Simulation and Projection of the Western Paci fi c Subtropical High in CMIP5 Mod

LIU Yunyun,LI Weijing,ZUO Jinqing,and HU Zeng-Zhen

1 Laboratory for Climate Studies,National Climate Center,China Meteorological Administration,Beijing 100081,China

2 Climate Prediction Center,NCEP/NWS/NOAA,College Park 20740,USA

Simulation and Projection of the Western Paci fi c Subtropical High in CMIP5 Models

LIU Yunyun1∗,LI Weijing1,ZUO Jinqing1,and HU Zeng-Zhen2

1 Laboratory for Climate Studies,National Climate Center,China Meteorological Administration,Beijing 100081,China

2 Climate Prediction Center,NCEP/NWS/NOAA,College Park 20740,USA

This work examined the performance of 26 coupled climate models participating in the Coupled Model Intercomparison Project Phase 5(CMIP5)in the simulation of the present-day temporal variability and spatial pattern of the western Paci fi c subtropical high(WPSH).The results show that most models are able to capture the spatial distribution and variability of the 500-hPa geopotential height and zonal wind fi elds in the western subtropical Paci fi c,but with underestimated mean intensity of the WPSH.The underestimation may be associated with the cold bias of sea surface temperature in the tropical Indian and western Paci fi c oceans in the models.To eliminate the impact of the climatology biases,the climatology of these models is replaced by that of the NCEP/NCAR reanalysis in the veri fi cation,and the models reproduce the WPSH’s enhancement and westward extension after the late 1970s.According to assessment of the simulated WPSH indices,it is found that some models(CNRM-CM5,FGOALS-g2,FIO-ESM,MIROC-ESM,and MPI-ESMP)are better than others in simulating WPSH.Then,the ensemble mean of these better models is used to project the future changes of WPSH under three representative concentration pathway scenarios(RCP8.5, RCP4.5,and RCP2.6).The WPSH enlarges,strengthens,and extends westward under all the scenarios, with the largest linear growth trend projected in RCP8.5,smallest in RCP2.6,and in between in RCP4.5; while the ridge line of WPSH shows no obvious long-term trend.These results may have implications for the attribution and prediction of climate variations and changes in East Asia.

western Paci fi c subtropical high(WPSH),simulation and projection,CMIP5,RCP scenarios

1.Introduction

Western Paci fi c subtropical high(WPSH)is one of the most important atmospheric circulation systems over East Asia.Its variability in the location and intensity has important impacts on the summer rainfall anomaly over China(Xu et al.,2001;Tao and Wei, 2006).Abundant moisture from the tropical oceans transports to eastern China through the southerly fl ow from its western boundary,then converges with the cold air from the high latitude and conforms the front, causing rain belts at the northwestern margin of the WPSH and hot and drought summer under its covered area(Han and Wang,2007).Clearly,its interannual variability causes droughts and fl oods over East China(Wu et al.,2002),while its interdecadal variability modulates the drought and fl ood pattern over East Asia(Hu,1997;Xiong,2001).Liu et al.(2013) noted that the biennial component of the WPSH in intensity and zonal position also has the obvious interdecadal transition in the late 1970s,with larger amplitudes during the recent 30 years.As a result,great attention has been given to variations of WPSH on the interannual and interdecadal timescales,since understanding the variations will help us to improve the forecast of summer climate anomaly in East Asia.

To simulate and project climate change,climate model becomes one of the main quantitative tools now. Due to a lot of uncertainties and biases in the climate models and in order to improve model simulations,the Coupled Model Intercomparison Project (CMIP)has been proposed by the World Climate Research Program(WCRP).The last phase of CMIP (CMIP5)involves about 30 climate modeling groups around the world,aiming to advance our knowledge of climate variability and climate change and projection, and to provide simulations for evaluation in the IPCC Fifth Assessment Report(AR5)(Taylor et al.,2012). Compared with the models used in the previous phase of CMIP simulations(such as CMIP3),most models in CMIP5 have been improved in many aspects,including physical processes and coupled carbon cycle. Thus,it will be interesting to assess the performance of CMIP5 models in the simulation of WPSH mean state and variability.Also,it is meaningful to project the possible changes of WPSH under di ff erent climate warming scenarios in the future.

In this work,the capacities of 26 CMIP5 models in simulating WPSH are evaluated from various aspects based on a set of reconstructed indices that describe the WPSH objectively(Liu et al.,2012).It is expected to provide some valuable information for the model development and improvement in the future by comparisons between the simulations and observations,and among the models.Finally,the models with good performance are identi fi ed and selected to project the possible evolution of WPSH in the future,which is expected to provide implication for future trends of the East Asian summer climate over China.

2.Data and CMIP5 models

Table 1 gives the basic description of the 26 coupled climate models participating in CMIP5. The simulation data used in this study include the data from two groups of model runs.(1)Historical runs, which are initiated from an arbitrary point of a quasiequilibrium control run and integrated longer than 156 yr(1850–2005).The historical run is forced by timeevolving greenhouse gases,ozone,aerosols,and a so-lar constant that are consistent with the observations, and for the fi rst time,time-evolving land cover/land use pattern is included(Taylor et al.,2012).(2)Future climate change projection runs,which are forced with three typical representative concentration pathways(RCPs),i.e.,RCP2.6,RCP4.5,and RCP8.5.Every emission scenario has a set of speci fi ed concentration of greenhouse gases,aerosols and other chemistry gases,and land cover/land use pattern(Moss et al., 2010;Xin et al.,2012).

Table 1.Description of the 26 coupled climate models participating in the CMIP5

Atmospheric monthly reanalysis(Kalnay et al., 1996)provided by NCEP/NCAR and the Extended Reconstructed SST dataset(ERSSTv3b)(Smith et al., 2008)from NOAA are used in this study.Hereafter we refer to these observation-based datasets as observations.The analysis period of the historical runs and observations is from 1951 to 2005,and the period of the future projection runs is from 2006 to 2099.The model output atmospheric fi elds are interpolated to the regular grids with a horizontal resolution of 2.5°×2.5°,and SST fi elds are interpolated to the regular grids with a horizontal resolution of 2°×2°for facilitating the comparison among CMIP5 models.

Normally,WPSH is represented in the weather chart as the region surrounded by the contour of 5880 gpm at 500-hPa level within 10°–45°N,110°E–180°. In order to quantitatively describe the variation of WPSH in the intensity and position,the monthly WPSH indices are computed based on NCEP/NCAR reanalysis,including the area,intensity,ridge line,and western boundary indices(Liu et al.,2012).It is well known that WPSH is strongest in summer,with signi fi cant impacts on the summer rainfall over East Asia (Nitta and Hu,1996;Wu et al.,2002).Therefore,we will examine the WPSH in summer(June,July,and August)in terms of its spatial distribution,amplitude variation,and interdecadal and interannual variations, while the entire time series of the monthly data is used only in power spectrum analysis.

To assess the model’s ability in simulating current climate,Taylor diagram(Taylor,2001)is used to centralize multi-model related information.The correlation coefficient between the model simulations and observations represents the model’s capability to simulate climatic variability in the concerned region. The root-mean-square error(RMSE)indicates deviation of the model from the observations(the closer to zero the RMSE is,the higher simulation capability the model has).The standard deviation ratio of model simulation relative to the observation indicates the model’s ability in simulating the amplitude of variability.These three assessment indicators displayed in a Taylor diagram can re fl ect the overall performance of the model.Therefore,Taylor diagram is used in this study to evaluate the 26 CMIP5 models,where both the self-correlation coefficient and the standard deviation of the observations are 1,and RMSE is 0.

3.Climatology of WPSH

3.1 Spatial distribution

According to the de fi nition of WPSH,the simulated geopotential height(H500)and zonal wind fi elds at 500 hPa(u500)by CMIP5 are examined fi rst. Note that the model biases present in the spatial pattern,coverage,as well as intensity of WPSH relative to the observations(Fig.1).The simulated H500 in CCSM4,CESM1-CM5,and FIO-ESM models is stronger than the observations,with a larger coverage area surrounded by the 5880-gpm contour.In the other CMIP5 models,however,the simulated H500 is much weaker than the observations,without the 5880-gpm isoline over the subtropical western Paci fi c in summer when WPSH is seasonally strongest in the observation;even no 5840-gpm isoline presents at H500 in HadGEM2-CC and IPSL-CM5A-LR models.Furthermore,the simulation of u500=0 m s−1is not good as well( fi gure omitted).Most of the models fail to capture the position of the WPSH ridge line except three models which reproduce the 5880-gpm contour. It is also noted that the simulated easterly wind belt on the south of WPSH is much weaker than the observation.This is consistent with the overall weak WPSH simulated by these models.

Fig.1.Climatological means of H500 in summer from the observation(NCEP/NCAR reanalysis)and CMIP5 model simulations.Light and dark shaded areas denote H500 larger than 5840 and 5880 gpm,respectively.

Systematic errors usually exist in the simulated atmospheric circulation in current global atmosphereocean coupled models(also known as the model climate drift;Sun and Ding,2008;Huang and Qu,2009; Feng and Li,2012).These biases in simulating the WPSH are common in most state-of-the-art GCMs,implying a challenge in simulating and predicting summer climate variability over East Asia(Zhang and Chen,2011a,b).Considering that the intensity and spatial pattern of WPSH are in fl uenced by the thermal anomalies from underlying surface,especially the sea surface temperature(SST)in the tropical Paci fi c and Indian Ocean(Nitta,1987;Huang and Sun,1994; Nitta and Hu,1996;Wu et al.,2002),we plot the observed and simulated climatological tropical SST in summer and the model biases relative to the observations in Fig.2.The result shows that the SST in both tropical Paci fi c warm pool and Indian Ocean is above 28°C in summer,which is favorable to the maintenance of the WPSH.However,compared to the observations,the simulated regions surrounded by the contour of SST=28℃ in most CMIP5 models are much smaller,indicating cold biases in the tropical Indian Ocean and Paci fi c warm pool.This consists with the underestimated H500 over the western Paci fi c in the simulations.Nevertheless,the three models(CCSM4, CESM1-CAM5,and FIO-ESM),which have produced realistic H500,also reproduce the SSTs in the tropical Indian and Paci fi c oceans with smaller cold biases. These results may suggest a connection in simulating the spatial distribution of tropical SST and in simulating the WPSH.

Fig.2.Climatological mean of SST(contours;°C)in summer from the observation(ERSSTv3b)and CMIP5 model simulations,with the model biases shaded.The contour interval is 2℃.The red line is isoline of 28℃.

3.2Climatology calibration

The Taylor diagram of the climatological H500 over the western Paci fi c in summer compared to the observation(Fig.3a)shows that the simulations of the spatial pattern of H500 are similar to the observation, with the correlation coefficient above 0.96,even 0.99 in 5 models.RMSEs of all models are within 1 gpm. The high correlations and small RMSEs indicate good simulations of the spatial distribution of H500 over the western Paci fi c.It is also noted that the amplitude of H500 is also well captured,with standard deviation of the models ranging from 0.6 to 1.6 gpm.Figure 3b is the Taylor diagram of the simulated climatological u500.All of the correlation coefficients with the observation are above 0.90,RMSEs are less than 0.5 m s−1,and the standard deviations are from 0.5 to 1.5 m s−1,which imply decent performances of the 26 CMIP5 models in capturing the spatial distribution and amplitude of u500.

We note that although there are signi fi cant model biases in simulating H500 and u500(Fig.1),possibly associated with cold SST biases in the tropical Indian and Paci fi c oceans(Fig.2),the spatial pattern and variability of H500 and u500 over the western Paci fi c are well reproduced in all of the 26 CMIP5 models (Fig.3).To correct the climatology biases in the simulations,the simulated climatology of H500 and u500 from all of the models is replaced by that from the observations.In other words,all the model outputs are calibrated to the same climatic state,and then superimposed with each model’s own variability of H500 and u500.For example,the calibration of H500 for a model(the same in u500)is:

wherehmodelrepresents an H500 simulation value from a speci fi ed model,hmodelrepresents the model climatology of H500,andhncepdenotes the climatology of H500 from the observation.In the following,the verifi cation of WPSH is based on the outputs from climatology calibrated runs.

4.Variability of WPSH

4.1 WPSH indices

Fig.3.Taylor diagrams of(a)simulated climatological H500 and(b)u500 over the western Paci fi c in summer compared to the observation.REF indicates the reference value of 1.The radial distance of the model code pointing from the origin is the standard deviation ratio of the model relative to the observation.The correlation coefficient of spatial pattern between model and observation is shown by the cosine of the azimuthal angle of model code point,and their root-mean-square error is given by the distance of model code pointing from the REF.

Fig.4.Time series of the WPSH indices in summer from the observation(thick black line)and calibrated model simulations(thin colored lines)for the period from 1951 to 2005.(a)Area index,(b)intensity index,(c)ridge line index, and(d)western boundary index.The thick blue line is the ensemble mean of the 26 CMIP5 models.

To characterize the variability of WPSH in intensity and position objectively,Fig.4 shows the 9-yr running mean time series of the WPSH indices,including the area,intensity,ridge line,and western boundary indices from both observation and calibrated model simulations. The observation curves show that the area and intensity indices of WPSH increase signi fi cantly and the western boundary index decreases after the late 1970s,i.e.,WPSH becomes stronger and more westward-extending during the recent 30 years.Such an interdecadal shift of WPSH has also been noted in previous works(Hu,1997;Huang et al.,2006;Zhao et al.,2007;Liu and Ding,2012).Interestingly,there is a slightly downward trend in the observed ridge line index(Fig.4c),suggesting that WPSH slightly shifts southward after the late 1970s, consistent with Nitta and Hu(1996)and Hu(1997).

Compared with the observations,all the calibrated model results capture the interdecadal shift of WPSH in the late 1970s.In these simulations,the observed greenhouse gases,ozone,aerosols,and solar constant,and the variability of land cover are all taken into account in the models(Taylor et al.,2012). Thus,it is suggested that these external forcing factors may have played an important role in causing the interdecadal variability of WPSH,although the internal variability of the atmospheric circulation over the western Paci fi c may also be in fl uenced by the local atmosphere-ocean interaction(Wang et al.,2005;Zhu and Shukla,2013).

4.2 Interdecadal variability of WPSH in spatial pattern

Fig.5.Distributions of the contour of(a,b)H500=5880 gpm and(c,d)u500=0 m s−1averaged in(a,c)1951–1960 and(b,d)1995–2005 from the observation(thick black line)and calibrated model simulations(thin colored lines).

To display visually the interdecadal shift of WPSH in its spatial pattern,distributions of the contours of H500=5880 gpm and u500=0 m s−1averaged in two periods of 1951–1960 and 1996–2005 are computed and compared(Fig.5).During the period of 1951–1960,WPSH is relatively weaker and eastward,with the observed 5880-gpm contour between 20°and 30°N,and its western boundary no more than 140°E(Fig.5a).The observed ridge line is near 25°N, with a northeastern-southwestern direction(Fig.5c). Most of the calibrated simulations are able to reproduce WPSH,i.e.,the 5880-gpm contour,much better than the model results before climatology calibration (Fig.1).Unfortunately,there are still 10 models failing to capture the 5880-gpm contours.Compared with simulation of the 5880-gpm contour line,the ridge lines of WPSH are better simulated,except near the coastland of China of about 115°E.

In the period of 1996–2005,WPSH becomes stronger and more westward,the area of the observed WPSH increases,with its southern and northern borders extending to 18°and 32°N,and the western boundary to 135°E(Fig.5b).The calibrated model results well simulate the interdecadal shift of WPSH. The area and intensity indices of WPSH increase signi fi cantly relative to the period of 1951–1960 in the model simulations(Figs.5b and 5d).

4.3 Linear trend and standard deviation of WPSH indices

In order to quantitatively assess the capabilities of the CMIP5 models in simulating WPSH indices,linear trends of each WPSH index from all the CMIP5 models are calculated,except for the ridge line index (trend too small)(see Fig.6).The observed results show that the linear ascending trend of WPSH area and intensity indices are more than 20%per decade, and the trend of the western boundary is –2.7%per decade,at the signi fi cance level of 0.01.Compared to the observations,all the models well simulate the tendency of enhancing and westward extension of WPSH during 1951–2005,but with large quantitative di ff erences.Some are larger than the observation in the intensity of WPSH,such as IPSL-CM5A-LR and GFDLESM2G,while others are weaker than the observations,such as GFDL-CM3 and HadGEM2-ES.Taking all the three WPSH indices into account,it is found that the simulations of nine models,i.e.,CESM1-CAM5-1-FV2,CNRM-CM5,FGOALS-g2,FIO-ESM, HadCM3,HadGEM2-CC,MIROC-ESM,MPI-ESMP,and NorESM1-M,are better than others in capturing the linear trend of WPSH indices.

Fig.6.Linear trend coefficients of the WPSH indices in summer from the observation(black bar)and CMIP5 model simulations(grey bars)for the period 1951–2005.(a)Area index,(b)intensity index,and(c)western boundary index.

The standard deviation of WPSH indices can be used to measure the capability of the models in simulating the interannual variability of WPSH.Figure 7 shows the simulated standard deviation ratio of the WPSH indices relative to the observations.The closer to 1 the ratio is,the better to simulate the interannual variability of WPSH indices the model is.The grey bars in Fig.7 denote the accumulated distance between the ratios of these four WPSH indices and the reference value of 1.The smaller the accumulated distance is,the closer to observations the simulated WPSH index is,which provides a measurement of the model performance in simulating the overall feature of WPSH.It is noted that most of the standard deviation ratios of the calibrated simulations to the observationsare larger than the reference value,indicating an overestimation of the amplitudes of WPSH indices in the models.GFDL-ESM2M,MIROC5,and INMCM4 models have relatively poor simulations in the interannual variability of WPSH,because of the too large accumulated ratios in the GFDL-ESM2M and MIROC5, and the too small one in INMCM4.According to Fig. 7,it is found that six models:ACCESS1-0, CanESM2,CNRM-CM5,FGOALS-g2,IPSL-CM5AMR,and MIROC-ESM,perform relatively better.

Fig.7.Standard deviation ratios of the simulated WPSH indices relative to the observation.The grey bars are the accumulated distance of the standard deviation ratio of the four WPSH indices from the reference value of 1.The smaller the grey bar is,the closer to the observation the simulated result is.

4.4 Interannual variability of WPSH

The interannual variability is another important feature to describe the characteristics of WPSH.Figure 8 shows the power spectrum of the monthly WPSH intensity index in the entire analysis period from the observation and 26 CMIP5 models.Considering that seasonal variability of WPSH is more signi fi cant than other timescales,a low-pass fi lter is fi rstly used to remove the seasonal variability(cycle of 11 months)from the raw data for highlighting the interannual signals.

It is known that signi fi cant tropospheric biennial oscillation(TBO)exists over the East Asian monsoon region on interannual timescales(Nitta and Hu, 1996).Being one of the dominant members of the East Asian monsoon system,WPSH has a TBO component on the interannual timescale as well(Liu et al.,2013).The observation shows that there are two interannual periods,i.e.,the quasi-four-year(36–60-month)and the quasi-biennial-year(24–36-month)period,and both pass the 0.05 signi fi cance level of the red noise test.The period of the quasi-biennial variation is consistent with the TBO component of WPSH and the monsoon rainfall in China(Chang and Li,2000; Chang et al.,2000;Ding,2007).The power spectra in the CMIP5 model simulations exhibit obvious differences from each other on the interannual timescales (Fig.8).To some extent,only ACCESS1-0,GFDLCM3,HadGEM2-CC,and HadGEM2-ES can capture both the quasi-four-year period and the TBO period.

Comparison of the 26 CMIP5 model simulations with the observations shows that most of the simulated H500 values in the western subtropical Paci fi c region are weaker than observation,and even the calibrated results of some models still fail to capture the spatial and temporal characteristics of WPSH. According to the overall simulations of the distributions of SST,H500,and u500,and the quantitative assessment of the WPSH indices,it is noted that CNRM-CM5,FGOALS-g2,FIO-ESM,MIROC-ESM, and MPI-ESM-P models are better than others.Due to the unavailability of the MPI-ESM-P model for theRCP scenarios,the ensemble mean results of the four models(CNRM-CM5,FGOALS-g2,FIO-ESM,and MIROC-ESM)are selected to project the evolution of WPSH in the future under di ff erent RCP scenarios.

Fig.8.Power spectra(red solid line)of the monthly WPSH intensity index from the observation and CMIP5 model simulations.Blue dashed line indicates the 0.05 signi fi cance level of the red noise test.

5.Possible changes of WPSH in di ff erent RCP scenarios

To investigate the possible changes of WPSH under three typical RCP scenarios in the 21st century, long-term integration(2006–2099)of H500 and u500 from the four selected models is averaged fi rstly as the climatological state,and then the anomaly of each model is added to calculate the time series of the WPSH indices;an approach similar to that in Section 2.2.Next,the 9-yr running mean of WPSH indices from 2006 to 2099 under di ff erent RCP scenarios from these four models’ensemble mean is used to project the possible changes of WPSH in the future.It is noted that all the simulated area,intensity,and western boundary indices of WPSH display signi fi cantly interdecadal variations in the di ff erent RCP scenarios (Fig.9).

Under the RCP2.6 scenario,both the WPSH area and intensity increase and extend westward obviously. After 2050,the linear trends of the WPSH area,intensity,and western boundary indices gradually approach zero.The linear ascending trends of WPSH under the RCP4.5 scenario are similar to those under the RCP2.6,but the period of signi fi cant growth lasts until about 2070,and then the growth trends weaken.Similar long-term variations of WPSH show up with faster growth in the period and a relatively smaller trend in the latter period under the RCP8.5 scenario.Interestingly,the western boundary index of WPSH maintains at 90°E since the late 2050s.In fact,the most western boundary of the 5880-gpm contour extends westward to the west of 90°E after the late 2050s.In that case,the de fi nition of the western boundary of WPSH is limited as 90°E(Liu et al., 2012).

Fig.9.Nine-yr running mean of the WPSH indices in summer from 2006 to 2099 under di ff erent RCP scenarios from four CMIP5 models’ensemble mean.(a)Area index,(b)intensity index,(c)ridge line index,and(d)western boundary index.Shaded areas represent one standard deviation from the multi-model mean.

Overall,WPSH enlarges,strengthens,and ex-tends westward under di ff erent RCP scenarios,with the largest linear growth trend projected in RCP8.5, weakest in RCP2.6,and in between in RCP4.5.The ridge line of WPSH has no obvious long-term trend in the three RCP scenarios.These results have implications for the attribution and projection of climate changes in East Asia in using the CMIP5 model output.

6.Summary and discussion

The performances of 26 coupled climate models participating in the Coupled Model Intercomparison Project Phase 5(CMIP5)in the simulation of the present-day temporal variability and spatial pattern of the WPSH are assessed.Then,the ensemble mean of four selected models is used to project the future possible changes of WPSH under three typical representative concentration pathways(RCPs).The results are as follows.

(1)The intensity of WPSH is underestimated in most CMIP5 model simulations,which may be associated with colder SST biases in the tropical Indian and western Paci fi c oceans in the models.Nevertheless, spatial distribution and variability of H500 and u500 are reasonably captured.To eliminate the impact of the model climatology biases,the model climatology is replaced with that of NCEP/NCAR reanalysis,which makes the model results more realistic,and without any change in temporal variability.

(2)The climatology-calibrated model results reproduce the observed interdecadal shift of WPSH (enhancement and westward extension after the late 1970s).According to overall assessment of the WPSH indices,it is identi fi ed that CNRM-CM5,FGOALS-g2,FIO-ESM,MIROC-ESM,and MPI-ESM-P have better performances than other models in simulating the WPSH.

(3)The selected models’simulations suggest that WPSH will enlarge,strengthen,and extend westward under the three RCP scenarios,with the highest linear growth trend projected in RCP8.5,in between in RCP4.5,and weakest in RCP2.6.The simulated ridge line of WPSH has no obvious long-term trend in the scenarios.

It is easily noticed that the model defaults,such as the systemic biases in the western subtropical regions,may a ff ect the credibility of these projections. Thus,it is necessary to explore the reasons resulting in the biases as well as their connection with the cold biases of the models in the tropical Indian and western Paci fi c oceans.It is also an interesting topic to examine possible changes of the East Asian summer rain belt under the projection of strengthening and westward extending WPSH associated with the increasing concentration of greenhouse gases in the future.Furthermore,the mechanism for the projected enhancing and westward-extending WPSH under the global warming scenarios is also worthy of further analysis.

Acknowledgments.Manythankstothe support of the National Innovation Team of Climate Prediction of the China Meteorological Administration. Thanks also go to the contribution of the WCRP’s Working Group on Coupled Modeling. NCEP/NCAR reanalysis and NOAA ERSSTv3b data are downloaded from NOAA website at http://www.esrl.noaa.gov/psd/.

REFERENCES

Chang,C.P.,and T.Li,2000:A theory for the tropical tropospheric biennial oscillation.J.Atmos.Sci., 57,2209–2224.

—–,Y.S.Zhang,and T.Li,2000:Interannual and interdecadal variations of the East Asian summer monsoon and tropical Paci fi c SSTs.Part I:Roles of the subtropical ridge.J.Climate,13,4310–4325.

Ding,Y.H.,2007:The variability of the Asian summer monsoon.J.Meteor.Soc.Japan,85B,21–54.

Feng Juan and Li Jianping,2012:Evaluation of IPCC AMIP models in simulating monsoon-like southwest Australian circulation.Climatic and Environ.Res., 17,409–421.(in Chinese)

Han Jinping and Wang Huijun,2007:Features of interdecadal changes of the East Asian summer monsoon and similarity and discrepancy in ERA-40 and NCEP/NCAR reanalysis.Chinese J.Geophys.,50, 1444–1453.(in Chinese)

Hu,Z.-Z.,1997:Interdecadal variability of summer climate over East Asia and its association with 500-hPa height and global sea surface temperature.J.Geo-phys.Res.,102(D16),19403–19412.

Huang Gang and Qu Xia,2009:Meridional location of West Paci fi c subtropical high in summer in IPCC AR4 simulation.Trans.Atmos.Sci.,32,351–359. (in Chinese)

Huang Ronghui and Sun Fengying,1994:Impacts of the thermal state and the convective activities in the tropical western warm pool on the summer climate anomalies in East Asia.Chinese J.Atmos.Sci.,36, 673–685.(in Chinese)

—–,Cai Rongshuo,Chen Jilong,et al.,2006:Interdecaldal variations of drought and fl ooding disasters in China and their association with the East Asian climate system.Chinese J.Atmos.Sci.,30,730–743. (in Chinese)

IPCC,2007:Climate Change 2007:The Physical Scienti fi c Basis.Solomon,S.,D.Qin,M.Manning,et al.,Eds.,Cambridge University Press,Cambridge, United Kingdom and New York,NY,USA,996 pp. Kalnay,E.,M.Kanamitsu,R.Kistler,et al.,1996:The NCEP/NCAR 40-year reanalysis project.Bull. Amer.Meteor.Soc.,77,437–472.

Liu,Y.Y.,Y.H.Ding,H.Gao,et al.,2013:Tropospheric biennial oscillation of the western Paci fi c subtropical high and its relationship with the tropical SST and atmospheric circulation anomalies.Chinese Sci. Bull.,58,3664–3672.

Liu Yunyun and Ding Yihui,2012:Analysis of the leading modes of the Asian-Paci fi c summer monsoon system.Chinese J.Atmos.Sci.,36,673–685.(in Chinese)

—–,Li Weijing,Ai Wanxiu,et al.,2012:Reconstruction and application of the monthly western Paci fi c subtropical high indices.J.Appl.Meteor.Sci.,23, 414–423.(in Chinese)

Moss,R.H.,J.A.Edmonds,K.A.Hibbard,et al.,2010: The next generation of scenarios for climate change research and assessment.Nature,46,747–756.

Nitta,T.,1987:Convective activities in the tropical western Paci fi c and their impact on the Northern Hemisphere summer circulation.J.Meteor.Soc. Japan,65,373–390.

—–,and Z.-Z.Hu,1996:Summer climate variability in China and its association with 500-hPa height and tropical convection.J.Meteor.Soc.Japan,74, 425–445.

Smith,T.M.,R.W.Reynolds,T.C.Peterson,et al., 2008:Improvements to NOAA’s historical merged land-ocean surfacetemperatureanalysis(1880–2006).J.Climate,21,2283–2296.

Sun Ying and Ding Yihui,2008:Validation of IPCC AR4 climate models in simulating interdecadal change of East Asian summer monsoon.Acta Meteor.Sinica, 66,765–780.(in Chinese)

Tao Shiyan and Wei Jie,2006:The westward,northward advance of the subtropical high over the West Paci fi c in summer.J.Appl.Meteor.Sci.,17,513–525.(in Chinese)

Taylor,K.E.,2001:Summarizing multiple aspects of model performance in a single diagram.J.Geophys. Res.,106(D7),7183–7192.

—–,R.J.Stou ff er,and G.A.Meehl,2012:An overview of CMIP5 and the experiment design.Bull.Amer. Meteor.Soc.,93,485–498.

Wang,B.,Q.H.Ding,X.Fu,et al.,2005:Fundamental challenges in simulation and prediction of summer monsoon rainfall.Geophys.Res.Lett.,32,L15711. doi:10.1029/2005GL022734.

Wu Guoxiong,Chou Jifan,Liu Yimin,et al.,2002:Dynamics of the Formation and Variation of Subtropical Anticyclones.Science Press,Beijing,1–20.(in Chinese)

Xin Xiaoge,Wu Tongwen,and Zhang Jie,2012:Introduction of CMIP5 experiments carried out by climate system model.Adv.Climate Change Res., 8,378–382.(in Chinese)

Xiong Anyuan,2001:Analysis on background of climatic variation for extremely rainy over the middle reaches of the Yangtze River in the 1990s.J.Appl.Meteor. Sci.,12,113–117.(in Chinese)

Xu Haiming,He Jinhai,and Zhou Bing,2001:The features of atmospheric circulation during Meiyu onset and possible mechanisms for westward extension (northward shift)of Paci fi c subtropical high.J. Appl.Meteor.Sci.,12,150–158.(in Chinese)

Zhang Hongfang and Chen Haishan,2011a:Evaluation of summer circulation simulation over East Asia by 21 climate models.Part I:Climatology.Scientia Meteor.Sinica,31,119–128.(in Chinese)

—– and —–,2011b:Evaluation of summer circulation simulation over East Asia by 21 climate models. Part II:Interannual variability.Scientia Meteor. Sinica,31,247–257.(in Chinese)

Zhao,P.,Y.N.Zhu,and R.H.Zhang,2007:An Asian-Paci fi c teleconnection in summer tropospheric temperature and associated Asian climate variability.Climate Dyn.,29,293–303.

Zhu,J.S.,and J.Shukla,2013:The role of air-sea coupling in seasonal prediction of Asian-Paci fi c summer monsoon rainfall.J.Climate,26,5689–5697.

:Liu Yunyun,Li Weijing,Zuo Jinqing,et al.,2014:Simulation and projection of the western Paci fi c subtropical high in CMIP5 models.J.Meteor.Res.,28(3),327–340,

10.1007/s13351-014-3151-2.

(Received October 19,2013;in fi nal form February 16,2014)

Supported by the National(Key)Basic Research and Development(973)Program of China(2010CB950501 and 2013CB430202), National Natural Science Foundation of China(41005037),and China Meteorological Administration Special Public Welfare Research Fund(GYHY201306024 and GYHY201306033).

∗Corresponding author:liuyuny@cma.gov.cn.

©The Chinese Meteorological Society and Springer-Verlag Berlin Heidelberg 2014