Interannual relationships between the tropical sea surface...

Interannual relationships between the tropical sea surfacetemperature and summertime subtropical anticycloneover the western North Pacific

Pei‐Hsuan Chung,1 Chung‐Hsiung Sui,2 and Tim Li3

Received 27 December 2010; revised 8 March 2011; accepted 6 April 2011; published 13 July 2011.

[1] The interannual variability of the Western North Pacific Subtropical High (WNPSH)in boreal summer is investigated with the use of the NCEP/NCAR Reanalysis Data. Themost significant change of the 500 hPa geopotential height field appears at the westernedge of the WNPSH, with dominant 2–3 year and 3–5 year power spectrum peaks.The 2–3 year oscillation of the WNPSH and associated circulation and sea surfacetemperature (SST) patterns possess a coherent eastward propagating feature, with a warmSST anomaly (SSTA) and anomalous ascending motion migrating from the tropicalIndian Ocean in the preceding autumn to the maritime continent in the concurrent summerof a strong WNPSH. A strong WNPSH is characterized by anomalous anticycloniccirculation and maximum subsidence in the western North Pacific (WNP). The anomalousWNPSH circulation has an equivalent barotropic vertical structure and resides in thesinking branch of the local Hadley circulation, triggered by enhanced convection over themaritime continent. A heat budget analysis reveals that the WNPSH is maintained byradiative cooling. The 3–5 year oscillation of the WNPSH exhibits a quasi‐stationaryfeature, with a warm SSTA (anomalous ascending motion) located in the equatorialcentral eastern Pacific and Indian Ocean and a cold SSTA (anomalous descending motion)located in the western Pacific. The anomaly pattern persists from the preceding winterto the concurrent summer of a high WNPSH. The greatest descent is located to thesoutheast of the anomalous anticyclone center, where a baroclinic vertical structure isidentified. The zonal phase difference and the baroclinic vertical structure suggest that theanomalous anticyclone on this timescale is a Rossby wave response to a negative latentheating associated with the persistent local cold SSTA. ECHAM4 model experimentsfurther confirm that the 2–3 year mode is driven by the SSTA forcing over the maritimecontinent, while the 3–5 year mode is driven by the local SSTA in the WNP.

Citation: Chung, P.‐H., C.‐H. Sui, and T. Li (2011), Interannual relationships between the tropical sea surface temperature andsummertime subtropical anticyclone over the western North Pacific, J. Geophys. Res., 116, D13111,doi:10.1029/2010JD015554.

1. Introduction

[2] The subtropical high over the North Pacific is a domi-nant component of atmospheric general circulation thatexperiences a distinct seasonal cycle with the maximumstrength and northernmost position occurring in the borealsummer. The evolution of the subtropical high over thewestern North Pacific (hereafter WNPSH) in summer isassociated with the onset and withdrawal of the Asian sum-

mer monsoon. The zonal shift of the western part of theWNPSH dominates the position of large‐scale frontal zonesin East Asia [Tao and Chen, 1987]. A southwestwardextension of the WNPSH would result in more water vaporconvergence and excessive rainfall along the Yangtze Rivervalley [Chang et al., 2000a, 2000b; Zhou and Yu, 2005]. Theprecipitation characters in Japan and Korea were also greatlyinfluenced by the movement of the WNPSH through chang-ing the transport of water vapor into the Baiu or Changmafrontal area [e.g., Tomita et al., 2004; Lee et al., 2005].[3] Traditionally, subtropical highs are regarded as the

descending branch of the Hadley circulation maintained byradiative cooling. This idealized, zonally symmetric Hadleycirculation model, however, would lead to a weaker(stronger) subtropical high in the summer (winter) hemi-sphere, which is contrary to observations [Hoskins, 1996].Moreover, the observed summertime subtropical highs are

1Department of History and Geography, Taipei Municipal University ofEducation, Taipei, Taiwan.

2Department of Atmospheric Sciences, National Taiwan University,Taipei, Taiwan.

3International Pacific Research Center, University of Hawai’i at Mānoa,Honolulu, Hawaii, USA.

Copyright 2011 by the American Geophysical Union.0148‐0227/11/2010JD015554

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 116, D13111, doi:10.1029/2010JD015554, 2011

D13111 1 of 19

http://dx.doi.org/10.1029/2010JD015554

in the form of high‐pressure cells rather than a zonallyuniform high‐pressure belt. The dynamics of the summer-time subtropical high is better understood in the context ofplanetary waves, as suggested by Ting [1994], who foundthat summertime subtropical stationary wave patterns aremaintained by the global diabatic heating.[4] Hoskins [1996] and Rodwell and Hoskins [2001]

applied the monsoon‐desert mechanism proposed byRodwell and Hoskins [1996] to argue that the remote diabaticheating over the North American monsoon region couldinduce a Rossby wave pattern to the west and contribute tothe lower tropospheric subtropical anticyclone in the NorthPacific. Rodwell and Hoskins [2001] further examined thedynamics of the summertime subtropical high in a globalnonlinear model and concluded that the combined effect ofthe topography and the monsoonal heating is of primaryimportance for the generation of the surface subtropicalhighs and subsidence aloft over the eastern Pacific. Thisdiffers from Chen et al. [2001], who argued that summertimesubtropical highs can be maintained by downstream energypropagation of stationary Rossby waves forced by convec-tion heating over the Asian monsoon region in a linear quasi‐geostrophic model. Their experiments, however, could notproduce the distinct meridional dependence of the observedvertical structures of the summertime planetary wave, pos-sibly due to the lack of other heating sources such as sensibleheating over the western portion of the continent and radia-tion cooling off the west coast of the continent. Liu et al.[2004] performed atmospheric GCM experiments to showthe importance of the global surface sensible heating andlongwave radiative cooling in the formation of the subtrop-ical anticyclones. Miyasaka and Nakamura [2005] demon-strated that the summertime subtropical high can beaccurately reproduced by the near‐surface thermal contrastbetween a cooler ocean and a hotter land to the east. The roleof local air‐sea interactions in generating the subtropicalanticyclone was investigated numerically by Seager et al.[2003]. All studies above suggest the important role of theland‐ocean‐atmosphere feedback in maintaining the sub-tropical high.[5] In addition to the annual cycle, the subtropical high in

the boreal summer also exhibits remarkable interannualvariations. Convective heating over the western Pacificwarm pool is regarded as an important forcing that affectsthe extent of the subtropical high in the boreal summer.According to Nitta [1987], convective activity in the trop-ical western Pacific during summer can influence theNorthern Hemispheric circulation; that is, Rossby wavesgenerated by the anomalous tropical heat source can prop-agate into higher latitudes. This leads to a north–southdipole between subtropics and middle latitude East Asia,known as the Pacific‐Japan (PJ) Pattern. Lu [2001] foundthat stronger convection over the warm pool is associatedwith more eastward and northward migration of the sub-tropical high. Lu and Dong [2001] performed numericalmodel experiments and showed that suppressed convectioncaused by a cold SSTA in the warm pool results in anom-alous anticyclonic circulation in the lower troposphere overthe subtropical western North Pacific (WNP), providing afavorable condition for the westward extent of the sub-tropical high.

[6] In addition to the warm pool heating, the El Nino‐Southern Oscillation (ENSO) may also influence the EastAsian climate through the modulation of low‐level windsover the WNP [e.g., Wang et al., 2000; Chang et al., 2000a,2000b; Lau and Nath, 2000; Kawamura et al., 2001; Lauand Wu, 2001; Chou, 2004]. An anomalous low‐levelanticyclone dominates the WNP during the mature phase ofan El Nino, whereas an anomalous cyclone occurs duringthe La Nina mature winter. The anomalous anticyclonepersists through the following spring and summer, leadingto a weak WNP monsoon in the ENSO decaying summer[e.g., Wang et al., 2000, 2003; Chou et al., 2003]. Theanticyclone in early summer favors the northward move-ment of the Mei‐yu/Baiu front and thus enhances (weakens)the East Asia (WNP) summer monsoon rainfall [Changet al., 2000a, 2000b].[7] The wintertime anomalous low‐level anticyclone in

WNP may be regarded as a Rossby wave response to thesuppressed convection over the western Pacific, which ispossibly caused both by the remote warm SSTA over theeastern Pacific [Wang et al., 2000; Chou, 2004; Chen et al.,2007] and the local cold SSTA [Lau and Nath, 2000; Wanget al., 2000]. The maintenance of the WNP anticyclone fromthe El Nino mature winter to its decaying early summer isattributed to a season‐dependent thermodynamic air‐seafeedback [Wang et al., 2000, 2003; Kawamura et al., 2001;Chou, 2004; Li and Wang, 2005]. During the El Ninodecaying summer, the SSTA in the eastern Pacific hasdiminished, yet the low‐level anticyclone still exists over theWNP. It is likely that both the cold SSTA in the WNP in latespring/early summer [Sui et al., 2007; Wu et al., 2010] andthe basin‐wide warming in the IO in summer [Yang et al.,2007; Xie et al., 2009; Wu et al., 2009a, 2009b] areresponsible for maintaining the WNPSH during the summer.[8] The Asian summer monsoon shows a strong quasi‐

biennial variability [e.g., Shen and Lau, 1995; Barnett,1991; Chang and Li, 2000; Li et al., 2001; Li and Zhang,2002], which is termed as the tropospheric biennial oscil-lation (TBO) in the Asian‐Australian monsoon region[Meehl, 1994, 1997]. The tropical Pacific and southeast IO(SEIO) SSTA associated with the Asian summer monsoonalso exhibits a biennial characteristic [Chang et al., 2000a;Wang et al., 2001]. A hybrid coupled GCM simulationshows that the TBO may exist even in the absence of ENSO[Li et al., 2006]. On the other hand, ENSO may be linked tothe monsoon TBO through the following three branches ofteleconnection: (1) WNP monsoon‐SEIO SSTA tele-connection or the Philippine‐Sumatra pattern [Li et al.,2002; Li et al., 2006; Wu et al., 2009b]; (2) IO convec-tion‐Central Eastern Pacific (CEP) SSTA teleconnection[Meehl, 1987; Chang and Li, 2000; Yu et al., 2002;Watanabe and Jin, 2002; Li et al., 2003]; and (3) Pacific‐East Asia teleconnection [Wang et al., 2000, 2003].[9] Because the WNPSH is regarded as one of the

important components of the Asian summer monsoon sys-tem, the objective of this study is to conduct an analysis ofthe interannual evolution of the WNPSH and investigatepossible mechanisms that link to the variability. The restof the paper is organized as follows. Section 2 describes thedata used in this study. In section 3 a WNPSH index isdefined and two distinctive spectrum peaks on quasi‐biennial

CHUNG ET AL.: TROPICAL SSTA AND SUBTROPICAL HIGH D13111D13111

2 of 19

(2–3 year) and lower‐frequency (3–5 year) periods areidentified. A regression analysis is subsequently performedto obtain the large‐scale circulation, SST, outgoing long-wave radiation, and heat and moisture budgets associatedwith the two oscillation periods. By synthesizing all features,possible mechanisms responsible for the two oscillations ofthe WNPSH are proposed. The idealized experiments of theECHAM4 AGCM are also examined in section 4. In section5 the seasonal lead‐lag patterns of regressed SSTA and cir-culation fields associated with the 2–3 year and 3–5 yearoscillations of the WNPSH are discussed. Section 6 com-prises a summary and discussions.

2. Data and Model Description

2.1. Reanalysis Data

[10] The primary data set used in this study is the NationalCenters for Environmental Prediction–National Center forAtmospheric Research (NCEP/NCAR) Reanalysis [Kalnayet al., 1996] that includes wind, geopotential height,stream function, vertical p velocity, sea level pressure,velocity potential, and outgoing longwave radiation flux ona 2.5 × 2.5 grid for 48 years (1958–2005). This is the lon-gest reanalysis product available. Another set of data used inthe study is the NOAA Extend Reconstructed Sea SurfaceTemperature version 2 (ERSST v2) developed on a 2° × 2°grid for the 48 years (1958–2005). The extended recon-struction of historical SST was using improved statisticalmethods from 1854 to the present. A full description of theERSST v2 is available in the work of Smith and Reynolds[2004].[11] Further, this study used the European Centre for

Medium‐Range Weather Forecasts (ECMWF) Reanalysis(ERA), which also contains the above variables at the sameresolution from 1957 to 2002. A full description of the ERAis available in the work of Gibson et al. [1996, 1997]. Theanalysis results from this study, based on NCEP/NCARdata, were compared against those from ERA and the majorfeatures were found to be essentially the same. Most of theresults discussed in this study are based on NCEP/NCARreanalysis, except the following.[12] In order to discuss the dynamic features of the cli-

mate oscillations in this study, the heat and moisture bud-gets were computed, the so‐called apparent heat source (Q1)and apparent moisture sink (Q2) [e.g., Yanai et al., 1973],using ERA. This decision is made due to the fact that theoverall simulation of the diabatic heating (and thereforethe moisture) field over the monsoon region in ERA is quiteclose to reality [Annamalai et al., 1999]. In addition, acomparison of the heat and moisture budgets based onERA, NCEP/NCAR R‐1 Reanalysis, and NCEP/DOEAMIP‐II (R‐2) Reanalysis [Kanamitsu et al., 2002] wascompleted. The results show that budgets computed basedon both ERA and R‐2 reanalysis data are quite similar to theobserved rainfall variability in the tropical and East Asianmonsoon regions. However, the R‐2 reanalysis data onlycover the satellite period from 1979 to present. Thereforethe Q1 and Q2 budgets were computed at each vertical level(23 pressure levels) using the 6‐hourly ERA data from 1958to 2001.

[13] According to Yanai et al. [1973], the large‐scale heatand moisture budget residuals, Q1 and Q2, respectively, aredefined and computed by

Q1 � cpp

p0

� �k @�

@tþ~� � r�þ !

@�

@p

� �ð1Þ

and

Q2 � �L@q

@tþ~� � rqþ !

@q

@p

� �ð2Þ

where � is the potential temperature, q is the water vapormixing ratio, ~� is the horizontal velocity, w is the vertical pvelocity. L is the latent heat of vaporization, k = R/cp with Rthe gas constant and cp the specific heat capacity at constantpressure of dry air, and p0 = 1000 hPa.[14] In equations (1) and (2),~�, �, and q denote large‐scale

fields that are mathematically defined as Reynolds averagedquantities (normally with an overbar, being neglected here)and are customarily assumed resolvable by reanalysis data.So Q1 and Q2 are computed by 6‐hourly data at each gridpoint from ERA products. The 6‐hourly estimates of Q1

and Q2 are averaged into monthly mean values and multi-plied by cp

−1 to be expressed in terms of equivalent warming/cooling rate (K/day).[15] The Q1 and Q2 can be interpreted by:

Q1 ¼ QR þ L c� eð Þ � r � s′v′ � @s′!′

@pð3Þ

and

Q2 ¼ L c� eð Þ þ Lr � q′v′ þ L@q′!′

@pð4Þ

where QR is radiative heating rate, c and e are condensa-tion and evaporation per unit mass of air respectively, s isthe dry static energy, and the prime denotes deviation fromthe Reynolds average. The primed variables denote subgridscale fluctuations that are assumed to be well separated fromthe resolvable large‐scale variable. Equation (3) representsthe total effect of radiative heating, latent heat released bynet condensation L(c − e), and the horizontal and vertical con-vergence of sensible heat fluxes due to subgrid‐scale eddiessuch as cumulus convection and turbulence. Equation (4),on the other hand, represents the total effect of net con-densation and divergence of eddy moisture fluxes due tosubgrid‐scale eddies. The contributions of the eddy hori-zontal transport terms r · s′v′ in equation (3) and r · q′v′ inequation (4) are generally small and may be negligible com-pared to the horizontal transports by the large‐scale motion[Yanai and Johnson, 1993].

2.2. Atmospheric General Circulation Model

[16] TheAtmospheric General CirculationModel (AGCM)used in the study is ECHAMversion 4.6 (hereafter ECHAM4),which was developed by the Max Planck Institute for Mete-orology [Roeckner et al., 1996]. The model was run at ahorizontal resolution of spectral triangular 42 (T42), with 2.8°grid and 19 vertical levels in a hybrid sigma pressure coor-dinate system. The climatological monthly SST constructed


3 of 19

by phase II of the Atmospheric Model IntercomparisonProject (AMIP II) was used as the model low boundaryforcing. The ECHAM4 model is one of the Top‐3 AMIP IImodels in simulating the interannual variability modes of theAsian‐Australian monsoon [Zhou et al., 2009a]. It showsexcellent performance in simulating the El Nino‐related tro-pospheric temperature anomalies [Zhou and Zhang, 2011].Thus the model is used in the study for investigating theinterannual variations of the summertime circulation in WNP.

3. Characteristics Associated Withthe Anomalous Subtropical High

3.1. WNPSH Index

[17] In this study the interannual variation of the summersubtropical high over theWNP is measured using the 500 hPageopotential height field (Z500). Although the summerPacific subtropical high can be detected in the lower tropo-sphere, the WNPSH is most pronounced in the middle tro-posphere. The center of the subtropical high in the lowertroposphere is located in the eastern Pacific, while the centerof the subtropical high in the middle troposphere resides overthe western North Pacific [Liu et al., 2001; Liu and Wu,2004]. For this reason, the variations of the WNPSH inthis study are measured by the Z500 variability, followingmany previous studies [e.g., Tao and Chen, 1987; Zhou andLi, 2002]. Figure 1 shows the standard deviation (~�Z) of thesummer (JJA) mean Z500 during the period of 1958–2005. Itis calculated from 2 to 7 year band‐pass filtered data and hasbeen normalized by its zonal mean value to illustrate thezonally asymmetric character. Figure 1 shows that the largestvariability occurs in the northwest Pacific region, southwestof the climatological mean subtropical high as depicted bythe contour line of Z500 = 5870 m. A WNPSH index isdefined as the area mean value of Z500 anomalies in JJA inthe region of (120°–140°E, 10°–30°N). This index is used tomeasure the strength of the interannual variability of the

WNPSH. For comparison, two additional indices are definedbased on 850 hPa stream function (y850) and geopotentialheight anomalies (Z850) in JJA averaged over the region of(120–150°E, 15–25°N), where the strongest interannualvariation of y850 and Z850 occurs (hereafter termed as y850

and Z850 indices). The y850 and Z850 fields were used by Leeet al. [2005] and Lu [2001] to study the subtropical highvariability over the WNP.[18] The power spectrums of the Z500, y850, and Z850

indices are examined. The Z500 index shows two dominantperiods at 2–3 years and 3–5 years, both of which pass the95% confidence level (Figure 2a). The y850 and Z850 indexalso presents two similar peaks although with the 3–5 yearband (2–3 year band) being above (slightly below or at) the95% confidence level (figures not shown). Additional powerspectrum analyses with 44‐year ERA40 Z500 data (figure notshown) show a similar result. Thus in the following, the Z500index will be used to represent the WNPSH quasi‐biennial(2–3 year) and lower‐frequency (3–5 year) variability.[19] Through a Fourier series decomposition of the

WNPSH index series, two band‐pass filtered time serieswere constructed by summing Fourier components withinthe 2–3 year and 3–5 year periods (Figure 2b). Hereafter, thetwo time series are referred to as the 2–3 year and 3–5 yearWNPSH indices. Positive indices represent the strong andwestward extending phase of the subtropical high, whilenegative indices represent the opposite. In the followinganalysis, the circulation and SST fields are regressed againstthe two WNPSH indices to investigate spatial and temporalstructures associated with the 2–3 year and 3–5 year oscil-lations of the WNPSH.

3.2. Circulation Patterns Associated With the 2–3 Yearand 3–5 Year Oscillations

[20] The structures related to 2–3 year and 3–5 yearoscillations of the WNPSH were investigated by examiningthe horizontal patterns of regressed fields of summer mean

Figure 1. The standard deviation of 2–7 year band‐pass filtered summer‐mean geopotential height at500 hPa (Z500) normalized by the corresponding zonal mean Z500 (shaded) and 5870‐m contour linefor a strong (red), normal (green), and weak (blue) WNPSH composite.


4 of 19

ERSST, outgoing longwave radiation (OLR), Z500, 500 hPavertical p velocity (w500), 200 hPa and 850 hPa streamfunction (y200 and y850), 200 hPa and 850 hPa velocitypotential (c200 and c850), and wind fields at 850 hPa and500 hPa (V850 and V500). Since the derived results of regres-sion can be applied to both the positive phase (represented bya strengthened and westward WNPSH) and the negativephase (represented by a weakened WNPSH), the interpreta-

tion and discussion of physical mechanisms afterward will bemerely based on the features of the positive phase.[21] The regressed Z500, V500, y850, and y200 are shown in

Figure 3. The WNPSH high anomaly of the 2–3 yearoscillation is associated with a meridional standing wavepattern with a zonally elongated positive Z500 anomalyand anticyclonic circulation in the WNP, a negative Z500anomaly over the northern East Asia, and a positiveanomaly over Eurasia (Figure 3b). The horizontal distribu-tion of y850 in Figure 3c shows a pair of anticyclonic cir-culations symmetric about 10 N. One is located in the WNP(overlapping with a positive Z500 anomaly center) and theother is located south of the Equator. The circulation patternof y200 (Figure 3a) shows that an anticyclonic circulationextends from southern to southeastern China and the WNP.[22] In the 3–5 year oscillation, an anticyclone and a

positive Z500 anomaly center (Figure 3e) are located nearTaiwan when the WNPSH extends westward. The positiveZ500 anomaly extends further to the tropical central easternPacific than that in 2–3 year oscillation. The positiveanomaly in Z500 and y850 fields (Figures 3e and 3f) acts asthe source of a Rossby wave train propagating from theWNP to North America. The distribution of y200 (Figure 3d)exhibits a cyclonic circulation pattern over China. In theWNP, y200 is in general out of phase with y850.[23] Figure 4 shows the regressed SST patterns for the 2–

3 year and 3–5 year modes. The SST field has been nor-malized at each grid point based on the climatologicalstandard deviation to capture the obvious SSTA variations.The SST distribution associated with the westward extendingWNPSH in the 2–3 year oscillation (Figure 4a) features awarm SSTA extending from the northern IO southeastwardto South China Sea, maritime continent and northeasternAustralia, and a cold SSTA in the central eastern equatorialand northeastern and southeastern Pacific. For the 3–5 yearoscillation (Figure 4b), however, a warm SSTA appears inthe tropical IO and the central eastern equatorial Pacific. TheSSTA in the central eastern Pacific is surrounded by coldSST anomalies to its north and south and in the maritimecontinent.[24] In the 2–3 year oscillation, the large‐scale divergent

circulation in upper and lower levels are shown by regressedc200 and c850 fields in Figures 5a and 5b. A broad over-turning circulation occurs in the summer of a strongWNPSH over the western Pacific, with the low‐level con-vergent (divergent) flow toward Java and the eastern IO (outof the tropical central eastern Pacific region) (Figure 5b).The regressed OLR field in Figure 5c shows enhancedconvection over the maritime continent and suppressedconvection over the Philippine Sea and the equatorial east-ern Pacific, consistent with the divergent wind in Figures 5aand 5b. A zonal band of enhanced convection is alsoobserved to the north of the WNPSH. The anomalous OLRpattern in the western Pacific resembles a negative phase ofthe PJ pattern [Nitta, 1987].[25] For the 3–5 year oscillation, the divergent circulation

pattern differs from that of the 2–3 year oscillation. Theregressed c200 and c850 fields (shown in Figures 5d and 5e)reveal two anomalous Walker cells in the tropics, with twoupper‐tropospheric divergence centers located in the tropicalcentral‐eastern Pacific and IO, respectively, and a conver-gence center located in the region extending from southern

Figure 2. (a) Power spectrums of the summer‐mean geo-potential height at 500 hPa (WNPSH index) and (b) decom-posed band‐pass filtered time series of the (top) 2–3 yearsand (bottom) 3–5 years WNPSH index. The pink curve inFigure 2a denotes the 95% confidence level.


5 of 19

Philippines to the maritime continent and eastern Australia(Figure 5d). The regressed OLR field (Figure 5f) shows theregions of suppressed convection in the maritime continentand subtropical North Pacific and the regions of enhancedconvection in the equatorial eastern Pacific and westerntropical IO, consistent with the features of the divergentflows shown in the c200 and c850 fields (Figures 5d and 5e).[26] The meridional overturning circulation of the 2–

3 year oscillation over the western Pacific is further revealedby the regressed zonally averaged vertical p velocity (w) ateach pressure level and latitude‐vertical divergent windprofile (v, w) along 110°E–140°E (Figure 6a). Consistentwith the regressed OLR and c fields, the flow is upwardover the maritime continent (0°–10°S) and East Asia nearJapan (30°–40°N), with the former being stronger than thelatter. The southern branch of upward motion causes upper‐tropospheric divergent flows toward the Philippine Sea (at10°–20°N) where the downward motion is pronounced. The

low‐level southward divergent flow to the south of theWNPSH may bring the moisture into the equatorial regionand enhance convection in the maritime continent. Theenhanced convection further reinforces the WNPSH subsi-dence via a positive feedback process.[27] For the 3–5 year oscillation, the regressed vertical

overturning circulation along 110°E–150°E (Figure 6b)shows two descending branches at 0°–10°S and 15°–25°N,coinciding with the two suppressed convective centers inthe OLR field. There is no direct linkage between the WNPand equatorial regions (Figure 5f). It can be seen from canalysis in Figures 5d and 5e that the equatorial branch ofthe descending motion within the western Pacific is a resultof the anomalous large‐scale Walker circulation. In the cfield, however, the convergent flow over the WNP is not asstrong as that in the maritime continent/eastern Australia(Figure 5d). The result suggests that unlike the 2–3 yearoscillation, the subsidence in the WNP is not directly

Figure 3. Regressed summer‐mean fields of stream function at (a, d) 200 hPa and (c, f) 850 hPa and(b, e) geopotential height with significant wind fields at 500 hPa for the 2–3 year (Figures 3a, 3c, and 3e)and 3–5 year (Figures 3b, 3d, and 3f) oscillations of WNPSH. Shaded (vectors) areas pass the 90% confi-dence level.


6 of 19

attributed to the equatorial forcing in the 3–5 year oscilla-tion. Rather, it may be forced through a Rossby waveresponse to anomalous heating associated with the localcold SSTA in Figure 4b [Sui et al., 2007; Wu et al., 2010].

3.3. Vertical Structures of the WNPSH Anomalies

[28] The analysis above indicates that the large‐scalecirculations associated with an enhanced summertimeWNPSH in the 2–3 year and 3–5 year oscillations are dis-tinctly different. The 2–3 year oscillation is accompanied byan anomalous local Hadley circulation that connects themaritime continent and the WNP, while the 3–5 yearoscillation is characterized by the anomalous Walker cir-culation and a cold SSTA to the southeast of the anomalousWNPSH. To reveal the specific processes that affectWNPSH, the horizontal and vertical structures of variousdynamic and thermodynamic variables associated with thetwo oscillations are further compared.[29] First, the relative zonal position between the anom-

alous anticyclone and descending motion centers during astrong WNPSH phase are examined by depicting the re-gressed y850 (contours) and w500 (blue shades) fields inFigure 7. Only the stronger (at the 95% significant level)portion of the regressed positive w500 field is shown inFigure 7. While the subsidence center (suppressed convec-tion in Figure 5c) in the 2–3 year oscillation is zonally in

phase with the anticyclone center, the maximum subsidence(suppressed convection in Figure 5f) in the 3–5 year oscil-lation is located to the southeast of the anomalous anticy-clone center. Such a phase difference suggests that theanomalous WNPSH in the former is possibly a directresponse to the local Hadley circulation anomaly, whereas inthe latter it is a baroclinic Rossby wave response to a neg-ative heating anomaly induced by the local cold SSTA(Figure 4b).[30] To reveal the vertical structure of the anomalous

WNPSH, the domains of the strongest downward motionand anomalous subtropical high center on the 2–3 year and3–5 year timescales respectively were chosen. The verticalprofiles of the domain‐averaged regressed fields of verticalp velocity, relative vorticity, Q1 and Q2 are computed ateach pressure level.[31] Figure 8a shows the regressed vertical p velocity

profiles over the area of strongest downward motion in the2–3 year and 3–5 year oscillations. The strongest subsidenceis located in the upper troposphere (300 hPa) in the 2–3 yearoscillation but in the middle troposphere (400–500 hPa) inthe 3–5 year oscillation. The magnitude of the subsidence inthe 3–5 year oscillation is weaker than that in the 2–3 yearoscillation. Figure 8b shows the vertical profiles of theregressed relative vorticity fields over the anomalous anti-cyclone center associated with the 2–3 year and 3–5 year

Figure 4. Regressed summer‐mean SSTA fields in (a) the 2–3 year and (b) 3–5 year oscillations ofWNPSH, normalized by climatological standard deviations at each grid point. Shaded areas pass the90% confidence level.


7 of 19

oscillations. While it exhibits a clear equivalent barotropicstructure in the 2–3 year oscillation, the vorticity has abaroclinic structure in the 3–5 year oscillation, changingfrom negative anomalies in the lower troposphere to apositive anomaly in the upper troposphere (above 250 hPa).This result is consistent with circulation patterns over theWNP area shown in Figures 3a, 3c, 3d, and 3f.[32] The vertical distributions of regressed Q1 and Q2

anomalies averaged over the strongest descent region areshown in Figures 8c and 8d. Note that the amplitude of thenegative Q1 anomaly in the middle‐upper troposphere isgreater in the 2–3 year oscillation than in the 3–5 yearoscillation (Figure 8c), while the amplitude of the negativeQ2

anomaly in the lower troposphere is greater in the 3–5 yearoscillation than in the 2–3 year oscillation (Figure 8d). Thisimplies that enhanced longwave radiative cooling due tothe reduction of the vertically integrated moisture caused bythe descent‐induced dry advection plays an important role

in the establishment of the anomalous WNPSH on the 2–3 year timescale. On the other hand, a peak of anomalousQ2 at 850 hPa in the 3–5 year oscillation implies strongnegative condensation heating at the top of the boundarylayer. The negative condensation heating is induced bythe local cold SSTA over the region of (150°–170°E, 15°–20°N) that is to the southeast of the anomalous anticycloneand is responsible for a baroclinic vertical structure.[33] In both the oscillations the warm SSTA under the

anomalous anticyclone may be regarded as a passiveresponse to the atmospheric forcing [Wang et al., 2005; Wuand Kirtman, 2005; Zhou et al., 2008]. The anomalousanticyclone leads to the suppression of convection and thusthe increase of downward solar radiation into the oceansurface. Meanwhile, less evaporation due to the weakeningwind speed may also increase SST.[34] The distinctive horizontal and vertical structures

above suggest that the physical mechanisms responsible

Figure 5. Regressed summer‐mean velocity potential and divergent wind at (a, d) 200 hPa and (b, e)850 hPa and (c, f) upward longwave radiation flux (similar to outgoing longwave radiation) for the2–3 year (Figures 5a, 5c, and 5e) and 3–5 year (Figures 5b, 5d, and 5f) oscillations of WNPSH. Thered (blue) contours denote positive (negative) velocity potential anomalies or enhanced (suppressed) con-vection. Shaded regions pass the 90% confidence level.


8 of 19

for the anomalous WNPSH may differ in the 2–3 year and3–5 year oscillations. According to the Gill [1980] model, aRossby wave response is located to the west of a heatingsource. This corresponds well to the 3–5 year oscillationcase, in which there is a phase shift between the anomalousanticyclone and the anomalous heating (or vertical motion).Further, the Rossby wave response to the anomalous heatingis baroclinic. These features are accurately observed in the3–5 year oscillation. Therefore the diagnosis result in thisstudy suggests that the variation of the WNPSH on the 3–5 year timescale is primarily caused by an atmosphericRossby wave response to a negative heating associated witha cold SSTA to the southeast of the anticyclone center. Incontrast, in the 2–3 year oscillation, the anomalous WNPSHcenter is zonally colocated with the vertical motion, and itsvertical structure is approximately barotropic. This impliesthat the Rossby wave response is unlikely and the down-ward motion is a major driving mechanism. That is, thestrengthening of the WNPSH on the 2–3 year timescaleis primarily attributed to the radiative cooling due to thedescent‐induced dry advection; the less clouds and watervapor in the atmosphere, the more longwave radiation is

emitted into space. The change of local vertical motion andcirculation in the WNP is induced by the remote forcing viaanomalous local Hadley circulation associated with enhancedconvective heating in the maritime continent.[35] The regressed results above are consistent with the

summertime subtropical LOSECOD diabatic heating patternidentified by Wu and Liu [2003] and Wu et al. [2009], whoshowed that the dominant diabatic heating sources over thenorthwestern Pacific region are convective heating andlongwave radiation cooling. The corresponding diagnoses ofFigures 5, 7, and 8 indeed demonstrate that in the 2–3 yearoscillation, the longwave radiation cooling associated withthe anomalous descent by the local Hadley circulation pre-vails over the negative convective heating; while in the 3–5 year mode, the negative convective heating due to the coldSSTA prevails over the longwave radiation cooling. Thenegative convective heating will induce an atmosphericRossby wave response to the west as the anomalousWNPSH. As a consequence, the subtropical anticycloneanomaly of the 2–3 year oscillation possesses a barotropicstructure, whereas in the 3–5 year oscillation possesses abaroclinic structure as presented in Figure 8b. This is

Figure 6. Regressed vertical velocity (contours) and divergent wind fields (a) averaged over 110°E–140°Efor 2–3 year and (b) averaged over 110°–150°E for 3–5 year oscillations of WNPSH. Shaded regions passthe 90% confidence level.


9 of 19

because a convective heating in the subtropics can generatea Sverdrup balance stream function pattern with anticyclone(cyclone) to the east and cyclone (anticyclone) to the westof the heating (negative heating) in the lower troposphereand the opposite in the upper troposphere, as discussed byWu and Liu [2003], Liu et al. [2004], and Wu et al. [2009].Therefore the distinct difference between the two oscillationmodes should be attributed to the different dominant dia-batic heating over the northwest Pacific.

4. Model Experiment Results

[36] To test the hypothesis that the WNPSH variability onthe 3–5 year timescale is primarily forced by local SSTA inthe WNP while the WNPSH variability on the 2–3 yeartimescale is attributed to the forcing over the maritimecontinent, the following control and sensitivity experimentshave been developed. In the control run (CTRL), ECHAM4was integrated for 20 years, forced by monthly climato-logical SST (based on the period of 1955–2000). In theWNP run, the model was forced by the climatological SSTplus a uniform SSTA of −1°C over a WNP domain (150°E–160°W, 10°–25°N) (blue box in Figure 9a). In the MC run,

the same simulation strategy was used as in the WNP runexcept with a +1°C SSTA specified in the maritime conti-nent region (90°–150°E, 20°S–5°N) (red box in Figure 9b).The amplitude of this specified SSTA in the WNP and MCregions is 2–3 times larger than the composite SSTA fieldassociated with the observed high WNPSH indices, in orderto obtain a robust response. The WNP and MC runs eachconsists of seven ensemble members which have initialconditions starting from 27 April 0000 UTC to 30 April0000 UTC, separating by 12 h apart. All experiments wereintegrated from May to August. The atmospheric responseto the local and remote SSTA forcing is estimated based onthe difference between the ensemble mean of the WNP/MCrun and the CTRL run (i.e., WNP‐CTRL and MC‐CTRL).[37] The anomalous summertime (JJA mean) precipitation

and y850 field responses to the WNP SSTA are shown inFigure 9a. The imposed local cold SSTA causes the decreaseof local precipitation to the southwest and an increase ofprecipitation in the tropical central western Pacific along5°N. An anomalous low‐level anticyclone forms to the east ofTaiwan in response to the cold SSTA forcing. The simulatedpatterns are qualitatively consistent with the observationsfrom the reanalysis data sets (Figures 3f, 5f, and 7b), but

Figure 7. Regressed summer‐mean stream function at 850 hPa (contour) and positive p velocity (onlymost significant descending region is indicated here by shading) at 500 hPa for (a) the 2–3 year and (b) 3–5 year oscillations of WNPSH.


10 of 19

the amplitude of the simulated precipitation and anticycloniccirculation is too strong, due to the greater SSTA specified.In the observation, the SSTA in the WNP decreases grad-ually from June to August [Wu et al., 2010].[38] The strong positive SSTA forcing in the MC induces

a positive precipitation anomaly over the northern part of theSSTA domain and a negative precipitation anomaly in thenorthern Philippines along 15°N (Figure 9b). An anomalousanticyclone is simulated near Taiwan and is located to thenorth of the negative precipitation anomaly center, similar to

the observed anticyclone‐vertical motion phase relation. Inthe tropical western Pacific, anomalous easterly flows in thesouthern part of the anticyclone extend to the north IO andconverge toward the MC region. The simulated patterns inthe MC run are quite similar to those observed shown inFigure 3c, 5c, and 7a but with less zonal extent and aslightly northward shift due to the neglect of tropical centralPacific cold SSTA forcing in Figure 4a.[39] It is seen from Figures 9a and 9b that the spatial

phase patterns between the anomalous precipitation center

Figure 8. Vertical profiles of regressed (a) vertical p velocity, (c) Q1, and (d) Q2 averaged over theregion of strongest descending motion and (b) relative vorticity averaged over the region of anomalousWNPSH center for the 2–3 year (red) and 3–5 year (blue) oscillations.


11 of 19

and the anticyclone center in both the WNP and MC runsare similar to those observed. In addition, the vertical pro-files of the vertical velocity and relative vorticity fieldsaveraged over the maximum precipitation and anticyclonicvorticity center in both the WNP and MC runs were eval-uated. In Figure 9c the anomalous descending motion in theMC run is larger than that in the WNP run. The strongestdescending motion is located in the middle tropospherearound 400–600 hPa in the MC run but at 500–700 hPa inthe WNP run. In the lower troposphere below 850 hPa,the amplitude of the descent in the WNP and MC run isslightly reversed as in the observation of NCEP/NCAR R‐1(Figure 8a). In addition to the p velocity, the vertical dis-tributions of relative vorticity over the anticyclone center inthe WNP and MC run also shows a distinct difference,similar to the observation. The anomalous anticyclone in theMC run (Figure 9d) is associated with the negative vorticitythroughout the troposphere that resembles the observedbarotropic structure in the 2–3 year oscillation (Figure 8b).In the WNP run (Figure 9d), the negative vorticity in thelower troposphere changes sign to a positive one at 400 hPa,consistent with the observed baroclinic structure in the 3–5 year oscillation (Figure 8b), although the baroclinicity inthe model is a little stronger due to neglecting other minorremote SSTA forcing in the simulation. The vertical distri-bution of the specific humidity (figure not shown) alsoresembles similar result in observed Q2 (Figure 8d) that theamplitude of the negative specific humidity anomaly in thelower troposphere is greater in the WNP run than in the MCrun.[40] The horizontal and vertical structures of the anoma-

lous circulation in the idealized numerical experimentsaccurately capture the observed characteristics in the 2–3 year and 3–5 year oscillations. Numerical experiments inthe study demonstrate the role of SSTA forcing in the WNPand MC regions in generating the anomalous anticycloneassociated with the westward extension of the WNPSHin the two oscillations. The baroclinic vertical structure inthe WNP run confirms that a Rossby wave response to thelocal cold SSTA is a key process involved in the 3–5 yearoscillation. Since the cold SSTA in the WNP weakens fromearly to late summer when El Nino is decaying in the 3–5 year oscillation, the SSTA forcing in the tropical IO mayalso contribute to the westward movement of the subtropicalhigh [Xie et al., 2009; Wu et al., 2010]. The numericalexperiment of Zhou et al. [2009b] has also shown that awarmer tropical IO may contribute to the westward exten-sion WNPSH in the interdecadal time scale. In an additionalidealized experiment with a +1°C SSTA specified in thetropical IO basin (20°S–20°N, figure not shown), ananomalous low‐level anticyclonic circulation is simulated inthe WNP, with a maximum center located slightly south-westward over the South China Sea. Since the simulatedanticyclone differs from the observed WNPSH in its loca-tion (a southwestward shift from the observed center) andvertical structure (the simulated baroclinic structure is muchweaker), the warm SSTA in the tropical IO may not be themajor forcing in causing the westward shift of the WNPSHin the 3–5 year oscillation. In the MC run, the simulatedhorizontal pattern and the vertical structure resemble theobserved in the 2–3 year oscillation. This confirms the roleof the remote warm SSTA forcing in the MC in inducing the

anomalous WNPSH through an anomalous local Hadleycirculation as discussed in section 3.

5. Temporal Evolution Features

[41] To better understand the interannual variability of theWNPSH, the seasonal evolution of anomalous SST andcirculation fields of w500, y850, and V850 associated with theWNPSH 2–3 year and 3–5 year oscillations are examined.More than one full cycle (from MAM(0) to JJA(2)) andnearly one full cycle (from MAM(0) to JJA(3)) of regressedfields are presented for 2–3 year and 3–5 year oscillations.Seasons of MAM(1), JJA(1), SON(1) denote the concurrentspring, summer and autumn of the westward WNPSH,whereas the numbers 0, 2, and 3 in parentheses denote thelead and lag years relative to the current year (1).

5.1. SSTA and Vertical Velocity

[42] The evolution of regressed SSTA for the 2–3 yearoscillation (Figure 10a) shows the following features.Starting from the preceding spring MAM(0), cold SSTanomalies dominated the northern IO, MC, and northernAustralia, while a weak warm SSTA appeared in the equa-torial central Pacific. The warm SSTA in the central Pacificmoved eastward and intensified from JJA(0) to SON(0),weakened afterward through MAM(1), and became nega-tive by the current JJA(1). After JJA(1), the cold SSTAstrengthened until D(1)JF(2) and then decayed from MAM(2)to JJA(2) when the cold SSTA turned into a weak warmSSTA. Accompanying the above evolution, a weak warmSSTA in the central IO in the previous summer JJA(0) grewthrough SON(0). By D(0)JF(1), the warm SSTA had spreadwidely to the northern IO, MC, and northern Australia. Thebroad‐scale warm SSTA further extended southeastwardfrom D(0)JF(1) to MAM(1) and eastward from MAM(1) toSON(1). By the following summer JJA(2), the SSTA hadevolved to a pattern similar to that in MAM(0). The aboveevolution feature resembles the quasi‐biennial oscillation ofthe tropical SSTA signal found in previous studies [e.g.,Barnett, 1991; Li et al., 2006].[43] Contrary to the 2–3 year oscillation, the evolution

of the 3–5 year oscillation shows a standing oscillation,with the SSTA persisting from MAM(0) to SON(1) andreversing its sign from D(1)JF(2) to JJA(3) (Figure 10b). Awarm SSTA developed in the central eastern Pacific fromMAM(0), reached its warmest phase in D(0)JF(1), and thenweakened in the following three seasons to D(1)JF(2) whena weak cold SSTA grew. The cold SSTA became strongerfrom MAM(2) to D(2)JF(3), weakened in MAM(3), andcontinued to last for two seasons until SON(3) (figure notshown). Accompanying the warm SSTA in the equatorialPacific, the surrounding cold SSTA to the northwest andsouthwest also persisted from JJA(0) to SON(1). In D(1)JF(2),the areas surrounding the equatorial Pacific were replacedby a warm SSTA that can persist to JJA(3). In addition, warmSST anomalies developed in the IO from D(0)JF(1) strength-ened and extended to the WNP and marginal seas along theEast Asia coast through MAM(1) to SON(1). In the followingseasons, weak cold SST anomalies appeared in the IO fromMAM(2), extended to the north and south from JJA(2) to D(2)JF(3), and reached the strongest intensity over the wholeocean basin in JJA(3). The evolution of the regressed SSTA


12 of 19

Figure 9. Anomalous precipitation (shaded, only areas passing the 90% confidence level plotted) and850 hPa stream function in (a) the WNP run and (b) the MC run, and vertical profiles of (c) anomalousvertical p velocity over the strongest descending region and (d) anomalous relative vorticity over theanticyclone center in the WNP (blue line) and the MC (red line) runs. The thick blue and red box inFigures 9a and 9b shows the region where uniform (either −1°C or +1°C) SSTA is specified.


13 of 19

patterns described above resembles a typical El Nino evo-lution with a long life cycle [e.g., Wang et al., 2000, 2003].The 3–5 year oscillation of the WNPSH may be modulatedby the remote low frequency El Nino SSTA forcing [Wu andZhou, 2008] via local air‐sea coupling in WNP [Wang et al.,2000; Sui et al., 2007].[44] The time‐longitude section of the regressed SST and

w500 fields along the Equator from MAM(0) to JJA(2) (toJJA(3) in 3–5 year oscillation) is shown in Figure 11. Forthe 2–3 year oscillations (Figures 11a and 11c), there is anobvious eastward propagation of the warm SSTA andupward motion from the IO to the western Pacific during theperiod from JJA(0) to JJA(2). The warm SSTA propagatedto the tropical western Pacific in D(0)JF(1) where theupward motion developed two seasons later in JJA(1). In thecentral eastern Pacific, warm SSTA (upward motion) existedfrom MAM(0) to MAM(1) and was followed by a coldSSTA (downward motion) from JJA(1) to JJA(2) with aslight eastward propagation. The center of the upwardmotion within (170°W–150°W) is located to the west of thewarm SSTA in the region (150°W–120°W) over the tropicalcentral eastern Pacific. However, over the tropical westernPacific, the two fields do not show a significant phasedifference.[45] The regressed fields for the 3–5 year oscillation

(Figures 11b and 11d) exhibit a standing sandwich patterncharacterized by a warm (cold) SSTA and upward (down-ward) motion in the IO, a cold (warm) SSTA and downward(upward) motion in the western Pacific, and a warm (cold)SSTA with upward (downward) motion in the central east-ern Pacific from MAM(0) to D(1)JF(2) [from MAM(2) toSON(3)]. Although warm (cold) SST anomalies are gener-ally accompanied by upward (downward) motion in theocean basins, this feature is not so clear in the IO where asignificant SSTA is only associated with weak verticalmotion. In the tropical western Pacific, the strongest des-cending (ascending) motion lagged the strongest cold(warm) SSTA by 1 month, developing in JJA(0) (MAM(2)),reaching the strongest intensity in D(0)JF(1) [D(2)JF(3)],and then persisting to JJA(1) (JJA(3)).[46] The spatial and temporal evolutions of the SSTA in

the 2–3 year and 3–5 year oscillations are consistent with thestudy by Barnett [1991], who found that the near‐equatorialcharacteristics of the quasi‐biennial SST/SLP mode arethose of a quasi‐progressive wave while the 3–7 year modeis closer to a standing wave.

5.2. Stream Function and Wind

[47] The examination of temporal evolution of theregressed c200 (figures not shown) and w500 (Figure 11c)from JJA(0) to SON(1) indicates that in the 2–3 year oscil-lation, a large‐scale upper‐level divergence (ascending) waslocated in the eastern Pacific and northern IO from JJA(0)to MAM(1), while a large scale upper‐level convergence(descending) appeared over the western Pacific. The patternwas nearly out of phase in JJA(1), with a dominant upper‐level divergence near Java and a large‐scale convergenceover the tropical central eastern Pacific (Figure 5a). It shouldbe noted that the c200 pattern has no significant signal in thetransition season MAM(1). This implies that the signal fromthe preceding boreal summer JJA(0) to winter D(0)JF(1) mayhave difficulty to pass through the spring season MAM(1)

on the 2–3 year oscillation. This differs from the 3–5 yearoscillation, in which the large‐scale upper‐layer convergentflow over the tropical western Pacific persisted from thepreceding summer JJA(0) to the current summer JJA(1).[48] The evolutions of the regressed y850 and V850 fields

(Figure 12) are, in general, consistent with the c200 field. Inthe 2–3 year oscillation (Figure 12a), a strong anomalousanticyclone existed over southern Asia in SON(0) andmoved to the Philippine Sea in D(0)JF(1). The anomalousanticyclone weakened and disappeared in MAM(1) butreemerged over the WNP in JJA(1), again indicating a springbarrier for the Philippine Sea anticyclone to pass through theboreal spring on this timescale. The local Hadley circulationin JJA(1), therefore, is essential for the development of thesummer anticyclonic circulation over the WNP on thistimescale. The evolution is also consistent with the verticalvelocity fields that descending motion within the tropicalwestern Pacific weakens in MAM(1), where the ascendingmotion becomes stronger and statistically significant in JJA(1) within the green box in Figures 11a and 11c.[49] In the 3–5 year oscillation (Figure 12b), however,

the anomalous anticyclone over the Philippine Sea andlow‐level northeasterly wind over WNP persisted fromD(0)JF(1) to JJA(1), consistent with the c200, SSTA andvertical velocity pattern in Figure 10b and Figures 11b and11d. A positive thermodynamic air‐sea feedback canexplain how the anomalous anticyclone is maintained fromthe preceding winter to the current summer [Wang et al.,2000, 2003; Sui et al., 2007]. Here the anomalous anticy-clone, initiated by a cold SSTA as a Rossby wave responseto a negative heat source in the boreal winter, may furtherreinforce the cold SSTA through the wind‐evaporation‐SST feedback under prevailing climatological northeasterlytrade winds to the spring. Although the seasonal changeof prevailing winds leads to a negative air‐sea feedbackand thus a weakening of the local cold SSTA in the borealsummer, the negative SSTA may still have a significantimpact on the WNP monsoon, particularly in early summer.Thus this local cold SSTA may have a large impact on thesummertime WNPSH variability.

6. Conclusion and Discussion

[50] The western edge of the summertime Pacific sub-tropical high in the Northern Hemisphere (WNPSH) exhibitsa significant zonal shift on the interannual timescale. Thelargest variability is found in the western North Pacific. Thisvariability is represented in this study by an index defined bygeopotential height anomalies at 500 hPa averaged in theregion where the largest WNPSH variability occurs. A powerspectrum analysis of the WNPSH index indicates twodominant peaks at 2–3 year and 3–5 year periods. Theoscillation characteristics of the WNPSH at the two periodsare investigated through a regression analysis of atmosphericand oceanic variables against the two WNPSH indices at the2–3 year and 3–5 year periods. The idealized numericalexperiments with the ECHAM4 model are also carried out toexamine the dominated SSTA and associated mechanismsfor the WNPSH variability on the two timescales. Since thelocation and strength of the subtropical anticyclone greatlychange from June to August, the conclusion obtained from


14 of 19

Figure 10. (a) Evolutions of the SSTA field from the preceding spring (MAM(0)) to the summer of thefollowing year (JJA(2)) regressed against the WNPSH index for the 2–3 year oscillation. (b) Evolutions ofthe SSTA field from the preceding spring (MAM(0)) to the summer of the year after next year (JJA(3))regressed against the WNPSH index for the 3–5 year oscillation.


15 of 19

this study may not be applicable to individual summermonths.[51] The 2–3 year oscillation of the WNPSH and the

associated SST and circulation possess a coherent eastwardpropagating feature. The primary signals include thesoutheastward spread of a warm SSTA and ascendingmotion from the Indian Ocean in the preceding autumn,SON(0), to the maritime continent in the current summerJJA(1) of the high WNPSH index. In the central and easternPacific, the SSTA changes its sign from positive in thepreceding summer through winter to negative in JJA(1),analogous to the La Nina development condition. In thecurrent summer, the WNPSH is characterized by an anom-alous anticyclonic circulation and descending motion northof the Philippines. The descent is a part of the anomalouslocal Hadley circulation, with enhanced convection and awarm SSTA appearing in the maritime continent. A heatbudget analysis reveals that the descent and anticyclone aremaintained primarily by radiative cooling, and the corre-

sponding vertical structure is equivalently barotropic. Theanticyclonic circulation in the WNP is accompanied by ananomalous cyclonic circulation to its north. As a result,the overall structure appears like a meridionally oriented PJtype pattern. The role of the remote SSTA forcing in themaritime continent in affecting the WNPSH variability isverified by the idealized ECHAM4 model experiment.[52] The 3–5 year oscillation of the WNPSH exhibits a

standing wave feature, that the warm SSTA in the equatorialcentral eastern Pacific, its surrounding cold SSTA to thenorthwest and southwest, and a warm SSTA in the IndianOcean and along the East Asian marginal seas, persistingfrom the preceding winter to the current summer of a highWNPSH index. Along with the evolution above, theascending motion over the tropical eastern central Pacificand Indian Ocean and the descending motion over thewestern Pacific also persist from the preceding autumn tothe current summer. These features resemble the tripoleSSTA pattern and associated overturning Walker circulation

Figure 11. (a, b) Longitude‐time section of regressed SSTA and (c, d) 500 hPa vertical p velocityanomaly fields averaged between 15°S and 15°N from MAM(0) to JJA(2) for 2–3 year oscillation(Figures 11a and 11c) and from MAM(0) to JJA(3) for 3–5 year oscillation (Figures 11b and 11d).Contours in warm (cold) colors indicate warm (cold) SSTA or downward (upward) motion. Shadedregions pass the 90% confidence level. Green boxes denote the current summer of a strengthenedWNPSH.


16 of 19

pattern frequently observed during the decaying phase ofEl Nino episodes [e.g., Wang et al., 2000, 2003]. For the 3–5 year oscillation, the maximum descent of the WNP isslightly east of the anomalous anticyclone center. Thegreatest descent appears in the lower troposphere, accom-panied by a negative latent heating and a cold SSTA to thesoutheast. The corresponding vertical structure of relativevorticity in the descent region bears a baroclinic nature,and the zonal phase difference between the anomalousanticyclone center and the descent implies a Rossby waveresponse to a negative heating anomaly in association withthe local cold SSTA. The local SSTA forcing mechanism isconfirmed by the idealized ECHAM4 model experiment.[53] Our findings of the 2–3 year and 3–5 year oscillations

of the WNPSH are consistent with previous studies ofBarnett [1991] and Wang et al. [2001] who found that there

are quasi‐biennial and low‐frequency variabilities in thetropical SST/SLP fields and in Asian monsoon circulationfields. Our findings are also consistent with the observa-tional fact that ENSO has both the quasi‐biennial and lower‐frequency spectral peaks. The consistency above indicatesthat the variability of the WNPSH connected with the AsianMonsoon system might be closely linked to the tropicalSSTA forcing in the interannual time scale. The possiblemechanisms that connect the subtropical high and tropicalSST include the monsoon‐warm ocean interaction associ-ated with the TBO [Li et al., 2006], ENSO‐East Asia tele-connection [Wang et al., 2000; Sui et al., 2007], and aseason‐dependent Indian Ocean impact on the WNP mon-soon [Wu et al., 2009b].[54] In addition to the tropical forcing, midlatitude pro-

cesses may also play a role in the interannual variability of

Figure 12. Regressed 850 hPa stream function and wind fields from the preceding summer (JJA(0)) tothe current autumn (SON(1)) for (a) the 2–3 year and (b) 3–5 year oscillations of WNPSH. Shadings andvectors denote regions passing the 90% confidence level.


17 of 19

the WNPSH [e.g., Enomoto, 2004]. Since the present studyfocus on the relationship between the WNPSH and tropicalforcing, the influence of the midlatitude processes have notbeen discussed here. Our ECHAM4 model simulationconfirms that the 2–3 year mode is driven by the SSTAforcing over the maritime continent, while the 3–5 yearmode is driven by the local SSTA in the WNP. Despite themodel results in support of the hypothesized physicalmechanisms in the 2–3 year and 3–5 year oscillations, therole of the air‐sea interaction in the WNP region, where theatmospheric feedback to the SSTA is regarded as animportant process, have not been illustrated. Further obser-vational analyses and coupled GCM modeling studies areneeded to investigate the relative importance of these factorsand their causality in causing the interannual variability ofthe WNPSH from dynamic and thermodynamic discussions.

[55] Acknowledgments. We appreciate very much for the con-structive comments from three anonymous reviewers. This researchwas supported by NSC 98‐2111‐M‐133‐002 and NSC 99‐2111‐M‐133‐001. C.H.S. was supported by NSC 97‐2111‐M‐008‐010‐MY2and NSC 99‐2745‐M‐002‐001‐ASP. T.L. was supported by ONR grantsN000140810256 and N000141010774 and the International PacificResearch Center that is sponsored by the Japan Agency for Marine‐Earth Science and Technology (JAMSTEC), NASA (NNX07AG53G),and NOAA (NA17RJ1230). This is SOEST contribution 8127 and IPRCpublication 773.

ReferencesAnnamalai, H., J. M. Slingo, K. R. Sperber, and K. Hodges (1999), The meanevolution and variability of the Asian summer monsoon: Comparisonof ECMWF and NCEP–NCAR reanalyses, Mon. Weather Rev., 127,1157–1186, doi:10.1175/1520-0493(1999)127<1157:TMEAVO>2.0.CO;2.

Barnett, T. P. (1991), The interaction of multiple time scales in the tropicalclimate system, J. Clim., 4, 269–285, doi:10.1175/1520-0442(1991)004<0269:TIOMTS>2.0.CO;2.

Chang, C.‐P., and T. Li (2000), A theory for the tropical tropospheric bien-nial oscillation, J. Atmos. Sci., 57, 2209–2224, doi:10.1175/1520-0469(2000)057<2209:ATFTTT>2.0.CO;2.

Chang, C.‐P., Y. Zhang, and T. Li (2000a), Interannual and interdecadalvariations of the East Asian summer monsoon and tropical Pacific SSTs.Part 1: Role of the subtropical ridge, J. Clim., 13, 4310–4325,doi:10.1175/1520-0442(2000)013<4310:IAIVOT>2.0.CO;2.

Chang, C.‐P., Y. Zhang, and T. Li (2000b), Interannual and interdecadalvariations of the East Asian summer monsoon and tropical Pacific SSTs.Part 2: Southeast China rainfall and meridional structure, J. Clim., 13,4326–4340, doi:10.1175/1520-0442(2000)013<4326:IAIVOT>2.0.CO;2.

Chen, P., M. P. Hoerling, and R. M. Dole (2001), The origin of the subtrop-ical anticyclones, J. Atmos. Sci., 58, 1827–1835, doi:10.1175/1520-0469(2001)058<1827:TOOTSA>2.0.CO;2.

Chen, J.‐M., T. Li, and J. Shih (2007), Fall persistence barrier of sea sur-face temperature in the South China Sea associated with ENSO, J. Clim.,20, 158–172, doi:10.1175/JCLI4000.1.

Chou, C. (2004), Establishment of the low‐level wind anomalies overthe western North Pacific during ENSO development, J. Clim., 17,2195–2212, doi:10.1175/1520-0442(2004)017<2195:EOTLWA>2.0.CO;2.

Chou, C., J. Y. Tu, and J. Y. Yu (2003), Interannual variability of the west-ern North Pacific summer monsoon: Differences between ENSO andnon‐ENSO years, J. Clim., 16, 2275–2287, doi:10.1175/2761.1.

Enomoto, T. (2004), Interannual variability of the Bonin High associatedwith the propagation of Rossby wave along the Asian jet, J. Meteorol.Soc. Jpn., 82, 1019–1034, doi:10.2151/jmsj.2004.1019.

Gibson, J. K., P. Kallberg, and S. Uppala (1996), The ECMWF Reanalysis(ERA) Project, ECMWF Newsl., 73, 7–17.

Gibson, J. K., A. Hernandez, A. Nomura, and E. Serrano (1997), Reanal.Proj. Rep. Ser. 1, 72 pp., Eur. Cent. for Medium RangeWeather Forecast.,Reading, U. K.

Gill, A. E. (1980), Some simple solutions for heat‐induced tropical circula-tion,Q. J. R. Meteorol. Soc., 106, 447–462, doi:10.1002/qj.49710644905.

Hoskins, B. J. (1996), On the existence and strength of the summer sub-tropical anticyclones, Bull. Am. Meteorol. Soc., 77, 1287–1292.

Kalnay, E., et al. (1996), The NCEP/NCAR 40‐Year Reanalysis Project,Bull. Am. Meteorol. Soc., 77, 437–471, doi:10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2.

Kanamitsu,M.,W. Ebisuzaki, J.Woollen, S.‐K.Yang, J. J. Hnilo,M. Fiorino,and G. L. Potter (2002), NCEP/DOE AMIP‐II Reanalysis (R‐2), Bull.Am. Meteorol. Soc., 83, 1631–1643, doi:10.1175/BAMS-83-11-1631.

Kawamura, R., T. Matsuura, and S. Iisuka (2001), Interannual atmosphere‐ocean variations in the tropical western North Pacific relevant to theAsian summer monsoon‐ENSO coupling, J. Meteorol. Soc. Jpn., 79,883–898, doi:10.2151/jmsj.79.883.

Lau, N.‐C., and M. J. Nath (2000), Impact of ENSO on the variability ofthe Asian‐Australian monsoons as simulated in GCM experiments,J. Clim., 13, 4287–4309, doi:10.1175/1520-0442(2000)013<4287:IOEOTV>2.0.CO;2.

Lau, K.‐M., and H. T. Wu (2001), Principal modes of rainfall‐SST variabil-ity of the Asian summer monsoon: A reassessment of the monsoon‐ENSO relationship, J. Clim., 14, 2880–2895, doi:10.1175/1520-0442(2001)014<2880:PMORSV>2.0.CO;2.

Lee, E. J., J. G. Jhun, and C. K. Park (2005), Remote connection of theNortheast Asian summer rainfall variation revealed by a newly definedmonsoon index, J. Clim., 18, 4381–4393, doi:10.1175/JCLI3545.1.

Li, T., and B. Wang (2005), A review on the western North Pacific mon-soon: Synoptic‐to‐interannual variabilities, Terr. Atmos. Ocean. Sci., 16,285–314.

Li, T., and Y. S. Zhang (2002), Processes that determine the quasibiennialand lower‐frequency variability of the south Asian monsoon, J. Meteorol.Soc. Jpn., 80, 1149–1163, doi:10.2151/jmsj.80.1149.

Li, T., Y. S. Zhang, C.‐P. Chang, and B. Wang (2001), On the relationshipbetween Indian Ocean SST and Asian summer monsoon, Geophys. Res.Lett., 28, 2843–2846, doi:10.1029/2000GL011847.

Li, T., Y. S. Zhang, E. Lu, and D. Wang (2002), Relative role of dynamicand thermodynamic processes in the development of the Indian Oceandipole, Geophys. Res. Lett., 29(23), 2110, doi:10.1029/2002GL015789.

Li, T., B. Wang, C.‐P. Chang, and Y. Zhang (2003), A theory for the IndianOcean dipole‐zonal mode, J. Atmos. Sci., 60, 2119–2135, doi:10.1175/1520-0469(2003)060<2119:ATFTIO>2.0.CO;2.

Li, T., P. Liu, X. Fu, B. Wang, and G. A. Meehl (2006), Tempo‐spatialstructures and mechanisms of the tropospheric biennial oscillation inthe Indo‐Pacific warm ocean region, J. Clim., 19, 3070–3087,doi:10.1175/JCLI3736.1.

Liu, Y., and G. Wu (2004), Progress in the study on the formation of thesummertime subtropical anticyclone, Adv. Atmos. Sci., 21, 322–342,doi:10.1007/BF02915562.

Liu, Y. M., G. X. Wu, H. Liu, and P. Liu (2001), Condensation heatingof the Asian summer monsoon and the subtropical anticyclone inthe Eastern Hemisphere, Clim. Dyn., 17, 327–338, doi:10.1007/s003820000117.

Liu, Y. M., G. X. Wu, and R. C. Ren (2004), Relationship between the sub-tropical anticyclone and diabatic heating, J. Clim., 17, 682–698,doi:10.1175/1520-0442(2004)017<0682:RBTSAA>2.0.CO;2.

Lu, R. (2001), Interannual variability of the summertime North Pacific sub-tropical high and its relation to atmospheric convection over warm pool,J. Meteorol. Soc. Jpn., 79, 771–783, doi:10.2151/jmsj.79.771.

Lu, R., and B. Dong (2001), Westward extension of North Pacific subtrop-ical high in summer, J. Meteorol. Soc. Jpn., 79, 1229–1241, doi:10.2151/jmsj.79.1229.

Meehl, G. A. (1987), The annual cycle and interannual variability inthe tropical Pacific and Indian Ocean region, Mon. Weather Rev., 115,27–50, doi:10.1175/1520-0493(1987)115<0027:TACAIV>2.0.CO;2.

Meehl, G. A. (1994), Coupled land‐ocean‐atmosphere processes andsouth Asian monsoon variability, Science, 266, 263–267, doi:10.1126/science.266.5183.263.

Meehl, G. A. (1997), The south Asian monsoon and the tropospheric bien-nial oscillation, J. Clim., 10, 1921–1943, doi:10.1175/1520-0442(1997)010<1921:TSAMAT>2.0.CO;2.

Miyasaka, T., and H. Nakamura (2005), Structure and formation mechan-isms of the Northern Hemisphere summertime subtropical highs, J. Clim.,18, 5046–5065, doi:10.1175/JCLI3599.1.

Nitta, T. (1987), Convective activities in the tropical western Pacific andtheir impact on the Northern Hemisphere summer circulation, J. Meteorol.Soc. Jpn., 65, 373–389.

Rodwell, M. J., and B. J. Hoskins (1996), Monsoons and the dynamics ofdeserts, Q. J. R. Meteorol. Soc., 122, 1385–1404, doi:10.1002/qj.49712253408.

Rodwell, M. J., and B. J. Hoskins (2001), Subtropical anticyclones andsummer monsoon, J. Clim., 14, 3192–3211, doi:10.1175/1520-0442(2001)014<3192:SAASM>2.0.CO;2.


18 of 19

Roeckner, E., et al. (1996), The atmospheric general circulation modelECHAM‐4: Model description and simulation of present‐day climate,Rep. 218, 90 pp., Max‐Planck‐Inst. for Meteorol., Hamburg, Germany.

Seager, R., R. Murtugudde, N. Naik, A. Clement, N. Gordon, and J. Miller(2003), Air‐sea interaction and the seasonal cycle of the subtropicalanticyclones, J. Clim., 16, 1948–1966, doi:10.1175/1520-0442(2003)016<1948:AIATSC>2.0.CO;2.

Shen, S., and K.‐M. Lau (1995), Biennial oscillation associated with theEast Asian summer monsoon and tropical Pacific sea surface tempera-tures, J. Meteorol. Soc. Jpn., 73, 105–124.

Smith, T. M., and R. W. Reynolds (2004), Improved extended reconstruc-tion of SST (1854–1997), J. Clim., 17, 2466–2477, doi:10.1175/1520-0442(2004)017<2466:IEROS>2.0.CO;2.

Sui, C.‐H., P. H. Chung, and T. Li (2007), Interannual and interdecadal vari-ability of the summertime western North Pacific subtropical high, Geophys.Res. Lett., 34, L11701, doi:10.1029/2006GL029204.

Tao, S., and L. Chen (1987), A review of recent research on the East Asiansummer monsoon in China, in Monsoon Meteorology, edited by C.‐P.Chang and T. N. Krishuamurti, pp. 60–92, Oxford Univ. Press, NewYork.

Ting, M. (1994), Maintenance of northern summer stationary waves in aGCM, J. Atmos. Sci., 51, 3286–3308, doi:10.1175/1520-0469(1994)051<3286:MONSSW>2.0.CO;2.

Tomita, T., T. Yoshikane, and T. Yasunari (2004), Biennial and lower‐frequency variability observed in the early summer climate in the westernNorth Pacific, J. Clim., 17, 4254–4266, doi:10.1175/JCLI3200.1.

Wang, B., R. Wu, and X. Fu (2000), Pacific‐East Asian teleconnection:How does ENSO affect East Asian climate?, J. Clim., 13, 1517–1536,doi:10.1175/1520-0442(2000)013<1517:PEATHD>2.0.CO;2.

Wang, B., R. Wu, and K.‐M. Lau (2001), Interannual variability of theAsian summer monsoon: Contrasts between the Indian and the westernNorth Pacific–East Asian monsoons, J. Clim., 14, 4073–4090,doi:10.1175/1520-0442(2001)014<4073:IVOTAS>2.0.CO;2.

Wang, B., R. Wu, and T. Li (2003), Atmosphere‐warm ocean interactionand its impact on Asian‐Australian monsoon variation, J. Clim., 16,1195–1211, doi:10.1175/1520-0442(2003)16<1195:AOIAII>2.0.CO;2.

Wang, B., Q. Ding, X. Fu, I.‐S. Kang, K. Jin, J. Shukla, and F. Doblas‐Reyes (2005), Fundamental challenge in simulation and prediction ofsummer monsoon rainfall , Geophys. Res. Lett . , 32 , L15711,doi:10.1029/2005GL022734.

Watanabe, M., and F. F. Jin (2002), Role of Indian Ocean warming in thedevelopment of Philippine Sea anticyclone during ENSO, Geophys. Res.Lett., 29(10), 1478, doi:10.1029/2001GL014318.

Wu, R., and B. P. Kirtman (2005), Roles of Indian and Pacific Ocean air‐sea coupling in tropical atmospheric variability, Clim. Dyn., 25, 155–170,doi:10.1007/s00382-005-0003-x.

Wu, G. X., and Y. M. Liu (2003), Summertime quadruplet heating patternin the subtropics and the associated atmospheric circulation, Geophys.Res. Lett., 30(5), 1201, doi:10.1029/2002GL016209.

Wu, B., and T. Zhou (2008), Oceanic origin of the interannual and interde-cadal variability of the summertime western Pacific subtropical high,Geophys. Res. Lett., 35, L13701, doi:10.1029/2008GL034584.

Wu, G. X., Y. Liu, X. Zhu, W. Li, R. Ren, A. Duan, and X. Liang (2009),Multi‐scale forcing and the formation of subtropical desert and monsoon,Ann. Geophys., 27, 3631–3644, doi:10.5194/angeo-27-3631-2009.

Wu, B., T. Zhou, and T. Li (2009a), Contrast of rainfall‐SST relationshipsin the western North Pacific between the ENSO developing and decayingsummers, J. Clim., 22, 4398–4405, doi:10.1175/2009JCLI2648.1.

Wu, B., T. Zhou, and T. Li (2009b), Seasonally evolving dominant inter-annual variability mode over the East Asia, J. Clim., 22, 2992–3005,doi:10.1175/2008JCLI2710.1.

Wu, B., T. Li, and T. Zhou (2010), Relative contributions of the IndianOcean and local SST anomalies to the maintenance of the western NorthPacific anomalous anticyclone during the El Nino decaying summer,J. Clim., 23, 2974–2986, doi:10.1175/2010JCLI3300.1.

Xie, S. P., H. Jan, T. Hiroki, D. Yan, T. Sampe, K. Hu, and G. Huang(2009), Indian Ocean capacitor effect on Indo–Western Pacific climateduring the summer following El Niño, J. Clim., 22, 730–747,doi:10.1175/2008JCLI2544.1.

Yanai, M., and R. H. Johnson (1993), Impact of cumulus convection onthermodynamic fields, in The Representation of Cumulus Convectionin Numerical Models, Meteorol. Monogr., vol. 46, edited by K. Emanueland D. J. Raymond, pp. 36–62, Am. Meteorol. Soc., Boston, Mass.

Yanai, M., S. Esbensen, and J. H. Chu (1973), Determination of bulk prop-erties of tropical cloud clusters from large‐scale heat and moisture bud-gets, J. Atmos. Sci., 30, 611–627, doi:10.1175/1520-0469(1973)030<0611:DOBPOT>2.0.CO;2.

Yang, J. L., Q. Y. Li, S. P. Xie, Z. Y. Liu, and L. X. Wu (2007), Impactof the Indian Ocean SST basin mode on the Asian summer monsoon,Geophys. Res. Lett., 34, L02708, doi:10.1029/2006GL028571.

Yu, J.‐Y., C. R. Mechoso, A. Arakawa, and J. C. Williams (2002), Impactsof the Indian Ocean on the ENSO cycle, Geophys. Res. Lett., 29(8), 1204,doi:10.1029/2001GL014098.

Zhou, T. J., and Z. Li (2002), Simulation of the East Asian summer mon-soon by using a variable resolution atmospheric GCM, Clim. Dyn., 19,167–180, doi:10.1007/s00382-001-0214-8.

Zhou, T. J., and R. Yu (2005), Atmospheric water vapor transport associatedwith typical anomalous summer rainfall patterns in China, J. Geophys.Res., 110, D08104, doi:10.1029/2004JD005413.

Zhou, T., and J. Zhang (2011), The vertical structures of atmospheric tem-perature anomalies associated with two flavors of El Nino simulated byAMIP II models, J. Clim., 24, 1053–1070, doi:10.1175/2010JCLI3504.1.

Zhou, T. J., R. Yu, H. Li, and B. Wang (2008), Ocean forcing to changes inglobal monsoon precipitation over the recent half‐century, J. Clim., 21,3833–3852, doi:10.1175/2008JCLI2067.1.

Zhou, T. J., B. Wu, and B. Wang (2009a), How well do atmosphericgeneral circulation models capture the leading modes of the interannualvariability of the Asian‐Australian monsoon?, J. Clim., 22, 1159–1173,doi:10.1175/2008JCLI2245.1.

Zhou, T., et al. (2009b), Why the western Pacific subtropical high hasextended westward since the late 1970s, J. Clim., 22, 2199–2215,doi:10.1175/2008JCLI2527.1.

P.‐H. Chung, Department of History and Geography, Taipei MunicipalUniversity of Education, Taipei 10042, Taiwan. ([email protected])T. Li, International Pacific Research Center, University of Hawai’i at

Mānoa, Honolulu, HI 96822, USA.C.‐H. Sui, Department of Atmospheric Sciences, National Taiwan

University, Taipei 10617, Taiwan.


19 of 19

Interannual relationships between the tropical sea surface...

Documents

Transcript of Interannual relationships between the tropical sea surface...