To clarify this point further, we define the subregion south (north) of 20°N as the southern (northern) WNP and calculated the trends of TC genesis frequencies over the two subregions (Table 2). It can be clearly seen that for weak TCs both the decreasing trend over the southern WNP and the increasing trend over the northern WNP are statistically significant, leading to significant poleward migration of the annual mean ϕLMI as shown in Fig. 2b. In contrast, intense TCs also show increasing trends, but they are much less pronounced than for weak TCs. The insignificant changes in genesis frequencies of intense TCs over the subregions are consistent with the small poleward shift of their annual mean ϕLMI. It should be mentioned that the sample numbers of all TCs, weak TCs, and intense TCs over the southern WNP are greater than those over the northern WNP. For the northern WNP, the sample numbers of TCs, weak TCs, and intense TCs are 216, 140, and 76, respectively, which also exhibit strong interannual variability (figure not shown).

The above results demonstrate that changes in TC exposure are closely related to the large reduction over the southern WNP and the increase of weak TCs over the northern WNP. It is these changes of weak TCs that predominantly lead to the observed significant poleward migration of the annual mean ϕLMI. Therefore, it is important to further understand what causes the regional variation of the trend in TC track density, in particular that for weak TCs, and thus the poleward migration of the annual mean ϕLMI. We will show in the next subsection that the regional variation in track density (genesis and track) trend results mainly from changes in the large-scale dynamical and thermodynamic conditions.

b. Changes in the large-scale environmental conditions

Changes in the large-sale environmental conditions are often considered being responsible for the trends and variability of TC activity in the literature [see the review of Zhan et al. (2012)]. Indeed, the spatial pattern of trend in TC exposure and thus the observed poleward migration of the annual mean ϕLMI discussed in section 3a are largely controlled by changes in the large-scale environmental conditions. We examined the changes in various large-scale environmental variables that are often used to understand the trends and variability of TC activity. Figure 5 shows the trends in SST, MPI, 850-hPa horizontal wind, 500-hPa geopotential height, 500-hPa vertical velocity, and air temperature at 150 hPa (approximately the outflow layer temperature of TCs) in JASON during 1980–2016 based on the ERA-Interim reanalysis data.

Fig. 5.

Trends in (a) SST (°C decade−1), (b) MPI (m s−1 decade−1), (c) 850-hPa horizontal wind (m s−1 decade−1), (d) 500-hPa geopotential height (gpm decade−1), (e) 500-hPa vertical velocity (Pa s−1 decade−1), and (f) 150-hPa temperature (°C decade−1) during 1980–2016. In (a) and (b), dots indicate areas where the trend is statistically significant at the 95% confidence level by the F test. In (c)–(f), orange (blue) color indicates areas where the positive (negative) trend is statistically significant at the 95% confidence level by the F test.

Fig. 5.

Trends in (a) SST (°C decade−1), (b) MPI (m s−1 decade−1), (c) 850-hPa horizontal wind (m s−1 decade−1), (d) 500-hPa geopotential height (gpm decade−1), (e) 500-hPa vertical velocity (Pa s−1 decade−1), and (f) 150-hPa temperature (°C decade−1) during 1980–2016. In (a) and (b), dots indicate areas where the trend is statistically significant at the 95% confidence level by the F test. In (c)–(f), orange (blue) color indicates areas where the positive (negative) trend is statistically significant at the 95% confidence level by the F test.

The local SST and MPI, as the key environmental factors affecting TC genesis and intensification, have been increasing considerably in most of the WNP. More strikingly, larger increasing trends occur in the northern WNP than in the southern WNP (Figs. 5a,b). This spatial pattern is consistent with changes in TC MPI in the work of Kossin et al. (2014). This suggests that the oceanic thermodynamic condition has been more favorable for TC genesis and intensification over most of the WNP, especially over the northern WNP. The similarity in the spatial patterns in SST and MPI trends in Figs. 5a and 5b, respectively, and in trend in TC exposure in Fig. 4 strongly suggests that changes in ocean must play an important role in the increase of TC genesis over the northern WNP. The spatial pattern in SST change has been linked to tropical expansion and results mainly from the anthropogenic climate change (Kossin et al. 2014). Since TCs forming over the northern WNP often experience shorter lifetime and move across relatively colder SSTs, it is not unreasonable to expect more weak TCs than intense TCs. On the other hand, the favorable SST and MPI conditions over the southern WNP are more likely to be masked by other unfavorable dynamical conditions.

The most striking feature in the changes in low-level winds (Fig. 5c) is the easterly trend over the entire equatorial western Pacific south of 12.5°N, which was thought to be linked to the enhancing Walker circulation in the studied period (Park et al. 2013). Associated with the easterly trend is a circulation trend coupled with an anticyclonic circulation trend over the southern WNP and a relatively small cyclonic circulation trend centered near Taiwan over the northwestern WNP. The latter is thought to be induced by the anticyclone trend to the southeast (Park et al. 2013). This spatial distribution in low-level circulation trend is unfavorable for TC genesis over the southern WNP but favorable for TC genesis over the north-northwestern WNP, consistent with the results based on low-level relative vorticity in Lin and Chan (2015).

The trends in the midtropospheric geopotential height (Fig. 5d) and vertical velocity (Fig. 5e) show a spatial pattern very similar to that in the low-level circulation tend (Fig. 5c). Relatively large positive trends in 500-hPa geopotential height are located south of 20°N and north of 30°N, while relatively small positive trends are just in between. Since the intensity of the WNP subtropical high shows a significantly negative correlation with TC genesis frequency (Liu and Chan 2013; Lin and Chan 2015), the large positive trend in 500-hPa geopotential height south of 20°N is unfavorable for TC genesis over the southern WNP. In contrast, the relatively small positive trend may have a much weaker impact on TC genesis over the northern WNP. Note that the overall positive geopotential height trends are consistent with the positive trends in SST. Kang and Elsner (2016) pointed out that high pressure in the middle and upper troposphere suppresses the energy transport by convection from the lower troposphere and may lead to fewer but stronger TCs. Positive trends in 500-hPa vertical velocity cover most of the region between 5° and 20°N, especially significant over the South China Sea and around 160°E, which suppresses TC genesis in the southern WNP. Note that the combination of subsidence and anticyclonic circulation trends over the South China Sea offsets the influence of warm SSTs and increased MPI shown in Figs. 5a and 5b, respectively, and thus leads to a decrease in TC genesis over the southern WNP. In contrast, the significantly negative trends in 500-hPa vertical velocity are located north of 20°N, which is favorable for TC genesis over the northern WNP.

The 150-hPa air temperature within the outflow layer of TCs shows significant cooling trends in the region south of 20°N over the WNP (Fig. 5f). Similar cooling trends in the outflow layer have been reported by Emanuel et al. (2013) and are used to explain the observed increasing trend in intense TCs in the past decades (Vecchi et al. 2013; Kossin 2015; Wing et al. 2015). Here we argue that this cooling trend in the outflow layer near the tropopause may have enhanced the genesis of intense TCs and thus contributed to insignificant reduction of intense TCs over the southern WNP.

To sum up, the changes in thermodynamic conditions are favorable for TC genesis and intensification over most of the WNP, especially over the northern WNP, and those in dynamical conditions also favor more TCs to form in the northern WNP. The larger number of TCs forming in the northern WNP generally have less time to intensify, leading to more weak TCs in the northern WNP. However, over the southern WNP, the changes in dynamical conditions largely offset the influence of thermodynamic conditions, leading to fewer but stronger TCs. Consequently, the trends in large-scale dynamical and thermodynamic environmental conditions are responsible for the trends in TC exposure discussed above and thus the poleward migration of the annual mean latitude in TC LMI.

5. Conclusions and discussion

a. Conclusions

In this study, the poleward migration of the mean location of TC lifetime maximum intensity (LMI) over the WNP since 1980 has been revisited. It is shown that this poleward migration is largely contributed by weak TCs (with maximum sustained surface wind speed less than 33 m s−1). The annual mean location of LMI of weak TCs is found to have migrated about 1° latitude poleward per decade since 1980. However, such a trend is much less pronounced for intense TCs. This is shown to be related to a significant decreasing trend of TC genesis in the southern WNP and a significant increasing trend in the northwestern WNP. Results from the analysis of changes in large-scale dynamical and thermodynamic conditions show that the increasing trends in SST and MPI are larger at higher latitudes than in the low latitudes, consistent with the SST trend associated with global warming. The associated changes in the large-scale circulation show an anticyclonic circulation trend over the southern WNP and a cyclonic circulation trend over the northern WNP in the lower troposphere and a high pressure trend south of 20°N and a cooling trend in the outflow layer near the tropopause over the northwestern WNP. These changes favor more TCs to form in the northern WNP and fewer but stronger TCs in the southern WNP. The more TCs forming in the northern WNP generally have less time to intensify, leading to more weak TCs in the northern WNP, which dominate the poleward migration of the annual mean location of TC LMI previously reported in the basin.

b. Discussion

El Niño–Southern Oscillation (ENSO) plays an important role in interannual variability in TC genesis location, intensity, lifespan, track, and so on [see the review by Zhan et al. (2012)]. In general, in an El Niño year, more TCs form in the southeastern WNP, which favors TCs reaching their LMI at relatively low latitudes. Recently, an El Niño event occurred in 2014, reached the super El Niño category in winter 2015/16, and dissipated during May 2016. As a result, the annual mean ϕLMI in both 2014 and 2015 dropped to the lowest values since 1990 (Fig. 2a), which distinctly lowered the rate of the overall migration trend (not shown). Nevertheless, the poleward migration of the annual mean ϕLMI is still statistically significant at 90% confidence level during 1980–2016, with a rate of 0.46° lat decade−1. It is consistent with the results in Kossin et al. (2016), who compared the migration rates with and without ENSO variability included. This suggests that the trend in the annual mean ϕLMI has not substantially affected by the ENSO variability. If these environmental changes continue, a concomitant continued poleward migration of the annual mean latitude where TCs achieve their LMI would have potentially profound consequences.

There is increasing evidence that climate change has widened the tropical belt over the past few decades [see the review by Seidel et al. (2008); Lucas et al. 2014]. This expansion leads to the larger warming trend in SST at higher latitudes than lower latitudes, and thus favors more TC genesis in the northern WNP (Fig. 4). In this sense, the expansion of tropical belt indeed has important implications for the poleward migration of the annual mean ϕLMI as proposed by Kossin et al. (2014).

TC activity is closely associated with both thermodynamic and dynamical conditions. In this study, it is shown that ocean conditions have become more favorable for TC genesis and intensification in the whole WNP basin, while the role of dynamical fields over the WNP is not consistent with that of ocean conditions, which are highly dependent on the horizontal distribution of SST over the Pacific. In recent years, the Walker circulation has experienced an unprecedented strengthening, leading to an easterly trend over the entire equatorial western Pacific and an anomalous anticyclonic circulation trend over the southern WNP. The interplay between the atmospheric conditions and ocean conditions leads to fewer but stronger TCs in the southern WNP, which mainly explains why weak TCs dominate the observed poleward migration of the annual mean ϕLMI since 1980 in this study.

Finally, it should be mentioned that although the poleward trend of intense TCs is considerably small it is still important. This is because in term of hazard exposure and mortality risk, although weak TCs might be migrating poleward more quickly, it is intense TCs that affect risk the most. In this sense, these two factors might best be considered together from an impact point of view. It should also be noticed that the new TC metric of the annual mean latitudinal location of LMI proposed by Kossin et al. (2014) is comparatively insensitive to some of the past data uncertainties, such as the determination of TC intensity based on satellite data. Although the stratification by intensity in this study introduces additional uncertainties, the three TC best-track datasets show consistent results, suggesting that the results from this study are robust and highly reliable. More importantly, our results provide a better understanding for several extra TC metrics as well.

Acknowledgments

The authors are grateful to Dr. James Kossin and two other anonymous reviewers for their helpful comments on the manuscript. This study has been supported by the National Natural Science Foundation of China (Grants 41375093, 41575052, and 41475082). Ruifen Zhan was supported in part by JAMSTEC through its sponsorship of the International Pacific Research Center (IPRC) in the School of Ocean and Earth Science and Technology (SOEST) at the University of Hawai‘i at Mānoa. Addition support has been provided by the Typhoon Scientific and Technological Innovation Group of Shanghai Meteorological Service. The monthly SST analysis data were downloaded from the Hadley Centre website http://www.metoffice.gov.uk/hadobs/hadisst/data/download.html. The JTWC best-track TC dataset was downloaded from https://metoc.ndbc.noaa.gov/web/guest/jtwc/best_tracks/western-pacific. The JMA best-track TC dataset was downloaded from http://www.jma.go.jp/jma/jma-eng/jma-center/rsmc-hp-pub-eg/besttrack.html. The CMA best-track TC dataset was downloaded from http://tcdata.typhoon.gov.cn/en/index.html. The ERA-Interim data were downloaded from http://apps.ecmwf.int/datasets/.

REFERENCES

REFERENCES
Choi
,
K.-S.
,
H.-D.
Kim
,
S.-D.
Kang
, and
C.-S.
Shim
,
2016
:
Decreasing trend of tropical cyclone genesis frequency in July–August over the western North Pacific in the last 20 years
.
Theor. Appl. Climatol.
,
125
,
241
251
, doi:.
Dee
,
D. P.
, and Coauthors
,
2011
:
The ERA-Interim reanalysis: Configuration and performance of the data assimilation systems
.
Quart. J. Roy. Meteor. Soc.
,
137
,
553
597
, doi:.
Elsner
,
J. B.
,
J. P.
Kossin
, and
T. H.
Jagger
,
2008
:
The increasing intensity of the strongest tropical cyclones
.
Nature
,
455
,
92
95
, doi:.
Emanuel
,
K. A.
,
2005
:
Increasing destructiveness of tropical cyclones over the past 30 years
.
Nature
,
436
,
686
688
, doi:.
Emanuel
,
K. A.
,
S.
Solomon
,
D.
Folini
,
S.
Davis
, and
C.
Cagnazzo
,
2013
:
Influence of tropical tropopause layer cooling on Atlantic hurricane activity
.
J. Climate
,
26
,
2288
2301
, doi:.
He
,
H. Z.
,
J.
Yang
,
D. Y.
Gong
,
R.
Mao
,
Y.
Wang
, and
M. N.
Gao
,
2015
:
Decadal changes in tropical cyclone activity over the western North Pacific in the late 1990s
.
Climate Dyn.
,
45
,
3317
3329
, doi:.
Kang
,
N.
, and
J.
Elsner
,
2016
:
Climate mechanism for stronger typhoons in a warmer world
.
J. Climate
,
29
,
1051
1057
, doi:.
Klotzbach
,
P. J.
,
2006
:
Trends in global tropical cyclone activity over the past twenty years (1986–2005)
.
Geophys. Res. Lett.
,
33
,
L10805
, doi:.
Klotzbach
,
P. J.
, and
C. W.
Landsea
,
2015
:
Extremely intense hurricanes: Revisiting Webster et al. (2005) after 10 years
.
J. Climate
,
28
,
7621
7629
, doi:.
Kossin
,
J. P.
,
2015
:
Validating atmospheric reanalysis data using tropical cyclones as thermometers
.
Bull. Amer. Meteor. Soc.
,
96
,
1089
1096
, doi:.
Kossin
,
J. P.
,
T. L.
Olander
, and
K. R.
Knapp
,
2013
:
Trend analysis with a new global record of tropical cyclone intensity
.
J. Climate
,
26
,
9960
9976
, doi:.
Kossin
,
J. P.
,
K. A.
Emanuel
, and
G. A.
Vecchi
,
2014
:
The poleward migration of the location of tropical cyclone maximum intensity
.
Nature
,
509
,
349
352
, doi:.
Kossin
,
J. P.
,
K. A.
Emanuel
, and
S.
Camargo
,
2016
:
Past and projected changes in western North Pacific tropical cyclone exposure
.
J. Climate
,
29
,
5725
5739
, doi:.
Lin
,
I.-I.
, and
J. C. L.
Chan
,
2015
:
Recent decrease in typhoon destructive potential and global warming implications
.
Nat. Commun.
,
6
,
7182
, doi:.
Liu
,
K. S.
, and
J. C. L.
Chan
,
2013
:
Inactive period of western North Pacific tropical cyclone activity in 1998–2011
.
J. Climate
,
26
,
2614
2630
, doi:.
Lucas
,
C.
,
B.
Timbal
, and
H.
Nguyen
,
2014
:
The expanding tropics: A critical assessment of the observational and modeling studies
.
Wiley Interdiscip. Rev.: Climate Change
,
5
,
89
112
, doi:.
Mei
,
W.
, and
S.-P.
Xie
,
2016
:
Intensification of landfalling typhoons over the northwest Pacific since the late 1970s
.
Nat. Geosci.
,
9
,
753
757
, doi:.
NDRRMC
,
2014
: Updates re the effects of Typhoon “Yolanda” (Haiyan). National Disaster Risk Reduction and Management Council, Republic of the Philippines, 2 pp. [Available online at http://www.ndrrmc.gov.ph/attachments/article/1329/Update_on_Effects_Typhoon_YOLANDA_(Haiyan)_17APR2014.pdf.]
Oouchi
,
K.
,
J.
Yoshimura
,
H.
Yoshimura
,
R.
Mizuta
,
S.
Kusunoki
, and
A.
Noda
,
2006
:
Tropical cyclone climatology in a global warming climate as simulated in a 20-km-mesh global atmospheric model: Frequency and wind intensity analyses
.
J. Meteor. Soc. Japan
,
84
,
259
276
, doi:.
Park
,
D.-S. R.
,
C.-H.
Ho
,
J.-H.
Kim
, and
H.-S.
Kim
,
2011
:
Strong landfall typhoons in Korea and Japan in a recent decade
.
J. Geophys. Res.
,
116
,
D07105
, doi:.
Park
,
D.-S. R.
,
C.-H.
Ho
,
J.-H.
Kim
, and
H.-S.
Kim
,
2013
:
Spatially inhomogeneous trends of tropical cyclone intensity over the western North Pacific for 1977–2010
.
J. Climate
,
26
,
5088
5101
, doi:.
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:.
Seidel
,
D. J.
,
Q.
Fu
,
W. J.
Randel
, and
T. J.
Reichler
,
2008
:
Widening of the tropical belt in a changing climate
.
Nat. Geosci.
,
1
,
21
24
, doi:.
Vecchi
,
G. A.
,
S.
Fueglistaler
,
I. M.
Held
,
T. R.
Knutson
, and
M.
Zhao
,
2013
:
Impacts of atmospheric temperature trends on tropical cyclone activity
.
J. Climate
,
26
,
3877
3891
, doi:.
Webster
,
P. J.
,
G. J.
Holland
,
J. A.
Curry
, and
H.-R.
Chang
,
2005
:
Changes in tropical cyclone number, duration, and intensity in a warming environment
.
Science
,
309
,
1844
1846
, doi:.
Wing
,
A. A.
,
K.
Emanuel
, and
S.
Solomon
,
2015
:
On the factors affecting trends and variability in tropical cyclone potential intensity
.
Geophys. Res. Lett.
,
42
,
8669
8677
, doi:.
Wu
,
L.
,
B.
Wang
, and
S.
Geng
,
2005
:
Growing typhoon influence on East Asia
.
Geophys. Res. Lett.
,
32
,
L18703
, 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:.
Yoshimura
,
J.
,
M.
Sugi
, and
A.
Noda
,
2006
:
Influence of greenhouse warming on tropical cyclone frequency
.
J. Meteor. Soc. Japan
,
84
,
405
428
, doi:.
Zhan
,
R. F.
,
Y.
Wang
, and
M.
Ying
,
2012
:
Seasonal forecasts of tropical cyclone activity over the western North Pacific: A review
.
Trop. Cyclone Res. Rev.
,
1
,
307
324
, doi:.

Footnotes

© 2017 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).