Responses of the East Asian Jet Stream to the North Pacific Subtropical Front in Spring

Studies using high-resolution satellite observations have revealed that midlatitude SST fronts exert a significant influence on the atmosphere above on both meso and micro timescales (Xie, 2004; Xu et al., 2008, 2010a). These midlatitude SST fronts affect sea surface winds and the atmospheric boundary layer via the SLP adjustment mechanism (Lindzen and Nigam, 1987; Small et al., 2003) and vertical mixing mechanism (Hayes et al., 1989; Wallace et al., 1989).

In addition to the atmospheric responses in the atmospheric boundary layer, midlatitude SST fronts also have marked impacts on the whole troposphere (Brayshaw et al., 2008; Nakamura and Miyama, 2014), through their effects on baroclinicity (Nakamura and Yamane, 2010; O'Reilly and Czaja, 2015) and convective activities (Kobashi et al., 2008). (Sampe et al., 2010) found that an SST front over the midlatitudes is conducive to the formation of the jet stream and storm tracks, and its displacement can shift the jet stream and storm tracks in the same direction (Ogawa et al., 2012; Nakamura and Miyama, 2014). Additionally, an SST front over the midlatitudes corresponds to a convective rainfall maximum (Minobe et al., 2008) and frequent lightning activities (Tokinaga et al., 2009)——a finding confirmed by numerical simulations (Kuwano-Yoshida et al., 2010; Xu et al., 2011).

Many studies have demonstrated the significant influences of SST fronts on the atmosphere over the western North Pacific, especially over the East China Sea (Xie et al., 2002; Xu et al., 2010b; Xu and Xu, 2015), the Kuroshio extension, and the Oyashio extension (Tanimoto et al., 2009; Kwon et al., 2010). In addition to these midlatitude SST fronts mentioned above, a maximum SST gradient exists over the Northwest Pacific around 25^°N in spring; namely, the North Pacific subtropical front (NPSTF; Fig. 1). Although many efforts have been made to elucidate the formation mechanism of the NPSTF in terms of intrinsic ocean dynamics (Yoshida and Kidokoro, 1967a, 1967b; Roden, 1975; Kubokawa, 1995, 1997, 1999; Kobashi and Kubokawa, 2012), little is known about its synoptic and climatic impacts on the atmosphere.

(Xie, 2004) suggested that the air-sea coupling effect is robust around the SST front over the world's oceans. Since the NPSTF is located farther south and has a higher SST with respect to the midlatitude SST front, it is likely that the NPSTF can significantly affect large-scale atmospheric circulation by changing atmospheric baroclinicity and convective activities. In fact, (Kobashi et al., 2008) demonstrated that the synoptic low above the SST front can be enhanced during April-May, due to increased atmospheric baroclinicity and condensation latent heat release caused by the NPSTF. However, little is known about the impacts of the NPSTF on the free atmosphere on the interannual timescale.

Figure 1.  Climatological spring SST (contours; units: ^°C) and SST gradient [color shading; units: ^°C (100 km)^-1] during 1982-2014: (a) AVHRR; (b) ERA-Interim; (c) HadISST. The amplitude of the SST gradient is measured by \(\sqrt{(\partial SST/\partial x)^2 +(\partial SST/\partial y)^2}\).

As is well known, the East Asian jet stream (EAJS) is an important circulation system over the North Pacific. The variability of the intensity and location of the EAJS also exerts a significant influence on the weather and climate over East Asia (Liang and Wang, 1998; Zhang et al., 2008; Lu et al., 2013). (Ye et al., 1958) suggested that the seasonal variability of the EAJS produces two significant abrupt changes in the atmospheric circulation over East Asia during June and December, leading to subsequent anomalous weather and climate. (Xie et al., 2015) found that the subtropical westerly jet stream exhibits a significant zonal pattern associated with its seasonal evolution, which affects both large-scale atmospheric circulation and precipitation over East Asia and the North Pacific. Moreover, numerous studies have shown that the EAJS can act as a waveguide and transmit the signals to remote regions, inducing anomalous climatic responses over these regions, e.g., the Arctic Oscillation (Gong and Ho, 2003), the North Atlantic Oscillation (Watanabe, 2004) and the Eurasian teleconnection (Wang and Zhang, 2015).

Therefore, exploring the linkage between the NPSTF and the EAJS should improve our understanding of the interactions among the circulation systems over East Asia and the western Pacific. Here, we focus on the responses of the EAJS to the intensity of the NPSTF during boreal spring, when the NPSTF reaches its maximum intensity.

The rest of the paper is organized as follows: We introduce the reanalysis datasets and methods used in this study in section 2. The statistical relationship between the intensity of the NPSTF and the EAJS is presented in section 3. We discuss the possible processes by which the NPSTF affects the EAJS in section 4. Numerical experiments are used to confirm the relationship between the NPSTF and the EAJS in section 5. A summary and discussion are given in section 6.

We use three SST datasets to identify the spatial and temporal characteristics of the NPSTF: HadISST, which provides monthly-mean SST at a resolution of 1^°× 1^° (Rayner et al., 2003); monthly-mean ERA-Interim SST at a resolution of 0.5^°× 0.5^° (Dee et al., 2011); and daily-mean AVHRR SST, from the NCDC of NOAA, at a resolution of 0.25^°× 0.25^° (Reynolds and Chelton, 2010). The NPSTF index is defined as the normalized anomalies of the amplitude of SST gradient over its maximum region.

The atmospheric data used in this study, expect for precipitation, are the 6-hourly ERA-Interim data, at a horizontal resolution of 1.5^°× 1.5^°, for wind, air temperature, geopotential height, and SLP. The 6-hourly fields are averaged to obtain daily-mean fields. In order to evaluate the high-frequency transient eddy feedback forcing, an 8-day high-pass filter is applied to these daily fields. Precipitation data are the monthly-mean data of CMAP, at a resolution of 2.5^°× 2.5^°. For consistency, all variables used in this study cover the period from January 1982 to December 2014.

Since the NPSTF index has a significant increasing tendency [Fig. S1 in electronic supplementary material (ESM)], the long-term linear trends are removed for all variables before further calculations. In order to obtain the interannual variability, we also perform Fourier analysis to remove the first three waves, which are usually related to interdecadal variability (Awan and Bae, 2016).

Baroclinicity is generally measured by a combination of static stability and horizontal temperature gradient. The latter is equivalent to vertical shear of the horizontal wind according to the thermal wind balance (Charney, 1947; Eady, 1949).

(Lindzen and Farrell, 1980) and (James, 1987) analyzed the baroclinic situation in the background of different air flows and showed that Eady's parameter could successfully estimate the maximum growth rate of baroclinic instability in the troposphere. We use Eady's parameter introduced by (Hoskins and Valdes, 1990) as the baroclinicity index, shown in Eq. (1): \begin{equation} \sigma_{\rm BI}=0.31gN^{-1}T^{-1}|\partial T/\partial y|=0.31f|\partial {V}/\partial z|N^{-1} ,(1) \end{equation} where N and f represent the static stability parameter and the Coriolis parameter, respectively; T is air temperature; z is height; and V is horizontal velocity. The larger the baroclinicity index value, the easier it is for synoptic activities to develop (Simmons and Hoskins, 1978). Ignoring the horizontal shears in the lower troposphere and moist processes, this index has been demonstrated to be a good indicator of atmospheric baroclinicity (Nakamura and Sampe, 2002; Nakamura and Yamane, 2010).

The atmospheric apparent heat source Q₁ consists of the heating due to radiation, the release of latent heat by net condensation, and the vertical convergence of the vertical eddy transport of sensible heat (Yanai et al., 1973; Ding, 1989). The method used here was first introduced in (Ding, 1989): \begin{equation} Q_1=C_p\left(\dfrac{p}{p_0}\right)^{R/C_p}\left(\dfrac{\partial\theta}{\partial t}+{V}\nabla\theta+\omega\dfrac{\partial\theta}{\partial p}\right) , (2)\end{equation} where p₀=1000 hPa, R is the gas constant, C_p is the specific heat capacity of dry air, p is pressure, θ is potential temperature, and ω is pressure vertical velocity. The vertical integration of the atmospheric heat source measures the sensible and latent heat fluxes from the surface into the atmosphere. The positive and negative values indicate the net heat gain and loss of the atmosphere, respectively.

TEFF denotes a low-frequency geopotential tendency due to the divergence or convergence of high-frequency transient eddy heat flux and that of vorticity flux. TEFF is vigorous, especially in the middle and high latitudes (Lau and Holopainen, 1984; Lau and Nath, 1991; Shi, 2013).

The feedback forcing can be represented by the geopotential tendency equation (Lau and Holopainen, 1984; Holopainen and Fortelius, 1987): \begin{equation} \left\{ \begin{array}{l} g\left [\dfrac{1}{f}\nabla^2+f\dfrac{\partial}{\partial p}\left(\dfrac{1}{\sigma}\dfrac{\partial}{\partial p}\right)\right ]\dfrac{\partial Z}{\partial t}=D_{\rm heat}+D_{\rm vort}+R\\[3mm] D_{\rm heat}=f\dfrac{\partial}{\partial p}\left(\dfrac{\nabla\cdot\overline{{V}'\theta'}}{\overline{S}}\right)\\[3mm] D_{\rm vort}=-\nabla\cdot\overline{{V}'\varsigma'} \end{array} \right. . (3)\end{equation} The boundary conditions at 1000 hPa and 100 hPa are \begin{equation} \left\{ \begin{array}{l} -\dfrac{gp}{R}\left(\dfrac{p_0}{p}\right)^{R/C_p}\dfrac{\partial}{\partial p}\left(\dfrac{\partial Z}{\partial t}\right)_{\rm heat}=-\nabla\cdot\overline{{V}'\theta'}\\[3mm] \dfrac{\partial}{\partial p}\left(\dfrac{\partial Z}{\partial t}\right)_{\rm vort}=0 \end{array} \right. , (4)\end{equation} where the prime and overbar represent the variables with high frequency and low frequency, respectively. In Eq. (3), Z is height; σ=-(α/θ)(∂θ/∂ p) is the static stability parameter, assumed to be a function of pressure only; α is the specific volume, V' is high frequency horizontal velocity; \(\overline{S}\) is the hemispheric mean of the quantity \(-\partial\overline\theta/\partial p\); and f is the Coriolis parameter at 43^°N (1.0× 10^-4 s^-1), to simplify the calculation (Holopainen and Fortelius, 1987). The term R in Eq. (3) represents all remaining components in the quasi-geostrophic potential vorticity balance, such as horizontal advection by the time-mean flow, diabatic effects and friction. The subscripts "heat" and "vort" indicate feedback forcing of the high-frequency transient eddy heat flux and vorticity flux, respectively. Other variables are commonly used meteorological variables. Following (Holopainen and Fortelius, 1987) and (Shi, 2013), the two-dimensional Laplace operator in Eq. (3) is calculated using a spherical harmonics function with T21 truncated.

Figure 2 shows the March-April-May mean SST gradient and its variance using three different SST datasets. The three datasets consistently show a pronounced large SST gradient between 20^°N and 30^°N, extending northeastward from eastern Taiwan to the central North Pacific, with a maximum up to 0.8^°C (100 km)^-1. A maximum variance is also located over this region of large SST gradient. Accordingly, the subtropical parallelogram region (22.5^°-33^°N, 139.5^°E-190.5^°W) is defined as the key region of the NPSTF (red box in Fig. 2) in this study. The NPSTF index is then defined as the normalized anomalies of the amplitude of the SST gradient over the key region. A higher (lower) NPSTF index indicates a larger (smaller) amplitude of the SST gradient over the key region and a stronger (weaker) NPSTF. Since the index values based on the three datasets are highly correlated (their correlation coefficients exceed the 99% confidence level; Fig. 3), we use the HadISST dataset in the rest of the paper and check the results using the other two datasets, which do not change qualitatively.

Figure 4 shows the regression coefficient of wind speed at 200 hPa onto the NPSTF index in spring. The NPSTF index shows a significant positive relationship with the 200 hPa wind speed around the center of the climatological EAJS in spring, which extends northeastward from East China to Northwest America via the central Pacific. This positive correlation suggests that the EAJS is significantly enhanced when a strong NPSTF is present. Figure 5 shows the regression coefficients of geopotential height and horizontal wind upon the NPSTF index. It can be clearly seen that when the NPSTF is developed, the related anomalous geopotential height exhibits an equivalent barotropic structure throughout the troposphere, characterized by a negative geopotential height anomaly over the central North Pacific to the north of the NPSTF (Kobashi et al., 2008) and two positive geopotential height anomalies over Northeast Asia and the central Pacific, respectively. Correspondingly, a significant southwesterly wind anomaly appears over the EAJS core around 30^°N, while a northeasterly wind anomaly is found south of the Bering Sea, which is consistent with the changes in the 200 hPa wind (Fig. 4).

Figure 2.  Climatological spring SST gradient [contours; only 0.7 and 0.8^°C (100 km)^-1 shown] and its variance (color shading). The red box indicates the key area of the NPSTF.

Figure 3.  The NPSTF index for the period 1982-2014 using three datasets.

Figure 4.  Regression coefficients of 200 hPa wind speed (color shading; units: m s^-1) onto the NPSTF index in spring. Contours indicate the climatological spring wind speed. Stippled areas represent statistically significant coefficients exceeding the 95% confidence level.

Figure 6a shows the vertical structure of the regression coefficient of air temperature onto the NPSTF index in spring. A significant anomalous atmospheric warming appears to the south of the front near 25^°N, while the opposite is found to the north of the front, resulting in an enhanced atmospheric temperature gradient in the whole troposphere near 30^°N (Fig. S2 in ESM). According to Eq. (1), an increase in horizontal temperature gradient can cause a pronounced increase in the atmospheric baroclinicity index (Fig. 6b), which occurs in the low levels (1000 hPa) over the front, near 20^°-30^°N, and extends up to the upper troposphere (∼250 hPa) to the north of the NPSTF. However, the atmospheric baroclinicity is markedly suppressed over the tropics (south of 20^°N) and over the middle and high latitudes (north of 45^°N).

Increased atmospheric baroclinicity usually induces enhanced synoptic-scale eddy activities (Simmons and Hoskins, 1978). In this study, the intensity of synoptic-scale transient eddy activities is measured by 850 hPa meridional eddy heat flux (v'T') (Blackmon et al., 1977). Figure 7 shows the regression coefficient of 850 hPa meridional eddy heat flux onto the NPSTF index in spring. We can see that changes in the meridional eddy heat flux correspond well to those in the atmospheric baroclinicity index. The enhanced transient eddy activities appear over the NPSTF and to its north, where the atmospheric baroclinicity index is relatively large (Fig. 6b), while the opposite is seen over the middle and high latitudes. (Lau and Holopainen, 1984) used the framework of a quasi-geostrophic equation to examine the transient eddy forcing effect on time-mean flow. From the solutions of the quasi-geostrophic equation [Eq. (3)] for a simplified system, they found that enhanced transient eddy activities can trigger an anomalous cyclonic circulation to the north and an anomalous anticyclonic circulation to the south via the anomalous convergence of transient vorticity flux and heat flux (Lau and Holopainen, 1984; Shi, 2013), which was confirmed by numerical simulations (Lau and Holopainen, 1984; Watanabe and Kimoto, 2000). Similarly, as shown in Fig. 8, a significant cyclonic height anomaly appears to the north of the enhanced transient eddy activities and displays a significant equivalent barotropic structure, with its center over the central Pacific and titled eastward with height. Thus, the related westerly wind anomalies over ∼30^°N benefit an intensified EAJS over the Northeast Pacific (east of ∼160^°E). Note that TEFF denotes a low-frequency geopotential tendency due to transient eddies. Although the amplitude of TEFF may be small for a single day, it cannot be ignored for a whole season (Shi, 2013).

Figure 5.  Regression coefficients of geopotential height (contours; units: m² s^-2) and horizontal wind (vectors; units: m s^-1) onto the NPSTF index in spring: (a) 200 hPa; (b) 500 hPa; (c) 850 hPa. Contour intervals at the three pressure levels are 50 and 100 m² s^-2 for positive and negative values, respectively. Light and heavy shading represents statistically significant coefficients exceeding the 95% and 99% confidence levels, respectively. Only the wind component exceeding the 95% confidence level in either direction is shown. The box indicates the key area of the NPSTF.

Figure 6.  Latitude-height section of regression coefficients of (a) air temperature (units: ^°C) and (b) atmospheric baroclinicity index (units: 10^-2 d^-1), zonally averaged over 139.5^°E-190.5^°W, onto the NPSTF index in spring. Light and heavy shading represents statistically significant coefficients exceeding the 95% and 99% confidence levels, respectively.

On the other hand, significantly suppressed transient eddy activities appear south of the Bering Sea (Fig. 7). This may be due to the weakened EAJS caused by the anomalous easterlies north of the cyclonic height anomalies (Ren and Zhang, 2006; Ren et al., 2010). In this scenario, the NPSTF may extend its influence even to mid- and high-latitude regions through anomalous transient eddy activities. As shown in Fig. 8, the anomalous anticyclonic circulation over Northeast Asia is enhanced by TEFF (Fig. 7). Moreover, the anticyclonic height anomalies can contribute to the maintenance of cyclonic height anomalies and further benefits the enhancement of the EAJS over the Northeast Pacific.

Figure 7.  Regression coefficients (contours) of 850 hPa meridional eddy heat flux [v’T’; units: ^°C (m s^-1)] onto the NPSTF index in spring. Light and heavy shading represents statistically significant coefficients exceeding the 95% and 99% confidence levels, respectively. The box indicates the key area of the NPSTF.

As mentioned above, a stronger NPSTF can excite an anomalous cyclonic circulation throughout the troposphere by enhancing the transient eddy activities over the subtropical region; the related westerly wind anomaly on the south side of this anomalous cyclone can enhance the EAJS over the Northeast Pacific (Fig. 4). On the other hand, the enhanced transient activities over the NPSTF act to weaken the transient activities in the middle and high latitudes, which in turn is favorable for the maintenance of the anomalous cyclonic circulation through the TEFF. Therefore, this coupled south-north response of transient eddy activities confirms that the NPSTF can influence the EAJS via changing baroclinicity and transient eddy activities in the free atmosphere. Note that the vertical mixing mechanism is important in the atmospheric adjustment to midlatitude SST fronts in which an increase (decrease) in SST enhances (reduces) the baroclinicity of the near-surface atmosphere and accelerates (decelerates) the surface wind (Wallace et al., 1989; Xie et al., 2002). In our results, the enhanced baroclinicity related to the NPSTF is located over the front in the lower troposphere and over the cold SST side in the upper troposphere. This is consistent with (Xie, 2004) and (Kobashi et al., 2008), who pointed out that the response of baroclinicity to the NPSTF appears different from the prevailing wind adjustment to SST fronts elsewhere, and suggested that some other mechanisms are responsible for the cyclonic circulation along the STF.

Figure 8.  Regression coefficients (contours) of low-frequency geopotential height due to TEFF (contour interval: 3.0 m² s^-2 d^-1) onto the NPSTF index in spring: (a) 200 hPa; (b) 500 hPa; (c) 850 hPa. Light and heavy shading represents statistically significant coefficients exceeding the 95% and 99% confidence levels, respectively. The box indicates the key area of the NPSTF.

Figure 9.  As in Fig. 7, except for precipitation (units: mm d^-1) in spring.

Figure 10.  The (a) latitude-height section of the regression coefficients of the atmospheric heat source (units: W m^-2) onto the NPSTF index in spring, and (b) its vertical profile and vertical difference against pressure. Regression coefficients are zonally averaged over 139.5^°E-190.5^°W in (a) and area-averaged over (24.5^°-35.5^°N, 139.5^°E-90.5^°W) in (b). The thick dashed box in (a) indicates the key area of the atmospheric heat source. Light and heavy shading in (a) represents statistically significant coefficients exceeding the 95% and 99% confidence levels, respectively.

(Ding and Liu, 2001) suggested that the intensified horizontal air temperature gradient and atmospheric baroclinicity in a cold surge can provide disturbances or mesoscale systems in the frontal zone and further lead to enhanced convective activities. Figure 9 shows the regression coefficient of precipitation onto the NPSTF index in spring. Precipitation is increased near the strong NPSTF, and the associated condensation latent heating should favor a strong atmospheric heat source. As shown in Fig. 10a, corresponding to the strong NPSTF, a significantly enhanced atmospheric heat source emerges north of the NPSTF throughout the whole troposphere. According to scale analysis and solely taking latent heat into account, (Liu et al., 1999) and (Wu et al., 1999) simplified the complete form of the vertical vorticity equation to β v∝ ∂ Q/∂ z∝-∂ Q/∂ p where v is meridional velocity. Therefore, the atmospheric heat source can trigger anomalous southerly wind (v>0) at the bottom of the maximum heat source (∂ Q/∂ p<0) and anomalous northerly wind (v<0) at the top of the maximum heat source (∂ Q/∂ p>0). Figure 10b displays the vertical profile of the regression coefficient of atmospheric heat source onto the NPSTF index in spring, together with its vertical difference against pressure. We can clearly see that the anomalous maximum center of the heat source is located at 200 hPa and ∂ Q/∂ p is almost negative in the troposphere. Therefore, an anomalous heat source associated with the stronger NPSTF index can excite negative ∂ Q/∂ p in the troposphere. According to the theory proposed by previous studies (Liu et al., 1999; Wu et al., 1999), negative ∂ Q/∂ p can favor the formation of an anomalous cyclonic circulation to its west and an anomalous anticyclonic circulation to its east due to the anomalous southerly wind. To confirm the above results, we define an atmospheric heat source index by vertically integrating the atmospheric heat source from 850 to 150 hPa over 24.5^°-35.5^°N (box in Fig. 10a). Figure 11 shows the regression coefficients of geopotential height and wind onto the heating index. At different pressure levels, an anomalous cyclonic circulation consistently occurs over the North Pacific and anticyclonic circulation anomalies are found over Northeast Asia and the central Pacific, respectively, which bear a close resemblance with those in Fig. 5 and are consistent with the findings of (Liu et al., 1999). Compared to the responses of TEFF (Fig. 8), the main difference is the farther westward location of the anomalous cyclonic circulation. Accordingly, the related significant westerly wind anomalies in the upper troposphere near 30^°N act to intensify the component of the EAJS over the Northwest Pacific (west of ∼180^°). Therefore, increased precipitation due to a strong NPSTF leads to an enhanced atmospheric heat source via condensation latent heating, which may excite an anomalous cyclonic circulation in the upper levels to the west of the source region and subsequently intensifies the EAJS over the Northwest Pacific.

Figure 11.  As in Fig. 5, except for regression coefficients onto the atmospheric heat source index.

Figure 12.  The (a) observed and (b) CTRL-simulated wind speed (units: m s^-1) at 200 hPa in spring.

In light of the above analysis, a strong NPSTF in spring could significantly enhance the 200 hPa wind via enhanced transient eddy activity and an enhanced atmospheric heat source. In this section, five numerical experiments based on CAM5.1 are used to verify the above conclusions.

CAM5.1 has a horizontal resolution of 1.9^°× 2.5^° (lat × lon) and a hybrid vertical coordinate with 30 levels, and is forced by the climatological monthly-mean SST during 1986-2005. All experiments are first integrated for five years and meteorological variables on 1 March in the sixth year are used as the initial conditions to run the following three months (March-April-May) with different SST gradients over the North Pacific subtropics. In the control experiment (CTRL), the model is run with the climatological SST. Figure 12 shows the wind speed at 200 hPa in spring from the ERA-Interim observations and CTRL. In CTRL, the 200 hPa wind speed has a maximum over southern Japan and extends northeastward from eastern China to the North Pacific, which is consistent with the observation except for a relatively weaker maximum center. On the whole, CAM5.1 simulates the atmospheric circulation in spring well (Deng et al., 2014; Deng and Xu, 2015; Zhao et al., 2015).

A set of sensitivity experiments with enhanced SST gradients over the NPSTF region are conducted to investigate the atmospheric responses of the NPSTF. In these sensitivity experiments, the SST gradients over the key area of the NPSTF are artificially amplified by a factor of 1.5 (SEN15), 2.0 (SEN20), 2.5 (SEN25), and 3.0 (SEN30). The SSTs and their gradients, zonally averaged over (139.5^°E-190.5^°W) in these sensitivity experiments, are shown in Fig. 13.

Figure 13.  Variation of (a) zonal-mean SSTs (units: ^°C) and (b) SST gradients [units: ^°C (100 km)^-1] zonally averaged over 139.5^°E-190.5^°W in the CTRL and sensitivity experiments.

Figure 14.  Differences in geopotential height (contours; units: m² s^-2) and horizontal wind (vectors; units: m s^-1) at 200 hPa in spring (a) between SEN15 and CTRL, (b) between SEN20 and CTRL, (c) between SEN25 and CTRL, and (d) between SEN30 and CTRL. Contour intervals are 20 m² s^-2 in (a, b) and 30 m² s^-2 in (c, d).

Figure 14 shows the differences of 200 hPa geopotential height and horizontal wind in spring between the sensitivity experiments and CTRL. For all sensitivity experiments, a negative geopotential height anomaly occurs over the central North Pacific north of the NPSTF, while a positive one is situated over the central Pacific. Correspondingly, a southwesterly wind anomaly appears over the EAJS core around 30^°N, which is consistent with our findings based on the observations (section 3). Moreover, amplitudes of geopotential height and wind anomalies are increased with an intensified NPSTF. Thus, these model simulations further confirm that anomalous westerly wind in the upper troposphere is indeed caused by the NPSTF.

This study investigates the relationship between the NPSTF and atmospheric circulation in spring on the interannual timescale. The results show that a strong NPSTF in spring can prominently enhance the EAJS. Both transient eddy activity and the atmospheric heat source play dominant roles in this process. The enhanced atmospheric temperature gradient due to a strong NPSTF increase atmospheric baroclinicity, resulting in an intensification of transient eddy and convection activities. On the one hand, the enhanced transient activities can excite an anomalous cyclonic circulation in the troposphere north of the NPSTF around 40^°N; and the related westerly wind anomalies to its south in the upper troposphere can intensify the EAJS over the Northeast Pacific. In addition, the enhanced transient eddy activities in the vicinity of the NPSTF act to suppress the transient eddy activities over the middle and high latitudes. The suppressed transient eddy activities can excite an anomalous anticyclonic circulation to its north, which in turn favors the maintenance of a cyclonic circulation anomaly. On the other hand, an enhanced atmospheric heat source appears over the NPSTF, which is related to increased rainfall. The enhanced heat source can excite an anomalous cyclonic circulation in the troposphere to its west, near 40^°N, which induces an enhancement of the EAJS over the Northwest Pacific. The results of a set of numerical experiments further confirm that anomalous westerly wind in the upper troposphere is indeed caused by the NPSTF. Note that as two types of behavior of an unstable atmosphere, the enhanced transient eddy activities and the atmospheric heat source associated with the NPSTF may not be independent. However, their influences on the atmospheric circulation prevail over different regions, suggesting that these two factors may combine to enhance the EAJS——a suggestion that needs further study.

Figure 15.  Latitude-height section of the regression coefficients of (a) zonal wind (contour interval: 0.5 m s^-1), (b) atmospheric baroclinicity index (contour interval: 0.5× 10^-2 d^-1), and (c) atmospheric heat source (contour interval: 0.1 W m^-2), onto the NPSTF index in spring. Zonal wind, atmospheric baroclinicity and the atmospheric heat source are zonally averaged over 140^°E-190^°W for the NCEP reanalysis dataset. Light and heavy shading represents statistically significant coefficients exceeding the 95% and 99% confidence levels, respectively.

As is well known, ENSO is the most dominant factor in air-sea interaction on the interannual timescale; it governs atmospheric circulation all over the world, especially over the North Pacific and East Asia (Zhang et al., 1999; Zhou et al., 2010). The question therefore naturally arises as to whether ENSO modulates the relationship between the NPSTF and the EAJS. We find that the results presented in this paper do not change (qualitatively) when the ENSO signals are filtered out in advance, as in (Gong et al., 2011) (Fig. S3 in ESM). In addition, there are stronger SST fronts in the midlatitudes of the North Pacific (Fig. 1), which also have significant influences on the atmospheric circulation over the North Pacific (Tanimoto et al., 2009; Kwon et al., 2010; Xu et al., 2010b; Xu and Xu, 2015). Thus, we define a midlatitude SST front index based on the normalized SST gradient over (31^°-37^°N, 140^°E-180^°) (Qiu, 2000; Wang et al., 2011) and recalculate the correlation between the NPSTF and the EAJS with the midlatitude SST front index filtered out in advanced. Results show that the NPSTF index still has a significant positive relationship with the 200 hPa wind speed around the center of the climatological EAJS (Fig. S4 in ESM), which is consistent with our findings presented in section 3. However, the relationship between the midlatitude SST fronts and the NPSTF over the North Pacific needs further investigation. Moreover, we perform a comparative analysis using NCEP-NCAR reanalysis data (Kalnay et al., 1996). The results are consistent with those using ERA-Interim. As shown in Fig. 15, the zonal winds are enhanced significantly in the whole troposphere around 30^°N due to a strong NPSTF, accompanying the intensification of atmospheric baroclinicity and the atmospheric heat source.

(Qiu and Chen, 2010) found that increased eddy kinetic energy along the North Pacific subtropical countercurrent (STCC), whose upward-titling thermocline is also called the NPSTF, is due to the enhanced baroclinic instability associated with the large vertical shear in the STCC background flow. Although they proposed that the interannual variation of the STCC is driven by a western Pacific pattern through Ekman transport convergence, they also suggested that there is positive feedback in the coupled air-sea interaction over the North Pacific in spring, which may be one reason why the NPSTF is most pronounced in spring. Our numerical results just confirm the significant influence of the NPSTF on the general circulation over the western Pacific, complementary to (Qiu and Chen, 2010). In addition to the interannual variability, the NPSTF and EAJS also exhibit significant intraseasonal variability (Kuang et al., 2007; Xie et al., 2015); further study is needed to investigate their relationship on this timescale.