SEARCH

SEARCH BY CITATION

Keywords:

  • low-frequency variability;
  • breaking waves;
  • tropical convection;
  • optimal growth

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Composite PNA anomaly evolution
  6. 4. Synoptic description of the PNA evolution
  7. 5. Conclusions and discussion
  8. Acknowledgements
  9. References

In this study, we use various diagnostic techniques to investigate the synoptic evolution of the Pacific–North American teleconnection pattern (PNA). National Center for Environment Prediction/National Center for Atmospheric Research reanalysis data are used. These data cover the years 1948–2008 for the months of November–March. It is found that the positive PNA is initiated by enhanced convection over the western tropical Pacific and weakened convection over the tropical Indian Ocean. The excitation of the negative PNA exhibits opposite features. For both phases, the response to tropical convection excites a small-amplitude PNA about 8–12 days prior to the pattern attaining its maximum amplitude. This is followed by slow, steady growth for about 5 days, after which driving by synoptic scale waves, via their eddy vorticity flux, together with stationary eddy advection lead to much more rapid growth and the establishment of the full PNA. For the positive PNA, the synoptic scale waves propagate eastward into the midlatitude northeastern Pacific, where they are observed to undergo cyclonic wave breaking. For the negative PNA, the synoptic scale waves first amplify over the midlatitude northeastern Pacific and then propagate equatorward into the Subtropics where they undergo anticyclonic wave breaking. Once established, for both phases, the PNA appears to be maintained through a positive feedback that involves a succession of wave breakings.

These results suggest that preconditioning may play an important role in the formation of the PNA. For the positive PNA, in its early development, the strengthening and eastward extension of the subtropical jet result in an increase in the cyclonic shear and a decrease in the meridional potential vorticity gradient, features that are known to favour cyclonic wave breaking. For the negative PNA, opposite changes were observed for the background flow, which favour equatorward wave propagation and anticyclonic wave breaking. The role of optimal growth is also discussed. Our results also suggest that the PNA is potentially predictable 1–2 weeks in advance. Copyright © 2011 Royal Meteorological Society


1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Composite PNA anomaly evolution
  6. 4. Synoptic description of the PNA evolution
  7. 5. Conclusions and discussion
  8. Acknowledgements
  9. References

The Pacific–North American teleconnection pattern (PNA) is one of the leading patterns of Northern Hemisphere midlatitude variability (Horel and Wallace, 1981; Wallace and Gutzler, 1981; Barnston and Livezey, 1987); it influences midlatitude predictability (Palmer, 1988; Sheng, 2002; Johnson and Feldstein, 2010) and Arctic sea ice (L'Heureux etal., 2008). The PNA can be characterized as a Rossby wave train with four centres of action: two of one sign located over the subtropical northeastern Pacific and northwestern North America and two of the opposite sign centred over the Gulf of Alaska and the southeastern USA. While the PNA exhibits substantial variability on interannual and interdecadal time-scales (Wallace and Gutzler, 1981; Overland etal., 1999) its intrinsic time-scale is about two weeks (Feldstein, 2000, 2002; Cash and Lee, 2001; Franzke and Feldstein, 2005).

Several different mechanisms have been proposed to account for the growth and maintenance of the PNA. The emerging consensus is that the following three processes are most important. These are (1) linear dispersion of a Rossby wave excited by tropical heating (Hoskins and Karoly, 1981; Simmons, 1982; Sardeshmukh and Hoskins, 1988; Branstator, 1985a, 1985b; Jin and Hoskins, 1995), (2) barotropic amplification due to the zonal asymmetry of the climatological flow (Frederiksen, 1983; Simmons etal., 1983; Branstator, 1990, 1992; Feldstein, 2002) and (3) driving by synoptic scale transient eddy vorticity fluxes (Egger and Schilling, 1983; Lau, 1988; Dole and Black, 1990; Schubert and Park, 1991; Branstator, 1992; Black and Dole, 1993; Ting and Lau, 1993; Higgins and Schubert, 1994; Feldstein, 2002; Franzke and Feldstein, 2005; Orlanski, 2003, 2005).

Our aim in this study is to investigate the relationship between tropical heating and the synoptic scale eddy driving of the PNA. For this purpose, we will examine PNA evolution from a synoptic perspective. We will focus on the impact of tropical convection on the strength and location of the subtropical jet and the subsequent influence of the subtropical jet on the occurrence of wave breaking. Breaking Rossby waves are typically identified by the reversal of the normally equatorward meridional potential temperature gradient. This investigation of the synoptic PNA evolution, particularly that of wave breaking, will allow us to present a simple kinematic description of the link between tropical convection and the synoptic scale eddy driving of the PNA.

An association between wave breaking and the PNA has recently been found by Martius etal. (2007). They found that relative to climatology the positive PNA phase coincides with an increase in the frequency of cyclonic wave breaking over the northeastern Pacific, and the negative PNA phase with both a reduction in the frequency of cyclonic wave breaking over the northeastern Pacific and an increase in anticyclonic wave breaking further to the south. However, as the focus of Martius etal. (2007) was on the climatology of wave breaking, they did not examine the timing, intensity and spatial structure of the wave breaking, nor its impact on the development and maintenance of the PNA.

Several studies have also addressed the question of what factors influence the phase of the PNA. For example, observational studies have found that the positive PNA phase is associated with enhanced convection over the western tropical Pacific and suppressed convection over the tropical Indian Ocean, whereas the negative PNA phase shows the opposite features (Higgins and Mo, 1997; Schubert and Park, 1991; Matthews etal., 2004; Mori and Watanabe, 2008; Johnson and Feldstein, 2010). These studies all find that the phase of the PNA is closely linked to the 30–60 day, eastward-propagating Madden–Julian Oscillation (MJO), the dominant pattern of intraseasonal variability within the Tropics (Madden and Julian, 1971, 1972). In this study, we will also focus on the relationship between tropical convection, wave breaking and the phase of the PNA.

The relationship between synoptic scale eddy energy and the PNA phase has been examined with a numerical model by Orlanski (2003, 2005). In those studies, he found that the positive PNA, which is associated with greater baroclinicity, coincides with more energetic eddies and cyclonic wave breaking. Opposite features were found for the negative PNA. He also showed that a moister model atmosphere results in more energetic eddies and cyclonic wave breaking. Consistently, over the North Atlantic, Riviere and Orlanski (2007) find that cyclonic wave breaking, which excites the negative phase of the North Atlantic Oscillation (NAO), is associated with stronger poleward moisture fluxes. However, although we will find that the PNA exhibits the same wave-breaking features as those in Orlanski (2003, 2005), we do not observe the same relationship between the amplitude of the eddies and the type of wave breaking. For the positive PNA, we found that this phase coincides with less energetic synoptic scale eddies over the northeastern Pacific, and vice versa for the negative PNA. A detailed examination of why this may be the case is beyond the scope of this study. In section 2, the data and methodology are presented. This is followed in section 3 by an examination of the composite evolution of the PNA and in section 4 by a detailed synoptic description and an analysis of the dynamical mechanisms of the PNA evolution. Conclusions are given in section 5.

2. Data and methodology

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Composite PNA anomaly evolution
  6. 4. Synoptic description of the PNA evolution
  7. 5. Conclusions and discussion
  8. Acknowledgements
  9. References

In this study, we analyse potential temperature, θ, on the two potential-vorticity-unit (PVU) surface (1 PVU = 10−6 m2 s−1 K kg−1), a surface that closely follows the tropopause poleward of about 25°N. This quantity serves as our ‘weather map’. Since θ is almost conserved on time-scales of several days, it can reveal dynamical processes such as wave breaking (Thorncroft etal., 1993; Lee and Feldstein, 1996; Morgan and Nielsen-Gammon, 1998; Benedict etal., 2004; Franzke etal., 2004; Feldstein and Franzke, 2006; Martius etal., 2007; Woollings et al., 2008).

We utilize daily (0000 UTC) National Center for Environment Prediction (NCEP)/National Center for Atmospheric Research (NCAR) reanalysis wind and temperature data (Kalnay etal., 1996; Kistler etal., 2001). These data cover the years 1948–2008 for the months of November–March (NDJFM). The calculations are performed on a grid corresponding to a rhomboidal 30 truncation. The value of θ on the 2 PVU surface is found by linear interpolation starting from the uppermost vertical level. A smoothed mean annual cycle is computed by taking the calendar mean for each day and applying a 20 day low-pass filter.

In this study, we use the PNA index of Franzke and Feldstein (2005) as a measure of the amplitude of the PNA pattern. The onset day for the positive (negative) phase of an event is defined as the first day on which the PNA index exceeds 1.25 (−1.25) standard deviations and subsequently stays above (below) that threshold for at least 5 consecutive days. The composites are centred around the day with the largest amplitude, hereafter denoted as the lag 0 day. We specify an event as corresponding to the 21 day period centred around the lag 0 day. Events that exhibit non-monotonic growth and decay, in which the index drops below the threshold and then rises back above, are discarded, as well as events that extend into non-NDJFM months. With this approach, 58 positive and 57 negative events are identified, the wave breaking characteristics of which we will investigate in section 4.

As in many studies, we use outgoing long wave radiation (OLR) data as a proxy for the deep convection in the Tropics. We use data provided by the National Oceanic and Atmospheric Administration for the years 1974–2008 on a 2.5° longitude × 2.5° latitude grid.

3. Composite PNA anomaly evolution

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Composite PNA anomaly evolution
  6. 4. Synoptic description of the PNA evolution
  7. 5. Conclusions and discussion
  8. Acknowledgements
  9. References

For both PNA phases, we display in Figure 1 composites of the anomalous (deviation from the mean annual cycle) 300 hPa stream function field spanning the time interval from lag −12 days to lag +8 days. The full PNA quadrupole anomaly pattern (a meridional dipole over the northeastern Pacific along with centres of action over western Canada and the southeastern USA) is most clear at lag 0 days, when the amplitude of the PNA is defined to attain its largest amplitude. For both phases, evidence for the occurrence of the PNA is apparent as early as lag −12 days, when weak anomalies of opposite sign are visible over the northeastern Pacific (Figure 1). Over the following 8 days, these two anomalies propagate slowly westward and continue to amplify. The two PNA anomalies over North America can be seen at lag −8 days, after which they also undergo steady growth. The most rapid growth of the PNA is observed to take place between lag −4 and lag 0 days. This is followed by the decay of each of the PNA anomalies over the following 8 days. Throughout the composite PNA life cycle, as indicated by the shading in Figure 1, each of the anomalies described above is found to be statistically significant above the 95% confidence level.

thumbnail image

Figure 1. Composites of positive (right column) and negative (left column) PNA events of 300 hPa stream function. The contour interval is 3 × 106 m2 s−1. Shading denotes anomalies that are significant at the 95% confidence level.

Download figure to PowerPoint

The lagged composite PNA index is shown in Figure 2. Consistent with Figure 1, for both phases, the PNA index has a small amplitude at lag −12 days and the full PNA life cycle takes between 2 and 3 weeks to complete. This is consistent with the results of Dole and Black (1990), Feldstein (2000, 2002), Cash and Lee (2001), and Franzke and Feldstein (2005). Furthermore, because the composite PNA index is very close to zero at both the beginning and the end of the life cycles, the intrinsic dynamics of the PNA must be an intraseasonal, not an interannual, time-scale phenomenon. Additional support for this view can be found in the teleconnection study of Johnson and Feldstein (2010). They show that each of the spatial patterns within the North Pacific continuum exhibits an e-folding time-scale of 5–10 days. These patterns include a range of positive and negative phase PNA-like patterns. This indicates that the intraseasonal time-scale property exhibited in Figure 2 is insensitive to the methodology used to define the PNA.

thumbnail image

Figure 2. Composites of the PNA index for (a) the positive and (b) the negative phase.

Download figure to PowerPoint

The difference between the positive and negative phase 2 PVU potential temperature composites shows a similar spatial pattern to the stream function anomalies (Figure 3). This indicates that 2 PVU potential temperature is also capable of describing the PNA life cycle.

thumbnail image

Figure 3. A composite of PNA events at lag 0 days of the potential temperature on the 2 PVU surface (difference between the positive and negative phases). The contour interval for the 2 PVU potential temperature is 2 K.

Download figure to PowerPoint

4. Synoptic description of the PNA evolution

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Composite PNA anomaly evolution
  6. 4. Synoptic description of the PNA evolution
  7. 5. Conclusions and discussion
  8. Acknowledgements
  9. References

In this section, we examine the composite life cycle for the two PNA phases. We also present one typical case for each PNA phase.

4.1. Positive-phase composite evolution

We first describe the composite evolution of the positive PNA phase (Figure 4). The anomalies relevant for the PNA evolution are statistically significant above the 95% confidence level (not shown). As can be seen, the PNA anomaly centres over the midlatitude northeast Pacific and western Canada are present in the 2 PVU potential temperature field at lag −8 days and by lag −4 days all four PNA anomaly centres can be observed. These results for the 2 PVU potential temperature field share most of the same features as those for the 300 hPa stream function field (Figure 1). When the anomalies first arise, their geographic centres are located about 10°–20° in latitude and/or longitude away from the location that they will attain when the full positive PNA forms. This is most apparent for the negative anomaly over the northeastern Pacific. Over the following 4 days, all four anomalies continue to strengthen. The occurrence of these anomalies corresponds to the subtropical jet undergoing a strengthening and eastward extension to about 150°W (Figure 5), the formation of a relatively deep trough over the midlatitude northeastern Pacific, an eastward shift in the longitude of the climatological ridge from the northeastern Pacific to western Canada and a deepening of the climatological trough over eastern North America. In terms of its influence on the 2 PVU potential temperature field, the establishment of the PNA pattern results in a widespread cooling over the northeastern Pacific, a warming over western Canada and Alaska and a cooling over the eastern USA.

thumbnail image

Figure 4. A composite of unfiltered potential temperature fields on the 2 PVU surface from all positive-phase PNA events. The contour interval is 5 K and light (dark) shading denotes anomalies with amplitudes greater (smaller) than 2.5 K (−2.5 K). The 285 K and 300 K isolines are emboldened. The unit arrow corresponds to a wind speed of 50 m s−1.

Download figure to PowerPoint

thumbnail image

Figure 5. Composites of 300 hPa zonal wind averaged over the 7 day interval from lag −10 through −4 days, for (a) the positive PNA phase and (b) the negative PNA phase. The contour interval is 5 m s−1. Zonal wind values above 40 m s−1 are lightly shaded and those above 50 m s−1 are heavily shaded, and solid (dashed) lines indicate positive (negative) values.

Download figure to PowerPoint

At the time of the initial PNA development, i.e., from lag −8 to lag −4 days, a synoptic scale trough off the east coast of Asia, near 140°E, intensifies and breaks cyclonically, forming a cutoff low (Figure 4). This is followed by eastward propagation of the trough toward the exit region of the subtropical jet, near 150°W. Similar synoptic eddy evolution has been observed by Dole and Black (1990) and Black and Dole (1993) for a teleconnection pattern that resembles the positive PNA phase. At lag −2 days, the synoptic scale trough undergoes a second episode of even more intense cyclonic wave breaking. This wave breaking coincides with a merging of the eastward-propagating synoptic scale trough with the weak trough associated with the PNA pattern that had already started to form between lag −12 and lag −8 days (Figures 1 and 4). This leads to an intensification of the PNA anomalies over the midlatitude northeastern Pacific and western Canada. This picture of cyclonic wave breaking over the northeast Pacific is consistent with the findings of Martius etal. (2007), as described in the Introduction. During the PNA decay, the breaking wave structure is maintained through to lag +4 days. Thus, cyclonic wave breaking plays a role in both the development of the PNA pattern and its maintenance. It is worth noting that the second cyclonic wave-breaking episode takes place on the poleward side of the subtropical jet, in a location where the cyclonic wind shear has been strengthened and the meridional potential vorticity gradient has been weakened. Consistently, Shapiro etal. (2001) find that cyclonic wave breaking occurs more frequently in a background wind field with enhanced cyclonic shear, and Postel and Hitchman (1999), Abatzoglou and Magnusdottir (2006), Scott and Cammas (2002) and Gong etal. (2010) all find that the strongest wave breaking tends to coincide with local minima in the background meridional potential vorticity gradient.

The picture suggested by Figures 1 and 4 is that the positive PNA is first triggered as much as 12 days before it attains its maximum amplitude, but that the most rapid growth of the PNA, which occurs between lag −4 and lag −1 days (see Figure 2), is associated with the second cyclonic wave-breaking event described above.

4.2. Negative-phase composite evolution

The composite evolution of the negative PNA phase is illustrated in Figure 6. As can be seen, two of the PNA anomalies are present at lag −8 days, followed by all four anomalies by lag −4 days. Again, as for the positive PNA, the geographic centres of these anomalies are located about 10°–20° in latitude and/or longitude away from the position that they will attain when the full negative PNA forms. The negative PNA amplifies most rapidly over the next several days. The presence of these four anomalies corresponds to a subtropical jet that is noticeably weaker and retracted westward beyond the Date Line (Figure 5), a westward displacement of the climatological ridge toward the north central Pacific, the formation of a weak trough over central Canada and the replacement of the climatological trough over eastern North America by a weak ridge over the southeastern USA. In terms of the 2 PVU potential temperature field, the establishment of the negative PNA phase results in changes that are opposite from those of the positive PNA, with warming over the northeastern Pacific and the eastern USA and cooling over western Canada.

thumbnail image

Figure 6. A composite of unfiltered potential temperature on the 2 PVU surface from all negative phase PNA events. The contour interval is 5 K and light (dark) shading denotes anomalies with amplitudes greater (smaller) than 2.5 K (−2.5 K). The 285 K and 300 K isolines are emboldened. The unit arrow corresponds to a wind speed of 50 m s−1.

Download figure to PowerPoint

During the initial growth of the negative PNA phase (lag −8 days), a positive anomaly is present off the east coast of Asia. This feature, which is opposite in sign to that for the positive PNA, indicates the occurrence of a weak trough upstream of the PNA region as the negative PNA is developing. During the next 6 days this positive anomaly expands eastward and merges with the positive anomaly over the northeastern Pacific associated with the negative PNA, which was already present as early as lag −12 to lag −8 days (Figures 1 and 6). This results in a further intensification of the ridge over the North Pacific. In contrast to the composite positive PNA, which exhibits strong cyclonic wave breaking over the midlatitude northeastern Pacific, the composite negative PNA shows no cyclonic wave breaking and just a hint that anticyclonic wave breaking may be taking place at lag 0 and +2 days over the subtropical northeastern Pacific near 20°N and 140°W. The anticyclonic wave breaking at these lags is alluded to by the equatorward extension of the 2 PVU potential temperature contours in this region. As will be discussed in section 4.6, case-to-case variability in the wave breaking longitude can obscure anticyclonic wave breaking in the negative PNA composite.

This picture of an absence of cyclonic wave breaking over the midlatitude northeastern Pacific together with anticyclonic wave breaking over the subtropical northeastern Pacific is also consistent with the results of Martius etal. (2007). The lack of cyclonic wave breaking for the composite negative PNA can also be understood in terms of a reduction in the cyclonic shear and an increase in the meridional potential vorticity gradient on the poleward side of the subtropical jet over the northeastern Pacific, opposite features from those described for the positive PNA, where strong cyclonic wave breaking in the composite was observed (Figure 4).

The occurrence of anticyclonic wave breaking near 20°N and 140°W for the negative PNA, and similarly the absence of this type of wave breaking for the positive PNA, can be understood as most likely arising from changes in the strength and zonal extent of the subtropical jet and the corresponding meridional potential vorticity gradient in this region. For example, in the MJO study of Matthews and Kiladis (1999), it was found that when the subtropical jet is strengthened and extended eastward, resulting in a larger meridional potential vorticity gradient in the subtropical northeast Pacific, high-frequency eddies propagate along the eastward extended subtropical waveguide (Hoskins and Ambrizzi, 1993), and do not propagate equatorward into the Tropics. In contrast, when the subtropical jet is weakened and retracted toward the west, thus leaving behind a smaller meridional potential vorticity gradient, Matthews and Kiladis (1999) found that high-frequency eddies readily propagate equatorward into the deep Tropics. This behaviour is consistent with quasi-geostrophic theory, in which it is found that waves tend to propagate in an equatorward direction when the zonal wind is weak and the meridional potential vorticity gradient is dominated by the planetary vorticity gradient. As these high-frequency waves reach lower latitudes, they approach their critical latitude (at which the phase speed of the wave equals the background zonal wind speed) and anticyclonic wave breaking can readily take place (Held and Phillips, 1987; Feldstein and Held, 1989; Thorncroft etal., 1993).

In summary, as for the positive PNA, the picture presented by Figures 1 and 6 suggests that the negative PNA is first excited about 12 days prior to it attaining its maximum amplitude and that the most rapid growth occurs between lag −4 and lag −1 days (see Figure 2). This rapid growth is associated with a suppression of cyclonic wave breaking over the midlatitude northeastern Pacific and the excitation of anticyclonic wave breaking over the subtropical northeastern Pacific.

4.3. Dynamical forcing processes of the PNA

To highlight the dominant processes that account for the growth and maintenance of the PNA, we calculate various terms in the stream function tendency equation (see Feldstein (2002), Franzke and Feldstein (2005), and Mori and Watanabe (2008) for more details and definitions). First, we examine the stationary eddy advection term:

  • equation image(1)

where u is the wind vector, ζ the relative vorticity, the superscript ‘S’ denotes the mean annual cycle, ‘L’ a 10 day low-pass filter with the mean annual cycle subtracted, ‘*’ indicates the deviation from the zonal mean and ‘C’ the climatological NDJFM mean. Consistent with previous studies, stationary eddy advection is found to be a key contributor toward the growth and maintenance of the PNA (Feldstein, 2002; Franzke and Feldstein, 2005; Mori and Watanabe, 2008). Figure 7 shows that the stationary eddy advection projects much more strongly on to the subtropical northeastern Pacific PNA anomaly than it does onto the corresponding midlatitude anomaly. This difference in the strength of the projections is most evident as the PNA is growing.

thumbnail image

Figure 7. Composites at lag −8 days (upper row) and lag 0 days (lower row) of 300 hPa stationary eddy advection (contour interval: 10 m2 s−2). Shading indicates the PNA composite patterns at lag 0 days.

Download figure to PowerPoint

We next examine the high-frequency eddy vorticity flux term:

  • equation image(2)

where the superscript ‘H’ denotes a 10 day high-pass filter. In Figure 8, for both PNA phases, it is shown that the driving by the high-frequency eddy vorticity fluxes takes on a monopole spatial structure, projecting strongly on to the PNA anomaly over the midlatitude northeastern Pacific. Compared with stationary eddy advection (Figure 7), it can be seen that as the PNA is growing the high-frequency eddy driving is dominant, whereas when the PNA is near its maximum amplitude the impact of these two terms over the northeastern Pacific is comparable. Thus the high-frequency eddies, which essentially correspond to synoptic scale eddies, also have an important impact on the growth and maintenance of the PNA. This monopole spatial structure for the eddy driving of the PNA contrasts with that for the NAO, where the eddy driving is found to take on a dipole spatial structure (Feldstein, 2003; Jin etal., 2006; Luo et al., 2007, 2008a). A plausible explanation for these differences between the NAO and PNA involves the latitude of the anomalies. For the NAO, the more southern member of the dipole is located in midlatitudes, where eddy driving is strong, whereas for the PNA the southernmost anomaly is located in the Tropics, where eddy driving is typically much weaker.

thumbnail image

Figure 8. Composites at lag −8 days (upper row) and lag 0 days (lower row) of 300 hPa high-frequency eddy forcing (contour interval: 5 m2 s−2). Shading indicates the PNA composite patterns at lag 0 days.

Download figure to PowerPoint

If we compare the high-frequency eddy driving for the two PNA phases over the midlatitude northeastern Pacific with that of the climatology (Figure 9), it can be seen that the extrema in the high-frequency eddy driving for the two PNA phases are located very close to the node in the climatological eddy driving. This indicates that the total eddy driving (climatology plus anomaly) is of similar strength but of opposite sign for the two PNA phases. Furthermore, these results show that the total high-frequency eddy driving of the positive (negative) PNA in this region is cyclonic (anticyclonic). For the positive PNA, the cyclonic driving is clear from Figure 4 (at lag −2 days) because of the strong northwest/southeast tilt of the cyclonically breaking eddy. In contrast, for the negative PNA (see Figure 6 at lag −2 days) there appears to be no horizontal tilt in the composite field. Our observation of strong anticyclonic driving for this phase suggests that there may be northeast/southwest tilted, high-frequency eddies present over the midlatitude northeastern Pacific, but the absence of these eddies in Figure 6 is likely due to case-to-case variability in the longitude of the eddies. As discussed in section 4.6, this does appear to be the case.

thumbnail image

Figure 9. Climatological 300 hPa high-frequency eddy forcing. The contour interval is 5 m2 s−2.

Download figure to PowerPoint

Further insight into dynamical processes can be obtained by examining the 300 hPa anomalous high-frequency eddy kinetic energy and 850 hPa high-frequency meridional heat flux (Figure 10). The largest anomalies in the 300 hPa high-frequency eddy kinetic energy are observed to occur over the midlatitude northeastern Pacific for both PNA phases. For all negative lags displayed, it can be seen that the negative PNA is associated with an increase in high-frequency eddy kinetic energy and vice versa for the positive PNA. This relationship between the eddy energy and the phase of the PNA is opposite from those found in the modelling study of Orlanski (2003, 2005), as discussed in the Introduction. These anomalies in the 300 hPa high-frequency eddy kinetic energy resemble those of the 850 hPa high-frequency meridional heat flux (Figure 10). This implies that the increase in high-frequency eddy kinetic energy associated with the negative PNA arises from enhanced baroclinic growth and the reduced high-frequency eddy kinetic energy of the positive PNA is due to reduced baroclinic growth. To some extent this result is to be expected, since over the northeastern Pacific (near 60°N) the positive (negative) PNA coincides with weakened (strengthened) westerlies (Figure 1).

thumbnail image

Figure 10. Composites at lag −2 days of positive PNA events of 850 hPa high-frequency meridional heat flux (contour interval: 2 m2 s−2; annual cycle of heat flux is subtracted) and 300 hPa high-frequency eddy kinetic energy (contour interval: 1 m2 s−2; annual cycle of kinetic energy is subtracted).

Download figure to PowerPoint

Another interesting feature in Figure 10 is the occurrence of 300 hPa high-frequency eddy kinetic energy and 850 hPa high-frequency meridional heat flux anomalies over central Siberia at negative lags. These anomalies take on opposite signs for the positive and negative PNA. This indicates that the synoptic features that occur along the east coast of Asia as the PNA is developing, as discussed in the previous two subsections, are preceded by synoptic scale anomalies that are located even further upstream. However, an examination of the relationship between these synoptic scale anomalies over central Siberia and the subsequent PNA evolution is beyond the scope of this study.

4.4. Impact of tropical convection

To illustrate briefly the relationship between tropical convection and PNA development, we show the composite OLR field averaged over the time period from lag −16 to lag −10 days (Figure 11). We select this particular time period because both modelling studies (Hoskins and Karoly, 1981; Jin and Hoskins, 1995; Matthews etal., 2004) and observational studies (Kiladis and Weickmann, 1992; Higgins and Mo, 1997) show that a large amplitude middle- and high-latitude response occurs about 10 days after the onset of tropical convection. As can be seen, the positive PNA phase is associated with increased convection (a negative OLR anomaly) over the western tropical Pacific and reduced convection (a positive OLR anomaly) over the tropical Indian Ocean. The negative PNA shows the opposite features. As discussed in the Introduction, these features resemble those of many previous observational studies of the PNA (Higgins and Mo, 1997; Schubert and Park, 1991; Matthews etal., 2004; Mori and Watanabe, 2008; Johnson and Feldstein, 2010).

thumbnail image

Figure 11. Composites of anomalous OLR for (a) positive and (b) negative PNA events averaged over the 7 day interval from lags −16 to −10 days. Zonal wave numbers larger than 7 have been neglected for plotting. The contour interval is 7 W m−2 and solid (dashed) lines indicate positive (negative) values. The shading indicates anomalies that are statistically significant above the 90% confidence level.

Download figure to PowerPoint

To evaluate whether the PNA corresponds to a robust response to the convective anomalies in Figure 11, we also calculate lagged composites of 300 hPa stream function (Figure 12) based on OLR indices that correspond to the amplitude of the two spatial patterns in Figure 11. These indices are calculated by projecting the daily OLR field on to the positive and negative phase OLR composite patterns (Figure 11). For those days on which these indices exceed one standard deviation we perform lagged composite calculations. As can be seen, for the positive PNA at lag −4 days two North Pacific anomalies are observed, by lag 0 a third anomaly is present over northwestern Canada and by lag +6 days the full positive PNA pattern is established. For the negative PNA, none of the PNA anomalies is present at lag −4 days, but by lag 0 three of the negative PNA anomalies are present and by lag +6 days the full negative PNA pattern is in place. This evolution of the stream function fields indicates that the quasi-stationary wave train triggered by the OLR anomaly of the negative phase of the PNA starts from a region of suppressed convection in the western Pacific. The observation that both the positive and negative PNA patterns are established at positive lags provides further support that the circulation anomalies associated with PNA arise as a response to tropical convection.

thumbnail image

Figure 12. Composites of anomalous 300 hPa stream function based on OLR indices (see text for details). The contour interval is 3 × 106 m2 s−1 and solid (dashed) lines indicate positive (negative) values. The shading indicates anomalies that are statistically significant above the 90% confidence level.

Download figure to PowerPoint

Additional evidence of the impact of tropical convection on the PNA is given by the joint probability density function (PDF) between the two OLR indices, discussed above, and high-frequency eddy kinetic energy (see Figure 13). For this calculation, we use the OLR indices and a high-frequency eddy kinetic energy index for each day within the lag −20 to lag +20 day interval for each PNA event. The kinetic energy index is defined as the spatial average of high-frequency eddy kinetic energy over the region 120°E–120°W, 30°N–60°N. Our results are found to be insensitive to the particular choice of boundaries for this region. In this PDF, the OLR indices are specified to lead the kinetic energy index by 5 days. For our discussion of Figure 13 we focus on positive values of the OLR indices, as it is these values that describe the tropical convection associated with PNA events. As can be seen from Figure 13(c), for positive OLR index values the difference PDF takes on a dipole structure, indicating that positive PNA events are more frequently associated with relatively small high-frequency eddy kinetic energy and negative PNA events with relatively large high-frequency eddy kinetic energy. These results are consistent with composites of high-frequency eddy kinetic energy shown in Figure 10. This finding suggests that the response to tropical convection associated with both PNA phases results in an enhancement of the North Pacific synoptic scale eddy energy for the negative PNA and a decline for the positive PNA.

thumbnail image

Figure 13. Joint PDFs of the two OLR indices and kinetic energy with the OLR index leading the kinetic energy index by 5 days: (a) positive PNA phase, (b) negative PNA phase and (c) difference between positive and negative PNA phase.

Download figure to PowerPoint

4.5. Positive-phase example

A typical example for the positive PNA phase is presented in Figure 14, which shows the event with a lag 0 day of January 14 1992. We start by looking at the total flow field at lag 0 days, verifying that this pattern has strong PNA-like features. This will be followed by an examination of both the development and the decay of this pattern. As we can see, at lag 0 days there is a strong trough centred over the northeast Pacific near 150°W, a large-amplitude ridge located over western North America and another strong trough observed over eastern North America. This pattern shows the dominant features of the positive PNA as illustrated in Figure 4.

thumbnail image

Figure 14. Unfiltered potential temperature fields on the 2 PVU surface for the positive PNA event from the 1991/1992 winter. Lag 0 corresponds to 14 January 1992. The contour interval is 5 K and shading is for values below 315 K. The wind vectors are denoted by arrows. The unit arrow corresponds to a wind speed of 50 m s−1. The corresponding lag days are shown above each panel. The letters that denote troughs and ridges are located directly below in the region where there is no shading.

Download figure to PowerPoint

We start by focusing on the eastward-propagating synoptic scale waves in the midlatitude northwestern Pacific during the time interval from lag −8 to lag −6 days. As can be seen, at lag −8 days there is a trough centred near 165°E and a ridge centred near the Date Line. This trough is denoted by a2 at lags −8, −7 and −6 days in Figure 14. The trough–ridge pair undergo eastward propagation, amplification and then strong cyclonic wave breaking. The location of a2, together with its features, suggests that it coincides with the northwestern Pacific synoptic scale trough shown in Figure 4 for the same time interval. Also, at lag −8 days there is a second trough in the northeastern Pacific, denoted by a1. The location of this trough, along with its relative weakness, suggest that it may correspond to the anomalous low in the northeastern Pacific associated with the PNA pattern that is observed in the early stage of its development (Figure 4). After the cyclonic wave breaking of the a2 trough, beginning at lag −6 days, one strong low forms over the northeastern Pacific, which we still denote by a2. At the same lag we begin to see evidence for the occurrence of the PNA pattern further downstream, denoted by the ridge B off the west coast of North America, followed by a trough C over eastern Canada at lag −1 days.

Beginning at lag −6 days and continuing through to lag +7 days there is a succession of upstream synoptic scale troughs that propagate eastward, undergo cyclonic wave breaking over the northeastern Pacific and then merge with and reinforce the midlatitude northeastern Pacific trough associated with the PNA pattern. These cyclonic wave breakings are also accompanied by an enhancement of the ridge further to the east over western Canada. These disturbances are denoted by the letters a3, a4 and a5. For example, at lag −8 days trough a3 is near 105°E. This trough propagates eastward and at lag −2 days undergoes cyclonic wave breaking, which strengthens the northeastern Pacific PNA trough. By lag +2 days this trough has substantially weakened. However a second upstream trough, a4, which also originated over eastern Asia six days earlier, propagates eastward and then undergoes cyclonic wave breaking over the northeastern Pacific, which again reinforces the northeastern Pacific trough associated with the PNA. A third synoptic scale trough, a5, undergoes the same evolution, as it too strengthens the PNA. This behaviour, consisting of a sequence of cyclonic wave breakings, suggests that the PNA is being maintained through a positive feedback process. On subsequent days, as the PNA pattern rapidly decays, only very weak disturbances are present over the North Pacific upstream of the PNA region. This behaviour suggests that it is the termination of an eddy feedback, and perhaps also the impact of Ekman pumping (Feldstein, 2002), that accounts for the decay of the positive PNA.

The replenishment of the potential temperature anomalies by upstream synoptic scale disturbances is not observed in the composite fields (Figure 4). This behaviour is to be expected, as there is substantial case-to-case variability in these disturbances.

4.6. Negative-phase example

For a typical example of the negative PNA, we examine the event with a lag 0 day of December 21 1970 (see Figure 15). We begin by verifying that the lag 0 day flow field does resemble the negative PNA phase. As can be seen, at lag 0 there is a strong ridge over the northeastern Pacific near 150°W, a trough centred over western Canada and another ridge along the east coast of North America. Similar features can be seen in the composite negative PNA pattern (Figure 6).

thumbnail image

Figure 15. As for Figure 14, except for the negative PNA event from the 1970/1971 winter. Lag 0 days corresponds to 21 December 1970.

Download figure to PowerPoint

We again begin by focusing on those disturbances along the east coast of Asia at lag −8 days. Over the time interval from lag −8 to lag −5 days, a trough centred close to 110°E propagates eastward and eventually undergoes cyclonic wave breaking near 150°E. If we define the strength of a trough by its meridional extent, it can be seen that this breaking wave is weaker than that observed for the positive PNA. Also, during the same time interval there are two amplifying ridges located over the North Pacific. These ridges are denoted by b1 and b2 in Figure 15. Between these two ridges, near the Date Line, there is a northeast/southwest tilted synoptic scale trough that is undergoing filamentation followed by mixing. Afterward the two ridges merge to form a single large-amplitude ridge near 150°W (denoted by B) that rapidly strengthens over the following 5 days. During the time interval from lag −3 to lag −1 days, along the west coast of North America, another northeast/southwest tilted synoptic scale trough (denoted by a) is seen to amplify rapidly and propagate equatorward from about 50°N to a latitude inside the Tropics. This trough a, together with the ridge B to its west, undergoes strong anticyclonic wave breaking that is observed to extend from the midlatitudes into the Subtropics. This anticyclonic wave breaking coincides with the further amplification of the ridge B over the northeastern Pacific and perhaps also the trough a further to the south in the Tropics. As was discussed in section 4.3, such northeast/southwest tilted disturbances are to be anticipated for individual negative PNA events, because it is this particular eddy tilt that results in anticyclonic driving.

Marked differences between the positive and negative PNA cases can also be seen after lag 0 days. For example, equatorward wave propagation together with strong anticyclonic wave breaking persists until lag +5 days. Furthermore, unlike the positive PNA, for which a succession of upstream disturbances appears to maintain the PNA pattern for that phase, for the negative PNA the flow upstream of the PNA region appears to be relatively quiet. Thus, it appears that the absence of upstream disturbances also contributes toward the maintenance of the negative PNA phase. The decay of the negative PNA, in this case, sets in at about lag +5 days as the equatorward wave propagation and anticyclonic wave breaking cease and upstream synoptic scale disturbances reappear.

5. Conclusions and discussion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Composite PNA anomaly evolution
  6. 4. Synoptic description of the PNA evolution
  7. 5. Conclusions and discussion
  8. Acknowledgements
  9. References

This study investigated the synoptic evolution of the PNA by examining the daily evolution of the 2 PVU potential temperature field along with other key variables. For both PNA phases, the earliest statistically significant PNA anomalies first occur over the northeastern Pacific almost 2 weeks before the PNA pattern attains its maximum amplitude. Our results are summarized in Figure 16, which illustrates the synoptic evolution of the positive and negative PNA life cycles as a schematic diagram.

thumbnail image

Figure 16. A schematic diagram of the generalized features of (a)–(c) positive and (d)–(f) negative PNA life cycles. The panels show the main features of the PNA evolution in 3–4 day increments. The thick northern (southern) contour corresponds to a value of 305 K (335 K) on the 2 PVU surface. The light (dark) shading indicates anomalous warm (cold) air.

Download figure to PowerPoint

Positive PNA events are first excited by an intensification of convection over the western tropical Pacific and a weakening of convection over the Indian Ocean. (The longitude of the convection is indicated by the thick black bar in Figure 16(a).) This results in the gradual development of a relatively small-amplitude positive PNA pattern. After this weak PNA pattern is established, an intense synoptic scale trough forms along the east coast of Asia (Figure 16(a)). This trough breaks cyclonically (Figure 16(b)), propagates eastward and then undergoes a second cyclonic wave breaking over the northeastern Pacific (Figure 16(c)). The latter wave breaking coincides with a much more rapid growth and the establishment of the full positive PNA pattern. These results suggest that it is this cyclonic wave breaking, together with stationary eddy advection over the subtropical northeastern Pacific, that accounts for the rapid growth of the positive PNA. The maintenance of the positive PNA appears to be accomplished through a positive feedback process, as a series of eastward-propagating synoptic scale disturbances undergo cyclonic wave breaking as they enter the northeastern Pacific. The termination of the positive PNA takes place when upstream disturbances are no longer present and the eddy feedback cannot be sustained.

The above results suggest that pre-conditioning plays an important role in the development of the positive PNA. That is, the main changes to the background flow field during the early development of the positive PNA are a strengthening and eastward extension of the subtropical jet, which results in an increase in the cyclonic shear and a decrease in the meridional potential vorticity gradient on the poleward side of the subtropical jet. It is these background flow features that are typically associated with cyclonic wave breaking. This suggests the following two-step process. First, anomalous tropical convection excites a small-amplitude PNA pattern that has a spatial structure over the northeastern Pacific favouring cyclonic wave breaking. Second, a synoptic scale disturbance propagates eastward into this region, where it undergoes cyclonic wave breaking, which contributes to the rapid formation of the full positive PNA pattern.

Negative PNA events are triggered by enhanced convection over the Indian Ocean and reduced convection over the western tropical Pacific. (The longitude of the convection is indicated by the thick black bar in Figure 16(d).) As for the positive PNA, following the anomalous tropical convection a small-amplitude negative PNA pattern slowly develops (Figure 16(d)). However, in contrast to the positive PNA for which large-amplitude, synoptic scale, cyclonically breaking disturbances are observed to propagate eastward into the PNA region, for the negative PNA synoptic scale disturbances are observed to undergo rapid baroclinic growth within the PNA region (Figure 16(e)) over the midlatitude northeastern Pacific, followed by equatorward propagation from the midlatitudes into the Subtropics, where anticyclonic wave breaking takes place (Figure 16(f)). (It is important to note that there is more case-to-case variability for the negative PNA and as a result the anticyclonic wave breaking is not apparent in the corresponding composite.) It is this equatorward wave propagation and anticyclonic wave breaking, together with stationary eddy advection in the Subtropics, that results in the rapid amplification of the negative PNA pattern. The maintenance of the negative PNA also appears to occur through a positive feedback, mostly through a sustained equatorward wave propagation and anticyclonic wave breaking along the west coast of North America, together with the absence of upstream synoptic scale disturbances. The decay of the negative PNA coincides with the termination of the equatorward wave propagation and the development of upstream synoptic scale disturbances.

As with the positive PNA, these results suggest that the tropical convection associated with the negative PNA generates a background flow that is preconditioned for the development of synoptic scale disturbances that further the development of the negative PNA. These background flow properties are opposite from those for the positive PNA, i.e. a weakened and retracted subtropical jet that corresponds to a decrease in the cyclonic shear and an increase in the meridional potential vorticity gradient on the poleward side of the subtropical jet. These results also suggest a two-step process for the development of the negative PNA, whereby tropical convection excites a small-amplitude negative PNA pattern followed by the equatorward propagation and anticyclonic wave breaking of a synoptic scale disturbance, which further amplifies the negative PNA.

Cash and Lee (2001) presented evidence that the full PNA pattern arises from non-modal growth of the optimal perturbation. An inspection of the stream function anomaly composites at lags −4 days (Figure 1) reveals that for both phases the anomalies over the North Pacific and eastern Asia take on a quadrupole spatial structure that closely resembles the optimal perturbation structure of the PNA (Cash and Lee, 2001; see their figure 3(a)). The results of this study can be used to offer an explanation for the quadruple optimal pattern for both PNA phases. For the positive PNA phase, the two downstream anomalies over the northeastern Pacific correspond to the response from the anomalous tropical convection. The two upstream anomalies, one negative along the east coast of Asia and the other positive over eastern Siberia, can be explained by the presence of an intense, cyclonically breaking, synoptic scale trough that is observed in the early stage of the positive PNA development. For the negative PNA phase, opposite features can account for its quadrupole optimal structure, i.e. the two downstream anomalies can be explained as a response to tropical convective anomalies of the opposite sign and the two upstream anomalies by the occurrence of a relatively weak trough in this region.

While most PNA events are associated with tropical convection (Higgins and Mo, 1997), as was supported by this study, there are also exceptions to this picture. For example, by examining the PNA within the framework of a continuum of North Pacific teleconnection patterns, Johnson and Feldstein (2010) showed that although most PNA-like patterns are associated with tropical convection there are some PNA-like patterns with an occurrence only weakly related to tropical convection. For the latter set of PNA-like patterns, their generation must be primarily through midlatitude eddy dynamics (Dole and Black, 1990; Black and Dole, 1993). In future research, we plan to investigate the dynamical processes that distinguish between convectively driven and non-convectively driven PNA events.

An outstanding question is that of which processes account for the approximate 2 week time-scale of the PNA. As shown by Luo etal. (2007) with a weakly nonlinear model of the NAO, dipole synoptic scale eddy driving with a period of 2 weeks will drive the NAO over the same 2 week time period. This result suggests that the time-scale of the NAO is primarily determined by synoptic scale eddy driving. In contrast, for the PNA, the findings of this study suggest that tropical convection also plays an important role in determining the time-scale of the PNA.

The results of this study can be used to suggest why the positive PNA phase occurs more frequently during El Niño and the negative PNA phase during La Niña (Johnson and Feldstein, 2010). Studies such as those of Shapiro et al. (2001), Postel and Hitchman (1999), Abatzoglou and Magnusdottir (2006), Scott and Cammas (2002) and Gong etal. (2010), briefly discussed in section 3, find that the dominant type of wave breaking observed during El Niño/Southern Oscillation (ENSO) events is also related to the strength of the horizontal wind shear and meridional potential vorticity gradient of the ENSO background flow. Therefore, it seems likely that changes in these quantities during ENSO play an important role in determining the phase preference for the PNA. Another possible mechanism that may link the phase of the PNA to ENSO involves the latitude of the subtropical jet. As shown in the weakly nonlinear model of the NAO (Luo etal., 2008b), northward jet excursions, when interacting with a topographically forced stationary wave, excite a negative over positive dipole pattern, and vice versa for southward jet excursions. The forcing by synoptic scale eddies then amplifies this dipole, which leads to the formation of the full NAO pattern. Since the PNA also takes on a dipole structure over the North Pacific, it is possible that this process may contribute to the preference for the positive PNA during El Niño and the negative PNA during La Niña.

For both PNA phases, the earliest statistically significant PNA anomalies first occur almost 2 weeks before the PNA pattern attains its maximum amplitude. This suggests that the PNA is potentially predictable at this extended-range time-scale. The recent findings of Johnson and Feldstein (2010) suggest that the PNA may be predictable 1–3 weeks ahead during active MJO events. In future research, we plan to investigate the extended range predictability of the PNA.

Acknowledgements

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Composite PNA anomaly evolution
  6. 4. Synoptic description of the PNA evolution
  7. 5. Conclusions and discussion
  8. Acknowledgements
  9. References

This study is part of the British Antarctic Survey Polar Science for Planet Earth Programme. It was funded by the Natural Environment Research Council and by the National Science Foundation Grants ATM-0649512 and ATM-0852379. We thank two anonymous reviewers for their helpful comments and the NOAA Earth Systems Research Laboratory for providing us with the NCEP/NCAR reanalysis and OLR data sets. Parts of the study were carried out while the first author was affiliated with the National Center for Atmospheric Research.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and methodology
  5. 3. Composite PNA anomaly evolution
  6. 4. Synoptic description of the PNA evolution
  7. 5. Conclusions and discussion
  8. Acknowledgements
  9. References
  • Following Thorncroft etal. (1993), cyclonic wave breaking is characterized by a southeast–northwest tilt of the trough/ridge pair. Anticyclonic wave breaking exhibits the opposite tilt.