Abstract

Summertime atmospheric circulation over the midlatitude western North Pacific (WNP) is influenced by anomalous convective activity near the Philippines. This meridional teleconnection, observed in monthly anomalies and known as the Pacific–Japan (PJ) pattern, is characterized by zonally elongated cyclonic and anticyclonic anomalies around the enhanced convection center and to its northeast, respectively, in the lower troposphere, with an apparent poleward phase tilt with height. The authors’ idealized two-layer linear model, whose basic state consists of a zonal subtropical jet and a pair of a monsoon system and a subtropical anticyclone, can simulate a PJ-like response against diabatic heating located between the pair. Each of the observed and simulated patterns can gain energy through barotropic and baroclinic conversions from the zonally varying baroclinic mean flow, in an efficiency comparable with that of energy generation due to the anomalous diabatic heating, indicating a characteristic of the pattern as a dry dynamical mode. In fact, the conversion efficiency is sensitive to the location of the anomaly pattern relative to the climatological-mean flow. Furthermore, the second-least damped mode identified in the idealized model bears certain resemblance with the observed PJ pattern, indicating its modal characteristics as well as a critical importance of these features in the mean field for the pattern. In addition to the PJ pattern, another meridional teleconnection pattern with high efficiency for its energy conversion is identified observationally in association with anomalous convection near the Bonin Islands.

The anomalous circulation of the PJ pattern, in turn, can intensify the anomalous convective activity near the Philippines through enhancing evaporation and moisture convergence and dynamically inducing anomalous ascent. It is thus hypothesized that the PJ pattern can be regarded as a moist dynamical mode that sustains itself both via dry energy conversion and interaction with moist processes.

1. Introduction

East Asian summer climate is influenced by two quasi-stationary surface anticyclones, the cool Okhotsk high and the subtropical Bonin (Ogasawara) high, and by the baiu/mei-yu stationary frontal system between them. Northward (southward) displacement of the baiu/mei-yu front associated with the anomalously intensified Bonin (Okhotsk) high brings a warmer (cooler) summer to Japan (e.g., Yasunaka and Hanawa 2006). Wakabayashi and Kawamura (2004) identified teleconnection patterns that can influence the summertime climate over Japan. One of them is in the form of a stationary Rossby wave train propagating along the polar front jet over the northern portion of the Eurasian continent, which is related to the development of the Okhotsk high (Nakamura and Fukamachi 2004). Meanwhile, the strength of the Bonin high is influenced by two other teleconnections. One is a wave-like anomaly pattern along the subtropical Asian jet (the Silk Road pattern; Enomoto et al. 2003; Enomoto 2004; Kosaka et al. 2009). The other is the Pacific–Japan (PJ) pattern identified by Nitta (1987). Analyzing monthly-mean fields for the June–September season over 6 yr (from 1978 to 1983), he found association between above- (below-) normal convective activity over the tropical western North Pacific (WNP) and anticyclonic (cyclonic) anomalies over the midlatitude Far East, giving rise to hot (cool) summer locally.

On the basis of the pioneering studies on the PJ pattern (e.g., Nitta 1987; Kurihara and Tsuyuki 1987) and a dynamical notion that stationary Rossby waves propagating from the tropics into midlatitudes must have equivalent barotropic structure (Hoskins and Karoly 1981), some studies (Tsuyuki and Kurihara 1989; Lau and Peng 1992; Grimm and Silva Dias 1995) used barotropic models linearized about upper-tropospheric time mean flow to argue that the PJ pattern may be manifested as a barotropically unstable mode that can develop in midlatitude westerlies. Among those studies, Tsuyuki and Kurihara (1989) further argued that the mode can be triggered by a Rossby wave excited by anomalous diabatic heating near the Philippines. However, the upper-tropospheric mean flow observed in boreal summer includes northeasterly flow over the (sub) tropical WNP (Fig. 1), which is unfavorable for Rossby waves to propagate northward (Z. Wang et al. 2005). This barotropic argument was modified substantially by Lu (2004), who considered how barotropic response can be generated by anomalous tropical convection that fundamentally acts as a baroclinic forcing. He stressed the importance of the vertically sheared climatological-mean zonal flow for the generation of the barotropic response, on a basis of theoretical studies by Lim and Chang (1986), Kasahara and Silva Dias (1986), Kato and Matsuda (1992), and Wang and Xie (1996). However, a vorticity budget analysis for a PJ-like pattern by Kawamura et al. (1996) shows the dominance of anomalous vorticity advection by the vertically sheared mean meridional flow, suggesting that the dynamics of the PJ pattern should be understood in a framework where the perturbations are embedded in the zonally varying baroclinic mean flow.

Fig. 1.

Climatological-mean JJA streamfunction (×106 m2 s−1; contoured) and horizontal winds (vectors; with scalings at the top) at the (a) 200- and (b) 850-hPa levels. (c) The corresponding climatology of vertically integrated mean moisture flux (vectors; with scaling at the top) and precipitable water (kg m−2; contoured). Contours are drawn with intervals of (a) 5 (±2.5, ±7.5, ±12.5, …), (b) 3 (±1.5, ±4.5, ±7.5, …), and (c) 5. In (c), horizontal divergence and convergence of the vertically integrated moisture flux greater than 0.04 g m−2 s−1 are indicated by light and heavy shading, respectively. Based on the JRA-25 dataset for the 1979–2007 period.

Fig. 1.

Climatological-mean JJA streamfunction (×106 m2 s−1; contoured) and horizontal winds (vectors; with scalings at the top) at the (a) 200- and (b) 850-hPa levels. (c) The corresponding climatology of vertically integrated mean moisture flux (vectors; with scaling at the top) and precipitable water (kg m−2; contoured). Contours are drawn with intervals of (a) 5 (±2.5, ±7.5, ±12.5, …), (b) 3 (±1.5, ±4.5, ±7.5, …), and (c) 5. In (c), horizontal divergence and convergence of the vertically integrated moisture flux greater than 0.04 g m−2 s−1 are indicated by light and heavy shading, respectively. Based on the JRA-25 dataset for the 1979–2007 period.

Recently, Kosaka and Nakamura (2006, hereafter KN06, 2008, hereafter KN08) discussed three-dimensional structure of the PJ pattern based on long-term reanalysis data. Figure 2 shows maps of precipitation and vorticity anomalies composited for the 38 strongest monthly events of the PJ pattern for its particular phase with enhanced convective activity in a tropical WNP domain [10°–20°N, 120°–130°E] observed in the June–August (JJA) season from 1979 to 2007. Concurrently with enhanced cumulus convection around the Philippines (Fig. 2a), a meridional dipole of anomalous vorticity tends to be observed in the lower troposphere (Fig. 2b), as a characteristic feature of the PJ pattern (Nitta 1987). Unlike in the aforementioned conventional picture, however, the vorticity anomalies exhibit an apparent poleward phase tilt with height (Figs. 2b,c; KN06; Hsu and Lin 2007), indicating the notable baroclinic component in its three-dimensional structure. The associated wave-activity flux, which is formulated by Takaya and Nakamura (2001) and parallel to local group velocity of stationary Rossby waves, is poleward and equatorward in the lower and upper troposphere, respectively, indicating that poleward Rossby wave propagation associated with the PJ pattern occurs mainly in the lower troposphere (cf. Tsuyuki and Kurihara 1989).

Fig. 2.

Horizontal maps of composited anomalies for the 38 strongest monthly events of the PJ pattern observed with enhanced tropical convection. (a) CMAP precipitation (mm day−1); vorticity (×10−6 s−1) at the (b) 850-hPa and (c) 200-hPa levels. Here, (b),(c) are based on the JRA-25 dataset. Contour intervals are (a),(b) 0.5 (±0.25, ±0.75, ±1.25, …) and (c) 1 (±0.5, ±1.5, ±2.5, …). Solid and dashed lines indicate positive and negative anomalies, respectively. Light and heavy shading represent the confidence levels of 90% and 95%, respectively, based on the t statistic. Also shown with arrows in (b),(c) is a wave-activity flux defined by Takaya and Nakamura (2001) for stationary Rossby waves, whose scaling is given at the top. The OLR anomaly center is indicated with triangles in (b),(c).

Fig. 2.

Horizontal maps of composited anomalies for the 38 strongest monthly events of the PJ pattern observed with enhanced tropical convection. (a) CMAP precipitation (mm day−1); vorticity (×10−6 s−1) at the (b) 850-hPa and (c) 200-hPa levels. Here, (b),(c) are based on the JRA-25 dataset. Contour intervals are (a),(b) 0.5 (±0.25, ±0.75, ±1.25, …) and (c) 1 (±0.5, ±1.5, ±2.5, …). Solid and dashed lines indicate positive and negative anomalies, respectively. Light and heavy shading represent the confidence levels of 90% and 95%, respectively, based on the t statistic. Also shown with arrows in (b),(c) is a wave-activity flux defined by Takaya and Nakamura (2001) for stationary Rossby waves, whose scaling is given at the top. The OLR anomaly center is indicated with triangles in (b),(c).

Kawamura et al. (1996), Fukutomi and Yasunari (2002), and Yasutomi (2003) showed kinetic energy (KE) gain of the zonally elongated anomalous circulation associated with PJ-like perturbations. Fukutomi and Yasunari (2002) also found that available potential energy (APE) gain for the perturbations through baroclinic energy conversion from the climatological-mean flow dominates in midlatitudes. The barotropic and baroclinic energy conversions alone can replenish the total energy (KE + APE) associated with the PJ pattern as efficiently as the APE generation due to anomalous convective heating (KN06; KN08). It is thus hypothesized that the PJ pattern can be regarded as a dry dynamical mode inherent in the climatological-mean flow between the Asian summer monsoon and the North Pacific subtropical anticyclone. Furthermore, KN06 showed that the PJ-associated anomalous circulation can be reinforced through its interaction with anomalous evaporation from the ocean and anomalous moisture convergence.

In contrast to the conventional recognition of the PJ pattern as a forced dynamical response of the dry atmosphere in which energy obtained from anomalous diabatic heating is simply dispersed away, the aforementioned results reveal a characteristic of the PJ pattern as a moist dynamical mode, where anomalous circulation can effectively extract energy from the mean state and enhances anomalous convective activity, which in turn reinforces the anomalous circulation. This modal characteristic is substantiated in the present study through data analysis and numerical experiments outlined in section 2. Section 3 discusses the dynamics of the PJ pattern, to suggest that the pattern is one of the dry dynamical modes in the three-dimensional climatological-mean flow in summer, which is verified by additional analysis with an idealized model in section 4. Then, feedback of the anomalous circulation on the tropical convective activity in association with the PJ pattern is examined through observational data analysis in section 5, where a possibility is presented that the PJ pattern may be a moist dynamical mode over the summertime WNP. These results imply that PJ-like anomaly patterns may be observed in other regions where the mean-flow structure is similar to that in the WNP, which is examined in a companion paper (Kosaka and Nakamura 2010). An empirical orthogonal function (EOF) analysis is conducted in section 6 to further examine behavior of the PJ pattern as a dynamical mode.

2. Data and a model

a. Data and analysis methods

In this study, we use monthly-mean data of the Japanese 25-yr reanalysis (JRA-25) of the global atmosphere (Onogi et al. 2007), which is known for better representation of tropical cyclones and precipitation than other reanalysis datasets. In fact, KN08 showed a more coherent structure of the PJ pattern in the JRA-25 than in other reanalysis datasets while suggesting uncertainties included in diabatic heating in reanalysis datasets. Horizontally smoothed fields of vorticity have been derived using a spectral expansion with T47 truncation, multiplying a spherical harmonic component of the total wavenumber n by exp{−K[n(n + 1)2]} (Hoskins 1980). The factor K has been set in such a way that amplitudes of the harmonic components with n = 24 are reduced by 50%. The smoothing has not been applied for other variables. We also use monthly gridded data of the National Oceanic and Atmospheric Administration (NOAA) interpolated outgoing longwave radiation (OLR) as a proxy for tropical convective activity and a key index for compositing. In addition, monthly data of the Climate Prediction Center (CPC) Merged Analysis of Precipitation (CMAP; Xie and Arkin 1997) are also utilized. We additionally use the best track data of severe tropical storms and typhoons provided by the Japan Meteorological Agency. Those four datasets are available over a 29-yr period from 1979 to 2007. The NOAA optimum interpolation sea surface temperature (SST) data, available only from December 1981, are also used. The horizontal resolution is 1° × 1° for the SST data and 2.5° × 2.5° for the other datasets.

A part of data analysis in this study is based on composite maps, which, unlike case studies, are suited for energetics (section 3), because both energy and its conversion are quadrature of the anomalies. As a key OLR index Iϕ,λ for our compositing, the strongest monthly OLR anomaly, either positive or negative, was identified within a 10° × 10° domain centered at (latitude ϕ, longitude λ) for each month. The months for which the identified anomaly exceeds a half the standard deviation for the particular calendar month were then used for compositing for enhanced (or suppressed) convection associated with the PJ pattern.

b. Linear quasigeostrophic model

To confirm essential dynamical features obtained from the composite analysis, a model analysis is also performed. The model is a steady, linear, quasigeostrophic two-layer model on a β plane. The model structure is kept as simple as possible to elucidate the fundamental dynamics of the PJ pattern. The quasigeostrophic vorticity and thermodynamic equations in pressure (p) coordinates may be expressed as

 
formula
 
formula

with u′ = −∂Φ′/f0y, υ′ = ∂Φ′/f0x, and ζ′ = ∇2Φ′/f0. Here, Φ denotes geopotential, u = (u, υ) denotes horizontal wind velocity, ζ denotes relative vorticity, ω denotes vertical p velocity, Q denotes diabatic heating rate per unit mass, κ denotes vorticity diffusion coefficient, α denotes Newtonian cooling rate, and denotes the horizontal gradient operator. Overbars and primes indicate basic-state quantities and perturbations, respectively, and f0 is the Coriolis parameter at the central latitude (45°N) of the β plane. In (2), S denotes static stability S = (R/p)(RT/CppdT/dp) with temperature T, the gas constant R, and the specific heat at constant pressure Cp. The value of κ is set in such a manner that a harmonic component with the maximum zonal wavenumber is damped with the e-folding time of 10 days, and α = (10 days)−1 is adopted.

In our two-layer model, (1) is applied separately to the 250- and 750-hPa levels (upper and lower levels, respectively) and (2) is applied to the 500-hPa level (middle level). Thus, Φ′, ζ′, u′, and υ′ are evaluated at both the upper and lower levels and ω′ and Q′ are evaluated only at the middle level, whereas u′ and υ′ used in (2) are linearly interpolated from those at the upper and lower levels. The model grid resolution is 6° in longitude and 2° in latitude. The model is bounded at the equator and the North Pole, at which Φ′ = 0 is assumed.

3. Dynamics of the PJ pattern in the real atmosphere

a. Energetics of the PJ pattern

We begin our discussion with an analysis based on the PJ pattern in our composite map shown in Fig. 2, which has been constructed for I15°N,125°E for JJA. The composited vorticity anomalies associated with enhanced convection around the Philippines (Fig. 2a) include a meridional dipole in the lower troposphere (Fig. 2b). The anomalies are characterized by their zonally elongated structure with a poleward phase tilt into the upper troposphere (Fig. 2c), as revealed in KN06 and KN08. The poleward phase tilt is an indication of eastward heat transport by the anomalies from the warm continental Asian monsoon to the cool maritime North Pacific subtropical anticyclone. In addition, the midlatitude anticyclonic anomalies show a westward phase tilt with height in their upstream portion, accompanying an upward wave-activity flux (KN06). This feature is indicative of poleward heat transport across the vertically sheared Asian jet. In fact, this eastward and poleward heat transport is recognized in most of the 38 strongest PJ events composited (figures not shown), assuring the robustness of the composited structure.

Figure 3 shows the distributions of barotropic energy conversion (Hoskins et al. 1983; Simmons et al. 1983),

 
formula

at the 850- and 200-hPa levels; vertically integrated baroclinic conversion of APE,1

 
formula

and the vertically integrated APE generation due to anomalous diabatic heating,

 
formula

In these equations, overbars and primes indicate climatological-mean quantities for JJA and the composited anomalies, respectively. Positive values indicate that the anomalies gain KE or APE from the climatological-mean state or anomalous diabatic heating.

Fig. 3.

Local barotropic energy conversion CK (×10−5 m2 s−3; solid and dashed contours) and the extended EP flux (arrows with scaling at the top) at the (a) 850- and (b) 200-hPa levels, based on the composited anomalies for the PJ pattern shown in Fig. 2. The corresponding (c) baroclinic energy conversion CP and (d) diabatic energy generation CQ, both integrated vertically from the surface to the 100-hPa level (×10−2 W m−2; heavy solid and dashed contours). Contour intervals are (a) 0.2 (±0.1, ±0.3, ±0.5, …), (b) 0.6 (±0.3, ±0.9, ±1.5, …), (c) 1 (±0.5, ±1.5, ±2.5, …), and (d) 2 (±1, ±3, ±5, …). Solid and dashed lines represent energy gain and loss, respectively, through the conversion/generation for the anomalies. In (a),(b), heavy shading indicates westerly regions of (a) > 5 m s−1 and (b) > 25 m s−1, whereas light shading in (a) indicates an easterly region of < −5 m s−1. In (c), the climatological-mean 400-hPa temperature is indicated with light contours for every 1 K, and shading indicates regions where the contribution of eddy zonal heat transport [i.e., the second term in Eq. (4)] exceeds 80% of total positive CP. The OLR anomaly center is indicated with triangles in (a),(b).

Fig. 3.

Local barotropic energy conversion CK (×10−5 m2 s−3; solid and dashed contours) and the extended EP flux (arrows with scaling at the top) at the (a) 850- and (b) 200-hPa levels, based on the composited anomalies for the PJ pattern shown in Fig. 2. The corresponding (c) baroclinic energy conversion CP and (d) diabatic energy generation CQ, both integrated vertically from the surface to the 100-hPa level (×10−2 W m−2; heavy solid and dashed contours). Contour intervals are (a) 0.2 (±0.1, ±0.3, ±0.5, …), (b) 0.6 (±0.3, ±0.9, ±1.5, …), (c) 1 (±0.5, ±1.5, ±2.5, …), and (d) 2 (±1, ±3, ±5, …). Solid and dashed lines represent energy gain and loss, respectively, through the conversion/generation for the anomalies. In (a),(b), heavy shading indicates westerly regions of (a) > 5 m s−1 and (b) > 25 m s−1, whereas light shading in (a) indicates an easterly region of < −5 m s−1. In (c), the climatological-mean 400-hPa temperature is indicated with light contours for every 1 K, and shading indicates regions where the contribution of eddy zonal heat transport [i.e., the second term in Eq. (4)] exceeds 80% of total positive CP. The OLR anomaly center is indicated with triangles in (a),(b).

As pointed out by Kawamura et al. (1996), Fukutomi and Yasunari (2002), KN06, and KN08, the lower-tropospheric anomalies gain KE over the South China Sea (SCS), to the east of the Philippines, and to the south of Japan (Fig. 3a). This positive CK is manifested in Fig. 3a as the extended Eliassen–Palm (EP) flux E = (υ2u2, −uυ′) pointing toward the stronger mean westerlies (or weaker easterlies), because CK can be approximated as CK ≈ E · u (Simmons et al. 1983). In the upper troposphere (Fig. 3b), areas of positive and negative CK are adjacent to one another around the exit of the Asian jet. As shown later, these positive and negative contributions cancel one another when integrated horizontally, yielding no significant net contribution.

In Fig. 3c, a large area of pronounced positive CP is found along the Asian jet over the midlatitude Far East, adjacent to a smaller area of negative CP in the jet exit to the east of Japan. These regions are characterized by the westward and eastward phase tilts of the midlatitude anticyclonic anomalies at their western and eastern flanks, respectively (KN06). The predominance of the westward tilt yields the net positive CP if integrated horizontally, as shown later. Weak positive CP to the south and southeast of Japan is related to the eastward heat transport (shaded in Fig. 3c) associated with the poleward-tilted vorticity perturbations embedded in the westward temperature gradient between the warmer Asian continent and the cooler North Pacific [i.e., the second term of Eq. (4)]. Although CQ is contributed to mainly by the anomalous latent heat release associated with the enhanced convection around the Philippines (Fig. 3d), positive and negative CQ are found associated with midlatitude diabatic heating anomalies, related in part to anomalous precipitation along the baiu/mei-yu front. Though inconspicuous energetically in our analysis, a possible contribution of the anomalous diabatic heating associated with the baiu/mei-yu precipitation anomalies to the formation of the PJ pattern has been pointed out by Lu and Lin (2009).

The contributions from energy conversions CK and CP (i.e., “dry” process) and the diabatic generation CQ (i.e., “moist” process) to the energy budget of the PJ pattern can be evaluated as time scales in which the area-integrated energy could be replenished by a particular conversion/generation integrated over a given domain: τCK = [KE]/[CK], τCP = [APE]/[CP], τdry = [KE + APE]/[CK + CP], τmoist = [KE + APE]/[CQ], and τtotal = [KE + APE]/[CK + CP + CQ], with square brackets indicating the spatial integrals. The time scales based on the composited anomalies shown in Fig. 2 are summarized in Table 1. The positive and negative signs of the time scales indicate the net energy gain and loss of the PJ-associated anomalies, respectively. Any process whose time scale for the energy gain is shorter than ∼30 days may be regarded as being effective for maintaining the monthly PJ pattern.

Table 1.

Time scales (days) with which horizontally integrated energy (KE for τCK, APE for τCP, and KE + APE for τdry, τmoist, and τtotal) could be replenished through the corresponding energy conversions (CK for τCK, CP for τCP, and CK + CP for τdry), diabatic energy generation (CQ for τmoist), and their sum (CK + CP + CQ for τtotal), for (composite) the composited PJ anomalies shown in Fig. 2, (model) the heat-induced response in the control experiment shown in Fig. 6, and the regressed anomalies shown in Figs. 14 and 15, (EOF1 and EOF2, respectively), based on the two leading modes of variability of the monthly 850-hPa horizontally smoothed vorticity over [0°–60°N, 100°–160°E] for JJA. The energy conversion/generation is integrated over the entire Northern Hemisphere for “net” and over subdomains as indicated, whereas energy is integrated over the entire Northern Hemisphere for all the cases. The vertical integrals from the surface to (composite and EOFs) the 100-hPa level and (model) the top of the model have been taken if indicated, before the horizontal integration. The time scales <30 days are highlighted in bold as an indication of efficient energy conversion/generation.

Time scales (days) with which horizontally integrated energy (KE for τCK, APE for τCP, and KE + APE for τdry, τmoist, and τtotal) could be replenished through the corresponding energy conversions (CK for τCK, CP for τCP, and CK + CP for τdry), diabatic energy generation (CQ for τmoist), and their sum (CK + CP + CQ for τtotal), for (composite) the composited PJ anomalies shown in Fig. 2, (model) the heat-induced response in the control experiment shown in Fig. 6, and the regressed anomalies shown in Figs. 14 and 15, (EOF1 and EOF2, respectively), based on the two leading modes of variability of the monthly 850-hPa horizontally smoothed vorticity over [0°–60°N, 100°–160°E] for JJA. The energy conversion/generation is integrated over the entire Northern Hemisphere for “net” and over subdomains as indicated, whereas energy is integrated over the entire Northern Hemisphere for all the cases. The vertical integrals from the surface to (composite and EOFs) the 100-hPa level and (model) the top of the model have been taken if indicated, before the horizontal integration. The time scales <30 days are highlighted in bold as an indication of efficient energy conversion/generation.
Time scales (days) with which horizontally integrated energy (KE for τCK, APE for τCP, and KE + APE for τdry, τmoist, and τtotal) could be replenished through the corresponding energy conversions (CK for τCK, CP for τCP, and CK + CP for τdry), diabatic energy generation (CQ for τmoist), and their sum (CK + CP + CQ for τtotal), for (composite) the composited PJ anomalies shown in Fig. 2, (model) the heat-induced response in the control experiment shown in Fig. 6, and the regressed anomalies shown in Figs. 14 and 15, (EOF1 and EOF2, respectively), based on the two leading modes of variability of the monthly 850-hPa horizontally smoothed vorticity over [0°–60°N, 100°–160°E] for JJA. The energy conversion/generation is integrated over the entire Northern Hemisphere for “net” and over subdomains as indicated, whereas energy is integrated over the entire Northern Hemisphere for all the cases. The vertical integrals from the surface to (composite and EOFs) the 100-hPa level and (model) the top of the model have been taken if indicated, before the horizontal integration. The time scales <30 days are highlighted in bold as an indication of efficient energy conversion/generation.

When CK is integrated over the entire Northern Hemisphere, τCK is shorter than a month in the lower troposphere, indicating the importance of the barotropic energy conversion. However, CK is negative in the upper troposphere or when integrated vertically, indicating that the net barotropic conversion acts to damp the monthly PJ anomalies. Nevertheless, there is a possibility that CK is still effective over the WNP. To check this possibility, the time scale is also evaluated with CK integrated only over the WNP (the “forcing sector,” 0°–60°N, 100°–150°E), whereas the domain for integrating KE is still the entire Northern Hemisphere. In this evaluation, CK contributes positively to the maintenance of the monthly anomalies with a time scale shorter than 1 month in the lower troposphere, whereas the corresponding time scale is about 2 months in the upper troposphere. It is thus suggested that the barotropic energy conversion over the WNP acts to maintain the monthly scale PJ anomalies fairly effectively, especially in the lower troposphere, whereas the converted KE is returned to the mean flow outside the WNP.

The replenishing time scale τCP for CP integrated throughout the depth of the troposphere is shorter than a week, indicative of the vital importance of the baroclinic energy conversion in the maintenance of the monthly PJ pattern. This high efficiency can also be confirmed from the time scale based on the vertically integrated CP solely for the forcing sector (τCP less than 2 weeks). The total energy conversion, mainly contributed by vertically integrated CP and additionally by the lower-tropospheric CK, can replenish total energy associated with the PJ pattern within a month (τdry in Table 1). This result suggests that the PJ pattern can be regarded as a dry dynamical mode inherent in the zonally asymmetric climatological-mean flow characterized by the Asian summer monsoon to the west, the North Pacific subtropical anticyclone to the east, and the vertically sheared Asian jet (KN06; KN08).

Table 1 also reveals that the contributions from the “dry” and “moist” processes over the forcing sector are comparable. It indicates that the dynamics of the monthly PJ pattern should not be discussed without moist processes including moisture supply into the convection region and the dynamical influence of convection, as addressed in section 5. One should keep in mind, however, that diabatic heating used in the evaluation of CQ is a product of reanalysis and therefore subject to a certain level of uncertainty (KN08).

b. Spatial dependence of energetics

If the PJ pattern is indeed a dynamical mode inherent in the climatological-mean field, the efficiency of the energy conversion should be maximized when the anomaly pattern is situated at the observed location or its vicinity. To assess the dependence of the energy conversion on the location of the PJ pattern relative to the climatological-mean field, the effective time scale of the energy conversion rate has been evaluated for the same anomaly pattern as in Fig. 2 but displaced artificially on a sphere relative to the fixed climatological-mean state. The shifting was performed every 10° in longitude and 5° in latitude.

At the 850-hPa level, CK is efficient around the observed location and to the east as τCK obtained is less than a month (Table 2a). A similar tendency can be found in the vertically integrated CK, though less efficient because of net negative CK in the middle or upper troposphere. At the 850-hPa level, positive CK observed in the exit of the monsoon westerly jet becomes weaker as the anomaly pattern is displaced eastward, but it is compensated by increasing positive CK in the exit of the trade winds (figures not shown). Our evaluation indicates that the importance of the exits of the trade winds and the monsoon westerly jet are comparable. Meanwhile, τCP (Table 2b) generally shows dominantly high efficiency of CP over CK, insensitive to the meridional shift of the pattern. When CK and CP are combined, the resultant total energy conversion remains highly efficient as long as the pattern shifting is in the zonal direction or slightly poleward (τdry in Table 2c), whereas the conversion efficiency lowers if the pattern is shifted meridionally too much (more than 5°). This sensitivity to the meridional position of the pattern arises mainly from the sensitivity of CK (Table 2).

Table 2.

Time scales (a) τCK, (b) τCP, and (c) τdry (days) with which horizontally integrated energy (KE for τCK, APE for τCP, and KE + APE for τdry) could be replenished through (a) barotropic energy conversion CK, (b) baroclinic energy conversion CP, and (c) their sum CK + CP when the anomaly pattern shown in Fig. 2 is shifted zonally and then meridionally. Evaluation of KE and APE is based on the original, nonshifted composite and integrated over the entire Northern Hemisphere, whereas CK and CP are integrated over [0°–60°N, 100°–150°E] for the nonshifted composite or the domain shifted with the anomaly pattern. All the energy and conversions are integrated from the surface to the 100-hPa level before integrated horizontally, but τCK evaluated solely at the 850-hPa level is also shown in right half of each of the cells in (a). Time scales (a),(c) <30 and (b) <15 days are highlighted with boldface.

Time scales (a) τCK, (b) τCP, and (c) τdry (days) with which horizontally integrated energy (KE for τCK, APE for τCP, and KE + APE for τdry) could be replenished through (a) barotropic energy conversion CK, (b) baroclinic energy conversion CP, and (c) their sum CK + CP when the anomaly pattern shown in Fig. 2 is shifted zonally and then meridionally. Evaluation of KE and APE is based on the original, nonshifted composite and integrated over the entire Northern Hemisphere, whereas CK and CP are integrated over [0°–60°N, 100°–150°E] for the nonshifted composite or the domain shifted with the anomaly pattern. All the energy and conversions are integrated from the surface to the 100-hPa level before integrated horizontally, but τCK evaluated solely at the 850-hPa level is also shown in right half of each of the cells in (a). Time scales (a),(c) <30 and (b) <15 days are highlighted with boldface.
Time scales (a) τCK, (b) τCP, and (c) τdry (days) with which horizontally integrated energy (KE for τCK, APE for τCP, and KE + APE for τdry) could be replenished through (a) barotropic energy conversion CK, (b) baroclinic energy conversion CP, and (c) their sum CK + CP when the anomaly pattern shown in Fig. 2 is shifted zonally and then meridionally. Evaluation of KE and APE is based on the original, nonshifted composite and integrated over the entire Northern Hemisphere, whereas CK and CP are integrated over [0°–60°N, 100°–150°E] for the nonshifted composite or the domain shifted with the anomaly pattern. All the energy and conversions are integrated from the surface to the 100-hPa level before integrated horizontally, but τCK evaluated solely at the 850-hPa level is also shown in right half of each of the cells in (a). Time scales (a),(c) <30 and (b) <15 days are highlighted with boldface.

These results suggest that the PJ pattern tends to be fixed meridionally to the climatological-mean field, mainly through the efficiency of the barotropic energy conversion with the lower-tropospheric monsoon westerlies and the trade winds and the upper-tropospheric Asian jet, although the efficient energy conversion into the PJ pattern is primarily through its baroclinic contribution. Interestingly, the high total efficiency for the westward shifted pattern arises from the effective CK in the upper troposphere, as opposed to the energetics for the observed PJ pattern. It should be noted that, embedded in the climatological-mean field that varies both zonally and meridionally, the artificially shifted anomalies no longer satisfy the vorticity or thermal balance, motivating us for another analysis in the next subsection.

c. Dependence of the energetics on the geographical location of anomalous convection

In addition to the aforementioned analysis based on the composited anomalies for I15°N,125°E, the same compositing has been repeated for each of the 55 OLR indices Iϕ,λ for PJ-like anomaly patterns with ϕ = 10°, 15°, 20°, 25°, and 30°N and λ = 105°, 110°, 115°, 120°, 125°, 130°, 135°, 140°, 145°, 150°, and 155°E. Then energy conversion/generation efficiencies for these PJ-like anomaly patterns have been evaluated on the basis of the composited anomalies with enhanced convection observed at various locations over the tropical/subtropical WNP (Table 3). Consistent with the results in the preceding subsection, the dry energy conversion (CK + CP) is particularly efficient when the enhanced convection is located around the northern Philippines (15°–20°N, 115°–125°E), as highlighted for τdry less than 30 days (Table 3a). The conversion efficiency remains fairly high against the longitudinal shift of the anomalous convection center as long as the center is at 15° or 20°N. Table 3a indicates a secondary domain of the highly efficient energy conversion near the Bonin Islands (27°N, 142°E) and to their east, for which a more detailed analysis is given in section 6. Again, the dry energy conversion is highly sensitive to the latitudinal position of enhanced convection, supporting the notion that the PJ pattern has characteristics of a dry dynamical mode in the climatological-mean field.

Table 3.

Time scales (a) τdry, (b) τmoist, and (c) τtotal (days) with which horizontally integrated KE + APE could be replenished through (a) the total energy conversion CK + CP, (b) diabatic generation CQ, and (c) their sum CK + CP + CQ for monthly composited anomalies associated with enhanced convection over the 55 domains whose centers are indicated with (ϕ, λ) in the table. The total energy (KE + APE) is integrated over the entire Northern Hemisphere, whereas the conversion/generation (CK + CP, CQ, and their sum) is integrated over [(ϕ − 15° to ϕ + 45°), (λ − 25° to λ + 25°)] for the composite based on the OLR index Iϕ,λ. The energy and its conversion/generation have been integrated vertically from the surface to the 100-hPa level before integrated horizontally. Time scales <30 days are highlighted with boldface. Time scales based on the composite with the enhanced convection over [10°–20°N, 120°–130°E], from which Fig. 2 and Table 1 have been produced, are indicated with boxes.

Time scales (a) τdry, (b) τmoist, and (c) τtotal (days) with which horizontally integrated KE + APE could be replenished through (a) the total energy conversion CK + CP, (b) diabatic generation CQ, and (c) their sum CK + CP + CQ for monthly composited anomalies associated with enhanced convection over the 55 domains whose centers are indicated with (ϕ, λ) in the table. The total energy (KE + APE) is integrated over the entire Northern Hemisphere, whereas the conversion/generation (CK + CP, CQ, and their sum) is integrated over [(ϕ − 15° to ϕ + 45°), (λ − 25° to λ + 25°)] for the composite based on the OLR index Iϕ,λ. The energy and its conversion/generation have been integrated vertically from the surface to the 100-hPa level before integrated horizontally. Time scales <30 days are highlighted with boldface. Time scales based on the composite with the enhanced convection over [10°–20°N, 120°–130°E], from which Fig. 2 and Table 1 have been produced, are indicated with boxes.
Time scales (a) τdry, (b) τmoist, and (c) τtotal (days) with which horizontally integrated KE + APE could be replenished through (a) the total energy conversion CK + CP, (b) diabatic generation CQ, and (c) their sum CK + CP + CQ for monthly composited anomalies associated with enhanced convection over the 55 domains whose centers are indicated with (ϕ, λ) in the table. The total energy (KE + APE) is integrated over the entire Northern Hemisphere, whereas the conversion/generation (CK + CP, CQ, and their sum) is integrated over [(ϕ − 15° to ϕ + 45°), (λ − 25° to λ + 25°)] for the composite based on the OLR index Iϕ,λ. The energy and its conversion/generation have been integrated vertically from the surface to the 100-hPa level before integrated horizontally. Time scales <30 days are highlighted with boldface. Time scales based on the composite with the enhanced convection over [10°–20°N, 120°–130°E], from which Fig. 2 and Table 1 have been produced, are indicated with boxes.

The moist energy generation (Table 3b) shows its high efficiency for the anomaly patterns associated with enhanced convection to the east or southeast of the Philippines (10°–20°N, 130°–150°E), where the dry conversion is generally not highly efficient (Table 3a). In contrast, the efficiency of the moist generation is much less effective for the convective events around the Philippines (including the eastern portion of the SCS) or around the Bonin Islands, where the associated dry energy conversion is particularly efficient (Table 3a). As a result, the total energy gain represented as the sum of the dry energy conversions and moist energy generation (CK + CP + CQ; Table 3c) is most efficient associated with enhanced convection over a broad domain that covers over and to the east of the Philippines, but the efficiency drops sharply for enhanced convective events that occur over the central or western portions of SCS or over the Chinese continent.

Major centers of vorticity anomalies for the 55 composites are plotted in Fig. 4. Obviously, the lower-tropospheric anticyclonic anomaly centers are tightly clustered around southern Japan (at ∼35°N), despite the corresponding convection centers scattered over the tropics, especially in the zonal direction (Fig. 4a). Though more scattered longitudinally than in the lower troposphere, the corresponding anticyclonic anomaly centers in the upper troposphere are confined to the vicinity of the Asian jet (Fig. 4b), which is consistent with the pronounced latitudinal dependence of the efficiency of the energy conversion and diabatic generation (Tables 2, 3). Specifically, Fig. 4b shows two clusters of the upper-tropospheric anticyclonic centers around 40°N near 150° and 170°E, the latter of which corresponds to the secondary cluster of the lower-tropospheric anticyclonic centers found around (38°N, 170°E) (Fig. 4a). Indeed, this eastern cluster represents the mode that typically accompanies anomalous convection near the Bonin Islands [around (25°N, 140°E)], whereas the primary cluster represents the mode with anomalous convection around the northern Philippines. This result is consistent with the energetic efficiencies shown in Table 3a. A more detailed discussion is given in section 6.

Fig. 4.

Locations of cyclonic (indicated with open circles) and anticyclonic (closed circles) anomaly centers at the (a) 850- and (b) 200-hPa levels, based on the 55 composited anomaly maps for monthly events of enhanced convection over the 55 subdomains within the tropical WNP. Each of the anomaly centers is defined as the local maximum of the composited vorticity anomaly, either cyclonic or anticyclonic, identified within a radius of 3000 km but excluded if its confidence level is lower than 90%. OLR anomaly centers for the individual composites are indicated by triangles. Sizes of the circles and triangles are proportional to magnitudes of the anomalies, with scaling at the top. Heavy shading indicates westerly regions of (a) > 5 m s−1 and (b) > 25 m s−1, whereas light shading in (a) indicates an easterly region of < −5 m s−1.

Fig. 4.

Locations of cyclonic (indicated with open circles) and anticyclonic (closed circles) anomaly centers at the (a) 850- and (b) 200-hPa levels, based on the 55 composited anomaly maps for monthly events of enhanced convection over the 55 subdomains within the tropical WNP. Each of the anomaly centers is defined as the local maximum of the composited vorticity anomaly, either cyclonic or anticyclonic, identified within a radius of 3000 km but excluded if its confidence level is lower than 90%. OLR anomaly centers for the individual composites are indicated by triangles. Sizes of the circles and triangles are proportional to magnitudes of the anomalies, with scaling at the top. Heavy shading indicates westerly regions of (a) > 5 m s−1 and (b) > 25 m s−1, whereas light shading in (a) indicates an easterly region of < −5 m s−1.

It is also obvious in Fig. 4 that upper-tropospheric cyclonic and anticyclonic anomaly centers are located systematically poleward of their lower-tropospheric counterpart. The poleward tilt of vorticity anomalies characterizes PJ-like anomaly patterns observed concurrently with enhanced convective activity that occurs in a broader domain extending from the northeastern portion of the SCS eastward to around the Bonin Islands. The zonal scattering of the upper-tropospheric anticyclonic anomaly centers may lead to a misinterpretation that a PJ-like pattern has eastward- or westward-tilted structure depending on the location of the anomalous convection. However, the vorticity anomalies in each of the composites are zonally elongated with westward and eastward tilts with height on only their western and eastern flanks, respectively.

In Fig. 4, cyclonic vorticity anomaly centers are also clustered over and to the west of the Sea of Okhotsk at the 850- and 200-hPa levels, respectively. Although their amplitudes are relatively small, they exhibit apparent northwestward phase tilt with height, as shown in Figs. 2b,c. It is suggested that the PJ pattern may be associated with the enhancement/weakening of the surface Okhotsk high, but the mechanisms for this meridional teleconnection are still unclear and beyond the scope of the present study.

4. PJ pattern in a simple dry model

Through our diagnosis in the preceding section, the critical importance of the Asian monsoon system, the North Pacific subtropical anticyclone, and the upper-level subtropical jet in the climatological-mean state has been suggested for the modal characteristics of the PJ pattern. In this section further analysis is carried out by using the quasigeostrophic model described in section 2b.

To eliminate any unnecessary complexity while retaining the essence of the dynamics of the PJ pattern, a simple, hypothetical basic field is prescribed in the model that consists only of a monsoonal low and a subtropical high as a low-level manifestation of the first baroclinic structure of the mean planetary wave and a vertically sheared, zonally uniform subtropical westerly jet. For our control experiment, the low-level basic state includes no zonally uniform structure, and only the pair of the cyclonic and anticyclonic eddies is present as the zonally varying component of the planetary wave (Fig. 5b). In the upper-level basic state, this pair of the cyclonic and anticyclonic circulations, after their signs have been reversed, is added to a zonally uniform westerly jet with its maximum wind speed of 20 m s−1 at 40°N (Fig. 5a). A weak tropical easterly wind is also added to the zonally symmetric component of the upper-level basic state, to suppress unrealistically strong propagation of Rossby waves into the tropics and their possible reflection at the equatorial boundary. Two sets of sensitivity experiments are also conducted, as described in section 4b. The static stability S = 3 × 10−6 J kg−1 Pa−2 is prescribed uniformly in each of the experiments.

Fig. 5.

Basic-state streamfunction (×106 m2 s−1; contoured) and horizontal wind u (vectors with scaling at the top) at (a) the upper and (b) lower levels for the control experiment of the model. Contours are drawn with intervals of (a) 5 (±2.5, ±7.5, ±12.5, …) and (b) 3 (±1.5, ±4.5, ±7.5, …). Solid and dashed lines indicate positive and negative values, respectively.

Fig. 5.

Basic-state streamfunction (×106 m2 s−1; contoured) and horizontal wind u (vectors with scaling at the top) at (a) the upper and (b) lower levels for the control experiment of the model. Contours are drawn with intervals of (a) 5 (±2.5, ±7.5, ±12.5, …) and (b) 3 (±1.5, ±4.5, ±7.5, …). Solid and dashed lines indicate positive and negative values, respectively.

a. Steady response to diabatic heating

Between the model monsoon and subtropical anticyclone, localized positive diabatic heating (Q′) is prescribed as the forcing (Fig. 6a), to which a steady response is investigated. At the lower level (Fig. 6c), the primary cyclonic response is centered to the immediate northwest of the Q′ center, whereas the primary anticyclonic response is located to its northeast. The upper-level response (Fig. 6b) is characterized by a pair of strong anticyclonic perturbations, one extending southwestward from the Q′ center and the other around the exit of the subtropical jet, with a cyclonic perturbation in between. The primary cyclonic and anticyclonic perturbations at the upper level are displaced poleward relative to their low-level counterpart. The associated wave-activity flux is pointing northward or northeastward at the lower level, whereas its equatorward component dominates at the upper level. These characteristics of the response are consistent with those in the composite analysis based on the observational data (section 3).

Fig. 6.

(a) The prescribed diabatic heating (K day−1), and the vorticity response (×10−6 s−1) at the (b) upper and (c) lower levels for the control experiment. Contours are drawn with intervals of (a) 0.5 (0.5, 1, 1.5, …) and (b),(c) 2 (±1, ±3, ±5, …). Solid and dashed lines indicate the positive (cyclonic) and negative (anticyclonic) values, respectively. Heavy shading indicates westerly regions of (b) > 15 m s−1 and (c) > 3 m s−1, whereas light shading in (c) indicates easterly regions of < −3 m s−1. Also shown with arrows in (b),(c) is a wave-activity flux (Takaya and Nakamura 2001), whose scaling is given at the top. The center of the diabatic heating is indicated with triangles in (b),(c).

Fig. 6.

(a) The prescribed diabatic heating (K day−1), and the vorticity response (×10−6 s−1) at the (b) upper and (c) lower levels for the control experiment. Contours are drawn with intervals of (a) 0.5 (0.5, 1, 1.5, …) and (b),(c) 2 (±1, ±3, ±5, …). Solid and dashed lines indicate the positive (cyclonic) and negative (anticyclonic) values, respectively. Heavy shading indicates westerly regions of (b) > 15 m s−1 and (c) > 3 m s−1, whereas light shading in (c) indicates easterly regions of < −3 m s−1. Also shown with arrows in (b),(c) is a wave-activity flux (Takaya and Nakamura 2001), whose scaling is given at the top. The center of the diabatic heating is indicated with triangles in (b),(c).

For the steady response shown in Fig. 6, barotropic KE conversion at the lower level (Fig. 7a) is distributed in a manner similar to that evaluated for the composited anomalies (Fig. 3a). Specifically, positive CK accompanied by westward-pointing E is found around the exits of the westerlies and easterlies in the tropics, whereas CK is weakly negative around the entrance of the midlatitude westerly jet. At the upper level (Fig. 7b), regions of positive and negative CK are aligned alternately around the exit of the subtropical jet as in the composite analysis (Fig. 3b). The net negative contribution apparently prevails for the model response, whereas in the composite analysis the positive and negative contributions are almost canceled out. Furthermore, the positive CK near the Q′ center and the negative CK to its east are not well correspondent to the CK distribution based on the composite analysis. These discrepancies perhaps arise from the oversimplification of the model basic state.

Fig. 7.

Local barotropic energy conversion CK (solid and dashed contours for ±3, ±9, ±15, … × 10−6 m2 s−3) with the extended EP flux (arrows with scaling at the top) at the (a) upper and (b) lower levels, and (c) local baroclinic energy conversion CP (heavy solid and dashed contours for ±5, ±15, ±25, … × 10−6 m2 s−3), based on the heat-induced response for the control experiment shown in Fig. 6. Solid and dashed lines represent energy conversion into the perturbations from the basic flow and vice versa. Heavy shading indicates westerlies of (a) > 3 m s−1 and (b) > 15 m s−1, whereas light shading in (a) indicates easterlies of < −3 m s−1. Superimposed with thin lines on (c) is the basic-state temperature at the middle level (every 1 K). The center of the diabatic heating is indicated with triangles.

Fig. 7.

Local barotropic energy conversion CK (solid and dashed contours for ±3, ±9, ±15, … × 10−6 m2 s−3) with the extended EP flux (arrows with scaling at the top) at the (a) upper and (b) lower levels, and (c) local baroclinic energy conversion CP (heavy solid and dashed contours for ±5, ±15, ±25, … × 10−6 m2 s−3), based on the heat-induced response for the control experiment shown in Fig. 6. Solid and dashed lines represent energy conversion into the perturbations from the basic flow and vice versa. Heavy shading indicates westerlies of (a) > 3 m s−1 and (b) > 15 m s−1, whereas light shading in (a) indicates easterlies of < −3 m s−1. Superimposed with thin lines on (c) is the basic-state temperature at the middle level (every 1 K). The center of the diabatic heating is indicated with triangles.

As in the composite analysis (Fig. 3c), positive CP is strong associated primarily with the westward-tilted perturbation embedded within the subtropical jet (Fig. 7c). Unlike in the composite, however, no negative CP is found to its east, because the anticyclonic perturbation exhibits no eastward phase tilt in the eastern flank. In Fig. 7c, another positive CP is found to the northeast of the Q′ center, associated with the poleward-tilted perturbation embedded in the zonal temperature gradient of the basic field [i.e., the second term of (4)]. Though overemphasized, this feature is also consistent with the composite analysis. Meanwhile, CP is strongly positive and negative also to the west and northwest of the Q′ center, respectively, with no counterpart in the composite analysis. Nevertheless, distribution of the moist diabatic generation (CQ > 0) is similar to that of the diabatic heating itself (figure not shown), as in the composite analysis.

Although a close comparison of the energetics between the model response and observed anomalies is not necessarily meaningful because of the oversimplification of the model, it is worthwhile to point out that the net contribution of each of the energy conversions and generation to the energy budget for the model response are qualitatively in good agreement with those for the composite analysis (Table 1). In the “Net” (over the hemispheric domain) and the “forcing sector” (0°–60°N, 155°E–155°W), the lower-level CK and CP are both positive, and so are CK + CP and CQ (Table 1). The most notable discrepancy between the energetics of the heat-induced response and those based on composite analysis is the stronger negative CK in the model upper layer. This may arise from the model’s constraint that the response must be stationary, which requires that all the energy generated and converted into the response should be dissipated or converted back into the basic flow elsewhere. Aside from this modest discrepancy, the model is nevertheless successful in simulating the essential features of the dynamics and energetics of the PJ pattern, including predominance of CP as well as comparable contributions of the dry energy conversion and moist diabatic generation.

b. Sensitivity to the configuration of the model basic field

The particular importance of the combined effects between the zonal asymmetry of the basic flow characterized by a monsoonal low and a subtropical anticyclone and the zonally symmetric subtropical jet can be confirmed through a set of sensitivity experiments, where the basic flow consists only of either the zonally symmetric or asymmetric component. For the purely zonally symmetric basic flow (Figs. 8a,c), the thermal response is obviously dissimilar to the PJ pattern. Unlike in the control experiment, the response is in the first baroclinic structure around the Q′ center. Unlike in the composite (Fig. 2) and control experiment (Fig. 6), poleward wave-activity propagation is evident only at the upper level as indicated by a northeastward-pointing wave-activity flux (Fig. 8a). It yields a weak cyclonic perturbation in midlatitudes to the north of the Q′ center, whose polarity is the opposite of that in the response for the control experiment.

Fig. 8.

Vorticity response (×10−6 s−1) at the (a),(b) upper and (c),(d) lower levels to the prescribed diabatic heating shown in Fig. 6a under the basic flow constructed only from the (a),(c) zonally symmetric and (b),(d) zonally asymmetric components of the basic flow shown in Fig. 5. Heavy and light shading in (b),(d) indicate westerlies of > 3 m s−1 and easterlies of < −3 m s−1, respectively, whereas heavy shading in (a) indicates a strong westerlies of > 15 m s−1. Contours, arrows, and triangles are as in Figs. 6b,c.

Fig. 8.

Vorticity response (×10−6 s−1) at the (a),(b) upper and (c),(d) lower levels to the prescribed diabatic heating shown in Fig. 6a under the basic flow constructed only from the (a),(c) zonally symmetric and (b),(d) zonally asymmetric components of the basic flow shown in Fig. 5. Heavy and light shading in (b),(d) indicate westerlies of > 3 m s−1 and easterlies of < −3 m s−1, respectively, whereas heavy shading in (a) indicates a strong westerlies of > 15 m s−1. Contours, arrows, and triangles are as in Figs. 6b,c.

In the purely zonally asymmetric basic flow without the subtropical jet (Figs. 8b,d), the response looks more similar to that for the control experiment and the observed PJ pattern. An anticyclonic perturbation forms at the lower level to the northeast of the Q′ center. The poleward phase tilt of the cyclonic perturbation to the northeast of the Q′ center indicates the importance of the vertically sheared meridional flow to the PJ-like response. Although the CK distribution in this experiment bears some resemblance to that in the control experiment shown in Fig. 7a, including the KE gain in the exits of the tropical westerly and easterly jets (figure not shown), the CP distribution is notably different from that in the control experiment shown in Fig. 7c. Although positive CP in midlatitudes can be found in association with the poleward-tilted perturbation embedded in the zonal temperature gradient of the basic field [i.e., the second term of (4)], no positive CP is found in midlatitudes as a contribution from the first term of (4) in the absence of the subtropical jet (figure not shown). Correspondingly, the midlatitude upper-level anticyclonic perturbation is weaker and associated with cool perturbation locally, which is in contrast to its counterpart for the control experiment shown in Fig. 6. As a result, the net energy conversion from the basic flow is less efficient, which elucidates the important role of the baroclinic subtropical (Asian) jet in the energetics of the PJ pattern.

Dependence of the horizontal structure of the PJ-like response, especially its meridional scale, on the basic field can be elucidated by varying the meridional scale of the monsoon and the subtropical anticyclone in the model synchronously with the axial latitude of the subtropical jet. A comparison is made between the experiments with the jet axis at 35°N (Figs. 9a,c) and 50°N (Figs. 9b,d). Although the primary lower-level cyclonic response is anchored near the Q′ center, whose position is unchanged between the two experiments, the primary anticyclonic response is displaced poleward and westward as the meridional width of the basic flow increases. In the case of the meridionally narrow monsoon, the primary lower-level anticyclonic center is located in the entrance of the lower-level midlatitude westerly jet to the northeast of the Q′ center, and the corresponding upper-level anticyclonic center is quite weak. Under the meridionally broader monsoon, in contrast, the primary upper-level anticyclonic response is situated in the exit of the subtropical westerly jet, and its lower-level counterpart is much stronger than in the case of narrower monsoon. It is thus suggested that the specific structure of the PJ pattern depends rather sensitively on the meridional scale of the basic flow, which seems consistent with our composite analysis that has demonstrated the sensitivity of the energy conversions to the meridional position of the pattern relative to the jets (Tables 2, 3). In addition, the zonal location of the anticyclonic response may also be sensitive to the horizontal scale of the perturbation, whose sensitivity to the basic flow structure arises from the vorticity balance.

Fig. 9.

Vorticity response (×10−6 s−1) at the (a),(b) upper and (c),(d) lower levels to the prescribed diabatic heating shown in Fig. 6a for the experiments with the meridionally (a),(c) “narrow” and (b),(d) “wide” basic flows. Contours are drawn with intervals of (left) 2 (±1, ±3, ±5, …) and (right) 3 (±1.5, ±4.5, ±7.5, …). Solid and dashed lines indicate positive and negative values, respectively. Heavy shading indicates westerlies of (a),(b) > 15 m s−1 and (c),(d) > 3 m s−1, whereas light shading in (c),(d) indicates easterlies of < −3 m s−1. Also shown with arrows is a wave-activity flux, whose scaling is given at the top. The center of the diabatic heating is indicated with triangles.

Fig. 9.

Vorticity response (×10−6 s−1) at the (a),(b) upper and (c),(d) lower levels to the prescribed diabatic heating shown in Fig. 6a for the experiments with the meridionally (a),(c) “narrow” and (b),(d) “wide” basic flows. Contours are drawn with intervals of (left) 2 (±1, ±3, ±5, …) and (right) 3 (±1.5, ±4.5, ±7.5, …). Solid and dashed lines indicate positive and negative values, respectively. Heavy shading indicates westerlies of (a),(b) > 15 m s−1 and (c),(d) > 3 m s−1, whereas light shading in (c),(d) indicates easterlies of < −3 m s−1. Also shown with arrows is a wave-activity flux, whose scaling is given at the top. The center of the diabatic heating is indicated with triangles.

c. Preferred modes in the model basic state

To further explore the possibility that the PJ pattern has characteristics of a dry dynamical mode, a modal analysis has been executed for the linear model with the control basic flow as shown in Fig. 5. The set of model Eqs. (1) and (2) can be expressed in the form of a linear algebra 𝗟x = f, where x includes the perturbation quantities and Q′ is included in f as the sole forcing term. The basic-state features are represented in the linear operator L, which is not self-adjoint in general for a system linearized about a zonally asymmetric basic flow. Its eigenvectors, therefore, do not constitute an orthogonal set, complicating their interpretation and mathematical application. Instead, an orthogonal set for the model system can be obtained through a singular value decomposition (SVD; Navarra 1993) as 𝗟 = 𝗨Σ𝗩T, where 𝗨 and 𝗩 are orthonormal matrices and Σ is a diagonal matrix of the singular values. Then, the stationary response x to the prescribed forcing f can be expressed as

 
formula

where ui and vi are the ith columns of the 𝗨 and 𝗩 matrices, respectively, and σi is the ith singular value. Equation (6) indicates that the solution x can be expressed as a linear combination of the v vectors, each weighted by the projection of the forcing f onto the corresponding u vector multiplied by a factor 1/σi. It follows that modes with the smallest singular values are dominant in a steady response to random forcing. In such a case with no mechanical forcing in (1) as in our application, f contains thermal forcing only, and only the corresponding elements of ui contribute to the weight (ui · f/σi) in (6).

For the basic state for the control experiment, the leading singular mode (SVD1), with the smallest singular value, is found to represent vorticity perturbations confined mostly to midlatitudes (figure not shown), probably related to baroclinically growing disturbances along the subtropical jet. Together with the third leading singular mode (SVD3; figure not shown), SVD1 represents wave-like perturbations along the subtropical jet, which may correspond to the Silk Road pattern, whose modal structure has been pointed out by Kosaka et al. (2009). As shown in Fig. 10c, the second singular mode (SVD2) is characterized by cyclonic and anticyclonic perturbations around 20°N and in midlatitudes, respectively, at the lower level. At the upper level (Fig. 10a), the mode is characterized by anticyclonic perturbations both in midlatitudes and the tropics and a cyclonic perturbation in between. These vorticity perturbations are tilted poleward with height, with their centers in good correspondence to their counterpart in the heat-induced model response in the control experiment (Fig. 6). The diabatic heating to excite SVD2 with the particular polarity as shown in Fig. 10 should be positive around the primary cyclonic center in the tropics (Fig. 10d). At the location where the positive Q′ center is placed for the control experiment, Q′ for exciting SVD2 with the same polarity as in the response for the control experiment is indeed positive, and the anomalous vertical motion associated with SVD2 is upward (Fig. 10b). In fact, the weight in the linear combination (6) assigned for SVD2 by the tropical Q′ centered at (20°N, 180°) is more than 2.5 times larger than the corresponding weight for any other SVD mode. This explains why the PJ-like SVD2 is excited by the tropical Q′ in the control experiment. These features, as well as the second smallest singular value that is equivalent to the corresponding damping time scale longer than a month, indicate that the SVD2 is the leading mode of variability that shapes the PJ-like heat-induced response, confirming the characteristic of the PJ pattern as a dry dynamical mode.

Fig. 10.

Vorticity perturbation (×10−6 s−1) at the (a) upper and (c) lower levels, and (b) vertical p velocity (×10−3 Pa s−1) in the response vector of SVD2 obtained for the control basic flow shown in Fig. 5. (d) Diabatic heating (K day−1) in the corresponding forcing vector. Contours are drawn with intervals of (a),(c),(d) 1 (±0.5, ±1.5, ±2.5, …) and (b) 2 (±1, ±3, ±5, …). Solid and dashed lines indicate positive and negative values, respectively. Heavy shading indicates westerlies of (a) > 15 m s−1 and (c) > 3 m s−1, whereas light shading in (c) indicates easterlies of < −3 m s−1. The corresponding singular value is (38.488 days)−1. The center of the diabatic heating in the control experiment is indicated with triangles.

Fig. 10.

Vorticity perturbation (×10−6 s−1) at the (a) upper and (c) lower levels, and (b) vertical p velocity (×10−3 Pa s−1) in the response vector of SVD2 obtained for the control basic flow shown in Fig. 5. (d) Diabatic heating (K day−1) in the corresponding forcing vector. Contours are drawn with intervals of (a),(c),(d) 1 (±0.5, ±1.5, ±2.5, …) and (b) 2 (±1, ±3, ±5, …). Solid and dashed lines indicate positive and negative values, respectively. Heavy shading indicates westerlies of (a) > 15 m s−1 and (c) > 3 m s−1, whereas light shading in (c) indicates easterlies of < −3 m s−1. The corresponding singular value is (38.488 days)−1. The center of the diabatic heating in the control experiment is indicated with triangles.

The particular characteristic of the heat-induced response (Fig. 6) as a preferred mode of this model atmosphere with the control basic state (Fig. 5) can be further elucidated in Fig. 11, which shows the centers of lower-level cyclonic and anticyclonic perturbations as responses to heating Q′ centered at 21 locations within the joint exit region of the lower-level westerly and easterly jets in the tropics. Most of the anticyclonic perturbation centers are clustered within the entrance region of the midlatitude lower-level westerlies (below the exit of the upper-level westerly jet). They are responses to Q′ located within the exit of the lower-level easterly jet or its confluence region with the lower-level westerlies, where the magnitude of Q′ to excite the SVD2 mode is particularly large (Fig. 10d). Among them, steady response to heating centered at (26°N, 162°W), near the maximum of positive Q′ to excite SVD2 (Fig. 10d), is most similar to the SVD2 pattern itself. In Fig. 11, another region of anticyclonic perturbations is located in the core of the midlatitude upper-level westerly jet (above the midlatitude lower-level easterlies), as a response to Q′ located on the northern flank of the exit of the tropical lower-level westerlies, where Q′ to excite SVD2 is weak or negative (Fig. 10d). These anticyclonic perturbations appear to be a manifestation of the excited SVD1 mode.

Fig. 11.

Locations of cyclonic (open circles) and anticyclonic (closed circles) perturbation centers at the lower level, based on the steady model response to diabatic heating (triangles) centered at each of 150°E, 160°E, 170°E, 180°, 170°W, 160°W, and 150°W in longitude and each of 15°, 20°, and 25°N in latitude with the basic state shown in Fig. 5. Size of a circle is proportional to the magnitude of a given perturbation at that point, with scaling at the top. Superimposed by a closed line is −1 × 10−5 (s−1) isolines of the vorticity perturbation in a given response. Dashed lines represent the correspondence of a pair of cyclonic and anticyclonic perturbations in a particular response. Heavy and light shading indicate westerlies of > 3 m s−1 and easterlies of < −3 m s−1, respectively, at the lower level.

Fig. 11.

Locations of cyclonic (open circles) and anticyclonic (closed circles) perturbation centers at the lower level, based on the steady model response to diabatic heating (triangles) centered at each of 150°E, 160°E, 170°E, 180°, 170°W, 160°W, and 150°W in longitude and each of 15°, 20°, and 25°N in latitude with the basic state shown in Fig. 5. Size of a circle is proportional to the magnitude of a given perturbation at that point, with scaling at the top. Superimposed by a closed line is −1 × 10−5 (s−1) isolines of the vorticity perturbation in a given response. Dashed lines represent the correspondence of a pair of cyclonic and anticyclonic perturbations in a particular response. Heavy and light shading indicate westerlies of > 3 m s−1 and easterlies of < −3 m s−1, respectively, at the lower level.

Interestingly, Fig. 10d indicates that the region of positive Q′ to excite SVD2 extends into the basic-state subtropical anticyclone, suggestive of a correspondence with the secondary domain of efficient energy conversion near the Bonin Islands in our composite analysis (Table 3a). A negative Q′ maximum is located on the northwestern side of the primary positive domain (Fig. 10d), suggesting a possible excitation of the PJ pattern by anomalous precipitation in the baiu/mei-yu frontal zone (Lu and Lin 2009). Other maxima of Q′ to excite SVD2 are found in the upstream of the subtropical jet (Fig. 10d). For Q′ placed at this particular location, a response is quite similar to the SVD2 pattern with a slight emphasis on midlatitude wave-like perturbations, whose polarities depend on the sign of Q′ assigned. Though in the real atmosphere deep cumulus convection is not active over central Asia, this result implies a possible connection between the PJ pattern and wavy anomalies along the Asian jet. In fact, Kosaka et al. (2009) indicated coexistence of the monthly Silk Road pattern with the PJ pattern, especially in July, as also hinted in our composite map (Fig. 2c). Because their land–sea contrasts that could control convective activity or orography that would modify low-level circulation over the Asian continent are missing in our highly idealized model, further discussion on the correspondence with the real atmosphere on this issue is beyond the scope of the present study.

5. Moist processes

In discussions in sections 3 and 4, diabatic heating associated with enhanced cumulus convection over the tropical WNP has been regarded as if it were an external forcing onto the anomalous circulation associated with the PJ pattern. However, the circulation anomalies associated with the PJ pattern can affect the convective activity through anomalous vertical motion, stratification, and moisture supply, which, if significant, may require us to regard the PJ pattern as a moist dynamical mode that interacts with the climatological-mean state through both dry dynamics and moist processes. In this section, several factors that are associated with moist processes and possibly contribute to the enhancement of the anomalous convective activity are investigated.

a. Interaction with the ocean and moisture transport

KN06 discussed the relationship between the anomalous convective activity around the Philippines and SST anomalies based on their composited anomalies. They found that the anomalous low-level circulation around the enhanced convection associated with the PJ pattern increases the surface wind speed, thereby enhancing evaporation from the ocean surface (see Figs. 14h,i). This process is similar to the wind-induced surface heat exchange (WISHE; Emanuel 1987) or wind–evaporation feedback (Neelin et al. 1987). It should be noted that the intensified surface wind speed and evaporation around the enhanced convection act to lower SST locally, as actually observed over SCS in our monthly statistics (B. Wang et al. 2005; KN06).

Fig. 14.

Horizontal distributions of (a) CMAP precipitation anomalies (mm day−1); vorticity anomalies (×10−6 s−1; contours) and wave-activity flux (arrows with scaling at the top of the panels) at the (b) 850-hPa and (c) 200-hPa levels; (h) anomalous wind (arrows with scaling at the top) and wind speed anomalies (m s−1; contours) at the 925-hPa level; and (i) latent heat flux anomalies (W m−2) from the surface, regressed onto the leading PC of horizontally smoothed vorticity over [0°–60°N, 100°–160°E] at the 850-hPa level for JJA, based on the JRA-25. Local barotropic energy conversion CK (×10−5 m2 s−3; contours) and the extended EP flux (arrows with scaling at the top) at the (d) 850- and (e) 200-hPa levels and (f) baroclinic energy conversion CP and (g) diabatic APE generation CQ integrated vertically from the surface to the 100-hPa level (×10−2 W m−2; heavy contours), all evaluated based on the regressed anomalies (a)–(c). Contour intervals are (a),(b),(d) 0.5 (±0.25, ±0.75, ±1.25, …); (c),(e),(f) 1 (±0.5, ±1.5, ±2.5, …); (g),(i) 2 (±1, ±3, ±5, …); and (h) 0.2 (±0.1, ±0.3, ±0.5, …). Solid and dashed lines indicate (a)–(c),(h),(i) positive and negative anomalies, respectively, and (d)–(g) energy conversion/generation into the anomalies and vice versa, respectively. The primary OLR anomaly center is indicated with triangles. Light and heavy shading represent (a)–(c),(h),(i) the confidence levels of 90% and 95%, respectively, based on the t statistic, and (d) the climatological-mean easterlies ( < −5 m s−1) and westerlies ( > 5 m s−1), respectively, whereas heavy shading in (e) indicates the climatological westerlies with > 25 m s−1. In (f), the climatological-mean 400-hPa temperature is indicated with light contours for every 1 K, and shading indicates regions where the contribution of eddy zonal heat transport exceeds 80% of total positive CP.

Fig. 14.

Horizontal distributions of (a) CMAP precipitation anomalies (mm day−1); vorticity anomalies (×10−6 s−1; contours) and wave-activity flux (arrows with scaling at the top of the panels) at the (b) 850-hPa and (c) 200-hPa levels; (h) anomalous wind (arrows with scaling at the top) and wind speed anomalies (m s−1; contours) at the 925-hPa level; and (i) latent heat flux anomalies (W m−2) from the surface, regressed onto the leading PC of horizontally smoothed vorticity over [0°–60°N, 100°–160°E] at the 850-hPa level for JJA, based on the JRA-25. Local barotropic energy conversion CK (×10−5 m2 s−3; contours) and the extended EP flux (arrows with scaling at the top) at the (d) 850- and (e) 200-hPa levels and (f) baroclinic energy conversion CP and (g) diabatic APE generation CQ integrated vertically from the surface to the 100-hPa level (×10−2 W m−2; heavy contours), all evaluated based on the regressed anomalies (a)–(c). Contour intervals are (a),(b),(d) 0.5 (±0.25, ±0.75, ±1.25, …); (c),(e),(f) 1 (±0.5, ±1.5, ±2.5, …); (g),(i) 2 (±1, ±3, ±5, …); and (h) 0.2 (±0.1, ±0.3, ±0.5, …). Solid and dashed lines indicate (a)–(c),(h),(i) positive and negative anomalies, respectively, and (d)–(g) energy conversion/generation into the anomalies and vice versa, respectively. The primary OLR anomaly center is indicated with triangles. Light and heavy shading represent (a)–(c),(h),(i) the confidence levels of 90% and 95%, respectively, based on the t statistic, and (d) the climatological-mean easterlies ( < −5 m s−1) and westerlies ( > 5 m s−1), respectively, whereas heavy shading in (e) indicates the climatological westerlies with > 25 m s−1. In (f), the climatological-mean 400-hPa temperature is indicated with light contours for every 1 K, and shading indicates regions where the contribution of eddy zonal heat transport exceeds 80% of total positive CP.

Anomalous surface convergence associated with the cyclonic anomalies around the enhanced convection center also tends to sustain or reinforce the anomalous convection (KN06). The anomalous moisture in the planetary boundary layer thus transported into the enhanced convection region further favors cumulus convection (moisture–convection feedback; Tompkins 2001). These processes suggest that the PJ-associated anomalous circulation serves to reinforce the associated anomalous convective activity around the Philippines.

b. Dynamically induced anomalous vertical motion

Dynamically induced large-scale ascent can trigger active convection. Using the composite based on I15°N,125°E (Fig. 2), a diagnosis with the linearized omega equation

 
formula

has been performed for the Northern Hemisphere. Again, overbars and primes indicate climatological-mean quantities for JJA and the composited anomalies, respectively. In solving (7), ω′ is set to be zero along the lateral boundaries and at the bottom (1000 hPa) and top (70 hPa) boundaries.

The composited anomalous vertical p velocity at the 400-hPa level associated with the PJ pattern (Fig. 12a) is in good correspondence with the composited anomalous precipitation (Fig. 2a). Though substantially weaker than its observational counterpart (Fig. 12a), the omega equation diagnoses an anomalous ascent around the center of the enhanced convection (Fig. 12b). Because the omega equation (7) does not include diabatic heating as the forcing, this ascent can be regarded as a direct contribution from dry processes through anomalous vorticity and thermal advection. This ascent is also consistent with our SVD analysis for the idealized model, where SVD2 accompanies ascent in the tropics where the prescribed diabatic heating is placed (Fig. 10b). From this diagnosis with the omega equation only with dry processes, in addition to the result in section 3 that the PJ pattern can be regarded as a dry dynamical mode in the climatological-mean field, it is hypothesized that the vorticity and/or thermal advection (i.e., dry process) associated with the PJ pattern can trigger or reinforce convective activity over the tropical WNP by inducing anomalous ascent around the enhanced convection, though verification through examining daily evolution may be needed to substantiate our hypothesis.

Fig. 12.

Anomalous vertical p velocity (×10−3 Pa s−1) at the 400-hPa level based on (a) the composite for the 38 strongest monthly events of the PJ pattern with enhanced convective activity in a tropical domain [10°–20°N, 120°–130°E] and (b) a diagnosis using the linearized omega equation with vorticity and thermal advections derived from the same composite. Contours are drawn with intervals of (a) 3 (±1.5, ±4.5, ±7.5, …) and (b) 1 (±0.5, ±1.5, ±2.5, …). Solid and dashed lines indicate anomalous descent and ascent, respectively. In (a), light and heavy shading represent the confidence levels of 90% and 95%, respectively, based on the t statistic. Light shading in (b) indicates that more than 50% of the diagnosed p velocity is induced solely by vorticity advection anomalies above the 400-hPa level. The OLR anomaly center is indicated with triangles.

Fig. 12.

Anomalous vertical p velocity (×10−3 Pa s−1) at the 400-hPa level based on (a) the composite for the 38 strongest monthly events of the PJ pattern with enhanced convective activity in a tropical domain [10°–20°N, 120°–130°E] and (b) a diagnosis using the linearized omega equation with vorticity and thermal advections derived from the same composite. Contours are drawn with intervals of (a) 3 (±1.5, ±4.5, ±7.5, …) and (b) 1 (±0.5, ±1.5, ±2.5, …). Solid and dashed lines indicate anomalous descent and ascent, respectively. In (a), light and heavy shading represent the confidence levels of 90% and 95%, respectively, based on the t statistic. Light shading in (b) indicates that more than 50% of the diagnosed p velocity is induced solely by vorticity advection anomalies above the 400-hPa level. The OLR anomaly center is indicated with triangles.

In addition to the anomalous tropical ascent, significant anomalous descent is found to the southeast of Japan (Fig. 12a). Although this descent is collocated with reduced precipitation (Fig. 2a), the northern portion of the anomalous descent is contributed to also by that accompanied by the anomalous vorticity and/or thermal advection (Fig. 12b), based on our diagnosis using the omega equation.

We have also solved the linearized omega equation (7) separately with the individual contributions to its RHS from the total anomalous vorticity advection, the total anomalous thermal advection, the mean zonal advection of the anomalies, the mean meridional advection of the anomalies, the advection of the mean variables by the anomalous winds, and all the anomalies but either above or below the 400-hPa level only. It turns out that the anomalous ascent around the OLR anomaly center and to its east is associated mainly with the anomalous vorticity advection [i.e., the first term on the RHS of (7)]. Furthermore, the contribution from the mean zonal advection is found negligible to the anomalous ascent, which, together with the dominance of the meridional vorticity advection, is consistent with the vorticity budget analysis by KN06. The mean meridional advection term is stronger in the upper troposphere than at the lower levels, therefore contributing more to the midtropospheric anomalous ascent. In fact, the diagnosed ascent is mainly induced by the upper-tropospheric vorticity advection anomalies (Fig. 12b).

c. Relationship with tropical cyclone activity

Nitta (1987) suggested that anomalous convective activity over the tropical WNP may be related to the anomalous activity of tropical cyclones. In fact, recent observational studies (Kawamura and Ogasawara 2006; Yamada and Kawamura 2007) indicated an excitation of PJ-like anomalies by typhoons. Figures 13a,b show tracks of tropical cyclones for the 38 strongest positive and negative events of the monthly PJ pattern associated with the most enhanced and suppressed convective activity, respectively, over [10°–20°N, 120°–130°E]. Indeed, the monthly events of enhanced convective activity (Fig. 13a) accompany much more tropical cyclones in the vicinity of the center of monthly OLR anomalies around the northern Philippines than the events of suppressed convection do (Fig. 13b), although slightly fewer cyclones are observed over the SCS. This result indicates that the monthly-mean tropical convective activity associated with the PJ pattern is, at least in part, a manifestation of modulated tropical cyclone activity. It is also suggested that a more thorough understanding of the PJ pattern requires an examination of shorter time-scale processes.

Fig. 13.

Tracks of severe tropical storms and typhoons (thin lines) observed in the 38 monthly (a) positive and (b) negative convection events of the PJ pattern for JJA with the most enhanced and suppressed monthly convective activity, respectively, over [10°–20°N, 120°–130°E]. Also indicated with thick contours are the corresponding monthly OLR anomalies (W m−2) composited for the corresponding 38 events. Contours are drawn with an interval of 4 (±2, ±6, ±10, …). Solid and dashed contours indicate positive and negative values, respectively. Light and heavy shading indicate the confidence levels of 90% and 95%, respectively, based on the t statistic for the OLR anomalies.

Fig. 13.

Tracks of severe tropical storms and typhoons (thin lines) observed in the 38 monthly (a) positive and (b) negative convection events of the PJ pattern for JJA with the most enhanced and suppressed monthly convective activity, respectively, over [10°–20°N, 120°–130°E]. Also indicated with thick contours are the corresponding monthly OLR anomalies (W m−2) composited for the corresponding 38 events. Contours are drawn with an interval of 4 (±2, ±6, ±10, …). Solid and dashed contours indicate positive and negative values, respectively. Light and heavy shading indicate the confidence levels of 90% and 95%, respectively, based on the t statistic for the OLR anomalies.

6. Two dominant patterns of variability over the WNP

a. Modal characteristics of two major anomaly patterns over the WNP

In this section, an EOF analysis is applied to monthly anomalies of 850-hPa vorticity over [0°–60°N, 100°–160°E] for JJA. The first (EOF1) and second (EOF2) modes are found to explain 22.5% and 14.8%, respectively, of the total variance, and they are thus well separated based on the criteria by North et al. (1982). These two EOFs become inseparable on the same criteria if the domain for the EOF analysis is widened eastward by 20° in longitude, though their structures remain qualitatively unchanged.

Structure of the EOF1 pattern, represented in anomalies regressed against the corresponding principal component (PC1) time series, is characterized by a meridional dipole of zonally elongated vorticity anomalies in the lower troposphere (Fig. 14b), with a notable poleward tilt of their phase lines into the upper troposphere (Fig. 14c). For the particular polarity shown in Fig. 14, enhanced precipitation (Fig. 14a) is observed slightly equatorward (∼15°N) of the 850-hPa primary cyclonic anomaly center (∼19°N), whereas a reduction of precipitation is significant to its northeast near Japan as well as its southwest over the Maritime Continent. These features are consistent with those of the PJ pattern in our composite analysis in section 3. However, the enhanced precipitation and lower-tropospheric cyclonic anomalies, as well as the associated diabatic energy generation (Fig. 14g) and the lower-tropospheric barotropic energy conversion (Fig. 14d), extend farther eastward into the date line in our EOF1 mode, resulting in higher efficiencies in the conversion and generation (Table 1). The efficient diabatic energy generation is also contributed to by anomalous diabatic cooling over the Maritime Continent (Fig. 14g), which is almost absent in the PJ pattern composited in section 3.

Figure 15 shows the structure of EOF2 based on monthly anomalies of 850-hPa vorticity. In addition to the meridional displacement by nearly a quarter wavelength from their counterpart in EOF1, the regressed anomalies in EOF2 (Fig. 15), including a positive precipitation anomaly near (20°N, 153°E), exhibit a noticeable eastward shift relative to the PJ pattern extracted in EOF1. This center of enhanced convection is located in the southern portion of the secondary domain over the tropical/subtropical WNP where anomalous convection accompanies circulation anomalies that can achieve efficient energy conversion from the mean state based on the composites (Table 3). In fact, the OLR anomaly center associated with EOF2 coincides with a secondary maximum of monthly OLR variance located in [15°–20°N, 150°–160°E], whereas the corresponding center for EOF1 is collocated with its primary maximum near the Philippines (Fig. 16). Furthermore, the midlatitude centers of vorticity anomalies associated with EOF1 (Fig. 14) and EOF2 (Fig. 15) also correspond well to the pairs of clusters of anomalous vorticity centers near (35°N, 140°E) and (38°N, 170°E), respectively, in the lower troposphere and to (40°N, 150°E) and (43°N, 170°E), respectively, in the upper troposphere (Fig. 4), based on the composites for the anomalous convective activity in the tropical/subtropical WNP.

Fig. 15.

As in Fig. 14, but for the anomalies regressed onto the second PC of horizontally smoothed vorticity over [0°–60°N, 100°–160°E] at the 850-hPa level for JJA, based on the JRA-25. In (d), contours are drawn with an interval of 0.1 (±0.05, ±0.15, ±0.25, …).

Fig. 15.

As in Fig. 14, but for the anomalies regressed onto the second PC of horizontally smoothed vorticity over [0°–60°N, 100°–160°E] at the 850-hPa level for JJA, based on the JRA-25. In (d), contours are drawn with an interval of 0.1 (±0.05, ±0.15, ±0.25, …).

Fig. 16.

Climatological variance of monthly OLR for JJA. Contours are drawn with an interval of 20 (W m−2)2. Light and heavy shading indicate values larger than 180 and 240 (W m−2)2, respectively. The OLR anomaly centers in EOF1 (Fig. 14) and EOF2 (Fig. 15) are indicated with triangles.

Fig. 16.

Climatological variance of monthly OLR for JJA. Contours are drawn with an interval of 20 (W m−2)2. Light and heavy shading indicate values larger than 180 and 240 (W m−2)2, respectively. The OLR anomaly centers in EOF1 (Fig. 14) and EOF2 (Fig. 15) are indicated with triangles.

Consistent with the energetics based on composite maps as summarized in Table 3a, the dry energy conversion for the anomalies associated with EOF2 is found nearly as efficient as its counterpart for EOF1 (Table 1). Compared with the EOF1 anomalies, the EOF2 anomalies tend to have barotropic conversion CK that is less efficient in the lower troposphere because of poleward displacement of the anomalous circulation relative to the axis of the trade winds (Fig. 15d). Nevertheless, CK is more efficient in the upper troposphere (Fig. 15e), yielding its net efficiency within the entire troposphere comparable to that of EOF1. Embedded in the exit of the Asian jet whose axis shows a noticeable southwest–northeast tilt to the east of Japan, upper-tropospheric zonal wind anomalies along the node of a meridional dipole of zonally elongated vorticity anomalies extracted in EOF2 yield efficient KE gain from the jet. In the jet exit, the anomaly pattern with EOF2 also yields baroclinic energy conversion (Fig. 15f) that is as efficient as that for EOF1 (Table 1). Magnitude of CP is even stronger locally than that associated with EOF1. Meanwhile, the enhanced convection is displaced zonally relative to a midtropospheric warm anomaly, resulting in positive and negative diabatic energy generations in the western and eastern portions, respectively, of the anomalous convection (Fig. 15g). Instead, strong diabatic generation is associated with a midtropospheric cold anomaly collocated with reduced rainfall around the equator (Fig. 15g). Despite this significant contribution, the net diabatic generation associated with the EOF2 anomalies is nevertheless much less efficient than the dry energy conversion (Table 1), qualitatively consistent with the result shown in Table 3b.

Anomalous lower-tropospheric moisture convergence has been obtained for both EOF1 and EOF2 based on monthly anomalies regressed on their respective PCs. A self-sustaining tendency is apparent only for anomalous convection associated with EOF1, with enhanced evaporation around the increased precipitation (Figs. 14h,i). This result is not surprising, because EOF1 well corresponds to the composited PJ pattern. For EOF2, in contrast, enhanced convective activity accompanies reduced evaporation (Fig. 15i) partly associated with the superposition of anomalous westerlies on the trade winds to the south of 20°N (Figs. 15h). This difference in the effectiveness of WISHE may be a factor that can contribute to the weaker precipitation anomalies associated with EOF2 (Fig. 15c) than those with EOF1 (Fig. 14c) and thereby less efficient diabatic generation for EOF2 (Table 1).

Overall, the energetics and moist processes related to the EOF1 and EOF2 patterns suggest the possible existence of two dynamical modes over the summertime WNP, one accompanying anomalous convection around the northern Philippines and the other near the Bonin Islands, as indicated in Table 3a. In other words, EOF1 well corresponds to the PJ pattern, whereas the EOF2 may be regarded as its “brother.” In fact, the variance of monthly OLR for the JJA season exhibits two local maxima over the WNP; the stronger maximum coincides with a center of anomalous OLR associated with EOF1, and the weaker maximum is collocated with the corresponding center for EOF2 (Fig. 16). It is conjectured that, for anomalous convection, if generated between these two centers of action, energy conversion/generation is not efficient enough for its substantial amplification.

b. Relationship with SST variability

Interestingly, the structure of EOF1 is similar to a 2–3-yr oscillation found in the intensity of the Bonin high found by Sui et al. (2007), whereas EOF2 may be related to the 3–5-yr oscillation of the high as indicated in their study. Kim et al. (2009) also found that two patterns of anomalous convection whose centers of action correspond to those in our EOF1 and EOF2. In the lag-regression maps of SST anomalies shown in Figs. 17a,c,e, La Niña and El Niño signals are found in the preceding and following winters, respectively, for the particular polarity of EOF1 shown in Fig. 14. In conjunction with the enhanced convection around the Philippines in JJA, the local SST tends to be lower than usual, especially over the SCS, indicating the upper-ocean response to the enhanced evaporation resulting from the increased surface wind speed (B. Wang et al. 2005; KN06). Rather, negative SST anomalies over the northern Indian Ocean that tend to be observed in summers after La Niña events (Fig. 17c) may trigger the PJ pattern, as suggested by Xie et al. (2009). Similar evolution of SST anomalies is found for lagged composite maps for I15°N,125°E (figure not shown).

Fig. 17.

Lagged SST anomalies (°C) for the (a),(b) preceding winter (DJF; lag of −6 months); (c),(d) concurrent summer (JJA; lag of 0 months); and (e),(f) following winter (DJF; lag of +6 months), regressed onto the (a),(c),(e) PC1 and (b),(d),(f) PC2 of horizontally smoothed vorticity anomalies over [0°–60°N, 100°–160°E] at the 850-hPa level for JJA. Contours are drawn with an interval of 0.1 (±0.05, ±0.15, ±0.25, …), with solid and dashed lines indicating positive and negative anomalies, respectively. The OLR anomaly centers for the 0-months lag are indicated with triangles in (c),(d). Light and heavy shading represent the confidence levels of 90% and 95%, respectively, based on the t statistic.

Fig. 17.

Lagged SST anomalies (°C) for the (a),(b) preceding winter (DJF; lag of −6 months); (c),(d) concurrent summer (JJA; lag of 0 months); and (e),(f) following winter (DJF; lag of +6 months), regressed onto the (a),(c),(e) PC1 and (b),(d),(f) PC2 of horizontally smoothed vorticity anomalies over [0°–60°N, 100°–160°E] at the 850-hPa level for JJA. Contours are drawn with an interval of 0.1 (±0.05, ±0.15, ±0.25, …), with solid and dashed lines indicating positive and negative anomalies, respectively. The OLR anomaly centers for the 0-months lag are indicated with triangles in (c),(d). Light and heavy shading represent the confidence levels of 90% and 95%, respectively, based on the t statistic.

Meanwhile, SST anomalies associated with EOF2 include a La Niña signal evolving from the preceding winter toward its peak time in the following winter (Figs. 17b,d,f). The La Niña signal is already evident in the concurrent summer (Fig. 17d). The enhanced convection centered at ∼20°N associated with EOF2 accompanies no significant SST anomalies locally (Fig. 17d), suggesting the importance of remote influence of the tropical SST anomalies on the anomalous convection. Further analysis is required to understand specific mechanisms that connect ENSO and our EOF2.

It is worthwhile to point out that both EOF1 and EOF2 represent signals of anomalous precipitation over Japan along the baiu/mei-yu front (Figs. 14a, 15a). In fact, the anomalous precipitation represented in EOF2 seems to be related to a particular aspect of the seasonal march related to the termination of the baiu/mei-yu season (Ueda et al. 1995). It is thus suggested that ENSO and Indian Ocean SST can remotely influence the summertime rainfall over East Asia and midlatitude WNP through atmospheric teleconnection extracted either in EOF1 or EOF2 (Alexander et al. 2004).

7. Summary and discussion

In the present study, our composite and EOF analyses of the monthly PJ pattern have shown that the associated vorticity anomalies are characterized by their zonally elongated structure with a poleward phase tilt with height, indicating eastward heat transport from the warmer Asian continent to the cooler North Pacific. The anomalies can efficiently gain KE and APE through conversion from the zonally asymmetric climatological-mean state and the diabatic generation resulting from the anomalous convective activity around the Philippines. Contributions from the dry energy conversion and moist diabatic generation are comparable, indicating dual characteristics of the PJ pattern as dry and moist dynamical modes. The efficiency, however, depends on the location of the anomalies relative to the climatological-mean field, because it maximizes if they are embedded in the mean field as observed. The conversion efficiency also maximizes for the circulation anomalies that accompany anomalous convection around the Bonin Islands. In fact, centers of vorticity anomalies associated with anomalous convection over various locations in the tropical/subtropical WNP and SCS tend to be clustered around those associated with anomalous convection in either of the two domains.

Our model analysis elucidates that the PJ pattern owes its existence to the climatological-mean field characterized by a summer monsoon system to the west, a subtropical anticyclone to the east near the surface, and an upper-level subtropical jet. The model basic flow that includes all of these features has been shown to support a least damped mode whose structure shares many features with the observed PJ pattern. The mode can efficiently be forced or maintained by tropical diabatic heating placed between the monsoon and the subtropical anticyclone. The maintenance is through the efficient energy conversions from the basic state, especially its baroclinicity. The observational and model results support our hypothesis that the PJ pattern can be regarded as a dynamical mode of the particular mean field.

Furthermore, the anomalous convective activity observed over the tropical WNP or SCS can generate APE associated with the PJ pattern as efficiently as the energy conversions from the climatological-mean state. Our diagnosis with an omega equation has indicated that anomalous vorticity and thermal advection associated with the “positive” PJ pattern can dynamically yield anomalous ascent over the region of the enhanced convection, acting to trigger or reinforce the anomalous convection. Combining with KN06’s analyses that indicate the enhanced moisture flux convergence due to the enhancement of surface evaporation and the confluent low-level mean jets, we hypothesize that the PJ pattern may also be self-sustaining through moist processes over the warm tropical WNP, where climatologically high SST and confluent low-level flows sustain abundant moisture necessary for convection. It should be noted that Lu and Lin (2009) suggested a possible role of anomalous diabatic heating over the baiu/mei-yu rainband in forcing and/or maintaining the PJ pattern. In summary, the PJ pattern may be a moist dynamical mode in which anomalous convective activity in the tropics can reinforce anomalous circulation that can be regarded as a dry dynamical mode, whereas the anomalous circulation can in turn reinforce anomalous precipitation in the tropics and along the baiu/mei-yu front. Key features of the mechanisms for the PJ patterns are summarized schematically in Fig. 11 of KN06.

The characteristic of the PJ pattern as a (moist) dynamical mode suggests that the pattern can be triggered by certain processes other than tropical convection. For example, an upper-tropospheric wave packet propagating from upstream along the Asian jet, a teleconnection called the Silk Road pattern, may enhance anomalous ascent around Philippines to trigger anomalous convection in the tropics, as can be justified by dominant contribution from upper-tropospheric vorticity advection by the mean meridional flow to induce the anomalous ascent (Fig. 12b; Kosaka et al. 2009). In fact, our composite map for the PJ pattern includes a signature of the Silk Road pattern that accompanies a wave-activity flux pointing southeastward in the jet exit (Fig. 2c). Additional analysis on daily evolution of the Silk Road and PJ patterns is needed to verify the possibility for the former pattern to trigger the latter.

For summertime seasonal forecast for the Far East, SST anomalies over the tropical WNP and ENSO-related indices have been used, because they may possibly influence the atmospheric circulation over East Asia through the PJ teleconnection, as postulated by Nitta (1987). In addition to the La Niña signal in the preceding winters, KN06 has found weak but significant warm SST anomalies to the east of the Philippines in the preceding month of a “positive” PJ event with enhanced convective anomaly around the Philippines, whose persistent signal can be inferred from Fig. 17c. The possibility that the PJ pattern is a (moist) dynamical mode, however, means that its behavior is not controlled completely by underlying SST anomalies. We have calculated correlation coefficients of monthly OLR anomalies for the JJA period averaged over [10°–20°N, 120°–130°E] with monthly SST anomalies averaged over [10°–20°N, 140°–160°E] for the preceding months [May–July (MJJ)] and over the Niño-3.4 region [5°S–5°N, 170°–120°W] 6 months earlier [December–February (DJF)]. The coefficients are −0.33 and +0.37, respectively, which exceed the 90% confidence level, to confirm their association. Consistent with the behavior of the PJ pattern as a (moist) dynamical mode, however, the SST anomalies can explain only 14% or less of the total variance of the OLR anomaly around the OLR anomaly center for the PJ pattern, which is certainly too low for practical forecasting.

Nevertheless, there still remains room to study the predictability of the seasonal PJ pattern based on SST anomalies on seasonal time scales, because many studies concerning the PJ pattern focus on seasonal-mean anomalies. Among them, Lu et al. (2006), using ensemble integrations by a general circulation model, extracted the leading mode of external variability that exhibits some common features with the PJ pattern. Xie et al. (2009) have proposed a possible mechanism on how ENSO affects the East Asian climate in the following summer via Indian Ocean SST anomalies and their possible triggering of a PJ-like anomaly pattern. In fact, our additional analysis using seasonal-mean data has revealed that the JJA-mean PJ pattern can be influenced more strongly by ENSO on seasonal time scales. Though much smaller in magnitude than the monthly anomalies, JJA-mean OLR anomalies over [10°–20°N, 120°–130°E] have a correlation coefficient of +0.60 with the Niño-3.4 SST for the preceding winter (DJF). The correlation coefficient is much higher than its counterpart for the monthly anomalies, and it exceeds the 99% confidence level. This time-scale dependence may be attributable to the intraseasonal nature of the PJ pattern, whose modal characteristic is more evident in its month-to-month variability.

Although the characteristics of the PJ pattern as a dry dynamical mode have been confirmed through our model analysis, our hypothesis that the pattern can also be regarded as a moist dynamical mode certainly requires further verification, for example, through analysis with (linear) dynamical models that include moist processes. Additionally, there is a need for further analyses of the PJ pattern on a wider frequency domain from daily to interannual time scales and its possible modulations under the warmed climate in future, to further confirm the relevance of our interpretation on its dynamics based on monthly anomalies and to thereby deepen our understanding of its mechanism and potential predictability.

Acknowledgments

The authors appreciate valuable comments and suggestions from Dr. R. Lu and other two anonymous reviewers. The authors also wish to acknowledge comments from Profs. M. Kimoto, H. Niino, K. Sato, and T. Yamagata (University of Tokyo). This study is supported in part by the Grant-in-Aid 18204044 and 22340135 by the Japanese Ministry of Education, Culture, Sports, Science and Technology and also by the Global Environment Research Fund (S-5) of the Ministry of the Environment, Japan. The JRA-25 reanalysis dataset used for this study is provided from the cooperative research project of the JRA-25 long-term reanalysis by the Japan Meteorological Agency (JMA) and the Central Research Institute of Electric Power Industry (CRIEPI). NOAA/OLR and optimum interpolation SST and CMAP precipitation datasets are provided by the NOAA/Cooperative Institute for Research in Environmental Sciences (CIRES) Climate Diagnostics Center, Boulder, Colorado, from their Web site (available online at http://www.cdc.noaa.gov/).

REFERENCES

REFERENCES
Alexander
,
M. A.
,
N-C.
Lau
, and
J. D.
Scott
,
2004
:
Broadening the atmospheric bridge paradigm: ENSO teleconnections to the tropical West Pacific-Indian Oceans over the seasonal cycle and to the North Pacific in summer.
Earth’s Climate: The Ocean-Atmosphere Interaction, Geophys. Monogr., Vol. 147, Amer. Geophys. Union, 85–103
.
Emanuel
,
K. A.
,
1987
:
An air-sea interaction model of intraseasonal oscillations in the tropics.
J. Atmos. Sci.
,
44
,
2324
2340
.
Enomoto
,
T.
,
2004
:
Interannual variability of the Bonin high associated with the propagation of Rossby waves along the Asian jet.
J. Meteor. Soc. Japan
,
82
,
1019
1034
.
Enomoto
,
T.
,
B. J.
Hoskins
, and
Y.
Matsuda
,
2003
:
The formation mechanism of the Bonin high in August.
Quart. J. Roy. Meteor. Soc.
,
129
,
157
178
.
Fukutomi
,
Y.
, and
T.
Yasunari
,
2002
:
Tropical-extratropical interaction associated with the 10–25-day oscillation over the western Pacific during the northern summer.
J. Meteor. Soc. Japan
,
80
,
311
331
.
Grimm
,
A. M.
, and
P. L.
Silva Dias
,
1995
:
Analysis of tropical–extratropical interactions with influence functions of a barotropic model.
J. Atmos. Sci.
,
52
,
3538
3555
.
Hoskins
,
B. J.
,
1980
:
Representation of the earth topography using spherical harmonics.
Mon. Wea. Rev.
,
108
,
111
115
.
Hoskins
,
B. J.
, and
D. J.
Karoly
,
1981
:
The steady linear response of a spherical atmosphere to thermal and orographic forcing.
J. Atmos. Sci.
,
38
,
1179
1196
.
Hoskins
,
B. J.
,
I. N.
James
, and
G. H.
White
,
1983
:
The shape, propagation and mean-flow interaction of large-scale weather systems.
J. Atmos. Sci.
,
40
,
1595
1612
.
Hsu
,
H-H.
, and
S-M.
Lin
,
2007
:
Asymmetry of the tripole rainfall pattern during the East Asian summer.
J. Climate
,
20
,
4443
4458
.
Kasahara
,
A.
, and
P. L.
Silva Dias
,
1986
:
Response of planetary waves to stationary tropical heating in a global atmosphere with meridional and vertical shear.
J. Atmos. Sci.
,
43
,
1893
1911
.
Kato
,
T.
, and
Y.
Matsuda
,
1992
:
External mode induced by tropical heating in the basic flow with vertical shear and its propagation.
J. Meteor. Soc. Japan
,
70
,
1057
1070
.
Kawamura
,
R.
, and
T.
Ogasawara
,
2006
:
On the role of typhoons in generating PJ teleconnection patterns over the western North Pacific in late summer.
SOLA
,
2
,
37
40
.
Kawamura
,
R.
,
T.
Murakami
, and
B.
Wang
,
1996
:
Tropical and mid-latitude 45-day perturbations over the western Pacific during the northern summer.
J. Meteor. Soc. Japan
,
74
,
867
890
.
Kim
,
J-E.
,
S-W.
Yeh
, and
S-Y.
Hong
,
2009
:
Two types of strong northeast Asian summer monsoon.
J. Climate
,
22
,
4406
4417
.
Kosaka
,
Y.
, and
H.
Nakamura
,
2006
:
Structure and dynamics of the summertime Pacific- Japan teleconnection pattern.
Quart. J. Roy. Meteor. Soc.
,
132
,
2009
2030
.
Kosaka
,
Y.
, and
H.
Nakamura
,
2008
:
A comparative study on the dynamics of the Pacific- Japan (PJ) teleconnection pattern based on reanalysis datasets.
SOLA
,
4
,
9
12
.
Kosaka
,
Y.
, and
H.
Nakamura
,
2010
:
Mechanisms of meridional teleconnection observed between a summer monsoon system and a subtropical anticyclone. Part II: A global survey.
J. Climate
,
23
,
5109
5125
.
Kosaka
,
Y.
,
H.
Nakamura
,
M.
Watanabe
, and
M.
Kimoto
,
2009
:
Analysis on the dynamics of a wave-like teleconnection pattern along the summertime Asian jet based on a reanalysis dataset and climate model simulations.
J. Meteor. Soc. Japan
,
87
,
561
580
.
Kurihara
,
K.
, and
T.
Tsuyuki
,
1987
:
Development of the barotropic high around Japan and its association with Rossby wave-like propagations over the North Pacific: Analysis of August 1984.
J. Meteor. Soc. Japan
,
65
,
237
246
.
Lau
,
K. M.
, and
L.
Peng
,
1992
:
Dynamics of atmospheric teleconnection during the northern summer.
J. Climate
,
5
,
140
158
.
Lim
,
H.
, and
C-P.
Chang
,
1986
:
Generation of internal- and external-mode motions from internal heating: Effects of vertical shear and damping.
J. Atmos. Sci.
,
43
,
948
957
.
Lu
,
R.
,
2004
:
Associations among the components of the East Asian summer monsoon system in the meridional direction.
J. Meteor. Soc. Japan
,
82
,
155
165
.
Lu
,
R.
, and
Z.
Lin
,
2009
:
Role of subtropical precipitation anomalies in maintaining the summertime meridional teleconnection over the western North Pacific and East Asia.
J. Climate
,
22
,
2058
2072
.
Lu
,
R.
,
Y.
Li
, and
B.
Dong
,
2006
:
External and internal summer atmospheric variability in the western North Pacific and East Asia.
J. Meteor. Soc. Japan
,
84
,
447
462
.
Nakamura
,
H.
, and
T.
Fukamachi
,
2004
:
Evolution and dynamics of summertime blocking over the Far East and the associated surface Okhotsk high.
Quart. J. Roy. Meteor. Soc.
,
130
,
1213
1233
.
Navarra
,
A.
,
1993
:
A new set of orthonormal modes for linearized meteorological problems.
J. Atmos. Sci.
,
50
,
2569
2583
.
Neelin
,
J. D.
,
I. M.
Held
, and
K. H.
Cook
,
1987
:
Evaporation-wind feedback and low- frequency variability in the tropical atmosphere.
J. Atmos. Sci.
,
44
,
2341
2348
.
Nitta
,
T.
,
1987
:
Convective activities in the tropical western Pacific and their impact on the Northern Hemisphere summer circulation.
J. Meteor. Soc. Japan
,
65
,
373
390
.
North
,
G. R.
,
T. L.
Bell
,
R. F.
Cahalan
, and
F. J.
Moeng
,
1982
:
Sampling errors in the estimation of empirical orthogonal functions.
Mon. Wea. Rev.
,
110
,
699
706
.
Onogi
,
K.
, and
Coauthors
,
2007
:
The JRA-25 reanalysis.
J. Meteor. Soc. Japan
,
85
,
369
432
.
Simmons
,
A. J.
,
J. M.
Wallace
, and
G. W.
Branstator
,
1983
:
Barotropic wave propagation and instability, and atmospheric teleconnection patterns.
J. Atmos. Sci.
,
40
,
1363
1392
.
Sui
,
C-H.
,
P-H.
Chung
, and
T.
Li
,
2007
:
Interannual and interdecadal variability of the summertime western North Pacific subtropical high.
Geophys. Res. Lett.
,
34
,
L11701
.
doi:10.1029/2006GL029204
.
Takaya
,
K.
, and
H.
Nakamura
,
2001
:
A formulation of a phase-independent wave-activity flux for stationary and migratory quasigeostrophic eddies on a zonally varying basic flow.
J. Atmos. Sci.
,
58
,
608
627
.
Tompkins
,
A. M.
,
2001
:
Organization of tropical convection in low vertical wind shears: The role of water vapor.
J. Atmos. Sci.
,
58
,
529
545
.
Tsuyuki
,
T.
, and
K.
Kurihara
,
1989
:
Impact of convective activity in the western tropical Pacific on the East Asian summer circulation.
J. Meteor. Soc. Japan
,
67
,
231
247
.
Ueda
,
H.
,
T.
Yasunari
, and
R.
Kawamura
,
1995
:
Abrupt seasonal change of large-scale convective activity over the western Pacific in the northern summer.
J. Meteor. Soc. Japan
,
73
,
795
809
.
Wakabayashi
,
S.
, and
R.
Kawamura
,
2004
:
Extraction of major teleconnection patterns possibly associated with the anomalous summer climate in Japan.
J. Meteor. Soc. Japan
,
82
,
1577
1588
.
Wang
,
B.
, and
X.
Xie
,
1996
:
Low-frequency equatorial waves in vertically sheared zonal flow. Part I: Stable waves.
J. Atmos. Sci.
,
53
,
449
467
.
Wang
,
B.
,
Q.
Ding
,
X.
Fu
,
I-S.
Kang
,
K.
Jin
,
J.
Shukla
, and
F.
Doblas-Reyes
,
2005
:
Fundamental challenge in simulation and prediction of summer monsoon rainfall.
Geophys. Res. Lett.
,
32
,
L15711
.
doi:10.1029/2005GL022734
.
Wang
,
Z.
,
C-P.
Chang
,
B.
Wang
, and
F-F.
Jin
,
2005
:
Teleconnections from tropics to northern extratropics through a southerly conveyor.
J. Atmos. Sci.
,
62
,
4057
4070
.
Xie
,
P.
, and
P. A.
Arkin
,
1997
:
Global precipitation: A 17-year monthly analysis based on gauge observations, satellite estimates, and numerical model outputs.
Bull. Amer. Meteor. Soc.
,
78
,
2539
2558
.
Xie
,
S-P.
,
K.
Hu
,
J.
Hafner
,
H.
Tokinaga
,
Y.
Du
,
G.
Huang
, and
T.
Sampe
,
2009
:
Indian Ocean capacitor effect on Indo–western Pacific climate during the summer following El Niño.
J. Climate
,
22
,
730
747
.
Yamada
,
K.
, and
R.
Kawamura
,
2007
:
Dynamical link between typhoon activity and the PJ teleconnection pattern from early summer to autumn as revealed by the JRA-25 reanalysis.
SOLA
,
3
,
65
68
.
Yasunaka
,
S.
, and
K.
Hanawa
,
2006
:
Interannual summer temperature variations over Japan and their relation to large-scale atmospheric circulation field.
J. Meteor. Soc. Japan
,
84
,
641
652
.
Yasutomi
,
N.
,
2003
:
Detection and dynamics of principal modes of Asian summer monsoon variability. Ph.D. thesis, University of Tokyo, 100 pp
.

Footnotes

Corresponding author address: Yu Kosaka, International Pacific Research Center, School of Ocean and Earth Science and Technology, University of Hawaii, 1680 East-West Road, Honolulu, HI 96822. Email: ykosaka@hawaii.edu

1

In KN06, the sign of the term that represents the contribution from uT ′ was incorrect, although the actual evaluation was performed correctly.