J. Meteor. Res.  2017, Vol. 31 Issue (1): 107-116   PDF    
http://dx.doi.org/10.1007/s13351-017-6088-4
The Chinese Meteorological Society
0

Article Information

ZHOU Botao, XU Ying . 2017.
How the “Best” CMIP5 Models Project Relations of Asian–Pacific Oscillation to Circulation Backgrounds Favorable for Tropical Cyclone Genesis over the Western North Pacific. 2017.
J. Meteor. Res., 31(1): 107-116
http://dx.doi.org/10.1007/s13351-017-6088-4

Article History

Received May 24, 2016
in final form December 7, 2016
How the “Best” CMIP5 Models Project Relations of Asian–Pacific Oscillation to Circulation Backgrounds Favorable for Tropical Cyclone Genesis over the Western North Pacific
Botao ZHOU1,2, Ying XU1     
1. National Climate Center, China Meteorological Administration, Beijing 100081;
2. Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters, Nanjing University of Information Science & Technology, Nanjing 210044
ABSTRACT: Based on the simulations of 32 models from the Coupled Model Intercomparison Project Phase 5 (CMIP5), the present study assesses their capacity to simulate the relationship of the summer Asian–Pacific Oscillation (APO) with the vertical zonal wind shear, low-level atmospheric vorticity, mid-level humidity, atmospheric divergence in the lower and upper troposphere, and western Pacific subtropical high (WPSH) that are closely associated with the genesis of tropical cyclones over the western North Pacific. The results indicate that five models can simultaneously reproduce the observed pattern with the positive APO phase accompanied by weak vertical zonal wind shear, strengthened vorticity in the lower troposphere, increased mid-level humidity, intensified low-level convergence and high-level divergence, and a northward-located WPSH over the western North Pacific. These five models are further used to project their potential relationship under the RCP8.5 scenario during 2050–2099. Compared to 1950–1999, the relationship between the APO and the vertical zonal wind shear is projected to weaken by both the multi-model ensemble and the individual models. Its linkage to the low-level vorticity, mid-level humidity, atmospheric divergence in the lower and upper troposphere, and the northward–southward movement of the WPSH would also reduce slightly but still be significant. However, the individual models show relatively large differences in projecting the linkage between the APO and the mid-level humidity and low-level divergence.
Key words: Asian–Pacific Oscillation     atmospheric circulation     tropical cyclone     CMIP5 evaluation and projection    
1 Introduction

Tropical cyclones (TCs) are a strong synoptic phenomenon generated over tropical oceans. Their occurrence is often accompanied by rainstorms and storm surges, resulting in huge economic losses and losses of human life across many countries. Thus, TC variability has long been a topic of prime interest in meteorological research.

TC activities over the western North Pacific, including source, number, track, and intensity, are determined by the dynamic, thermal, and environmental backgrounds associated with TC genesis and development (Gray, 1968;Chen and Ding, 1979;Ding and Reiter, 1983;Lander, 1996;Zhang and Peng, 2003;Zhou and Cui, 2008;Sun and Chen, 2011). Atmospheric oscillations have been recognized to play important roles through changing local atmospheric and oceanic conditions in the western North Pacific (e.g.,Liebmann et al., 1994;Ho et al., 2005;Fan, 2007;Wang and Fan, 2007;Wang et al., 2007;Fan and Wang, 2009;Zhou and Cui, 2014). As a dominant zonal atmospheric oscillation, the Asian–Pacific Oscillation (APO), characterized by a seesaw variation of the temperature between Asia and the North Pacific in the upper troposphere (Zhao et al., 2007), is known to exert a significant effect on TC activity over the western North Pacific (Zhou et al., 2008;Cui et al., 2010). The positive phase of summer APO is concurrent with reduced vertical zonal wind shear, anomalous convergence in the lower troposphere and divergence in the upper troposphere, and a northward-located western Pacific subtropical high (WPSH), providing a favorable environment for TC genesis and leading to an increased frequency of TCs over the western North Pacific. The case is reversed when the summer APO is in its negative phase.

The Fifth Assessment Report of the IPCC (IPCC, 2013) indicated that further warming in the future will change all components of the climate system. This conclusion derives from the projections of the Coupled Model Intercomparison Project Phase 5 (CMIP5) models under the Representative Concentration Pathway (RCP) scenarios (Moss et al., 2010;Taylor et al., 2012). In the context of global warming scenarios, the APO is projected to weaken towards the end of the 21st century (Zhou, 2016). As such, the question naturally arises as to whether or not the relationship between the APO and the circulation background associated with TC genesis over the western North Pacific will change in response to the decrease in APO intensity in the future.

With this question in mind, this study first assesses the ability of CMIP5 models to simulate the observed relationship between summer APO and the atmospheric circulation background related to TC genesis, since the realistic performance of models in producing the present state is vitally important for reliable projections of climate change (Kidston and Gerber, 2010). Moreover, based on this evaluation, we extract the models that are “best” at capturing the current relationship and project the potential relationship during what remains of the 21st century.

2 Data and method

The data of the historical and RCP8.5 simulations by 32 CMIP5 models (Table 1) are employed in this study. The historical simulation represents the 20th century climate, and the RCP8.5 represents a high radiative forcing scenario where the radiative forcing reaches 8.5 W m–2 in 2100. Readers can browse the CMIP5 website ( http:// cmippcmdi.llnl.gov/cmip5/availability.html) for more details.

Table 1 Basic information on the 32 CMIP5 models used in this study
Model Modeling group Atmospheric resolution (lon. × lat.)
ACCESS1.0 Commonwealth Scientific and Industrial Research Organization (CSIRO) and Bureau of Meteorology (BoM), Australia 192 × 145
ACCESS1.3 CSIRO and BoM, Australia 192 × 145
BCC_CSM1.1 Beijing Climate Center (BCC), China Meteorological Administration (CMA), China 128 × 64
BCC_CSM1.1(m) BCC, CMA, China 320 × 160
BNU-ESM Beijing Normal University, China 128 × 64
CanESM2 Canadian Centre for Climate Modeling and Analysis, Canada 128 × 64
CCSM4 National Center for Atmosphere Research, United States 288 × 192
CMCC-CM Euro-Mediterranean Center on Climate Change (CMCC), Italy 480 × 240
CMCC-CMS CMCC, Italy 192 × 96
CNRM-CM5 Centre National de Recherches Météorologiques–Centre Européen de Recherche et de Formation Avancée en Calcul Scientifique, France 256 × 128
CSIRO Mk3.6.0 Queensland Climate Change Centre of Excellence and CSIRO, Australia 192 × 96
FGOALS-g2 State Key Laboratory of Numerical Modeling for Atmospheric Sciences and Geophysical Fluid Dynamics, Institute of Atmospheric Physics, Chinese Academy of Sciences, China 128 × 60
FIO-ESM First Institute of Oceanography, China 128 × 64
GFDL CM3 NOAA Geophysical Fluid Dynamics Laboratory (GFDL), United States 144 × 90
GFDL-ESM2G NOAA GFDL, United States 144 × 90
GFDL-ESM2M NOAA GFDL, United States 144 × 90
GISS-E2-H NASA Goddard Institute for Space Studies (GISS), United States 144 × 90
GISS-E2-R NASA GISS, United States 144 × 90
HadGEM2-AO UK Met Office (UKMO) Hadley Centre, United Kingdom 192 × 144
HadGEM2-CC UKMO Hadley Centre, United Kingdom 192 × 144
HadGEM2-ES UKMO Hadley Centre, United Kingdom 192 × 144
INM-CM4.0 Institute for Numerical Mathematics, Russia 180 × 120
IPSL-CM5A-LR Institute Pierre-Simon Laplace (IPSL), France 96 × 96
IPSL-CM5A-MR IPSL, France 144 × 143
IPSL-CM5B-LR IPSL, France 96 × 96
MIROC5 Atmosphere and Ocean Research Institute (University of Tokyo), National Institute for Environmental Studies, and Japan Agency for Marine-Earth Science and Technology, Japan 256 × 128
MIROC-ESM Atmosphere and Ocean Research Institute (University of Tokyo), National Institute for Environmental Studies, and Japan Agency for Marine-Earth Science and Technology, Japan 128 × 64
MIROC-ESM-CHEM Atmosphere and Ocean Research Institute (University of Tokyo), National Institute for Environmental Studies, and Japan Agency for Marine-Earth Science and Technology, Japan 128 × 64
MPI-ESM-LR Max Planck Institute for Meteorology, Germany 192 × 96
MRI-CGCM3 Meteorological Research Institute, Japan 320 × 160
NorESM1-M Norwegian Climate Centre, Norway 144 × 96
NorESM1-ME Norwegian Climate Centre, Norway 144 × 96

The observational data adopted to validate the behavior of the CMIP5 models include the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis data, with a resolution of 2.5° × 2.5° (Kalnay et al. 1996), and the TC frequency record from the Joint Typhoon Warning Center. Because of different resolutions for different models (see Table 1), the simulation data are interpolated onto the same grid as the NCEP–NCAR data before analysis.

In this study, we focus on the interannual relationship between the APO and TC-related atmospheric circulation, including vertical zonal wind shear, atmospheric vorticity, mid-level humidity, convergence and divergence, and the WPSH in the summer season (June to August). The 50-yr average, i.e., 1950–1999 for the historical simulation and 2050–2099 for RCP8.5, is chosen as the climate state. We use correlation analysis to investigate the relationship between a pair of variables. Before calculating the correlation, the linear trend of all the grid data is removed. The statistical significance is determined by using the Student's t-test.

3 Evaluation of the CMIP5 models' perfor-mance

Before evaluating the performance of the CMIP5 models in simulating the relationship between summer APO and the TC-related atmospheric background over the western North Pacific, we first present the observed correlation between the TC frequency over the western North Pacific and the vertical zonal wind shear (Fig. 1a), vorticity at 850 hPa (Fig. 2a), specific humidity averaged from 700 to 500 hPa (Fig. 2c), divergence at 1000 (Fig. 3a) and 150 hPa (Fig. 3c), and horizontal winds at 850 hPa (Fig. 4a). In summary, in association with an above-normal frequency of TCs occurring over the western North Pacific, the vertical zonal wind shear tends to be weaker than the normal in the major TC genesis region (Fig. 1a). Positive vorticity anomalies at the low level (Fig. 2a) and more moisture at the mid-level (Fig. 2c), concurrent with anomalous convergence in the lower troposphere and divergence in the upper troposphere (Figs. 3a and 3c), dominate the Pacific within 10°–30°N. Furthermore, a cyclonic circulation anomaly prevails in the western Pacific south of 40°N (Fig. 4a), indicative of a northward-shifted WPSH (Zhang et al., 2003). The combined effect of these factors provides favorable conditions for TC genesis over the western North Pacific, which has been reported in previous studies.

Fig. 1 Observed correlations of vertical zonal wind shear with (a) tropical cyclone frequency over the western North Pacific and (b) Asian–Pacific Oscillation. Heavy (Light) shading indicates areas above the 95% (90%) confidence level. The dashed rectangle represents the key regions selected.
Fig. 2 Observed correlations of (a, b) 850-hPa vorticity and (c, d) 700–500-hPa averaged specific humidity with tropical cyclone frequency over the (a, c) western North Pacific and (b, d) Asian–Pacific Oscillation. Heavy (light) shading indicates areas above the 95% (90%) confidence level.
Fig. 3 Observed correlations of atmospheric divergence at (a, b) 1000 hPa and (c, d) 150 hPa with TC frequency over the (a, c) western North Pacific and (b, d) Asian–Pacific Oscillation. Heavy (light) shading indicates areas above the 95% (90%) confidence level.
Fig. 4 Observed correlations of 850-hPa horizontal winds with (a) tropical cyclone frequency over the western North Pacific and (b) Asian–Pacific Oscillation, which are shown as arrows. Heavy (light) shading indicates areas above the 95% (90%) confidence level.

Correspondingly, we plot the correlation of summer APO with the aforementioned atmospheric background for TC genesis in the NCEP–NCAR reanalysis data. According to the definition of Zhao et al. (2007), the APO is measured by the arithmetic difference in regionally averaged eddy air temperature in the upper troposphere (500–200 hPa) between (15°–50°N, 60°–120°E) and (15°–50°N, 180°–120°W). As shown in the figures, the APO is negatively correlated with the vertical zonal wind shear (Fig. 1b) and the low-level divergence (Fig. 3b), but positively correlated with the low-level vorticity (Fig. 2b), mid-level humidity (Fig. 2d), and high-level divergence (Fig. 3d) over the lower latitudes of the North Pacific. In addition, a cyclonic anomaly prevails in the tropical western Pacific (Fig. 4b). The APO-related pattern generally resembles that of the TC frequency over the western North Pacific, demonstrating that the APO can affect TC frequency through changing the vertical zonal wind shear, vorticity, humidity, and divergence over the western North Pacific and the location of the WPSH.

Next, we evaluate the ability of the CMIP5 models to simulate the relationship between the APO and these physical variables. To facilitate quantitative evaluation, we define six indices to represent the above-mentioned key factors. The indices for the vertical zonal wind shear (VWS), vorticity at 850 hPa (VOR850), humidity averaged from 700 to 500 hPa (HUM), divergence at 1000 hPa (DIV1000) and divergence at 150 hPa (DIV150) are the averages of the variables over the regions (10°–20°N, 140°–165°E), (17.5°–27.5°N, 140°E–180°), (12.5°–25°N, 135°E–180°), (15°–25°N, 140°E–170°W) and (15°–25°N, 140°E–170°W), respectively. The index (UV850) defined as the difference in 850-hPa zonal wind between (10°–20°N, 120°–170°E) and (25°–35°N, 120°–170°E) is used to measure the shift in the WPSH. The key regions selected are outlined by the dashed rectangle in Figs. 14.

Figure 5 displays the correlation coefficients of the APO with the six indices in the CMIP5 historical simulation and the observation. For the observation, the correlation coefficients of the APO with VWS, VOR850, HUM, DIV1000, DIV150, and UV850, during 1950–1999, are –0.39, 0.47, 0.39, –0.47, 0.55, and 0.43, respectively. They are all significant at the 99% level. For the TC frequency over the western North Pacific, its correlation coefficients with the six key variables are –0.47, 0.47, 0.37, –0.57, 0.55, and 0.64, respectively, also above the 99% confidence level.

Fig. 5 Correlation coefficients of the Asian–Pacific Oscillation with (a) VWS, (b) VOR850, (c) HUM, (d) DIV1000, (e) DIV150, and (f) UV850 (see Section 3, paragraph 3, of the main text for definitions of these indices) in the historical simulation and observation. Correlations of tropical cyclone frequency over the western North Pacific with the six indices are also shown.

In the historical simulations, the CMIP5 simulated APO–VWS correlation coefficients range from –0.37 (CCSM4) to 0.52 (CMCC-CM). Among the 32 individual models, 18 models produce an out-of-phase relationship between the two variables, while only 6 models yielding negative correlations significant above the 90% level. The APO–VOR850 correlation coefficients are between –0.19 (BCC_CSM1.1) and 0.71 (BNU-ESM), among which 16 models simulate positive correlations exceeding the 90% confidence level. The APO–HUM correlation coefficients vary from –0.27 (INM-CM4.0) to 0.71 (FIO-ESM) and there are 11 models that can produce positive correlations higher than the 90% confidence level. The APO–DIV1000 correlation coefficients are in the range of –0.63 (FIO-ESM) to 0.21 (INM-CM4.0). Twenty-seven out of 32 models can capture the inverse relationship, among which 11 models simulate correlations significant above the 90% confidence level. All the models except INM-CM4.0 and MRI-CGCM3 can reproduce the observed APO–DIV150 relation. Moreover, 21 models simulate positive correlations passing the 90% confidence level. The simulated APO–UV850 correlation coefficients range from –0.33 (MRI-CGCM3) to 0.80 (BNU-ESM). Twenty-seven models simulate an in-phase relationship that is consistent with the observation, 14 of which exhibit correlations above the 90% confidence level.

If the models' behaviors in these six aspects are synthetically considered, it is found that there are five CMIP5 models (ACCESS1.3, BNU-ESM, CCSM4, GISS-E2-H, and NorESM1-ME) showing good capacity to simultaneously reproduce the observed significant correlations (higher than the 90% confidence level). These five models are then selected as the “best” models for the projection. The multi-model ensemble (MME) in this study is calculated as the average of the five “best” models with equivalent weight. For the MME simulation, the correlation coefficients of the APO with VWS, VOR850, HUM, DIV1000, DIV150, and UV850 are –0.27, 0.47, 0.51, –0.44, 0.51, and 0.49, respectively.

The MME simulated spatial distributions of correlations of the summer APO with the vertical zonal wind shear (Fig. 6a), 850-hPa vorticity (Fig. 7a), 700–500-hPa averaged humidity (Fig. 7c), divergence at 1000 (Fig. 8a) and 150 hPa (Fig. 8c), and 850-hPa horizontal winds (Fig. 9a), are further shown. Clearly, the MME simulation bears a general resemblance to the observation. That is, the MME can capture the strong APO-related pattern corresponding to weaker vertical zonal wind shear, positive vorticity anomalies at the low level, enhanced humidity at the mid-level, and anomalous low-level convergence and upper-level divergence over the lower latitudes of the North Pacific, and the cyclonic circulation anomaly over the tropical western Pacific. Therefore, the use of the MME for future projection is justified.

Fig. 6 MME simulated correlations between the Asian–Pacific Oscillation and vertical zonal wind shear in the (a) historical and (b) RCP8.5 simulations. Heavy (light) shading indicates areas above the 95% (90%) confidence level.
Fig. 7 MME simulated correlations between the Asian–Pacific Oscillation and (a, b) vorticity at 850 hPa and (c, d) specific humidity averaged from 700 to 500 hPa, in the (a, c) historical and (b, d) RCP8.5 simulations. Heavy (light) shading indicates areas above the 95% (90%) confidence level.
Fig. 8 MME simulated correlations between the Asian–Pacific Oscillation and atmospheric divergence at (a, b) 1000 and (c, d) 150 hPa, in the (a, c) historical and (b, d) RCP8.5 simulations. Heavy (light) shading indicates areas above the 95% (90%) confidence level.
Fig. 9 MME simulated correlation between the Asian–Pacific Oscillation and 850-hPa horizontal winds (shown as arrows) in the (a) historical and (b) RCP8.5 simulations. Heavy (light) shading indicates areas above the 95% (90%) confidence level.
4 Projection

The MME results of the correlations between the APO and the vertical zonal wind shear, 850-hPa vorticity, 700–500-hPa averaged specific humidity, 1000-hPa divergence, 150- hPa divergence, and 850-hPa winds, during 2050–2099 under the RCP8.5 scenario, are shown in Figs. 6b, 7b, 7d, 8b, 8d, and 9b, respectively. Their spatial distributions under the RCP8.5 scenario are in general similar to their counterparts in the historical MME simulation, but there are also some discrepancies. For example, the correlation between the APO and the vertical zonal wind shear is reduced obviously in the major source region for TC genesis. Its correlation with the divergence at 150 hPa is also somewhat weakened over the Pacific within 10°–30°N.

To quantitatively measure such changes in the future, the correlation coefficients of the APO with the six indices for the five individual models and the MME simulation under the RCP8.5 scenario are calculated. As shown in Fig. 10a, the MME result shows that the APO–VWS correlation coefficient decreases from –0.27 during 1950–1999 (significant at the 95% level) to –0.17 during 2050–2099 (below the 90% confidence level). That is, if future emissions follow the RCP8.5 path, the APO will only account for 2.9% of the variance of the vertical zonal wind shear during 2050–2099; whereas, during 1950–1999, it can explain 7.3% of the variance. All of the five individual models project a consistent weakening tendency. Such a weakening passes the 95% confidence level, as indicated by the t-test for the difference in the correlation coefficients. Therefore, the linkage between the APO and the vertical zonal wind shear will weaken during the second half of the 21st century.

Fig. 10 Correlation coefficients of the Asian–Pacific Oscillation with (a) VWS, (b) VOR850, (c) HUM, (d) DIV1000, (e) DIV150, and (f) UV850 (see Section 3, paragraph 3, of the main text for definitions of these indices). Red bars represent 2050–2099 under RCP8.5 and blue bars represent 1950–1999 in the historical simulation.

A slight weakening in the APO–VOR850 relationship is also projected by the MME. During 1950–1999, the APO–VOR850 correlation coefficient is 0.47. During 2050–2099, it decreases to 0.41 (significant at the 99% level). Four models (ACCESS1.3, BNU-ESM, CCSM4, and GISS-E2-H) show the same projection as the MME (Fig. 10b). Similarly, the MME projects a slight reduction of the APO–HUM relationship, with the correlation coefficient decreasing from 0.51 during 1950–1999 to 0.43 during 2050–2099 (significant at the 99% level). However, large differences exist among the individual models. Two models (BNU-ESM and CCSM4) show decreased correlations and three models (ACCESS1.3, GISS-E2-H, and NorESM1-ME) show increased correlations (Fig. 10c).

For the APO–DIV1000 relationship (Fig. 10d), the MME result suggests a slight decrease during 2050–2099 under the RCP8.5 scenario. The correlation coefficient is –0.41 (significant above the 99% level), close to that (–0.44) during 1950–1999. There are also substantial differences among the individual models. Three models (BNU-ESM, CCSM4, and GISS-E2-H) project suppressed correlations, while two models (ACCESS1.3 and NorESM1-ME) project intensified correlations. The magnitudes of increase or decrease in correlation coefficients for the individual models are in general greater than the change in the MME result.

For the APO–DIV150 correlation (Fig. 10e), the MME indicates that it will decrease from 0.51 during 1950–1999 to 0.38 during 2050–2099 (significant at the 99% level). Compared to the historical simulation, the variance contribution of the APO to the upper-level divergence is reduced by 11.6% under the RCP8.5 scenario. For the individual models, all except NorESM1-ME project such a weakening correlation.

The relationship between the APO and UV850 is also projected by the MME to decline (Fig. 10f). The correlation coefficient is 0.59 during 1950–1999 but 0.49 during 2050–2099 (significant at the 99% level). That is, the variance of the APO contributing to the northward–southward movement of the WPSH would decrease by 8.8% under the RCP8.5 scenario. Such a decrease is projected consistently by four models (ACCESS1.3, BNU-ESM, CCSM4, and GISS-E2-H).

5 Conclusion and discussion

In this study, we evaluate the capacity of 32 CMIP5 models to simulate the linkage of the summer APO to the vertical zonal wind shear, low-level vorticity, mid-level humidity, low- and high-level atmospheric divergence and location of the WPSH that are associated with the TC frequency over the western North Pacific. It is found that five models (ACCESS1.3, BNU-ESM, CCSM4, GISS-E2-H, and NorESM1-ME) can simultaneously reproduce the significant relationship apparent in the observation. Further, the five “best” models are selected as the MME members to project their potential relationship during 2050–2099 under the RCP8.5 scenario. The main results can be summarized as follows:

(1) The MME projects that the relationship between the APO and the vertical zonal wind shear will weaken. Such a weakening tendency is projected consistently by all of the individual models.

(2) The MME also projects that the APO will still be significantly correlated to the low-level vorticity, mid-level humidity, and high- and low-level atmospheric divergence over the western North Pacific, as well as the northward–southward shift of the WPSH, but their correlation is inclined to reduce to differing degrees.

(3) The weakening in the relationship of the APO with the low-level vorticity, high-level divergence, and northward–southward shift of the WPSH is consistently projected by 80% of the individual models. However, relatively large discrepancies exist among individual models with respect to the projection of the linkage between the APO and the mid-level humidity and low-level divergence.

Therefore, it can be deduced that the summer APO will still exert impacts on the TC frequency over the western North Pacific, but their relationship during 2050–2099 will become slightly weaker than at present. It is important to acknowledge, however, that this conclusion is an estimate based only on the current state-of-the-art models. Whilst the ensemble simulations of the five “best” models (those that can reproduce the observed relationship well) justify the projection, a number of uncertainties still exist in the projection due to the uncertainty of the emissions scenario and the model differences. For example, both the MME and the individual models project that the relationship between the APO and vertical zonal wind shear will weaken as a response to future warming. Therefore, the warming scenario is one main driver responsible for the weakening of their relationship. In contrast, the association of the APO with low-level atmospheric divergence is projected by three models to decline, and by two models to increase, thereby resulting in no significant change as projected by the MME. This discrepancy may be related to the differences among climate models and their response to the forcing. It also reflects the uncertainty due to model differences. Certainly, detailed investigation is needed in future research.

In addition, the present study concentrates on the future relationship of the APO with the atmospheric background favorable for TC genesis over the western North Pacific on the interannual timescale.Chen (2009) proposed from the interdecadal viewpoint that the strength and movement of the WPSH can modulate TC frequency in the northwestern quadrant of the western North Pacific. Therefore, the nature of the linkage of TC genesis location to the northward–southward movement of the WPSH, and whether there is a decadal shift, under the future warming scenario, are important questions to resolve. Due to comparatively coarse horizontal resolutions, the CMIP5 models do not output data on TCs. Therefore, these issues need to be studied by using high-resolution regional climate models, which will be the focus of our group's work in the near future.

References
Chen Lianshou and Ding Yihui, 1979:Summary of Tropical Cyclones over the Western North Pacific. Science Press, Beijing, 491 pp. (in Chinese)
G. Chen ,2009: Interdecadal variation of tropical cyclone activity in association with summer monsoon, sea surface temperature over the western North Pacific. Chinese Sci. Bull. , 54 , 1417–1421. DOI:10.1007/s11434-008-0564-2
X. Cui, B. T. Zhou, K. Fan ,2010: Linkage between Asian-Pacific oscillation and the large-scale atmospheric circulations related to the tropical cyclone frequency over the western North Pacific in Bergen climate model. Climatic Environ. Res. , 15 , 120–128.
Y. H. Ding, E. R. Reiter ,1983: Large-scale circulation influencing the typhoon formation over the western Pacific. Acta Oceanol. Sin. , 5 , 561–574.
K. Fan ,2007: New predictors and a new prediction model for the typhoon frequency over western North Pacific. Sci. China: Earth Sci. , 50 , 1417–1423. DOI:10.1007/s11430-007-0105-x
K. Fan, H. J. Wang ,2009: A new approach to forecasting typhoon frequency over the western North Pacific. Wea. Forecasting , 24 , 974–986. DOI:10.1175/2009WAF2222194.1
W. M. Gray ,1968: Global view of the origin of tropical disturbances and storms. Mon. Wea. Rev. , 96 , 669–700.
C. H. Ho, J. H. Kim, H. S. Kim, et al ,2005: Possible influence of the Antarctic Oscillation on tropical cyclone activity in the western North Pacific. J. Geophys. Res. , 110 , D19104. DOI:10.1029/2005JD005766
IPCC, 2013:Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 1535 pp.
E. Kalnay, M. Kanamitsu, R. Kistler, et al ,1996: The NCEP/NCAR 40-yr reanalysis project. Bull. Amer. Meteor. Soc. , 77 , 437–471.
J. Kidston, E. P. Gerber ,2010: Intermodel variability of the poleward shift of the austral jet stream in the CMIP3 integrations linked to biases in 20th century climatology. Geophys. Res. Lett. , 37 , L09708. DOI:10.1029/2010GL042873
M. A. Lander ,1996: Specific tropical cyclone track types and unusual tropical cyclone motions associated with a reverse-oriented monsoon trough in the western North Pacific. Wea. Forecasting , 11 , 170–186.
B. Liebmann, H. H. Hendon, J. D. Glick ,1994: The relationship between tropical cyclones of the western Pacific and Indian Oceans and the Madden–Julian oscillation. J. Meteor. Soc. Japan , 72 , 401–412.
R. H. Moss, J. A. Edmonds, K. A. Hibbard, et al ,2010: The next generation of scenarios for climate change research and assessment. Nature , 463 , 747–756. DOI:10.1038/nature08823
J. Q. Sun, H. P. Chen ,2011: Predictability of western North Pacific typhoon activity and its factors using DEMETER coupled models. Chinese Sci. Bull. , 56 , 3474–3479. DOI:10.1007/s11434-011-4640-7
K. E. Taylor, B. J. Stouffer, G. A. Meehl ,2012: An overview of CMIP5 and the experiment design. Bull. Amer. Meteor. Soc. , 93 , 485–498. DOI:10.1175/BAMS-D-11-00094.1
H. J. Wang, K. Fan ,2007: Relationship between the Antarctic Oscillation in the western North Pacific and typhoon frequency. Chinese Sci. Bull. , 52 , 561–565. DOI:10.1007/s11434-007-0040-4
H. J. Wang, J. Q. Sun, K. Fan ,2007: Relationships between the North Pacific Oscillation and the typhoon/hurricane frequencies. Sci. China: Earth Sci. , 50 , 1409–1416. DOI:10.1007/s11430-007-0097-6
Q. Y. Zhang, J. B. Peng ,2003: The interannual and interdecadal variations of East Asian summer circulation and its impact on the landing typhoon frequency over China during summer. Chinese J. Atmos. Sci. , 27 , 97–106.
Q. Y. Zhang, S. Y. Tao, L. T. Chen ,2003: The inter-annual variability of East Asian summer monsoon indices and its association with the pattern of general circulation over East Asia. Acta Meteor. Sinica , 61 , 559–568.
P. Zhao, Y. N. Zhu, R. H. Zhang ,2007: An Asian–Pacific teleconnection in summer tropospheric temperature and associated Asian climate variability. Climate Dyn. , 29 , 293–303. DOI:10.1007/s00382-007-0236-y
B. T. Zhou, X. Cui ,2008: Hadley circulation signal in the tropical cyclone frequency over the western North Pacific. J. Geophys. Res. , 113 , D16107. DOI:10.1029/2007JD009156
B. T. Zhou, X. Cui ,2014: Interdecadal change of the linkage between the North Atlantic Oscillation and the tropical cyclone frequency over the western North Pacific. Sci. China: Earth Sci. , 57 , 2148–2155. DOI:10.1007/s11430-014-4862-z
B. T. Zhou ,2016: The Asian–Pacific Oscillation pattern in CMIP5 simulations of historical and future climate. Int. J. Climatol. , 36 , 4778–4789. DOI:10.1002/joc.4668
B. T. Zhou, X. Cui, P. Zhao ,2008: Relationship between the Asian–Pacific Oscillation and the tropical cyclone frequency in the western North Pacific. Sci. China: Earth Sci. , 51 , 380–385. DOI:10.1007/s11430-008-0014-7