Mean FlowStorm Track Relationship and RossbyWave Breaking in Two Types of El-Nio

The El-Niño has been documented to have substantial impacts on global climate through atmospheric teleconnection excited by vigorous convection over the tropical central and eastern Pacific Ocean (Trenberth et al., 1998; Wang et al., 2012). Over the East Asia sector, northerlies are weakened during the winter of El-Niño, resulting in less frequent cold surges occurring in that region (Zhang et al., 1996; Tao and Zhang, 1998; Li et al., 2005). The summertime precipitation anomaly in East Asia is also closely related to El-Niño (Chou et al., 2002; Arai and Kimoto, 2008; Zhou et al., 2008; Li and Zhou, 2011). During the developing stage of El-Niño, flooding tends to occur over the Yangtze River and drought may occur in South China and North China; while in the decaying stage of El-Niño, above-normal precipitation tends to occur south of the Yangtze River and over North China, and below-normal precipitation often occurs over the lower reaches of the Yangtze and Huaihe rivers (Huang and Wu, 1989; Huang et al., 2004). During the winters of El-Niño years, the northern part of the USA and the southern part of Canada feature dryer and warmer conditions, while southern USA is characterized by wetter and colder conditions (Ropelewski and Halpert, 1986; Kiladis and Diaz, 1989; Livezey et al., 1997; Patten et al., 2003). Besides large-scale circulation, synoptic-scale transient eddy activities, which are organized as storm tracks over the mid-latitude oceanic regions, are significantly affected by El-Niño events. During El-Niño events, the storm track over the North Pacific is strengthened and extends downstream, in agreement with the strengthening and downstream extension of the subtropical jet over the North Pacific (Rasmusson and Wallace, 1983; Hoerling and Ting, 1994; Chen and van den Dool, 1999; Zhu et al., 2001; Yang et al., 2002; Ren et al., 2008; Seager et al., 2010).

A new type of El-Niño was recently found to have different characteristics compared with the canonical El-Niño in terms of the SST pattern during the mature phase. It was named the “Central Pacific El-Niño” (Kao and Yu, 2009), but is also known as “El-Niño Modoki” (Ashok et al., 2007), the “dateline El-Niño” (Larkin and Harrison, 2005), or the “Warm Pool El-Niño” (Kug et al., 2009). Despite the various names given by different researchers, they in essence represent the same phenomenon; that is, a maximum SST warming occurring in the central Pacific that roughly coincides with the Niño4 area (Wang et al., 2012). Recently, many studies have compared the impacts of the Central Pacific El-Niño (CP El-Niño, hereafter) with the canonical Eastern Pacific El-Niño (EP El-Niño, hereafter) on large-scale circulation and regional climate. (Feng and Li, 2011) found that spring rainfall over South China decreases during CP El-Niño, while during EP El-Niño it increases slightly. (Weng et al., 2008) documented the distinct large-scale circulation patterns related to the two types of El-Niño and the resulting climate anomaly conditions along the North Pacific rim. (Graf and Zanchettin, 2012) documented the different impacts of the two types of El-Niño on the North Atlantic Oscillation (NAO) and European climate. Besides the large-scale atmospheric circulation anomaly, (Ashok et al., 2009) documented the storm track anomaly in the Southern Hemisphere during CP El-Niño.

Thus far, most efforts undertaken have focused on the anomaly of large-scale circulation and its influences on regional climate during the two types of El-Niño. However, no study to our knowledge has investigated the storm track-mean flow relationship over the North Pacific region associated with CP El-Niño. Synoptic-scale transient eddy activities are considered as an internal atmospheric candidate for maintaining or invigorating the mean flow variability (Hoskins and Pearce, 1983; Lau and Holopainen, 1984 ). It has been shown in the literature that there exists a symbiotic relationship between storm track and mean flow due to their dynamical connections (Lau, 1988; Luo et al., 2007; Ren et al., 2008; Ren et al., 2010; Li and Wettstein, 2012; Xiang and Yang, 2012). On the one hand, the mean flow anomaly can cause anomalous eddy activities via altering baroclinicity; while on the other hand, the induced anomalies of eddy activities reinforce the anomalous circulation by eddy momentum flux forcing (Lorenz and Hartmann, 2003; Robinson, 2006; Gerber and Vallis, 2007; Ren et al., 2008), demonstrating a positive feedback mechanism between eddy activities and mean flow. Therefore, we hypothesized that the anomalies of large-scale circulation under CP El-Niño conditions are also accompanied by transient eddy activity anomalies over the North Pacific. In this paper, we attempt to demonstrate the features of the storm track and its dynamical connection with anomalous mean flow under two-type El-Niño.

Recently, Rossby wave breaking (RWB) has been shown as responsible for the strong dynamical linkage between transient eddies and mean flow (Woollings et al., 2008; Strong and Magnusdottir, 2008; Song et al., 2011 ). RWB usually occurs during the late stage of a baroclinic eddy life cycle (Thorncroft et al., 1993; Hitchman and Huesmann, 2007), characterized by strong eddy momentum flux. There are two types of RWB—namely, anticyclonic wave breaking (AWB) and cyclonic wave breaking (CWB)—according to the shape of the overturning of potential vorticity (PV) contours (McIntyre and Palmer, 1983, 1984). Previous studies have suggested that AWB is associated with poleward momentum flux and acts to push the jet poleward, while CWB favors the equatorward shift of the jet (Kunz et al., 2009; Franzke et al., 2011; Woollings et al., 2011). Many studies have discussed their roles in maintaining the large-scale circulation anomaly. For example, (Benedict et al., 2004) and (Franzke et al., 2004) related positive (negative) NAO to the upper tropospheric anticyclonic (cyclonic) wave breaking events occurring in the North Atlantic sector. (Gong et al., 2010) documented the role of RWB in maintaining the Southern Annular Mode (SAM) under El-Niño conditions. In the wave breaking stage, the tilted eddies have the crucial role of increasing the eddy momentum flux convergence/divergence, leading to the dynamical effect on the mean flow. As such, RWB can be a key process during which intensive wave-mean flow interaction takes place. Possible feedback may exist between RWB and mean flow during the two types of El-Niño. In the present study, we also focused on the RWB processes that could help us understand the essential characteristics of storm track-mean flow interaction associated with the two types of El-Niño. Furthermore, such work contributes to our understanding of how certain anomalous large-scale circulations are maintained during the two types of El-Niño.

The paper is organized as follows. The dataset and analysis methods are described in section 2. In section 3, the large-scale atmospheric circulation, storm track anomaly and anomalous occurrence of RWB related to the two types of El-Niño are presented. In section 4, we show the dynamical linkage between the observed large-scale circulation anomaly and the anomalous occurrence of RWB. Finally, a discussion and conclusions are given in section 5.

The atmospheric datasets used were from the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) reanalysis dataset. We used daily and monthly mean fields, with a horizontal resolution of 2.5°×2.5°. The SST data were from the HadISST (Hadley Centre Sea Ice and Sea Surface Temperature) dataset, with a horizontal resolution of 1°×1°. We used data covering the years 1979-2010 for two reasons. First, the data for this time period are more reliable because they incorporate satellite observations. Secondly, the El-Niño experienced a substantial change in characteristics around the mid-1970s (Wu, 2003; An, 2004). Prior to that time, warm water first occurred near the coast of Peru, and then migrated westward to the eastern and central Pacific. Since that time, the warm SST first occurs in the central Pacific and then migrates eastward. The winter mean in this study refers to the average over January-February-March (JFM), during which the difference in SST pattern between the two types of El-Niño is most apparent.

To study the activity of synoptic-scale transient eddy activities, a bandpass-filtered technique (Murakami, 1979) was applied to extract synoptic-scale disturbances with periods of 2.5-8 days from the NCEP-NCAR dataset. The transient eddy activities (storm tracks) were measured by synoptic bandpass-filtered at 300 hPa. The horizontal Eliassen-Palm vector (E), which is defined as (Hoskins et al., 1983), was calculated at 300 hPa to diagnose the interaction between transient eddies and large-scale atmospheric mean flow. The overbar denotes the average over winter (JFM in this study), while the prime denotes the 2.5-8 day bandpass-filtered fields. The convergence (divergence) of E corresponds to a forcing of decreasing (increasing) westerly mean flow (Hoskins et al., 1983).

Potential temperature on a 2 PVU surface was also calculated using NCEP/NCAR 6-hourly data to detect RWB. PV is defined as PV=(ζ+f) ∂θ/∂p; namely, the absolute vorticity multiplied by static stability measured by the vertical gradient of potential temperature θ, which itself is calculated at each pressure level. ζ is the relative vorticity and f is the Coriolis parameter. Potential temperature on the 2 PVU surface was derived by a linear interpolation in the vertical direction. We identified a RWB event when large-scale overturning of potential temperature contours occurred. In other words, when there was an inverse in climatology-mean meridional gradient of potential temperature. This method has been used to represent RWB in many studies (e.g., Thorncroft, 1993; Frantzke et al., 2004; Ndarana2011) and has been proven to be equivalent to another widely used method in which RWB is detected by the inverse of the meridional PV gradient on an isentropic surface (Hitchman2007; Gabriel2008; Strong2008). The former method was chosen because most tropospheric RWB events occur near the tropopause, and a 2 PVU surface is a good representation of the tropopause in the middle to high latitudes.

A contour configuration identification technique following (Strong and Magnusdottir, 2008) was applied to record the warm tongue of a breaking event (Fig. 1). An AWB event that occurred on 5 January 1980 and a CWB event that occurred on 6 January 1998 are shown in Figs. 1a and b, respectively. The centroid, area, onset time and duration were recorded for the two types of breaking events (i.e., AWB and CWB). When the distance between two RWB centroids at 6 h interval was less than 10° longitude or latitude, we attributed them to the same event. The time-mean occurrence of AWB:

and that of CWB:

were calculated. The λ,φ, and t denote longitude, latitude and time, respectively. The RWB occurrence functionβ(λ,φ, t ) would be unity when a grid point (λ,φ) was within the coverage of an RWB event (e.g., the shaded region in Fig. 1) at a certain time t, while it would be zero if the grid point(λ,φ) was without the coverage of any RWB event. T denotes the length of time over which the average was made, and the subscript “a” and “c” represent AWB and CWB, respectively.

Figure 1.  The configuration of potential temperature contours on a 2 PVU surface when (a) an AWB event occurred on 5 Jan 1980; (b) a CWB event occurred on 6 Jan 1998. When considering the occurrence of RWB events, only grids within the shaded areas were taken into account. The solid circles indicate the centroids of RWB events.

Figure 2.  Composite SST anomalies (K) during EP El-Niño (left) and CP El-Niño (right) years: (a) and (b) winter-mean (JFM) SST anomalies; (c) and (d) time-longitude sections of monthly mean SST anomalies. An ordinate value of 0 indicates a developing year, while 1 indicates a decaying year.

To obtain the two types of El-Niño events, we classified each event as having either winter-mean Niño3 or Niño4 index values larger than one standard deviation as an El-Niño event (Kug et al., 2009). By applying subject observation, we identified four EP events (1982/83, 1986/87, 1991/92, 1997/98) and four CP events (1991/92, 1994/95, 2002/03, 2004/05). Others were considered as “mixed type”. The composite fields of winter-mean SST, geopotential height, wind fields, storm track, Eady growth rate maximum (Eady, 1949) and frequency of occurrence of AWB/CWB were generated. The two-tailed student t-test was utilized to test the significance of anomalies at the 90% confidence level.

To diagnose the large-scale circulation related to individual RWB events, daily atmospheric anomaly data were used to produce lead-lag composites for AWB/CWB events. We only included RWB events that lasted for more than one day. Fields were centered on wave-breaking centroids prior to their inclusion into the composites. All anomalies were obtained by subtracting the annual cycle. For monthly mean fields, the annual cycle was simply the climatological mean for each month. For daily fields, a 20-day running average was applied for the calendar mean for each day to obtain the annual cycle.

Figure 3.  Upper panels: composite winter mean (JFM) geopotential height anomalies (shaded, gpm) at 500 hPa during (a) EP El-Niño years and (b) CP El-Niño years. Solid (dashed) lines indicate positive (negative) anomalies significant at the 90% confidence level. (c, d) The same as (a, b), but for wind (m s^-1) at 300 hPa over the North Pacific-North America region. Dark (light) shadings indicate positive (negative) zonal wind anomalies significant at the 90% confidence level.

Figure 4.  Composite winter mean (JFM) storm track anomalies (color shading, gpm) at 300 hPa and Eady growth rate maximum(σ_B) anomalies (contours, d^-1) between 850-700 hPa during (a) EP El-Niño years and (b) CP El-Niño years.

Figure 2 displays the composite SST anomaly related to the two different types of El-Niño. It is evident that the SST anomaly for EP El-Niño is much stronger than that of CP El-Niño, which is a phenomenon also reported in other studies (Ashok et al., 2007; Kug et al., 2009; Yeh et al., 2009). Figures 3a and b show the composite winter mean (JFM) geopotential height anomaly at 500 hPa (Z500) in the Northern Hemisphere during the two types of El-Niño. The most significant signals are mainly confined to the North Pacific-North America sector during both EP and CP El-Niño. Therefore, our discussion in the following part focuses on that region (15°-65°N, 140°E-80°W).

The anomalies of large-scale atmospheric circulation of the Pacific-North America (PNA) pattern related to EP El-Niño in the North Pacific-North America region (Fig. 3a) have been well documented, and our results are consistent with others (Trenberth et al., 1998; Larkin and Harrison, 2005; Wang et al., 2012). The corresponding wind field at 300 hPa in Figs. 3c is characterized by a local strengthening and downstream extension of the East Asia-North Pacific jet. The intensified westerly covers an area ranging from 20°-40°N. The strengthened westerly in the 20°-30°N band is probably the poleward branch of an anticyclone, which is part of a Gill-type response to the convective heating anomaly located in the eastern and central tropical Pacific (Matsuno1966; Gill1980). The northern part of the westerly anomaly (30°-40°N) is very possibly related to the transient eddy activity anomaly, as will be shown in Figs. 4 and 7. In the high latitudes of the North Pacific in Fig. 3c, there is a significant easterly anomaly with amplitude similar to the westerly anomaly to the south. Over North America, consistent with the dipole anomaly in the height field in Fig. 3a, a tripolar anomaly of zonal wind is evident in Fig. 3c. The large-scale atmospheric pattern is very different from a wave train shown in a heating experiment using linear models (Hoskins and Karoly, 1981), due to some nonlinear processes playing a role in determining such an anomalous pattern in the middle latitudes (Sheng et al., 1998).

As for CP El-Niño in Figs. 3b and d, the amplitude of anomalies is somewhat weaker due to its significantly weaker SST anomaly compared to that for EP El-Niño, as shown in Fig. 2. The anomaly of the Z500 field has one negative center in the south and two positive ones in the north over the North Pacific (Fig. 3b), which is distinct from the monopole low anomaly over the North Pacific in EP El-Niño years (Fig. 3a). The corresponding zonal wind anomaly exhibits a north-south dipole straddling at the latitude of the mean position of the East Asia-North Pacific jet (Fig. 3d). The southern lobe of the zonal wind dipole anomaly is probably related to the diabatic heating caused by the anomalous convection in the Central Pacific (Matsuno, 1966; Gill, 1980). Such a pattern means that, during CP El-Niño, the downstream region of the jet would be nudged toward the south.

Figure 5.  Zonal mean (160°E-120°W) of Eady growth rate maximum anomalies (lines with solid circles, d^-1), components of vertical shear (lines with triangles, d^-1), and static stability (lines with rectangles, d^-1) anomalies during (a) EP El-Niño and (b) CP El-Niño years.

Figure 4 shows the composite fields of storm track and Eady growth rate maximum anomalies. The storm track is represented by the standard deviation of geopotential height at 300 hPa on the synoptic timescale, and the Eady growth rate maximum, indicating baroclinicity, was calcuated in the low-level troposphere according to σ_BI=0.31fV_ZN^-1, where N is the Brunt-Väisälä frequency, and V_z is the vertical gradient of absolute horizontal wind. For the EP El-Niño case, a north-south dipole pattern of baroclinicity anomaly straddles at the latitude of 42°N over the North Pacific, while another dipole straddles at the latitude of 30°N over continental North America (Fig. 4a). During CP El-Niño, a baroclinicity dipole straddles at the latitude of 37°N over the North Pacific (Fig. 4b). The mid-latitude storm track activity can be primarily attributed to the atmospheric baroclinicity. As a result, the pattern of anomalous storm track activity is almost identical to that of anomalous baroclinicity for both EP and CP El-Niño.

The anomaly of baroclinicity is mainly due to the anomaly of vertical wind shear V’_z and the anomaly of Brunt-Väisälä frequency N’. To identify the relative contribution of V’_z and N’ to the anomaly of baroclinicity, we linearized the formula of Eady growth rate maximum about the climatological mean and obtained the expression: . The overbar denotes the climatological mean of winter mean, while the prime denotes the anomaly of winter mean. σ’_BI is the baroclinicity anomaly, which is the sum of two parts—one represents the contribution of anomalous vertical wind shear and the other the static stability . Comparing the two parts demonstrates that the baroclinicity anomaly is dominated by the vertical wind shear anomaly over the North Pacific (Figs. 5a and b). Therefore, strengthened (weakened) upper-troposphere zonal wind (Figs. 3c and d) is overlapped with the enhanced (reduced) baroclinicity (Figs. 4a and b), and as such the invigorated (suppressed) storm track (Figs. 4a and b). To summarize, for both EP and CP El-Niño, the patterns of their respective anomalous storm track activity are almost identical to their respective baroclinicity and upper-level zonal wind anomalies.

Figure 6.  Climatological frequency of occurrence (shading, d^-1) of (a) AWB and (b) CWB. Grey contours represent the climatology of zonal wind at 300 hPa. The interval between contours is 15 m s^-1, with the contour of 30 m s^-1 emboldened.

Figure 7.  Composite winter mean (JFM) anomalous frequency of occurrence of AWB (shading, d^-1) during (a) EP El-Niño years and (b) CP El-Niño years. (c, d) The same as (a, b), but for CWB. The grey contours represent the composite zonal wind anomalies at 300 hPa during the two types of El-Niño. The interval between contours is 3 m s^-1, with the dashed contours being negative.

The above analysis has demonstrated the symbiotic relationship between large-scale circulation and storm track variability. Next, we present the occurrence patterns of RWB associated with the two types of El-Niño. Figure 6 shows the climatological frequency of occurrence of the two types of RWB. The preferential area of CWB locates in higher latitudes than that of AWB. Obviously, AWB occurs more frequently than CWB in the North Pacific sector. The ratio of the total number is roughly 3:2 (AWB: CWB). Such discrepancy between the two types of RWB is far smaller than in the Atlantic, where AWB is much more dominant (Gabriel and Peters, 2008; Strong and Magnusdottir, 2008). In terms of RWB identification, (Gabriel and Peters, 2008) used the ERA-40 (European Centre for Medium-Range Weather Forecasts 40-yr Reanalysis) dataset and showed the climatological spatial distribution of RWB events in their Fig. 9 as we do in our Fig. 6. Through comparison, we find the general features are similar between our results and theirs. Such similarity indicates that the identification of RWB events is not sensitive dataset selection. Our results concerning climatological RWB frequency are slightly different from those in (Strong and Magnusdottir, 2008), in that we detected RWB along the tropopause, while they did so along a 350 K isentropic surface, which extended to the lower stratosphere in the middle to high latitudes.

The composites of anomalous AWB/CWB frequency of occurrence related with two-type El-Niño cases are shown in Fig. 7, overlapped with the corresponding zonal wind anomaly. The number of AWB events is reduced over the Northeast Pacific for both EP and CP types. For EP El-Niño cases, the robustly decreased signal of AWB events spans over the whole Northeast Pacific basin, where AWB events occur climatologically. In contrast, the signal of AWB events for CP El-Niño cases is over a small region with weak amplitude. The anomalous patterns of CWB associated with EP and CP El-Niño are distinctly different. During EP El-Niño, more CWB events occur in most areas of the USA and in the southern part of Canada; while during CP El-Niño, the increase in frequency of occurrence of CWB is located over the middle to high latitudes of the North Pacific, especially to the southeast of the Aleutian Islands. We also redetected the wave breaking by the method introduced by (Gabriel and Peters, 2008). The results were similar to those shown in Figs. 6 and 7 (data not shown), indicating such anomalous patterns are robust and method-independent.

Figure 8.  Normalized time series of indices of (a) AWB1, U₁ and ST1; (b) CWB2, U₂ and ST2; and (c) CWB3, U₃ and ST3 (see section 4 for the definition of these indices). Orange lines represent AWB1/CWB2/CWB3; blue lines represent U₁ U₂ U₃; and grey bars represent ST1/ST2/ST3.

The grey contours in Fig. 7 display the composite zonal wind anomalies at 300 hPa. When we consider the locations of positive/negative zonal wind anomalies and anomalous frequency of occurrence of AWB/CWB events, it can be seen that, in the upper two panels, the areas with decreased AWB coincide with the straddle point of the dipole of anomalous zonal wind. The zonal wind increased to the south and decreased to the north. In the bottom two panels, the areas with increased CWB are located at the negative lobe of the dipole of the zonal wind anomaly. The zonal wind is strengthened to the south of the increased CWB area.

In studies concerning the impacts of the two types of El-Niño on winter climate, the defined El-Niño years and the wintertime months are slightly different (Larkin and Harrison, 2005; Ashok et al., 2007; Yeh et al., 2009; Graf and Zanchettin, 2012). To verify how the difference in defining El-Niño events would affect our results, we carried out some sensitivity tests. It was found that slight changes in the selection of El-Niño years and wintertime months for composition do not significantly affect the major results reported above (data not shown).

In section 3, we demonstrated the anomalous pattern of mean flow, storm track and RWB during the two types of El-Niño. In both EP and CP El-Niño years, the mean flow anomalies (Figs. 3c and d) indicate a close conjunction with the storm track anomalies (Figs. 4a and b), and with RWB frequency anomalies (Fig. 7). To verify the robustness of the relationships shown in section 3, we defined three groups of indices for the anomalies of mean flow (upper-troposphere zonal wind), storm track and RWB. All the indices were defined in the regions with significantly large anomalies during the two types of El-Niño. The wintertime frequency of occurrence of AWB averages over the region (30°-50°N, 170°-130°W) (green box in Fig. 7a) was defined as AWB1. Similarly, the frequency of occurrence of CWB averages over the regions (35°-50°N, 120°-80°W) and (40°-55°N, 170°-140°W) (green boxes in Figs. 7c and d) were defined as CWB2 and CWB3, respectively. The zonal wind indices named U₁, U₂, and U₃, were defined as U₁=U_NI-U_SI, U₂=U_M2-U_N2-U_S2, , and U₃=U_S3-U_N3 at 300 hPa, respectively. U means the zonal wind. The above subscript N1 and S1 indicate the north and south blue boxes in Fig. 3c, respectively. N2, M2, and S2 correspond to the red boxes over the northern, middle and southern regions in North America in Fig. 3c, respectively. N3 and S3 indicate the red boxes over the northern and southern regions in Fig. 3d. The storm track indices of ST1, ST2, and ST3 were defined using the same formulas, boxes and levels as U₁, U₂, and U₃.

Figure 9.  Composite Z500 geopotential height anomalies (gpm) during the lifecycle of AWB events occurring in the North Pacific-North America sector in winter. Fields are centered on each wave breaking centroid (as described in section 2) prior to their inclusion in the composites. Abscissa and ordinate values are relative longitude and latitude, respectively. Longitude/latitude value of zero represents the centroid, which is marked by a cross in each figure. The other longitudes and latitudes is the distance relative to the centroid. (a) Average from 3 days before onset to 3 days after onset; (b) 3 days before onset; (c) onset; and (d) 3 days after onset.

Figure 10.  The same as Fig. 9, but for CWB. Abscissa and ordinate values are relative longitude and latitude, respectively.

Figure 8 shows the time series of the indices defined above. The correlation coefficients between U₁ and ST1, U₂ and ST2, and U₃ and ST3 are 0.85, 0.53 and 0.76, respectively. For mean flow and storm track indices defined over the North Pacific (U₁, and U₃ and ST1, ST3), the correlation coefficients are significantly high (0.85, and 0.76), verifying again the close conjunction of mean flow with storm track over the North Pacific. The three pairs of RWB and zonal wind indices—AWB1 and U₁, CWB2 and U₂, and CWB3 and U₃—also correlate well with each other, with correlation coefficients of 0.84, 0.82 and 0.77, respectively. All of the six correlation coefficients mentioned above exceed the threshold for the 99% confidence level (α_0.01=0.44), verifying the close relationship among them.

The close relationship between the mean flow and the storm track under two-type El-Niño has been discussed in section 3. Some statistical relationships between occurrence of RWB and large-scale circulation anomalies associated with two-type El-Niño have also been highlighted above. To reveal the basis for such statistical relationships, and the possible dynamical linkage between RWB and large-scale circulation, we produced lead-lag composites of multiple variables using daily fields with respect to all the RWB events occurring in the North Pacific-North America region. There were two reasons why we used all the RWB events occurring in the region rather than subsets related to the two types of El-Niño. Firstly, we wanted to try to figure out the general dynamical linkage between individual RWB events and large-scale circulation anomalies, especially Z500 and zonal wind. Secondly, those composites made only for El-Niño years would inevitably be contaminated by circulation anomalies associated with physical processes other than RWB, such as diabatic heating in the tropics. By clarifying such a linkage, we can reach an understanding of the dynamical linkage between the anomalous occurrence of RWB and large-scale circulation under the two types of El- Niño conditions.

Figure 9 shows the evolution of the Z500 anomaly during the life cycle of AWB events. Three days before the onset of AWB, a relatively weak high anomaly occurs slightly upstream of the centroid of breaking (Fig. 9b). Three days later (the onset day), the high anomaly is intensively strengthened and a weak low anomaly occurs to the northwest of the high (Fig. 9c). The extraordinarily strong high anomaly, with a central value of close to 100 gpm, is located at about the centroid of an AWB event. The amplitude of the positive anomaly indicates that blocking-like patterns may constantly form during the mature phase of AWB events. Thus, the jet may split when crossing the blocking-like pattern. Three days after the onset, the anomaly is weakened and the pattern rotates clockwise so that the high anomaly becomes slightly east-west elongated (Fig. 9d). The anomalous highs and lows all tilt forward during life cycles of AWB events, suggesting a poleward momentum flux (Hoskins et al., 1983). During a CWB event (Fig. 10), an anomalous low occurs upstream and an anomalous high occurs slightly downstream. At the onset time of an event (Fig. 10c), both the anomalous high and low are strengthened significantly and are of comparable amplitude. The anomalous high and low tilt backward so that westerly momentum is transported equatorward. Three days after the onset, both anomalous centers tilt farther backward so that they turn into a north-south dipole (Fig. 10d).

As addressed previously, the configuration of AWB/CWB events and related Z500 anomalies exhibited in Figs. 9 and 10 can be used to understand the relationship between anomalous occurrence of RWB (Fig. 7) and Z500 anomalies (Figs. 3a and b) associated with the two types of El-Niño. As shown in Fig. 9, the occurrence of AWB events is always companied by a monopole high anomaly at the centroid of breaking events. Accordingly, when AWB events occur less frequently over the North Pacific region under EP El-Niño conditions (Fig. 7a), local Z500 decreases harmoniously (Fig. 3a). In general, the Z500 anomaly related to CWB events is a dipole pattern of high anomaly to the northeast and low anomaly to the southwest (Fig. 10). When EP El-Niño occurs, the significant increased occurrence of CWB (Fig. 7c) is located at the straddle point of a northeast-southwest aligned dipole pattern for the Z500 anomaly (Fig. 3a) over North America. When CP El-Niño occurs, the significant increased occurrence of CWB (Fig. 7d) over the North Pacific coincides with the straddle point of the northeast-southwest aligned dipole pattern for the Z500 anomaly with the high and low anomalies to the east and to the south of the Aleutian Islands, respectively (Fig. 3b).

Figure 11.  Composite zonal wind anomalies (shading, m s^-1) and E anomalies (gray arrows, m² s^-2) averaged from 1 day before onset to 2 days after onset during (a) AWB and (b) CWB events. Abscissa and ordinate values are relative longitude and latitude, respectively. Time-latitude cross section for zonal wind (shading, m s^-1) and ▽`E anomalies (gray contours, m s^-2) during (c) AWB and (d) CWB events. Fields in (c) and (d) are averaged from -20°-20°of relative longitude. The interval between contours is 0.5×10^-5 m s^-2, with dashed contours being negative. The negative (positive) abscissa values indicate days before (after) the onset day. E was calculated using 2.5-8 day bandpass-filtered fields.

Figure 12.  Composite latitude-height cross sections for zonal wind anomalies (shading, m s^-1) averaged from 1 day before onset to 2 days after onset during (a) AWB and (b) CWB events. Abscissa and ordinate values are relative longitude and latitude, respectively. The fields are averaged from -5°-5° of relative longitude.

Figures 11a and b show the composite zonal wind anomaly related to RWB events. To shed light on the dynamical linkage between RWB and wind, composite E anomalies have also been plotted. For AWB events in Fig. 11a, the related positive and negative zonal wind anomalies are respectively located to the north and south of the centroid, indicating the zonal wind increases to the north and decreases to the south of the centroids of AWB events. The E anomalies emanate from the center of the positive zonal wind anomaly, curve clockwise and then converge into the negative center. Since the divergence (convergence) of E indicates the convergence (divergence) of momentum flux, such a pattern of anomalous E is consistent with the corresponding zonal wind anomaly. E is calculated using 2.5-8 day bandpass-filtered fields. Therefore, it represents the contribution of synoptic-scale disturbances to the establishment of such a zonal wind anomaly. For CWB, the westerly decreases around the centroid of wave-breaking events. To the south of the negative center is the positive zonal wind anomaly with comparable strength. To the north of the negative center is a weak westerly anomaly. The E anomaly diverges (converges) at the positive (negative) center of the zonal wind anomaly. Unlike AWB, which shifts the jet at the centroid poleward, CWB decreases the westerly near the centroid and increases the westerly to the south. Corresponding to the spatial patterns of geopotential height shown in Figs. 9 and 10, the patterns of zonal wind anomaly linked with AWB (CWB) also exhibit a forward (backward) tilt.

As briefly stated above, the configurations of AWB/CWB and related zonal wind shown in Fig. 11 can be used to understand the relation between the anomalous occurrence of AWB/CWB and the zonal wind anomalies under the two types of El-Niño conditions (Fig. 7). In the upper two panels of Fig. 7, where the number of occurrences of AWB linked with the two types of El-Niño both reduce, the zonal wind increases (decreases) to the south (north) of it, consistent with the general fact that AWB has a close conjunction with the increasing (decreasing) zonal wind to the north (south) of it, which is demonstrated well in Fig. 11a. In the bottom panels of Fig. 7, where CWB occurs more frequently, the zonal wind decreases there and increases to the south of it. This is also consistent with the connection of individual CWB events with the zonal wind anomalies in Fig. 11b.

To examine the lead-lag relationship between RWB and large-scale circulation, we present latitude-time cross sections of upper-tropospheric zonal wind and divergence of E anomalies in Figs. 11c and d for AWB and CWB events, respectively. The composites are averaged west 20° (denoted as -20°, hereafter) to east 20° (denotes east 20°, hereafter) longitude relatively centered at each wave breaking centroid. The composites only include RWB events with a lifetime longer than one day, and thus day 0-1 could be considered as the mature phase of the events, and day 0 could be seen as the onset day. The divergence/convergence of the E anomaly, representing anomalous eddy forcing, leads the zonal wind anomaly by about one day, and reaches a maximum around the onset of AWB/CWB events (Figs. 11c and d). In the case of AWB, the effective forcing on zonal wind begins about three days before the onset and the positive forcing attains a maximum one day before the onset. About half a day after the onset, the zonal wind anomaly peaks at the mature phase of wave-breaking events. The anomalous zonal wind pattern fades out four days after the onset. In the case of CWB, the anomalous eddy momentum forcing starts later than in the case of AWB, but the magnitude of anomalous forcing is about twice as large. Therefore, the resultant zonal wind anomaly is comparable to that associated with AWB during the mature phase, due to the much faster growth around and before the onset day. Since before the onset of RWB events there is weak but detectable anomalous eddy momentum forcing and anomalous zonal wind (Figs. 11c and d), it is hard to tell whether RWB drives zonal wind or the opposite is true, especially for AWB. Nevertheless, the prominent and persistent anomalous eddy momentum forcing and zonal wind anomalies during and around the mature phase of RWB truly demonstrate the strong dynamical linkage between RWB and mean flow. Furthermore, they indicate that anomalous RWB events act to reinforce and invigorate the circulation anomalies over the North Pacific-North America region that occur during the two types of El-Niño.

The above analysis has demonstrated the dynamical relationship between RWB and zonal wind in the upper level. The composite vertical structures of zonal wind anomalies for RWB are plotted in Fig. 12. It illustrates the eddy-mean flow cooperation in the whole troposphere. The composite profiles for AWB and CWB events are both dominated by equivalent barotropic structure, with the maximum anomaly located around 300 hPa. For AWB, the anomalies tilt slightly poleward with the increase of height, especially for the positive lobe. For CWB, the negative anomaly tilts slightly equatorward with the increase of height. During AWB events, baroclinicity would increase (decrease) to the north (south) of the centroid. During CWB events vertical wind shear (baroclinicity) would decrease near the centroid and increase to the south of it. Such connection between individual RWB events and wind shear bears close resemblance with that between the anomalous RWB occurrence (Fig. 7) and the winter-mean composite baroclinicity, as well as the storm track activity anomalies during the two types of El-Niño (Fig. 4). Thus, the dynamical linkage between RWB and zonal wind exists not only in the upper troposphere, but also in the mid-lower troposphere.

In the present study, we have documented the dynamical connections between storm track and mean flow under two types of El-Niño, through investigating the accompanying RWB occurrence anomalies. During EP El-Niño, the North Pacific jet stream is strengthened downstream; meanwhile, the zonal wind anomaly over continental North America shows a tripole structure, corresponding to the PNA pattern. During CP El-Niño, the jet over the North Pacific is nudged southward slightly. Under both types of El-Niño conditions, the storm track anomalies are largely determined by baroclinicity anomalies that are dominated by local vertical wind shear. Therefore, the anomalous distributions of baroclinicity and storm track are consistent with those of upper-level zonal wind for both EP and CP-Niño, indicating the impacts of mean flow on storm track variability.

The total number of AWB events tends to decrease over the North Pacific to the east of the dateline, and that of CWB tends to increase over continental North America under EP El-Niño cases. The anomalies of wave breaking related to CP El-Niño exhibit an above-normal number of CWB events over the North Pacific north of 40°N, and a weak decreasing of AWB events over the central North Pacific. Identical relations between large-scale circulation anomalies and anomalous RWB occurrence were found for both types of El-Niño. When AWB events are below-normal, Z500 decreases locally, and zonal wind is strengthened (weakened) to the south (north). When CWB events are above-normal, Z500 increases (decreases) to the northeast (southwest), and zonal wind is weakened locally while strengthened to the south. Such relations between mean flow and RWB are shown to be robust for both types of El-Niño.

To understand the connection between RWB occurrence and anomalous mean flow pattern under El-Niño conditions, daily composites for Z500 and zonal wind anomalies were produced with respect to all AWB/CWB events. We found that AWB events are always accompanied by a monopole high anomaly. Therefore, when AWB occurs less frequently over the North Pacific during EP El-Niño, Z500 significantly decreases there and zonal wind increases (decreases) to the south (north). CWB events are always related to a northeast-southwest dipole Z500 anomaly. As such, when CWB events occur more frequently over the North Pacific under CP El-Niño conditions and over North America under EP El-Niño conditions, Z500 increases (decreases) to the northeast (southeast) and zonal wind is weakened locally and strengthened to the south. The vertical composites of zonal wind anomalies show that the dynamical linkage between RWB and zonal wind exists in both the upper troposphere and the mid-lower troposphere.

Composites of E anomalies and lead-lag composites of ▽`E anomalies with respect to RWB show that RWB events act to maintain and reinforce the circulation anomalies that occur during the two types of El-Niño over the North Pacific-North America region. From a spatial aspect, the positive (negative) eddy momentum forcing overlaps with increasing (decreasing) zonal wind. From a temporal aspect, the maximum anomalous eddy momentum forcing leads the maximum zonal wind anomaly by about one day, both being prominent around the mature phase of RWB events, especially for CWB cases.

The zonal wind anomaly and RWB also cooperate with each other, suggesting the possible influence of mean flow on the occurrence of RWB. (Riviere, 2009) proposed that the poleward (equatorward) shift of the jet would favor the occurrence of AWB (CWB). This idea is partly supported by our results. From what has been stated above, we propose that RWB cannot be considered as an independent forcing on mean flow, since both RWB and related large-scale circulation anomalies develop, peak and decay synchronously, as shown in our study. There are two possibilities concerning the relationship between RWB and large-scale circulation: one is that a two-way impact may exist between certain flow anomaly patterns and the occurrence of RWB; and another is that RWB is only an alternative description of such mean flow anomalies, rather than the cause, as suggested by (Luo et al., 2008).