The East Asian trough (EAT) is a distinct component of the boreal winter circulation whose strength corresponds to the amplitude of the Northern Hemispheric stationary waves. In this study, the mechanism and climatic impacts of the intraseasonal variations of the EAT’s strength are investigated through composite analysis and dynamical diagnostics. The significant roles played by the low-frequency Rossby wave (RW) and synoptic transient eddy (TE) are revealed. Before the peaks of strong EAT events, an upper-tropospheric RW train propagates across northern Eurasia and interacts with preexisting surface cold anomalies over central Siberia. This pattern intensifies the Siberian high and causes RW convergence toward the EAT, leading to 30% of the EAT’s amplification directly via the RW-induced feedback forcing. Meanwhile, RW weakens the background baroclinicity and reduces TE activities near the entrance region of the North Pacific storm track. The TE-induced feedback forcing leads to another 30% of the EAT’s amplification. The evolution and dynamical processes of the weak EAT events generally resemble those of the strong events with opposite signs. These results are consistent with the knowledge on the mechanism of the strong and weak EAT events regarding the role of RWs with additional quantitative description and provide new insights regarding the role of TEs. Variations of the EAT’s strength exert significant climatic impacts on East Asia and its downstream region. Near-surface air temperature is below (above) normal over East Asia during the growth and peak stages of the strong (weak) EAT events and above (below) normal over North America afterward.
The East Asian trough (EAT), also known as the Far East trough, is one of the most significant time-mean zonally asymmetric circulation features in the wintertime Northern Hemisphere, which is characterized by the strongest negative deviation from the zonal mean over East Asia (Zhang et al. 1997; Wang et al. 2009a). As a quasi-stationary coastal trough, it is excited by large-scale topography such as the Himalaya (Held et al. 2002; Holton 2004) and the large-scale thermal contrast between the warm Pacific Ocean and the cold Eurasian continent (Smagorinsky 1953; Nakamura et al. 2010). To the rear of the EAT, large-scale sinking motion causes strong radiative cooling and contributes to the buildup of the Siberian high (Ding and Krishnamurti 1987; Zhang et al. 1997). In front of the EAT, part of the cold monsoonal flow encounters warm air and results in a tight meridional temperature gradient over the western North Pacific, which is crucial for the Pacific storm track activity (Nakamura et al. 2002; Ren et al. 2008, 2010). The development of the Aleutian low is also suggested to be closely related to the EAT (Jhun and Lee 2004).
The variability of the EAT exerts substantial influences on the weather and climate over East Asia and beyond (Yang et al. 2002; Huang et al. 2012). On the synoptic time scale, a cold wave (also referred to as cold surge or cold-air outbreak) constitutes one of the most significant weather events occurring in the Northern Hemisphere (Ding and Krishnamurti 1987). Staff Members of Academia Sinica (1958) suggested that such cold waves are often related to the development and rebuilding of the EAT in association with short wave trains over the Eurasian continent. After the Siberian high and EAT reach certain intensity, an upper-level short wave develops from the western Eurasian continent. When this wave propagates eastward, it shows remarkable deepening and replaces the original coastal trough (i.e., the EAT; Staff Members of Academia Sinica 1958), resulting in a southward-moving surface anticyclone and outbreaks of cold waves (Zhang et al. 1997). Accompanying the occurrence of cold waves, synoptic disturbances develop over downstream regions especially over the North Pacific as a result of baroclinic growth (Lau and Lau 1984). On the interannual time scale, a deep EAT facilitates a strong East Asian winter monsoon (EAWM), which leads to cold anomalies over East Asia (Cui and Sun 1999; Chen et al. 2005; Wang et al. 2009a; Huang et al. 2012), a weak North Pacific storm track (Lee et al. 2010), and less (enhanced) precipitation over southern China (the Maritime Continent) (Wang and Chen 2010; Wang and Feng 2011). In addition, the decadal weakening of the EAT after the mid-1980s is related to the weakened EAWM and the resultant East Asian warming (Wang et al. 2009b; Lee et al. 2013a,b; Kim et al. 2013).
Variations of the EAT and the attendant outbreak of cold waves could be influenced by many factors, among which the atmospheric dynamical process is suggested to play a crucial role. For example, successive upper-level Rossby wave trains along the polar front jet over the northern Eurasian continent often precede the deepening of the EAT and the outbreak of East Asian cold waves (Staff Members of Academia Sinica 1958; Joung and Hitchman 1982; Park et al. 2014). The wave train features a barotropic structure over most of the Eurasian continent. When it reaches East Asia, it changes into a baroclinic structure and interacts with the surface cold anomalies, contributing substantially to the amplification of the Siberian high and the outbreak of cold waves (Takaya and Nakamura 2005b). Meanwhile, Rossby wave trains from upstream could also propagate along the waveguide in the upper-level subtropical Asian jet to East Asia (Hoskins and Ambrizzi 1993) and lead to a strong EAT and outbreaks of cold waves (Hong et al. 2009). On the other hand, Takaya and Nakamura (2005a) found that the feedback forcing from the synoptic transient eddies of the North Pacific storm track could also give rise to the amplification of the Siberian high and occurrence of cold waves. Although they did not specifically look at the EAT, their results imply the possible feedback forcing on the EAT from downstream regions in addition to the upstream incoming Rossby waves.
It is noteworthy that most of the above studies treated the EAT as a factor that influences the weather and climate of East Asia and focused on the behavior of the EAT during cold waves or strong and weak EAWM winters. However, the variability and mechanism of the EAT itself were not specifically investigated. Meanwhile, these studies mainly concerned the strength of the EAT, while it should be emphasized that as a part of quasi-stationary wave the EAT could also vary in its phase besides its amplitude (Wang et al. 2009a; Wang 2014). By performing empirical orthogonal function (EOF) analysis on the winter mean 500-hPa geopotential height over the trough region of the EAT, Wang et al. (2009a) proposed that the first two leading EOFs (Figs. 1a,b) represent the strength and the horizontal tilt of the seasonal mean EAT, respectively, which are manifestations of the amplitude and phase of the stationary planetary waves (Wang et al. 2009a; Wang 2014). Wang et al. (2009a) further revealed that the climatic influences of the two EOFs are both remarkable and linked to the seasonal mean intensity and pathway of the EAWM, respectively. Despite these studies, the mechanism of the EAT variability associated with the two leading modes remains to be understood.
In this study, we focus on the EAT variability and investigate the dynamics and climate impacts related to the strength of the EAT on the intraseasonal time scale. This paper consists of six sections. Section 2 introduces the datasets and our methods. Section 3 presents an overview of the temporal and spatial features of the strong and weak EAT events on the intraseasonal time scale. Sections 4 and 5 reveal the involved mechanism from the perspective of Rossby wave dynamics and eddy-induced geopotential height tendency. Section 6 shows the climate impacts of the strong and weak EAT events on East Asia and North America, and section 7 summarizes the key findings.
2. Data and methodology
The 40-yr European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-40; Uppala et al. 2005) dataset is used in this study. This dataset has a horizontal resolution of 2.5° × 2.5° and 23 vertical layers extending from 1000 to 1 hPa, covering 45 years from September 1957 to August 2002. Daily results averaged from the 6-hourly dataset are used. The daily climatology was calculated by averaging the daily results of a particular day over the 45 years and was smoothed by a 31-day running mean. In this study, the intraseasonal time scale is considered, so an 8–90-day Butterworth bandpass filter was applied to the daily anomalies to remove the synoptic and interannual components. We focus on the intraseasonal variability of the EAT in boreal winter (December–February), where each winter consists of 90 days by omitting 29 February of leap years.
b. Identification of EAT events
Through EOF analysis of the normalized winter mean 500-hPa geopotential height over the trough region of the EAT, Wang et al. (2009a) identified that the first EOF (EOF1) explains 47% of the variance and indicates the intensity of the EAT, while the second EOF (EOF2) explains 22% of the variance and presents the tilt of the EAT axis (Figs. 1a,b). Accordingly, we refer to the EOF1 mode as strong (EOF1+) or weak (EOF1−) EAT cases, and the EOF2 mode as vertical (EOF2+) or flat (EOF2−) cases. To obtain the typical EAT events on the intraseasonal time scale, the normalized wintertime 8–90-day bandpass-filtered daily 500-hPa geopotential height field in the region 25°–50°N, 100° E–180° is projected onto the first two EOF modes (Figs. 1a,b) to obtain a daily time series. Specifically, defining Z to be the observed normalized 500-hPa geopotential height field, we projected Z onto the EOF spatial pattern e to obtain the principal component (PC) time series y (Baldwin et al. 2009):
After we get the daily PC time series of the two EOFs, we define typical EAT events in the following way, taking a strong EAT event as an example. First, we rank the values of first PC (PC1) time series in descending order to find the date with the largest PC1 (i.e., the peak day). If the PC1 values on at least three days centered on the peak day all exceed one standard deviation, then this peak day is marked as day 0 of a strong EAT event. Once a day 0 is identified, no day within 21 days of the central date (day 0) can be defined as a strong EAT event. This procedure prevents the algorithm from counting the same strong EAT event repeatedly. Second, we repeat the above procedure until the values of PC1 cannot meet the above criteria to guarantee that all the strong EAT events are identified. This method is similar to that in Franzke et al. (2011), who studied the Pacific–North American teleconnection.
Figure 1c is the scatter graph of normalized PC1 versus PC2, which shows a uniform distribution of dots in phase space, indicating that the EOF1 and EOF2 are orthogonal to each other on the intraseasonal time scale. The correlation coefficient between PC1 and PC2 is only 0.008, confirming that the strength and tilt of the EAT are simultaneously independent from each other. In this study, we investigate the “pure” cases of the EAT variability. That means when the strength of the EAT is examined the possible influences from the tilt of the EAT should be avoided. Thus, only the strong and weak EAT events with neutral tilt are analyzed. Taking these criteria, we obtained 32 (34) strong (weak) EAT events and 29 (33) vertical (flat) EAT events, whose day 0 results are marked with large blue-filled dots in Fig. 1c. The days in the central part of the scatter graph (i.e., the open green circles) were treated as the “normal” days that serve as a reference. Composite analysis is used to illustrate the anomalous patterns associated with the strong and weak EAT events. Composite anomalies are obtained by subtracting the sample of selected strong or weak EAT events from the sample of normal days, which are defined as days in the central square of Fig. 1c. A two-tailed Student’s t test is applied to evaluate the confidence levels.
c. Diagnostic tools
To illustrate the propagation of Rossby waves associated with the intraseasonal variations of the EAT, the wave activity flux that is independent of wave phase and parallel to the local group velocity on a zonally varying basic flow (Takaya and Nakamura 2001) is employed. The wave activity flux is defined as
where is the climatology of winter mean horizontal wind velocity averaged over the 1971–2000 period, V′ = (u′, υ′) is the perturbation horizontal geostrophic wind velocity, ψ′ is the perturbation geostrophic streamfunction, f0 is the Coriolis parameter at 45°N, R is the gas constant of dry air, N2 is the square of the buoyancy frequency, p0 is the pressure divided by 1000 hPa, T′ is the perturbation air temperature, and H0 is the scale height. Primes denote the intraseasonal (i.e., low frequency) anomalies associated with the EAT. Subscripts x and y denote partial derivatives in the zonal and meridional directions, respectively.
Atmospheric eddies (i.e., waves) can lead to geopotential height tendencies (∂Z/∂t) of quasi-stationary flow via horizontal convergence of eddy vorticity and eddy heat fluxes. To diagnose the possible feedback forcing of eddies on the intraseasonal variations of the EAT, the eddy-induced ∂Z/∂t is given by Lau and Holopainen (1984):
where Z, g, and f are geopotential height, gravitational acceleration, and Coriolis parameter, respectively; σ = −(α/θ)(∂θ/∂p) is the static stability; θ is the potential temperature, assumed to be a function of pressure p only; and α is the specific volume. The eddy forcing term D can be decomposed into that which arises from the eddy heat flux DHEAT and from the eddy vorticity flux DVORT:
where the overbars indicate a time average and the primes represent the deviation from the corresponding time-averaged quantity. Here, is the hemispheric mean of the quantity , and ζ′ is the perturbation relative vorticity. In this study, the ∂Z/∂t induced by both the low-frequency Rossby waves and the high-frequency transient eddies are evaluated, where the transient eddies are extracted by applying a 2–8-day bandpass Lanczos filter with 161 weights on the 6-h data. By comparing the pattern, magnitude, and time evolution of the observed and eddy-induced ∂Z/∂t, the contribution of eddies to the observed ∂Z/∂t can be quantified.
3. Overview of the strong and weak EAT events
a. Temporal features
Figure 2 shows the evolution of the composite of normalized PC time series for the EAT events identified in section 2b. The PC values are very small from day −10 to day −6, and they reverse signs on day −6 and grow rapidly till day −1. After the peak (day 0), they decay rapidly from day +1 to day +6 and reverse signs on day +6. From day +6 to day +10, the PC values remain small till day +10. Overall, the life cycles are approximately 2 weeks for both the strong and weak events (Fig. 2a) and the vertical and flat events (Fig. 2b). Because the composite PC time series are very close to zero at both the beginning and the end of all the life cycles, the intrinsic dynamics of the EAT variability should be subject to an intraseasonal time-scale phenomenon. Moreover, these results suggest that it is enough to look at the 21-day period when the intraseasonal evolution and dynamics of the EAT are investigated. In the following part of this paper, the strong and weak EAT events (Fig. 2a) are analyzed; while the vertical and flat EAT events (Fig. 2b) will be analyzed in a following paper.
b. Three-dimensional structure
First, it is necessary to show the structure of the EAT during the peaks of strong and weak events so that the basic features and physical meanings of the two types of events can be better revealed. Figures 3a and 3b present the 500-hPa geopotential height fields during the peaks (averaged from day −2 to day +2) of the two cases. The EAT is deep and wide in the strong EAT events and shallow and narrow in the weak EAT events. Hence, the attendant meridional height gradient is intense (loose) over East Asia and the western North Pacific in the strong (weak) EAT events. These features can be further illustrated in the longitude–altitude plot of stationary wave field along 37.5°N (Figs. 3c,d), where the stationary wave is defined as the departure from the zonal mean. In the climatological mean field, a deep trough (i.e., the EAT) is observed at approximately 135°E. The trough shows an obvious westward tilt with height (Figs. 3c,d). In the strong EAT events (Fig. 3c), a negative geopotential height anomaly is observed from 120°E to 180°. The negative anomaly overlaps the climatological mean trough, suggesting that the amplitude of the climatological stationary wave is enhanced in the strong EAT events. In the weak EAT events, a similar anomaly with opposite sign is observed, but the positive anomaly in the EAT trough region is located slightly eastward compared with the strong EAT events (Fig. 3d). This pattern indicates significantly reduced amplitude and a slightly eastward-shifted phase of the climatological stationary wave. On the one hand, this result suggests that the variations of the strength of the EAT correspond to changes in the amplitude of the climatological stationary waves on the intraseasonal time scale, consistent with the situation on the interannual time scale (Wang et al. 2009a; Wang 2014). On the other hand, it suggests that the EOF1 mode in Wang et al. (2009a), which is obtained from the winter mean field, is also appropriate for the intraseasonal time-scale analysis.
Figure 4 shows the 500- and 1000-hPa geopotential height anomalies during different stages of the strong and weak EAT events. During the growth stage (from days −6 to −1) of the strong EAT events, a wave train–like signature is situated across the Eurasian continent at 500 hPa, resembling the Eurasian pattern (Wallace and Gutzler 1981; Liu et al. 2014). The structure of this wave train is mostly barotropic, and the phase tilts slightly westward with height in the regions to the east of 50°E (Fig. 4a). Meanwhile, the anticyclonic center of this wave train extends from Siberia to Alaska, showing a blocking-like pattern over the North Pacific. On the peak day (day 0), the lower reaches of the wave train amplify, and the blockinglike pattern shifts westward slightly. Consequently, a strong meridional dipole is observed over the western Pacific at both 1000 and 500 hPa (Fig. 4b), resembling the North Pacific Oscillation (NPO) and its tropospheric embodiment, the western Pacific (WP) pattern (Wallace and Gutzler 1981; Linkin and Nigam 2008; Wang et al. 2011). The southern cyclonic center of this dipole coincides with the trough region of the climatological EAT (Figs. 3c and 4b), indicating the strengthening of the EAT and the climatological stationary waves. After the peak day, the wavy and dipolar structures are both diminished, while a short wave train toward North America is observed along the rim of the North Pacific (Fig. 4c), implying a plausible dispersive mechanism for the decay of the strong EAT events.
The circulation patterns of the weak EAT events are roughly symmetric to those of the strong EAT events with opposite signs. A barotropic Eurasian pattern–like wave train whose phase tilts slightly westward with height in the regions to the east of 50°E is observed before the peak of the weak EAT events (Fig. 4d). An NPO- and/or WP-like dipolar structure is located over the North Pacific region, implying that both the upstream and the downstream anomalies could contribute to the weak EAT events. On the peak day, the NPO- and/or WP-like dipole is amplified and its southern positive center is located in the trough region of the climatological EAT (Figs. 3d and 4e), indicating the weakening of the EAT and the climatological stationary waves. In the decaying stage, a wave train–like anomaly emanates from the southern center of the dipole toward North America along the East Asian jet stream (Fig. 4f). This trend resembles the pattern found for strong EAT events and implies a plausible dispersive mechanism for the decay of the weak EAT events. Compared with the strong EAT events, the strength of the circulation anomalies is slightly weaker in the weak EAT events.
4. Mechanism: Analysis of Rossby wave dynamics
a. Strong EAT events
The analyses in section 3b suggest a possible role for the upstream Rossby wave trains for the strong and weak EAT events, so the daily evolution of the low-frequency 300-hPa geopotential height anomalies and the associated wave activity flux (Takaya and Nakamura 2001) were examined. Upstream of the EAT, clear Rossby wave activities emanate from a cyclonic center to the east coast of North America toward the Scandinavian Peninsula on day −9 of the strong EAT events (Fig. 5a). They evolve into a well-defined wave train that can reach East Asia on day −6 (Figs. 5b and 6a). The path of this wave train generally coincides with the upper-tropospheric polar front jet, which serves as a waveguide over the North Atlantic and the Eurasian continent because of the enhanced meridional potential vorticity gradients of the jet (Hsu and Lin 1992; Hoskins and Ambrizzi 1993; O’Rourke and Vallis 2013). From day −6 to day 0, the wave train propagates southeastward along the waveguide and converges into the regions of Siberia and the western portion of the EAT (Figs. 5b–d and 6). Meanwhile, significant cyclonic anomalies develop and peak over the trough region of the EAT. This process resembles that of the amplification of the Siberian high (Takaya and Nakamura 2005b), but the peak (day 0) of the Siberian high (see right column of their Fig. 3) corresponds to approximate day −2 of the EAT (Fig. 7a). The appearance of this delay could be understood from two aspects. First, the EAT is located in the downstream region of the Siberian high, so the upstream wave train first reaches the Siberian high region and then reaches the EAT region. Second, as will be discussed in the following paragraphs, the peak of the EAT needs to lag the Siberian high dynamically. In addition to the wave train along the polar front jet, a second wave train is observed along the subtropical Asian jet in the upper troposphere (Figs. 5a–c).
To reveal the possible cause–effect relation between the incoming low-frequency Rossby wave train and the deepening of the EAT, it is helpful to compare the evolution of the three-dimensional convergence of the wave activity flux (−∇3 ⋅ W) and that of the tendency of the wave activity density (∂A/∂t) over the key region of the EAT because the strong EAT events are featured with enhanced cyclonic disturbances over the location of the climatological EAT (Fig. 5d). When the wave and the basic flow are both conservative in the Wentzel–Kramers–Brillouin (WKB) approximation, the deepening of the EAT could be interpreted to be a result of the incoming Rossby wave if the ∂A/∂t in the key region of the EAT is proportional to the in situ −∇3 ⋅ W. Figure 7b shows the time series of the low-frequency ∂A/∂t and the −∇3 ⋅ W over the trough region of the EAT (30°–50°N, 110°–170°E). The ∂A/∂t is close to zero before day −4, and it amplifies to its peak on day −1, corresponding to the growth of the strong EAT events. Subsequently, it shrinks and turns negative on day +1, corresponding to the decay of the strong EAT events. The −∇3 ⋅ W evolves in a quite consistent manner with the ∂A/∂t although discrepancies exist (Fig. 7b). These discrepancies may be caused by several factors. For example, the basic flow and the disturbances embedded in it are not as conservative as in theory. The phase speed of the Rossby wave is not zero in reality (Takaya and Nakamura 2001), while it is assumed to be zero when we calculate W with Eq. (2). Despite these discrepancies, the results shown here suggest that the amplification of the anomalous cyclonic center near the Sea of Japan, and thereby the strengthening of the EAT, may be attributed to the incoming Rossby wave activity.
A closer inspection indicates that the convergence of the wave activity flux before the peak of the strong EAT is mainly contributed by the vertical component emanating from the lower boundary near the Siberian high and the coast of East Asia. This somewhat unexpected result can be understood by comparing our Figs. 5 and 6 with Takaya and Nakamura (2005b). When the equivalent barotropic wave train propagates toward East Asia, its structure becomes baroclinic to the east of approximately 90°E (Fig. 6). After day –4 [approximately corresponding to day 22 of Takaya and Nakamura (2005b)], the preexisting surface cold anomaly in central Siberia (Fig. 6b) can act as an anticyclonic potential vorticity anomaly to interact with the upper-level wave train by modifying the stratification (Hoskins et al. 1985; Takaya and Nakamura 2005b), and thereby reinforcing a downstream cyclonic center to strengthen the EAT [Figs. 6b–d, also see Fig. 8b of Takaya and Nakamura (2005b)]. This process, which can be traced back to as early as day −6 (Fig. 6a), can also be interpreted from the southward transport of cold air (i.e., northward heat flux) and the attendant upward Rossby wave activity flux to the east of 90°E (Fig. 6), similar to Figs. 9b and 10b in Takaya and Nakamura (2005b). Therefore, these results suggest that the intensification of the EAT does not result simply from the convergence of an incoming upstream Rossby wave train. Instead, the interaction of this wave train with the preexisting surface cold anomalies and the resultant secondary upward wave activity from the Siberian high region play an important role.
In the downstream region of the EAT, an anticyclonic blockinglike center is observed over the North Pacific as early as on day −9 (Fig. 5a). It amplifies, retrogrades, and extends westward from day −6 to day 0 (Figs. 5b–d), resembling the Pacific-origin-type amplification of the Siberian high (Takaya and Nakamura 2005a) and the occurrence of a blocking-type cold surge (Park et al. 2014). During this process, it merges with the anticyclonic center of the upstream wave train and forms an NPO- and/or WP-like dipole over the western North Pacific. Meanwhile, northeastward wave activity emanating from the cyclonic center around Japan may help to amplify and maintain the northern blockinglike center (Figs. 5b–d). As discussed in the previous paragraph, the amplified Siberian high and the attendant upward-propagating wave activity are important for the strong EAT events; therefore, the blockinglike anomalies over the North Pacific may also contribute constructively for the strengthening of the EAT.
After day 0, a clear wave train originating from the eastern Siberia is observed to propagate toward the tropical eastern Pacific (Figs. 5e–g), consistent with those reported in Lau and Lau (1984). Meanwhile, a second wave train is observed along the North Pacific rim toward North America (Figs. 5e–g), implying possible influences on the North American climate. Both of the two dissipative processes transmit Rossby wave energy out of the EAT and contribute to the decay of the strong EAT events (Fig. 7b).
b. Weak EAT events
In the weak EAT events, an anomalous anticyclonic center is observed near Japan on day 0 (Fig. 8d), which is located slightly eastward compared with its cyclonic counterpart in the strong EAT events (Figs. 5d and 3c,d). Despite the opposite signs of the geopotential height anomalies to the strong EAT events (Figs. 5b–d and 8b–d), the ∂A/∂t is positive from day −8 to day 0 of the weak EAT events (Fig. 7c). This is because the wave activity density A is roughly proportional to the squared eddy potential vorticity, and its positive tendency could result from both the amplified cyclonic and anticyclonic eddies (anomalies). The patterns of evolution of ∂A/∂t and −∇3 ⋅ W in this anticyclonic center coincide very well with each other over the trough region of the EAT (30°–45°N, 120°–170°E), although the magnitude of −∇3 ⋅ W is smaller than that of ∂A/∂t (Fig. 7c). Again, this finding suggests the importance of the incoming Rossby wave train on the weakening of the EAT. In this case, a short wave train begins to be observed over the North Atlantic and Europe on day −9 (Fig. 8a). This pattern persists and transmits Rossby wave activity toward the EAT along the polar front jet waveguide over the following days (Figs. 8b,c).
A closer inspection suggests that the convergence of the wave activity flux into the anticyclonic center in the trough region of the EAT before day 0 is also mainly impacted by the vertical component emanating from the lower boundary over East Asia to the east of about 100°E. The explanation to this phenomenon is similar to that of the strong EAT events despite opposite signs of circulation anomalies. That is, the eastern portion of the northern wave train has a baroclinic structure to the east of about 100°E, and the phase line of the anticyclonic anomaly around 125°E tilts westward with height (Fig. 9). In the presence of warm anomalies to the west of this anomalous anticyclone, this configuration facilitates the northward transport of warm air (i.e., northward heat flux). This process converts the background available potential energy (APE) into eddy APE by upward wave activity flux, reinforcing the upper-tropospheric anticyclonic center near Japan. Therefore, these results suggest that similar to the strong EAT events, the weakening of the EAT does not result simply from the accumulation of incoming upstream Rossby wave train, either. The interaction of an upstream wave train with the surface warm anomalies over the East Asian region also plays an important role.
In addition to the wave train over the Eurasian continent, a meridional dipole is observed over the eastern North Pacific on day −9 (Fig. 8a). It shifts westward and merges with the upstream wave train over the following days (Figs. 8b–d), contributing constructively to the NPO- and/or WP-like pattern on day 0 (thereby the weak EAT events). After day 0, the decay of the anticyclonic center near Japan is attributable to the eastward emanation of wave activity along the East Asian jet stream toward North America and the tropical eastern Pacific (Figs. 8d–f), and the meridional-oriented wave activity out of the anticyclonic center may also contribute.
5. Mechanism: Analysis of geopotential height tendency
a. Strong EAT events
The importance of the low-frequency Rossby wave to the amplification of the EAT is suggested in terms of the convergence of the Rossby wave flux in section 4a, but these analyses are qualitative. To gain a quantitative evaluation of the contribution of the Rossby wave, the geopotential height tendency induced by the eddy heat and vorticity fluxes of the Rossby wave is investigated following the method of Lau and Holopainen (1984). Meanwhile, synoptic transient eddies develop downstream of the climatological EAT, form a well-defined storm track estimated by the variance of the 2–8-day bandpass-filtered 300-hPa geopotential height (see Fig. 12c; Blackmon et al. 1977; Nakamura et al. 2002), and exert substantial forcing on the time mean flow (Lau and Holopainen 1984; Lau and Nath 1991) in boreal winter. Moreover, the retrogression of a blockinglike anticyclone (Fig. 5) resembles the Pacific-origin-type amplification of the Siberian high (Takaya and Nakamura 2005a), thereby implying possible forcing of the transient eddies on the intraseasonal evolution of the EAT. Hence, the geopotential height tendency induced by the transient eddies is also analyzed.
Figure 10 shows the observed and eddy-induced geopotential height tendency (∂Z/∂t) averaged from days −6 to −1. A cyclonic ∂Z/∂t is observed in the trough region of the EAT at approximately 40°N around Japan (Fig. 10a), indicating the deepening process of the EAT before day 0. The feedback forcing of the low-frequency Rossby wave train contributes constructively to this cyclonic ∂Z/∂t (Fig. 10a), consistent with the analysis of the Rossby wave convergence (section 4a). In addition, the net forcing from transient eddies appears to be more evident (Fig. 10b). An inspection of the time evolution reveals that the transient eddy–induced ∂Z/∂t is strong from days −8 to −4 and the Rossby wave–induced ∂Z/∂t is prominent afterward over the trough region of the EAT (30°–50°N, 110°–170°E) (Fig. 11a). During the growth stage of the strong EAT events, the two effects together explain approximately 60% of the observed magnitude of ∂Z/∂t (Fig. 11a). These results suggest that both the low-frequency Rossby wave and the high-frequency transient eddies give rise to the intraseasonal deepening of the EAT substantially via their feedback forcing. On the one hand, this conclusion is consistent with the knowledge that the deepening of the EAT could result from the development of upstream Rossby wave trains (e.g., Staff Members of Academia Sinica 1958; Zhang et al. 1997; Takaya and Nakamura 2005b). On the other hand, it proposes the importance of Pacific transient eddies (i.e., storm tracks) for the deepening of the EAT. During the decaying stage of the strong EAT, the Rossby wave is the predominant factor and explains approximately 50% of the observed anticyclonic ∂Z/∂t (Fig. 11a), indicating that the Rossby wave dispersion shown in section 4a is the main mechanism driving this pattern. The transient eddies also contribute with less importance (Fig. 11a).
To understand why the synoptic transient eddies contribute to the deepening of the EAT, the anomalous background baroclinicity and the resultant transient eddy activities are examined. The baroclinicity is measured by the maximum Eady growth rate (Hoskins and Valdes 1990), defined as BI = 0.31(f/N)d|V|/dz, where f is the Coriolis parameter, N is the static stability, z is the vertical coordinate, and V is the horizontal wind vector. In the climatological mean sense, the wintertime baroclinicity peaks along approximately 35°N over the western North Pacific (Fig. 12a), facilitating in its downstream regions the genesis of transit eddies (Fig. 12b) and the formation of a well-organized storm track (Fig. 12c). Before the peak of the strong EAT events, the lower-tropospheric baroclinicity is significantly reduced along approximately 55°N from 90° to 150°E (Fig. 12a). Consequently, the genesis of transient eddies in these regions is significantly weakened as measured by the lower-tropospheric meridional transient eddy heat flux (Fig. 12b). This process may decrease the upstream seeding for the Pacific storm track (Robinson et al. 2006; Penny et al. 2010, 2013). As a result, the North Pacific storm track is weakened both in its entrance region near the Sea of Japan and in its core region over the central North Pacific (Fig. 12c). In addition, the amplified Siberian high and the attendant EAWM (Fig. 7a) can also weaken the downstream storminess and give rise to weakened Pacific storm track (Nakamura 1992; Nakamura et al. 2002). In this situation, the weakened transient eddy activity leads to cyclonic ∂Z/∂t to its south over East Asia (Fig. 10b) by anomalous southward vorticity flux (Lau and Holopainen 1984). Although the eddy heat flux tends to offset the effects of the eddy vorticity flux, the latter is much stronger than the former in the upper troposphere (not shown; Lau and Nath 1991). Thus, the net transient eddy–induced ∂Z/∂t remains cyclonic and contributes constructively to the observed ∂Z/∂t (Fig. 10b).
The changes of the baroclinicity and the attendant synoptic transient eddy activity may have at least two causes. First, they may result from the preexisting cold anomaly in central Siberia (Figs. 6b and 13a), which could reduce the near-surface temperature gradient and baroclinicity to the north of the cold anomaly. Second, they are very likely a consequence of the incoming Rossby wave train discussed in section 4a. When the Rossby wave train propagates into East Asia, it leads to a meridional dipole straddling approximately 55°N and centered around 120°E in the upper-tropospheric geopotential height field (Figs. 4a and 5c). The dipole is also evident in the lower troposphere (Fig. 4a), and it weakens the climatological meridional pressure gradient and thereby the westerly wind along approximately 55°N from 90° to 150°E. The magnitude of the dipole is larger in the upper than in the lower troposphere, so the easterly wind anomalies increase with height. This process reduces the westerly shear and, thereby, the lower-tropospheric baroclinicity and transient eddy activity (Figs. 12a–c). This result suggests that, for the upstream Rossby wave train, it contributes constructively to the deepening of the EAT not only directly through its wave energy convergence and its feedback forcing, but also indirectly through its modulation on the activity and feedback forcing of transient eddies. Moreover, the spatial pattern of ∂Z/∂t induced by the high-frequency transient eddies resembles the observation more than that induced by the low-frequency Rossby waves does (Figs. 10a,b). This implies that the indirect effects of the low-frequency Rossby wave should be an essential ingredient in the deepening of the EAT. Despite the above physical picture, it should be noted that the low-frequency Rossby wave and high-frequency transient eddies can only explain about 60% of the observed magnitude of ∂Z/∂t. The remaining 40% of the magnitude may be attributable to the complex scale interactions and energy cascades between low- and high-frequency eddies (e.g., Jiang et al. 2013; Nie et al. 2013).
b. Weak EAT events
In the weak EAT events, an anticyclonic ∂Z/∂t is observed in the trough region of the EAT around Japan (Fig. 10c), indicating the weakening process of the EAT before day 0. Analyses of the geopotential height tendency show that both the low-frequency Rossby wave and the high-frequency transient eddies contribute directly to the anticyclonic ∂Z/∂t (Figs. 10c,d). The net forcing from transient eddies seems to be stronger and it corresponds better to the observed spatial pattern of ∂Z/∂t (Figs. 10c,d). The analysis of the time evolution suggests that the transient eddy–induced ∂Z/∂t over the trough region of the EAT (30°–45°N, 120°–170°E) is always anticyclonic from days −8 to −4, consistent with the observed ∂Z/∂t, and that it explains approximately 30% of the observed magnitude during the growth stage (Fig. 11b). In contrast, the Rossby wave–induced ∂Z/∂t is cyclonic before day −6, and it then turns to anticyclonic and explains approximately 30% of the observed magnitude from days −5 to −1 (Fig. 11b). Hence, it suggests that similar to the situation in the strong EAT events, both the Rossby wave and the transient eddies result in a direct net constructive contribution to the development of the weak EAT events. During the decaying stage, Rossby waves facilitate the observed cyclonic ∂Z/∂t, consistent with the analysis in section 4b, and transient eddies contribute with less importance (Fig. 11b). The net effect of the two is constructive but it only explains about 30% of the observed magnitude for the decay of the weak EAT. This result indicates that other factors such as interactions between transient eddies and low-frequency Rossby waves (Jiang et al. 2013; Nie et al. 2013) or diabatic process may be involved.
Like in the strong EAT events, the background baroclinicity and the resultant transient eddy activities are investigated to understand why transient eddies can be organized to exert feedback forcing on the EAT. Before the peak of the weak EAT events, the lower-tropospheric baroclinicity is significantly enhanced along approximately 50°N from 90° to 160°E (Fig. 12d). It may increase the upstream seeding for the Pacific storm track (Robinson et al. 2006; Penny et al. 2010, 2013). Consequently, the genesis of transient eddies is enhanced over northeastern China and downstream regions as measured by the lower-tropospheric meridional transient eddy heat flux (Fig. 12e). In addition, the weakened Siberian high and the attendant EAWM (Fig. 7a) can strengthen the downstream storminess (Nakamura 1992; Nakamura et al. 2002). Both of these mechanisms can lead to significant reinforcement of the North Pacific storm track (Fig. 12f). Hence, anticyclonic transient eddy–induced ∂Z/∂t is observed from East Asia to the North Pacific along approximately 40°N, which coincides with the location of the observed ∂Z/∂t (Fig. 10d). Again, this forcing effect is dominated by the eddy vorticity flux over the eddy heat flux (not shown).
Similar to the strong EAT events, the observed variations of the transient eddies can also be attributed to the upstream incoming Rossby wave train. When the Rossby wave train propagates to East Asia, it leads to a meridional dipole straddling at approximately 50°N over East Asia centered around 140°E in the upper-tropospheric geopotential height field (Figs. 4d and 8c). The northern pole of the dipole is evident and features cyclonic anomalies in the lower troposphere (Fig. 4d). This configuration strengthens the climatological meridional pressure gradient and westerly wind. The westerly wind anomalies increase with height because the anomalous pressure gradient is tighter in the upper troposphere. Hence, the enhanced westerly shear increases the background baroclinicity along approximately 50°N from 90° to 160°E, giving rise to enhanced transient eddy activity.
6. Impacts on East Asian and North American climate
a. Impacts on East Asia
Figure 13 shows the 850-hPa temperature anomalies on selected days during the life cycle of the strong and weak EAT events. On day −6 of the strong EAT events, an anomalous cold center of approximately −1°C is observed around Lake Baikal (Fig. 13a). It progresses southeastward, expands, and covers almost the whole of East Asia during days −3 to +3 (Figs. 13b–d), with the amplitude of coldness being approximately −3°C. This process is mainly contributed by the advection of basic-state temperature by anomalous low-frequency wind (not shown) that occurs concurrently with the deepening of the EAT. Note that the preexistence of this surface cold in central Siberia is an important ingredient for the vertical coupling that leads to the amplification of the Siberian high (Takaya and Nakamura 2005b) and the convergence of Rossby waves toward the EAT (section 4a), which then produces low-frequency wind anomalies and cold advections. Therefore, this drop in East Asian temperature can be regarded to some extent as a positive feedback process. From day +6 onward, the main body of the cold center moves out of the Asian continent and decays significantly (Figs. 13e,f). The time series of 850-hPa temperature anomalies over East Asia (20°–50°N, 110°–150°E) captures the above-mentioned temperature evolution well with the area-averaged temperature minimum −2.1°C being observed on day 0 (Fig. 14a). Recall that the intensity of the Siberian high reaches its peak on day −2 (Fig. 7a). Hence, the 850-hPa temperature minimum over East Asia occurs almost simultaneously with the peak of the strong EAT events, whereas both of them lag the peak of the Siberian high by approximately 2 days.
In the weak EAT events, no clear temperature anomalies are observed over East Asia on day −6 (Fig. 13g). From days −3 to 0, warm anomalies emerge and amplify over the coastal region of East Asia, and expand significantly southward and eastward (Figs. 13h,i). The amplitude of the warmness reaches approximately 3°C, similar to that of the strong EAT events (Fig. 13c), but the spatial scale of the warm center is smaller than its cold counterpart in the strong EAT events (Figs. 13b,c). This feature can be well illustrated in the time series of the area-averaged temperature over East Asia over the same domain as in the strong EAT events (Fig. 14b). The warm anomaly evolves almost symmetrically to that in the strong EAT events, except it has smaller magnitude (approximately 1.3°C on day 0) arising from its smaller spatial scale (Fig. 13i). In addition, the warmness is not very persistent. It decays rapidly on day +3 and switches to clear cold anomalies on day +6 (Figs. 13j–l). This feature can also be seen clearly in Fig. 14a. These results suggest that the impacts of weak EAT events on the East Asian temperature are slightly weaker than those of the strong EAT events.
b. Impacts on North America
From day −6 to day 0 of the strong EAT events, an anomalous anticyclone (cyclone) is located over Alaska (the Great Lakes) (Figs. 5b–d), which enhances the climatological ridge (trough) over the west (east) coast of North America. Therefore, the cold advection from the Arctic toward North America is enhanced (not shown), leading to significant cold anomalies over the northern part of North America (Figs. 13a–c). These cold anomalies, however, cannot be regarded simply as influences of the EAT because they occur concurrently with the strong EAT events. After day 0, the Rossby wave train propagates from the EAT region toward North America (Fig. 5), indicating potential impacts on the climate downstream of East Asia. From days +3 to +9, the wave train emanating from eastern Siberia features a cyclonic anomaly over eastern North Pacific around Alaska (Figs. 5e–g). The latter weakens the climatological ridge west of the Rocky Mountains (Fig. 3a) so that the cold advection from the Arctic toward North America is reduced (not shown). Hence, the northern parts of North America are featured with significant warming from day +3 onward (Figs. 13d–f and 14b). The warmness peaks on day +7 with a magnitude of 2.5°C and persists till approximately day +16 (Fig. 14b). This may be good news for North America in that approximately two weeks of mild weather could be expected after the severe weather of East Asia. Before the peak of the weak EAT events, the northern part of North America is warmer than normal, but the magnitude of the warmth is very small (Figs. 13g–i). After day 0, an anomalous cold center is observed over the northeastern Pacific, and it shifts northeastward, leading to significant coldness over North America after day +6 (Figs. 13j–l and 14b). Nevertheless, the magnitude of the temperature anomalies during the weak EAT events is relatively small compared with the strong EAT events, indicating the diminished impacts of the weak EAT events on North American temperature.
7. Summary and discussion
This study investigates the evolution, mechanisms, and climatic impacts of the strength of the EAT on the intraseasonal time scale during boreal winter by compositing strong and weak EAT events identified from the 45-yr ERA-40 dataset via use of the EOF method. Both the strong and weak EAT events have life cycles of approximately 2 weeks, suggesting that the intrinsic dynamics of the EAT variability is on the intraseasonal time scale. During the peaks of their life cycles, the deepening and shallowing of the EAT feature significant increases and decreases in the amplitude of the stationary waves, respectively.
The strong EAT events are preceded by an upstream low-frequency Rossby wave train that originates in the North Atlantic and propagates along the waveguide of the North Atlantic and Eurasian polar front jets. The wave train has a barotropic (baroclinic) structure to the west (east) of approximately 90°E. It interacts with the preexisting surface cold anomalies over central Siberia and strengthens the Siberian high, consistent with Takaya and Nakamura (2005b). Concurrently, the lower-tropospheric northward low-frequency eddy heat flux associated with the amplified Siberian high leads to significant upward Rossby wave activity fluxes that converge toward the trough region of the EAT. Analyses of the geopotential height tendency show that the Rossby wave can amplify the EAT directly and explain about 30% of the total magnitude of the deepening of the EAT. Meanwhile, the low-frequency Rossby wave can change the low-frequency flow and weaken the background baroclinicity, thereby facilitating weakened transient eddy activity over the entrance region of the North Pacific storm track. Then, the transient eddies exert strong feedback forcing on the low-frequency flow via eddy vorticity and heat fluxes and contribute approximately 30% of the magnitude of the observed amplification of the EAT. On the one hand, the dynamical processes associated with low-frequency Rossby waves are consistent with our previous understanding of the evolution of the strong EAT events (e.g., Staff Members of Academia Sinica 1958; Zhang et al. 1997) and provide a quantitative description compared with earlier studies. On the other hand, the essential role of high-frequency transient eddies is further proposed and it provides new insight into this issue. With the amplification of the EAT, the 850-hPa temperature of East Asia drops significantly as a result of cold advection. After peaks in the strong EAT events, two Rossby wave trains emanate from the EAT and dissipate toward North America and the tropical eastern Pacific, respectively, leading to the decay of the strong EAT events. The former weakens the stationary ridge to the west of the Rocky Mountains and leads to persistent mild weather over North America.
The evolution, mechanisms, and climatic impacts of the weak EAT events generally resemble those of the strong events except they have opposite signs of circulation and climatic anomalies. The low-frequency Rossby waves and high-frequency transient eddies contribute directly to about 30% of the magnitude of the weakening of the EAT, respectively. The decay of the weak EAT events is attributable to a Rossby wave train that propagates out of the EAT toward North America along the East Asian jet stream, and other processes such as frictional or diabatic damping may also be involved. This wave train is less persistent than its counterpart in the strong EAT events. Hence, the resultant cold over central North America is modest.
It is interesting to note that the upstream low-frequency Rossby wave is shown to be important for both the strong and weak EAT events, but its role is not realized simply through the accumulation of wave activity from upstream incoming waves. On the one hand, the upward Rossby wave activity flux emanating from the lower boundary of East Asia, which arises from the altered strength of the Siberian high and the attendant northward eddy heat flux is the main contributor to the wave activity convergence. The variations in the strength of the Siberian high involve vertical interactions between the upper-tropospheric Rossby wave train and the surface temperature anomalies (section 4; Hoskins et al. 1985; Takaya and Nakamura 2005b). Hence, the incoming Rossby wave train experiences a secondary baroclinic growth over East Asia. On the other hand, the Rossby wave train could modulate the activity of transient eddies by altering the lower-tropospheric baroclinicity. The transient eddies then lead to the development of the strong or weak EAT events via its vorticity and heat forcing. This indirect effect explains approximately 30% of the observed magnitude during the growth stage of both the strong and weak EAT events, and this is comparable to that induced directly by the vorticity and heat forcing of the low-frequency Rossby wave.
In addition to the upstream Rossby wave train, an NPO- and/or WP-like dipole is observed over the North Pacific during the growth and peak periods of both the strong and weak EAT events (Figs. 5b–d and 8c–d). The anticyclonic (cyclonic) anomaly over the Bering Sea is especially prominent. Its evolution resembles the Pacific-origin-type evolution of the Siberian high (Takaya and Nakamura 2005a), especially in the strong EAT events. All these results suggest the importance of the North Pacific circulation on the strength of the EAT (thus the EAWM), consistent with the analysis on the interannual time scale (e.g., Li et al. 2007; Wang et al. 2011; Takaya and Nakamura 2013; Wang and Chen 2014).
We thank the three anonymous reviewers for their constructive comments that lead to improvements to the manuscript. This work was supported by the National Natural Science Foundation of China (41422501 and 41275058) and the Excellent Young Scientists Project of the Chinese Academy of Sciences.