SEARCH

SEARCH BY CITATION

Keywords:

  • Arctic Oscillation;
  • Arctic precipitation;
  • Modoki;
  • decadal variability;
  • tropical Pacific

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Concluding Remarks
  7. Acknowledgments
  8. References

[1] The temporal and spatial characteristics of decadal-scale variability in the Northern Hemisphere (NH) cool-season (October–March) Arctic precipitation are identified from both the Climate Prediction Center (CPC) Merged Analysis of Precipitation (CMAP) and the Global Precipitation Climatology Project (GPCP) precipitation data sets. This decadal variability is shown to be partly connected to the decadal-scale variations in tropical central Pacific sea surface temperatures (SSTs) that are primarily associated with a decadal modulation of the El Niño–Southern Oscillation (ENSO), i.e., transitions between periods favoring typical eastern Pacific warming (EPW) events and periods favoring central Pacific warming (CPW) events. Regression and composite analyses reveal that increases of central Pacific SSTs drive a stationary Rossby wave train that destructively interferes with the wave number-1 component of the extratropical planetary wave. This destructive interference is opposite to the mean effect of typical EPW on the extratropical planetary wave. It leads to suppressed upward propagation of wave energy into the polar stratosphere, a stronger stratospheric polar vortex, and a tendency toward a positive phase of the Arctic Oscillation (AO). The positive AO tendency is synchronized on the decadal scale with a poleward shift of the NH storm tracks, particularly in the North Atlantic. Storm track variations further induce changes in the amount of moisture transported into the Arctic by synoptic eddies. The fluctuations in the eddy moisture transport ultimately contribute to the observed decadal-scale variations in the total Arctic precipitation in the NH cool season.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Concluding Remarks
  7. Acknowledgments
  8. References

[2] The accumulation of snow and the growth of ice cover in the Arctic during the Northern Hemisphere (NH) winter are important to the Earth's climate and depend significantly on the amount of precipitation that falls within the Arctic. Since Arctic winters are characterized by low temperatures and broad sea ice cover that limit local evaporation, moisture influx from lower latitudes provides the most important source of moisture for Arctic precipitation. The meridional transport of moisture into the Arctic is known to be primarily associated with synoptic-scale eddies [Peixoto and Oort, 1992; McBean et al., 2005] that tend to organize themselves into distinct storm tracks in winter [Blackmon, 1976]. Significant correlations are found between the activity of midlatitude cyclones (i.e., surface signatures of synoptic eddies) and the total moisture transport into the Arctic and by extension, total Arctic precipitation [e.g., Sorteberg and Walsh, 2008]. It is also well established that synoptic eddy activity within major storm tracks, such as the ones over the North Pacific and North Atlantic, is dynamically coupled to various atmospheric low-frequency modes, including the North Atlantic Oscillation (NAO) and the Arctic Oscillation (AO) [e.g., Thompson and Wallace, 1998]. Both the NAO and AO variability prove to be crucial in modulating winter cyclone activity north of 60°N [Serreze et al., 1997] and determining the variability in sea ice cover and sea ice motion across the Arctic Ocean [Deser et al., 2000; Dickson et al., 2000; Rigor et al., 2002].

[3] Recent studies have uncovered connections between the NAO/AO variability and tropical sea surface temperatures (SSTs). For example, the seasonality of the AO and decadal-scale changes in the NAO have been linked to SST variability in the western tropical Pacific (between 140°E and 170°W) [Jia et al., 2009; King and Kucharski, 2006; Kucharski et al., 2006]. Jia et al. [2009] and Kucharski et al. [2006] attribute these connections to the direct propagation of anomalous Rossby waves from the tropical Pacific to the AO/NAO action center in the North Atlantic and Arctic oceans. Another potential mechanism, in which the anomalous Rossby wave propagation has an indirect effect on the AO/NAO, can be distilled from studies that linked El Niño–Southern Oscillation (ENSO) variability to the increased occurrence of sudden stratospheric warmings (SSWs) in the NH polar region during boreal winter [e.g., Taguchi and Hartmann, 2006; Taguchi, 2010]. These studies suggest that SST anomalies in the tropical Pacific that are due to El Niño result in an increased upward propagation of wave energy associated with the wave number-1 component of the extratropical planetary wave. The enhanced wave forcing in the stratosphere leads to the breakdown of the NH stratospheric polar vortex and an increase in the occurrence of SSWs. Although the index of the AO is typically defined from loading patterns in the sea level pressure (SLP) and tropospheric geopotential height field, the variability of the AO has been shown to reflect the changes in the NH stratospheric polar vortex [Thompson and Wallace, 1998; Baldwin and Dunkerton, 1999]. The stratospheric pathway of the ENSO impact thus provides a second explanation of the linkage between the tropical SSTs and AO/NAO variability, at least on interannual time scales.

[4] However, not all warm ENSO events lead to a weakened polar vortex. The response depends on the phase of the Quasi-Biennial Oscillation (QBO) and ultimately the specific type of extratropical teleconnection patterns excited by the tropical Pacific SST anomalies. Garfinkel and Hartmann [2008] showed that the weakening effect of El Niño on the polar vortex is maximized when the stationary Rossby wave forced by the SST anomalies projects onto the Pacific North America (PNA) pattern [Wallace and Gutzler, 1981]. The deepening of the Aleutian Low (AL), characteristic of the positive PNA phase, significantly enhances the climatological wave number-1 component of the planetary wave and the subsequent upward propagation of planetary wave energy into the stratosphere. The projection of the atmospheric response to the SST anomaly onto other extratropical teleconnection patterns, including the western Pacific (WP) and tropical Northern Hemisphere (TNH) patterns, does not have the same weakening effect on the polar vortex because of the lack of a significant AL signal that constructively interferes with the climatological wave number-1 pattern.

[5] It remains unclear how the tropical-extratropical connections discussed above manifest themselves in the Arctic hydrological cycle, in particular the snow accumulation and land and sea ice growth in the NH cool season (October–March). Given the essential role of the Arctic ice in the Earth's climate system, it is important to better understand the local and remote factors that control the Arctic ice recovery in the cool season across various time scales. As an initial effort on this topic, this study focuses on precipitation and aims to quantify the temporal and spatial characteristics of decadal-scale variations in the NH cool-season Arctic precipitation and to identify tropical forcing factors behind such variations. The direct and indirect effects of tropical SST variability on the Arctic climate, as discussed above, are explored further and intercompared. Following the introduction, section 2 describes the data and methods used in the analysis. Major results are reported in section 3, where the significance of the exact character of the warm ENSO events is emphasized. Section 4 gives the concluding remarks.

2. Data and Methods

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Concluding Remarks
  7. Acknowledgments
  8. References

[6] Monthly precipitation data from both the Climate Prediction Center (CPC) Merged Analysis of Precipitation (CMAP) [Xie and Arkin, 1997] and the Global Precipitation Climatology Project (GPCP) Version 2.1 data sets [Adler et al., 2003] are used to calculate the total Arctic precipitation during each cool season from 1979–1980 to 2008–2009. A total precipitation index is computed as the area-weighted average over all the grid points in each data set north of the Arctic Circle (66.5°N). To isolate the decadal-scale signals, a 7 year moving average is applied to the Arctic precipitation index as well as to other standard atmospheric fields being analyzed, all of which come from the NASA Modern Era Retrospective Analysis for Research and Applications (MERRA) reanalysis data set on a 0.5° latitude × 0.67° longitude grid [Bosilovich, 2008] (see also http://www.earthzine.org/2008/09/26/nasas-modern-era-retrospective-analysis/). The 7 year moving average has been used in previous studies that address decadal-scale variability [e.g., King and Kucharski, 2006], and the results obtained in our analysis are qualitatively similar despite small changes in the definition of the filter, e.g., 5 year versus 7 year moving average.

[7] The application of the 7 year moving average reddens the time series under consideration and reduces the effective number of degrees of freedom [e.g., Garfinkel et al., 2010; Naoe and Shibata, 2010]. A Monte Carlo approach is adopted to derive statistical significance for all the correlation and regression results that utilize smoothed data [Woollings et al., 2010]. Using the correlation calculation as an example, for each of the two smoothed time series of observation, we create a hypothetical time series of equal sample size with its elements drawn randomly and independently from a Gaussian distribution characterized by the same mean and standard deviation of the original, unsmoothed time series of observation. The two hypothetical time series are then smoothed by the same 7 year moving average filter and the correlation coefficient between these two time series is calculated. This drawing and smoothing process is repeated 5000 times to obtain an empirical probability distribution function (PDF) of the correlation coefficients. The p value of the correlation coefficient calculated from the original, smoothed time series of observation is finally obtained based on the percentiles of this empirical distribution.

[8] To quantify the effect of storm track activity and eddy moisture transport on Arctic precipitation, the meridional moisture transport associated with synoptic eddies, equation image, and the synoptic eddy kinetic energy (SEKE), equation image, are both derived from the daily MERRA data, where u, v, and q are the zonal wind, meridional wind, and specific humidity, respectively. The overbar indicates averaging over the cool season and the prime corresponds to synoptic-scale fluctuations obtained through a 2–6 day Butterworth band-pass filter. The curly braces indicate mass-weighted vertical averaging between 1000 and 600 mbars.

[9] To identify the remote (tropical) forcing factors of the Arctic precipitation change, we examined the 7 year smoothed SST field from the Hadley Center Sea Ice and Sea Surface Temperature (HadISST) data set [Rayner et al., 2003] and the outgoing longwave radiation (OLR) field from the NOAA Interpolated OLR data set [Liebmann and Smith, 1996]. The two-dimensional wave activity flux [Plumb, 1985; Takaya and Nakamura, 2001],

  • equation image

calculated at 250 mbars with MERRA winds, is adopted to reveal the horizontal propagation of stationary Rossby waves from the Tropics as excited by the SST and associated diabatic heating anomalies. p and ψ are the pressure (in millibars) and the stream function, respectively. The primes in this equation represent the deviation from the zonal mean. Based on monthly MERRA winds and temperature, the vertical propagation of planetary wave energy is diagnosed through the quasi-geostrophic form of the Eliassen-Palm (EP) flux that includes spherical geometry considerations [Edmon et al., 1980, equations 3.1a and 3.1b]:

  • equation image

The primes represent the deviation from the zonal mean, and the overbars represent the zonal mean. f is the Coriolis parameter, and ϕ, r0, and θ are the latitude, the mean radius of Earth, and the potential temperature, respectively. The subscript “p” is the vertical derivative with respect to pressure. A fast Fourier transform (FFT) is applied to the 500 mbar geopotential height field at each latitude to quantify the changes in the planetary wave power spectrum across different zonal wave numbers. Similarly calculated are the contributions from the individual wave number components to the total anomalies of vertical wave propagation.

3. Results

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Concluding Remarks
  7. Acknowledgments
  8. References

3.1. Decadal-Scale Variations in the Cool-Season Arctic Precipitation

[10] Figure 1a shows the cool-season Arctic precipitation anomalies (bars) for the period 1979–1980 to 2008–2009 in GPCP, CMAP and MERRA data sets. The 30 season mean values removed from the three data sets are substantially different; the mean cool-season total Arctic precipitation in GPCP (188.9756 mm) is about 1.93 (1.2) times larger than the corresponding CMAP (MERRA) mean. Despite the differences in the magnitude of the climatology, distinct decadal-scale variations (smoothed solid curves) exist in all three data sets with above normal precipitation found during 1990–1994 and below normal precipitation occurring in 1997–2001. The correlation coefficients among the smoothed curves are in the range of 0.70–0.78 and are all significant at the 90% level, demonstrating the robustness of the decadal-scale signal in the Arctic precipitation. In addition, the correlation coefficients between the unsmoothed Arctic precipitation values of the three data sets range from 0.73 to 0.78 and are all significant at the 99% level, demonstrating the similarity of the representation of interannual variations of the Arctic precipitation in all three data sets, despite the differences in the magnitude of their respective climatologies. All three precipitation time series are positively correlated with the smoothed AO index (r = 0.5345 for GPCP, r = 0.6356 for CMAP, r = 0.4744 for MERRA), with the correlation coefficients for GPCP, CMAP, and MERRA significant at the 85%, 90%, and 80% levels, respectively. Although the length of the observational record severely limits the level of statistical significance derived here, it is fairly evident that increasing Arctic precipitation is typically accompanied by a tendency toward the positive phase of the AO on decadal time scales. In the remaining part of this paper, only results based on the GPCP are presented since the choice of the precipitation data does not affect the conclusions qualitatively.

image

Figure 1. (a) Annual anomalies of the total cool-season (October–March) Arctic precipitation in (blue) GPCP, (green) CMAP, and (red) MERRA (in mm). The bars represent the unsmoothed precipitation anomalies, and the solid curves represent precipitation anomalies smoothed by a 7 yr running mean filter. Also plotted is the AO index (purple dashed curve), smoothed by a 7 yr running mean filter. (b) Distribution of the interdecadal standard deviation of the (blue) GPCP Arctic precipitation (in mm) and (red) meridional eddy moisture flux (in (m/s)(g/kg)) across different longitude sectors.

Download figure to PowerPoint

[11] The breakdown of the signal in the total Arctic precipitation into different longitudinal sectors (Figure 1b) reveals that amplitudes of the decadal-scale variations in precipitation and in the meridional moisture flux associated with synoptic eddies at the Arctic Circle are both greatest in the Atlantic sector of the Arctic (i.e., the Greenland, Norwegian, and Barents seas). By absolute value the largest standard deviation of the 7 year smoothed precipitation occurs in the sector 0°–30°E, corresponding to the Norwegian Sea. By percentage of the sector mean precipitation, the largest standard deviation is found in the sector 30°E–60°E, corresponding to the Barents Sea. The collocation of the maximum interdecadal standard deviations of precipitation and the synoptic eddy moisture transport demonstrates the importance of synoptic eddies, as part of the North Atlantic storm track, in modulating the decadal-scale variability in Arctic precipitation. Figures 2a and 2b show the regression of the 850 mbar SEKE and the lower tropospheric synoptic eddy moisture transport, respectively, onto the Arctic precipitation index. Associated with high values of Arctic precipitation are an elevated level of synoptic eddy activity north of the Arctic Circle and an increase of poleward eddy moisture transport in the Atlantic sector of the Arctic. Thus the variability of cool-season Arctic precipitation is directly tied to the storm track variability on decadal time scales, and the former's connection to the AO, as shown in Figure 1a, is established through the coupling between AO variability and storm tracks. The northward shift of the zonal wind jets that accompanies the positive phase of the AO tends to steer more cyclones into the Arctic and leads to a poleward shift of the storm tracks.

image

Figure 2. (a) 850 mbar synoptic eddy kinetic energy (SEKE, in J/kg) and (b) lower troposphere moisture transport by synoptic eddies (in (m/s)(g/kg)) regressed onto the Arctic precipitation index. The moisture transport is vertically averaged over 1000–600 mbars. Areas with values significant at the 90% level are hatched.

Download figure to PowerPoint

[12] The fact that the decadal-scale variability in Arctic precipitation and the associated synoptic eddy moisture transport are more pronounced in the Atlantic sector of the Arctic compared with the Pacific sector is largely due to the differences in the mean latitudinal position and intensity of the Pacific and Atlantic storm tracks. The Pacific storm track is primarily zonally oriented with an exit region in the Gulf of Alaska south of the Arctic Circle. On the other hand, the Atlantic storm track has a distinct southwest-northeast tilt, with an exit region extending northward across the Arctic Circle. In addition, the climatological Pacific storm track is weaker than the Atlantic storm track in boreal winter [e.g., Chang et al., 2002; Deng and Mak, 2005, 2006]. When, for example, a zonally symmetric mode, such as AO, shifts toward a positive phase, it is coupled to the poleward movement of both the Atlantic and Pacific storm tracks. However, more (intense) cyclones and lower-latitude moisture can reach the interior Arctic in the Atlantic sector because of a stronger and more poleward positioned Atlantic storm track. As a result, the strongest coupling among the Arctic precipitation, synoptic eddy moisture transport, and storm track occurs in the Atlantic sector rather than in the Pacific sector.

3.2. Connections Between the Decadal-Scale Variations in the Arctic Precipitation and in the Tropical Pacific SSTs

[13] Figure 3a shows the coefficients of regression between the tropical SST and the 7 year smoothed Arctic precipitation index. Statistically significant positive values are found across the tropical central Pacific (east of 165°E), centered on the equator and extending northeastward into the coast of Baja California. A nearly identical pattern is found when the tropical SST is regressed onto the smoothed October–March mean AO index. This SST regression pattern across the tropical Pacific bears some similarity to the SST anomaly pattern typical of an El Niño Modoki event [e.g., Ashok et al., 2007, Figure 5b; Weng et al., 2009, Figure 3b]. The extension of positive SST anomalies toward Baja California, closely matches the SST anomaly pattern associated with the central Pacific-type of warming discussed by Yu and Kao [2007] and Kao and Yu [2009]. Sun and Yu [2009] reported a decadal-scale modulation of ENSO variability that is characterized by a SST pattern also similar to that in Figure 3a. Under this modulation, weak (strong) ENSO periods correspond to a situation in which central (eastern) Pacific warming events are favored and the ascending branch of the Walker circulation shifts toward the central (eastern) Pacific. The western and central Pacific dipoles in the OLR regression field (Figure 3b) are also consistent with this shift of the Walker circulation.

image

Figure 3. (a) Sea surface temperature (SST, in K) and (b) outgoing longwave radiation (OLR, in W/m2) regressed onto the Arctic precipitation index. The purple box in 3a corresponds to the region over which the central Pacific warming (CPW) index is defined. Areas with values significant at the 90% level are hatched.

Download figure to PowerPoint

[14] The SST and OLR regression results suggest a connection between the cool-season Arctic precipitation and the modulation of the ENSO variability on decadal time scales, with the AO likely acting as a dynamical medium. Specifically, an increase (decrease) of the cool-season Arctic precipitation on decadal time scales is associated with more frequent central (eastern) Pacific warming events and weak (strong) ENSO intensity periods according to the ENSO modulation discussed by Sun and Yu [2009]. For example, the early 1990s are characterized by above normal Arctic precipitation, and central Pacific warming (CPW) events were much more common in this period. According to Yu and Kim [2010], every year during the period 1990–1995 can be identified as a central Pacific (CP) El Niño. Conversely, during the late 1990s, the Arctic precipitation was below normal and strong eastern Pacific warming (EPW) events, such as the 1997–1998 El Niño, occurred. The connection between the Arctic precipitation and the central Pacific SSTs is further confirmed by the co-occurrences of the peak values of Arctic precipitation and low values of an index describing the ENSO modulation [Sun and Yu, 2009, Figure 1c]. In the next step, we examine in detail the potential dynamical pathways through which the central Pacific SST anomalies project onto the variability of the AO, which is in turn coupled to the variability in storm track activity and synoptic eddy moisture transport into the Arctic. It is also important to recognize that regression results presented here do not exclude the possibility of a cold-phase ENSO (i.e., La Niña) contributing to the observed Arctic precipitation variability; however, results of the composite analysis shown in section 3.4 indicate that La Niña does not play a critical role here.

3.3. Direct Projection of the Tropical Central Pacific SST Forcing Onto the AO Variability

[15] To illustrate the dynamical implications of CPW for the extratropical circulation, we define a CPW index as the SSTs averaged over the purple box (10°S–15°N, 165°E–130°W) in Figure 3a. The correlation between the smoothed CPW index and the Arctic precipitation index (October–March mean AO index) is 0.78 (0.61) and significant at the 95% (85%) level. In other words, warm SSTs over the tropical central Pacific are associated with increased total Arctic precipitation and a positive tendency of the AO.

[16] The coefficients of regression between the 7 year smoothed 250 mbar stream function and the CPW index are displayed in Figure 4a. A stationary Rossby wave train emanating from the tropical central Pacific is the primary feature of Figure 4a. The origin of the wave train is located between 180°W and 150°W, consistent with the longitudes where the maximum SST signals are identified (Figure 3a). The wave train appears to propagate into the Arctic and project onto a positive AO pattern, i.e., negative stream function anomalies over the Arctic. Consistent with the stream function anomalies is a poleward shift of the upper tropospheric zonal jets, as shown in Figure 4b. The regression of the stationary wave activity flux onto the CPW index (Figure 4c), however, indicates a horizontal pathway of wave propagation originating from the tropical central Pacific and diminishing around the Aleutian Islands, without reaching the interior region of the Arctic. This suggests that the direct stationary Rossby wave response in the North Pacific is not necessarily the most important mechanism responsible for projecting the tropical Pacific SST forcing onto the AO variability.

image

Figure 4. (a) 250 mbar stream function (in m2/s), (b) 250 mbar zonal wind (in m/s), and (c) 250 mbar wave activity flux (in m2/s2) values regressed onto the CPW index. Areas of values significant at the 90% level are hatched in 4a and 4b, and vectors with either component significant at the 90% level are shown in Figure 4c.

Download figure to PowerPoint

3.4. Indirect Projection of the Tropical Central Pacific SST Forcing Onto the AO Variability

[17] Following the discussion in section 1, the positive AO phase can also be a tropospheric response to a strengthened NH stratospheric polar vortex [e.g., Baldwin and Dunkerton, 1999; Black, 2002]. As a prominent feature of the high-latitude atmospheric circulation in boreal winter, the stratospheric polar vortex is characterized by an area of strong zonal mean westerly winds and low zonal mean temperatures in the polar stratosphere. Fluctuations in the strength of this vortex appear as anomalies in the zonal mean zonal winds and zonal mean temperature. When regressed onto the CPW index defined in the previous section, the zonal mean zonal wind field shows positive values, i.e., strengthened westerly winds throughout the stratosphere and troposphere north of 50°N (Figure 5a). The increase in zonal wind speed is collocated with the climatological position of the NH winter stratospheric polar vortex, thus representing a strengthening of the vortex. The regression of the zonal mean temperature onto the CPW index (Figure 5b) indicates negative anomalies below the midstratosphere north of 60°N, consistent with a poleward shift of the large meridional temperature gradient and the strengthening of zonal mean zonal winds.

image

Figure 5. (a) October–March averaged zonal mean zonal wind (in m/s) regressed onto the OM CPW index. (b) OM average zonal mean temperature (in K) regressed with the OM CPW index. Areas of values significant at the 90% level are hatched in both figures.

Download figure to PowerPoint

[18] Since an important source of variability in the stratospheric polar vortex is the vertical propagation of Rossby wave energy into the stratosphere [e.g., Polvani and Waugh, 2004], we next examine the impact of the tropical CPW on the upward propagation of tropospheric planetary waves. The vertical propagation of waves is quantified in terms of the EP flux vector described in section 2. Here we compare the composite EP flux for a 5 yr period (1990–1994) of relatively high values of the smoothed CPW index with that of a 5 yr period (1997–2001) with relatively low values of the smoothed CPW index. The selection of the compositing periods is independent of the specific smoothed index used (i.e., CPW or Arctic precipitation) because of the high correlation between the two. Figure 6a shows the differences in the EP flux vectors between the two periods (i.e., 1990–1994 minus 1997–2001), and the vectors have been scaled by a factor of 5 above the 100 mbar level following the plotting convention of Garfinkel and Hartmann [2008]. Pronounced downward pointing vectors exist in the lower and middle stratosphere between 50°N and 75°N. Since this region is occupied by upward pointing EP flux vectors in the cool-season climatology (not shown), the downward pointing vectors in Figure 6a indicate suppressed propagation of tropospheric planetary waves into the stratosphere at the NH high latitudes during periods of elevated tropical central Pacific SSTs (i.e., increased occurrence frequency of CPW).

image

Figure 6. (a) Difference of the composite EP flux vectors (y component in (m2/s2)m, z component in (m2/s2)Pa) between a 5 yr period (1990–1994) of above normal Arctic precipitation and a 5 yr period (1997–2001) of below normal Arctic precipitation. Only vectors with at least one component significant at the 90% level, based on a Welch's t test, are plotted. (b) The composite difference of the power spectrum of the cool-season 500 mbar geopotential height (in m, averaged over 45°N–90°N) between the two 5 yr periods defined in 6a, shown as a function of zonal wave number.

Download figure to PowerPoint

[19] The zonal wave number is a key factor that determines whether vertical propagation of waves is plausible in a westerly environment characteristic of the upper troposphere and lower stratosphere in the NH cool season [Charney and Drazin, 1961]. Since waves with the largest zonal wavelength and smallest zonal wave number (typically 1 and 2) have the best chance of propagating into the stratosphere, we examine further which component of the planetary wave is responsible for the suppressed propagation given in Figure 6a. Figure 6b shows the differences of the composite power spectrum of the monthly cool-season 500 mbar geopotential height north of 45°N between the periods 1990–1994 and 1997–2001. What stand out are a significant reduction of the wave number-1 amplitude and a moderate increase of the wave number-2 amplitude in association with the tropical CPW. Extraction of the wave number-1 and -2 components from the total composite EP flux anomalies (Figure 6a) indeed shows suppressed upward propagation of wave number-1 waves (Figure 7a) and enhanced upward propagation of wave number-2 waves (Figure 7b). Since the amplitude of decrease of the former is significantly larger than the amplitude of increase of the latter, the net result (sum of the two, Figure 7c) is suppressed upward propagation of tropospheric planetary waves and a strengthened stratospheric polar vortex accompanying the tropical CPW.

image

Figure 7. Difference of the composite EP flux vectors (same units as in Figure 6a) associated with (a) wave number-1 and (b) wave number-2 components of the planetary wave between the period 1990–1994 and the period 1997–2001. (c) Sum of Figures 7a and 7b. All vectors are plotted.

Download figure to PowerPoint

3.5. Differences in Effects of CPW and Canonical ENSO Warming on the NH Stratospheric Polar Vortex

[20] The effect of CPW on the polar vortex discussed above is the exact opposite of what previous studies [e.g., Taguchi and Hartmann, 2006] have suggested for canonical warm ENSO events, in which eastern Pacific warming (EPW) results in a weakening of the polar vortex. Given the close link discovered between the Arctic precipitation and the relative occurrence frequency of CPW and EPW events (i.e., the decadal modulation of the ENSO variability), it is necessary to investigate why these two types of warming events leave different dynamical fingerprints in the polar vortex. As noted by Garfinkel and Hartmann [2008], the ability of canonical ENSO to affect the polar vortex ultimately depends on the specific type of extratropical teleconnection excited by the tropical SST anomalies. The deepening of the AL and effective projection onto the PNA pattern prove to be the most effective way of enhancing the wave number-1 component of the extratropical planetary wave, increasing its upward propagation into the stratosphere and thus weakening the polar vortex. Here we compare the 500 mbar geopotential height anomalies in December–January–February (DJF) associated with classical EPW (Figure 8a) and CPW (Figure 8b) events. The deepening of the AL is the most pronounced feature in the EPW height composite, while in the CPW composite, a positive anomaly south of Alaska indicates a weakening of the AL.

image

Figure 8. Composite anomalies of the DJF 500 mbar geopotential height (in m) corresponding to a classic (a) eastern Pacific warming (EPW) event and (b) central Pacific warming (CPW) event. EPW winters used in compositing include 1982–1983, 1987–1988, and 1997–1998. CPW winters include 1990–1991, 1994–1995, 2002–2003, and 2004–2005.

Download figure to PowerPoint

[21] The composite anomalies shown in Figure 8 are decomposed into their wave number-1 and wave number-2 components and displayed in Figure 9 together with the NH winter climatological planetary waves. It is clear that the deepening of the AL in the EPW case creates wave number-1 anomalies (Figure 9a) that interfere constructively with the climatological wave number-1 pattern (Figure 9c), while the CPW generates wave number-1 anomalies (Figure 9b) that interfere destructively with the climatological wave number-1 pattern. This discrepancy is consistent with the opposite effects CPW and EPW exert on the vertical propagation of waves and the resulting changes in the strength of the polar vortex. On the other hand, the wave number-2 components of the composite height anomalies (Figures 9d and 9e) tend to reduce (enhance) the amplitude of the climatological wave number-2 waves in the EPW (CPW) case.

image

Figure 9. (a–c) Wave number-1 and (d–f) wave number-2 components of the composite anomalies of the DJF 500 mbar geopotential height (in m) corresponding to the EPW (Figures 9a and 9d) and CPW (Figures 9b and 9e) cases. The wave number-1 and -2 components of the climatological (1979–2009) DJF planetary wave are shown in Figures 9c and 9f, respectively.

Download figure to PowerPoint

[22] Considering a composite of canonical CPW winters, based on the same set of CPW winters used in deriving the height composite in Figure 8b, we display in Figure 10 the CPW winter anomalies with respect to a 30 winter (1979–1980 to 2008–2009) climatology in zonal mean zonal wind, zonal mean temperature, power spectrum of 500 mbar height, and EP flux vectors. The results are generally consistent with those in Figures 5 and 6, and they show that accompanying the occurrence of a canonical CPW event are increased zonal mean zonal winds north of 50°N (Figure 10a), decreased zonal mean temperatures in the stratosphere north of 60°N (Figure 10b), an increase (decrease) of wave number-1 (-2) planetary wave amplitude (Figure 10c), and a net decrease in the upward propagation of wave energy into the lower stratosphere (Figure 10d). The anomalies shown in Figures 10c and 10d further confirm the destructive and constructive interference of CPW with the climatological wave number-1 and -2 components of the extratropical planetary wave, respectively. The similarity between the canonical CPW winter anomalies and the previously discussed composite differences between two periods indicates that the occurrence of multiple CPWs during the period of 1990–1994 is at least partly responsible for the signals seen in Figures 5 and 6. The increased occurrence frequency of individual CPW events during one period thus contributes positively to an increase in the Arctic precipitation through accumulated effects of individual CPWs on the stratospheric polar vortex.

image

Figure 10. Composite anomalies associated with a canonical CPW winter (October–March) in (a) zonal mean zonal winds (in m/s), (b) zonal mean temperature (in K), (c) the power spectrum of 500 mbar geopotential height (in m) and (d) the EP flux vector (y component in (m2/s2)m, z component in (m2/s2)Pa). The CPW winters used in constructing the composites are the same as those in Figures 8 and 9.

Download figure to PowerPoint

[23] The differences between the impact of the CPW (as discussed in section 3.4) and EPW (as discussed in previous studies) on the stratospheric polar vortex and AO variability are ultimately a result of the different teleconnection patterns that the CPW and EPW excite in the extratropics. The transition from one period (e.g., 1990–1994), when the CPW events are more common, to another period (e.g., 1997–2001), when the EPW events occur more often, characterizes the decadal-scale modulation of ENSO variability and contributes significantly to the transition from one period of a strengthened polar vortex (positive AO) to another period of a weakened polar vortex (negative AO). With the additional coupling between AO variability and storm track activity, the tropical Pacific SSTs are able to leave distinct dynamical fingerprints on the Arctic precipitation.

4. Concluding Remarks

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Concluding Remarks
  7. Acknowledgments
  8. References

[24] Pronounced decadal-scale variations in the cool-season Arctic precipitation are identified in two observation-based precipitation data sets (GPCP and CMAP) and one high-resolution reanalysis product (MERRA). Despite the discrepancies in the magnitude of their climatological values of Arctic precipitation, all three data sets clearly show above normal precipitation in the period 1990–1994 and below normal precipitation in the period 1997–2001. This decadal-scale oscillation, significantly correlated with the decadal-scale variations in the AO, is partly driven by the tropical central Pacific SST changes across decadal time scales. The tropical Pacific SST anomalies tied to the Arctic precipitation variability closely resemble the SST anomalies representative of the decadal modulation of ENSO variability [Sun and Yu, 2009], suggesting that the more frequent occurrence of CPW events might have contributed to the increase of Arctic precipitation in the early 1990s and after 2003 (Figure 1a).

[25] Dynamically, the diabatic heating induced by the CPW drives a stationary Rossby wave train that extends northward into the North Pacific. The direct projection of the CPW forcing onto the AO variability is limited, given that the wave activity originating from the tropical central Pacific does not reach the interior of the Arctic. However, the SST-forced stationary Rossby wave destructively interferes with the wave number-1 component of the extratropical planetary wave, leading to suppressed upward propagation of waves into the polar stratosphere, a strengthened stratospheric polar vortex, and a positive tendency in the AO index. The tropical Pacific SSTs thus project indirectly onto the AO variability through modifying the structure of extratropical planetary waves. The effects of CPW and EPW on the polar vortex (AO variability) are exactly opposite, with the wave number-1 component of the planetary wave strengthened in an EPW event. The decadal-scale modulation of ENSO variability, characterized by transitions between periods favoring CPW and those favoring EPW, ultimately generates discernable dynamical fingerprints in the decadal-scale AO variability.

[26] The positive AO tendency partly induced by CPW is coupled with a poleward shift of the upper tropospheric zonal wind jets and storm tracks. The elevated synoptic eddy activity north of the Arctic Circle increases the amount of moisture transported into the Arctic by synoptic eddies and contributes directly to the decadal-scale variations in Arctic precipitation. The strongest signal of the storm track response with regard to the Arctic precipitation variability is found over the Atlantic sector of the Arctic, a region characterized by the greatest decadal variations in both precipitation and eddy moisture flux into the Arctic. On decadal time scales, the AO and North Atlantic storm track thus act as a bridge connecting the tropical Pacific SSTs to the strength of the cool-season Arctic hydrological cycle. However, we recognize that the direct and indirect (stratospheric) pathways discussed here account for only a portion of the decadal-scale variance in the AO and in the Arctic precipitation index, and the length of the observational record, particularly of the precipitation data sets, places a limit on the level of statistical significance we may derive here. Future investigations will focus on (1) testing the hypothesis formulated in this study with controlled general circulation model (GCM) experiments and (2) identifying other dynamical processes contributing to the tropical Pacific-Arctic connection. Of particular importance for the latter is the need to quantify the synoptic eddy (e.g., vorticity and heat flux) feedback onto the interdecadal AO variability at the storm track regions. The preliminary analysis of the Chemistry-Climate Model Validation Activity (CCMVal, http://www.pa.op.dlr.de/CCMVal/) model output shows that the connection described in this paper is captured in GCM simulations forced with the observed SST field. Details of the model analysis will be reported in the future.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Concluding Remarks
  7. Acknowledgments
  8. References

[27] We thank three anonymous reviewers for their thoughtful comments and suggestions that led to major improvements in the manuscript. The CMAP, GPCP precipitation data and the NOAA Interpolated OLR data used in this study were provided through NOAA/OAR/ESRL (http://www.esrl.noaa.gov/psd/). The HadISST data set was provided by the UK Met Office (www.metoffice.gov.uk/hadobs). MERRA reanalysis data were acquired from the NASA Global Modeling and Assimilation Office (GMAO) and the GES DISC. Bradley M. Hegyi was supported by the NASA Earth and Space Science Fellowship under grant NNX10AO64H, and Yi Deng was supported by NASA Energy and Water Cycle Study (NEWS) under grant NNX09AJ36G.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Concluding Remarks
  7. Acknowledgments
  8. References