A Spectral Analysis of Wave Activities and Blocking in the Northern Hemisphere Winter

A systematic analysis of wave acthities and blocking in the Northern Hemisphere winter is performed with the daily 500 mb height fields for the 1956-1991 period and the monthly sea surface temperatures (SSTs) for the 1970-1991 period. The power spectra in the wavenumber-frequency do­ main are partitioned into westward propagating, eastward propagating and standing variances. Regional indices are used to classify the seasonal circu­ lation into fou r categories of circulation patterns according to the domi­ nant blocking activities or the season: Pacific blocking, Atlantic blocking, double blocking, and no blocking. The distinctions among categories of blocking are characterized in the standing, westward propagating and east­ ward propagating variances. The distinctions are also renected in the ki­ netic energy and nonlinear wave-wave interactions of the planetary-scale and the synoptic-scale waves. A com)l-Ositc study of dominant circulation patterns is also presented in this study. The results show that the wave-wave interaction of the plan­ etary-scale and the synoptic-scale waves plays a crucial role in the forma­ tion and maintenance of Pacific blocking, and that Atlantic blocking and double blocking might be forced and maintained by baroclinic processes.


l. INTRODUCTION
In studying the low-frequency atmospheric variability, much anention has been focused I n the winter blocking phenomena. Blocki ng circulation is important for the long-range weather orecasting for its relatively high foreca sting skills and the local synoptic signifi c ance (see hul da. 1981; Kung et al., 1990 and. Although most single blocking episodes have ubmonthly durations. monthly and even seasonal mean values of temperature and precipita on are otien greatly influenced by strong, persistently recurring blockings (Rex, 1950;Namias, 70 TA O, Vol. 8, No. /,March 1997Quiroz. 1987). Many studies have been done concerning the mechanisms responsi ble for the amplification of quasi-stationary waves associated with blocking flow.
The studies of observed winter blocking have been done by analyzing the characteristics of single blocking flows in the Pacific and the Atlantic (Hartmann and Ghan, 1980;Metz, 1986;Mullen, 1987;Nakamura and Wallace, 1990). Metz (1986) analyzed the forcing of planetary flow by cyclone-scale eddy vorticity fluxes with the observation and model simula tions. Mullen ( 1987) evaluated the effects of transient eddy heal and vorticity fluxes to the mean flow. Dole (1986 and1989) comprehensively examined the persistent anomaly pauems in the Northern Hemisphere winter circulations. He suggested that the quasi-horizontal energy dispersion is likely to account for the development of downstream anomalies. More recently, Higgins and Schubert (1994) studied the model simulated persistent Pacific anomalies. They indicated that the role of the time-mean flow and synoptic-scale eddies in the development of the persistent Pacific anomalies is different at different stages. However, the concurrent occur rence of blocking flow in both the Pacific and the Atlantic (i.e., double blocking), which has a significa nt impact on the regional weather over North Amer i ca and Europe, possesses some what different character i stics from the occurrence of single blocking (Dacamara et al., 1991: Min, 1995. Kung and Baker (1986) and Kung et al. (1989), in their studies of observed and simulated winter blocking episodes. showed that the development of winter blocking is associated with the nonlinear wave-wave transfer of kinetic energy from synoptic-scale disturbances to plan etary waves, and the successful blocking simulation depends on adequate amplif i cation of planetary waves. Tanaka and Kung (1988) Nakamura and Wallace ( 1990) showed evidence of the enhancement of baroclinic wave activity during the onset of blocking. Lejentis and Madden (1992) reported that 20-40% of blocks were related to travel ing wave I, whereas the percentage was small for wave 2. The statistical study of winter blocking by Hartmann and Ghan ( 1980) indicated that the mechanisms sustaining bloc king ridges are different in the Atlantic and the Pacific regions. By studying the monthly mean circulation palterns in the Northern Hemisphere winter during a 34-year period, Dacamara et al. ( 1991) reported that the distinction between the Pacif i c and Atlantic blockings appears in the stationary energy component of zonal wavenumber one. Their results also demonstrated that winter blocking activities are associated with different sea surface temperature (SST) anomaly patterns.
Since Deland ( 1964) introduced cross-spectral analysis into the study of large-scale propa gating waves, numerous studies of large-scale propagating waves may be found in literature (e.g., Hayashi, 1977;Fraedrich andSonger, 1978: Speth andKirk, 1981;Speth and Madden, 1983;Hansen et al .. 1989). However, most of the studies are based on a relatively shorter period and are restricted to a narrow latitude belt The main purpose of this study is to obtain the climatological characteristics of midlatitude wave activity in the wavenumber and frequency domain, and to explore the relationship of wave activity and winter blocking circulations in the Northern Hemisphere. A systematic analy-sis of the midlatitude wave activity and blocking circulations in the Northern Hemisphere winter is performed with the daily 500 mb height fields for the 1956-1991 period. The winter seasons are grouped into four different categories of circulation patterns on the basis of inten s i ty and location of blocking activities during the season. The large-scale circulation patterns are examined with space-time spectra, time spectra, kinetic energy, and nonlinear transfer of kinetic energy among planetary-scale and synoptic-scale waves. Further, a composite approach of winter blocking episodes during the 36 winter seasons is attempted to examine the roles of planetary-scale and synoptic-scale waves in different stages of blocking flow.

Data
Daily National Meteorological Center (NMC) octagonal grid analyses of the Northern Hemisphere 500 mb height fields are used in this study. The dataset covers 36 winter seasons from 1956 to 1991. The winter season is defined as December and the following January and February. The year of the winter season is idemified by the year in which January and Febru ary occur. Daily observational analyses were used at 1500 GMT from 1956 to 1957 and at 1200 GMT for the remaining 34 years. The available data were bilinearly interpolated from the 1977-point octagonal grid to the 4° x 5" latitude-longitude spherical grid from J 8°N to the pole. The missing data were interpolated linearly in time at each spherical grid point. Fourier transforms of the 500 mb height for zonal wavenumbers were then computed at the latitude circle on a daily basis. Wind components were obtained by the geostrophic approximation. The resulting time-dependent cosine and sine coefficients of height are treated as separate . time series for the 36 winter seasons, starting at November 15 and lasting for 120 days. Tem poral Fourier transform was utilized to determine 60 harmonic coefficients. Next, the power spectrum, the co-spectrum and the quadrature are estimated from the temporal Fourier coeffi cients. The spatial harmonics were cut off at wavenumber n= 12, beyond which the remaining zonal variance was verified to be negligibly small. A running average over five adjacent fre quencies was applied to each spectral estimate of power spectra, co-and quadrature spectra with a resulting frequency resolution of 5/120 days·•. Monthly SST analyses were obtained from the NMC real-time analyses (Reynolds, 1988) for the period of 1970-1991. The seasonal mean SST for the winter season was obtained as the average of monthly means of December, January and February (D.IF). Cross correlation analyses wilh SST were based on the 22-ycar datasets.
Hansen e1 al. ( 1989) compared several approaches of the partitioning techniques of the space-time spectral into standing and traveling parts. He pointed out that these techniques are qualitatively comparable. By following Hayashi' s approach of partitioning of the space-time spectra into standing and traveling wave parts (sec Hayashi, 1977), the standing variance ST(11. (J)) and the traveling variances TV(11. ± (J)) are respectively expressed as.
The procedure for identifying the blocking patterns follows that of Lejeniis and 0kland (1983) and Kung el al. (1989Kung el al. ( , 1990Kung el al. ( . 1992Kung el al. ( , 1993: In the current study, we attempt to utilize the re giona l circulation indices to classify the wi nter season circulation patterns. The focus of this classification is on the characteristics or dominant circulation patterns. Two regional indices, 11,, c and l,,,. i • are introduced separncely for the Pacific (160"E·120"W) and Atlantic (60"W-20"E) sectors. Monthly mean indices arc ob tained as the monthly averages of the daily values. The regional monthly indices are thereafter normalized with respect to their means and standard deviations during the 36-year period. The seasonal indices are obtai ned as the average of DJF normalized I e.c and i,m, for the 36 years.
The evaluation of the kinetic energy at wavenumber n, K(n), and the nonlinear transfer of kinetic energy from wavenumber m to 11, L(n,m), follows DaCamara et al. ( 1991) and Kung et al. (1993), and is detailed in Dacamara (1991). A comprehensive energetic description is not po s sible with the restricted dataset in this study except for kinetic energy and nonlinear wave wave interactions at 500 mb level, and the venical flux term is not involved in computati on of the nonlinear wave-wave interaction terms.

WAVENUMBER AND FREQUENCY SPECTRA
In this study, emphasis is placed on the midlatitude wave activities in the Nonhem Hemi sphere. The following presentations are for the computational results averaged in the 42°-62"N lati1ude belt, unless otherwise specified.
The long-term mean spectral density of 500 mb Z is shown in Figure I. The spectra are  ( 1989). In order to make equal areas on these diagrams represent equal variance, the spectral densities are normalized by their wavenumbers and fre quencies. The results of the 36-year mean spectral densities generally agree with those by Hansen et al. (1989) for a period of 16 years. The westward propagating variances show a spect ral peak at the ultralong waves (n=2-3) with a period of 20-40 days (Figure I a), and the eastward propagating variances show two spectral peaks: one near n=6-7 at roughly a 5-day period and another near n=5 at about the 10-day period. The peak of the westward propa gating variances occurs at n=2-3 instead ofn=I, as in Madden (1983) and Speth and Kirk (1981), because of the nor ma lizatio n. The standing variance (Figure I c) shows its largest power at n=2-4 and reaches its peak at n=4 at about a 20-day period. Table I gives the mean power spectra of 500 mb Z for different wavenumbers and fre quency bands at three lat it udinal belts of 18 ° -38 °N, 42 ° -62 °N and 66 ° -86 °N. It is seen that the power spectra of the short waves n=I0-14, compared with other waves, are neg lig ibl y small and the power spectra of the planetary-scale waves n=l-3 increase as the latitude in creases poleward. For the planetary-scale waves, the westward propagat i ng variances are more pronounced at the low-frequency band {T=20-60 days) in the mid and high latitudes. The synoptic-scale waves (n=4-9) show domina nt eastward motion in the midla titudes at the short period (T=3-6 days). The standing variances have the largest power for n= 1-3 at the low- frequency band in all latitude belts.

75
The standard deviation of 500 mb height field from the long-term seasonal mean field for C data period of 1956-1991 is display in Figure 2a. It shows three centers of large variability the Nonhern Hemisphere circulation: the eastern Nonh Pacific. the North Atlantic, and the tic region of the eastern hemisphere. The locations of the three centers are in general agree lnent with those of Blackmon et al. (1976) although slight differences exist. These differences \nay be related to the relatively shon datasets and the removal of the seasonal cycles in their �tudies. As Blackmon et al. (1986) and Metz (1986) demonstrated, these centers are also the preferential regions of winter blackings. The center in the North Pacific is extensively located between the dateline and J 50°W, indicating the importance of the Northern Pacific region in the variation of the Northern Hemisphere winter circulation. The long-term mean time spectra of 500 mb height for the low-frequency planetary-scale waves (n= 1-4, T=20-60 days), the medium-range synoptic-scale waves (n=4-9, T=8-15 days) and the short-period synoptic-scale waves (n=4-9, T=3-6 days) are presented in Figures 2b-d, respectively. Three maxima of the time spectra of the planetary-scale waves are clearly seen (Figure 2b). The locations of these maxima are very close to the standard deviation maxima (Figure 2a), indicating that the vari ability of the winter circulation in the Northern Hemisphere is mainly attributable to the low frequency planetary-scale waves. For the medium-range synoptic-scale waves (Figure 2c), the time spectra show maximum values in the midlatitudes over the eastern Pacific and western Atlantic, respectively, indicating the contribution of long-lived cyclones and/or anti-cyclones to the atmospheric variability. A noteworthy characteristic of the time spectra of the short period synoptic-scale waves is the two maxima over the western Pacific and the western At lantic along the paths of the winter storm track where the baroclinic instability is the larges!. These maxima are closely associated with the winter baroclinic wave activity in the midlatitudes.
It is generally recognized that the SST anomalies exert a major forcing on the winter circulation through heat release in the ocean-atmosphere system (e.g., Namias, 1964;Shukla. 1986;Kung et al., 1990). The GCM study by Mechoso et al. (1987) has suggested that wann equatorial Pacific SST anomalies associated with the ENSO may act to enhance high fre quency, intermediate scale wave activity in the Northern Hemisphere midlatitudes. However, the low-frequency atmospheric variability is also linked to slow changes in external forcing (e.g., Horel and Wallace, 1981;DaCamara et al., 1991;Kung et al., 1993). Figure 3 shows the correlation of SSTs with the power spectra of 500 mb Z for the westward propagating plan etary-scale waves (n= 1-3) with T=20-60 days, and the eastward propagating synoptic-scale waves (n=4-9) with T=3-6 days for the period of 1970-1991. The correlation coefficient of 0.40 is at 5% significance level. It is seen that the relationships between SST forcing and the planetary-scale wave activity are quite different among wavenumbers. The westward propa gating planetary-scale waves are associated not only with the equatorial SSTs but also with the midlatitude SSTs. However, the eastward propagating synoptic-scale waves show a high posi tive correlation with the SSTs in the western and central Pacific and negative correlations near the west coast of North America. A broad region of negative values is found in the North Atlantic sector. These may suggest that the synoptic-scale wave activity in the northern midlatitudes is associated with the SST anomalies over the midlatitude ocean. Table 2 presents the classification of the 36 winter seasons for 1956-1991, according to the specified categories of circulation: PAC, ATL, DBL, and NBL. The normalized values of I r.c and l,n are also given. It is cautioned that the classification of seasonal circulation pat terns is based on the blocking index. not on the actual occurrence of blocking episodes al- (c) westward propagating variance (n=3, T=20-60 day s) and (d) east ward propagating variances (n=4-9, T=3-6 days). Contour interval is 0.1.

77
though there is a general agreement between them. The anomaly patterns for PAC, A TL and DBL clearly show the dipo le structures in the respectively preferred regions of blocking for-  (Figure 4). For NBL, the anomaly field shows a reversed weak dipole structure with negative anomalies in the north and positive anomalies in the south. Madden ( 1983) found that the interference of stationary and propagating waves of the same longitudinal scale causes ti me variations in the large-scale circulation. Quiroz (1987) and Lejenas and Madden (1992) exam.ined the relationship between blocks and large-scale westward-propagating waves. They suggested that the reinforcement of ridges by retrogress ing n=l played an important role in the occurrence of blocking. Figure 5 shows the differences of power spectral density of 500 mb height from NBL for PAC. A TL and DBL. It is noted that the major difference occurs at the low-frequency planetary-scale and the high-frequency syn opti c-scale waves. For PAC, strong positive values are seen in the westward propagating vari ances at wavenumbers n= 1-4 with a larger than I 0-day period, indicating the amplification of the westward propagating planetary-scale waves during the occurrence of PAC. The eastward propagating variances show large negative values at n ;:: 5 with a less than 15-day period, indi cating the auenuati on of the synoptic-scale wa\'C activities when PAC dominates in the winter season. For the standing variance, slight increases can be seen in the low-frequency planetary s cale waves.
For ATL, the wes1ward propagating variance only slightly increase a1 the low-frequency portion (T;:: 20 days) with maximum at n=3 and T;:: 40 days (Figure 5b). For the eastward propagating variance, a large positive value is seen at n=6 with a period around 5 days, indicat-  A TL and DBL) and no blocking situations (NBL) for the low-fr equency plane t a ry waves (n= 1-4, T=20-60 days) and the high-frequency synoptic waves (n=4-9, T=3-6 days). For the Pacific blocking, the time spectra show strong positive values near Alaska, suggesting that Pacific blocking is closely associated with the large scale low-frequency oscillations in the winter circulation. It is also noted that large negative values are found in the two winter storm track centers (Figure 6a, right), indicating the weakening of the winter storm activities in both the Pacific and the Atlantic sectors when Pacific blocking dominates in the winter season.
In the case of A TL (Figure 6b ), the time spectra of the low-frequency planetary waves in   (Figure 6c). However, the time spectra for the high-frequency synoptic-scale waves are identical to that for PAC. This, in conjunction with the wavenumber-frequency spectra, suggest that in the occurrence of Pacific blocking (both PAC and DBL cases) the baroclinic activities in the midlatitudes ar e weak- ened. The revealed enhancement of high-frequency synoptic-scale wave activity during the Atlantic blocking is consistent with Metz's ( 1986) study of the observed and modeled blockings.
His results indi cated that the forcing of planetary flow by cyclone-scale eddy vorticity fluxes was able to induce and maintain blocking highs in the Atlantic sector.   of PAC possesses the largest val ue among categories. Por n=2 and 3, DBL shows distinctively large values. The mean K(4-9) of ATL is distinctively larger than the winter mean while K(4-9) of PAC is well below the winter se. 'l �on mean. This is consistent with the analyses of the power spectra and the time spectra.

Kinetic Energy
Although the balance of kinetic energy depends on a number of processes, they are not all directly related to the development of blocking patterns. Kung and Baker (1986), however. pointed out that the development of a single blocking in the Pacific or Atlantic was consis tently associated with an upscale kinetic energy input to n=I through wave-wave interaction. In the situation of double blocking, the energy input is observed at n=2. In their study of monthly circulation patterns, Dacamara el al. ( 1991) indicated that the monthly values of nonlinear transfers were generally comparable with winter means without discernible monthly variation. However, the year-to-year variations of these seawnal mean values are large. as it is shown in Table 4. The mean nonlinear wave-wave transfer of kinetic energy among the plan etary-scale waves for the four categories agrees with the study by Dacamara el (II. ( 1991) and the major case study by Kung el al. ( 1993 ). The winter seasonal mean of the nonlinear transfer between the planetary-scale waves a nd the synoptic-scale waves, as listed in Table 4, shows that generally n= I and 3 gains kinetic energy from synoptic-scale waves (n=5·9) and n=2 loses kinetic energy to n=5-9. Although these values are not significant in view of large standard deviation, the differences among categories are clearly seen. In the case of PAC, L( 1,5-9) shows a distinctively large negative value and its displacement is more than one standard deviation from the ensemble mean, L(2,5· 9) contr.L�tly shows a large positive value. For ATL, the nonlinear transfer of ki netic energy to the planetary-scale waves (n=l-3) from the synoptic-scale waves (n=5-9) shows consistently positive values. indicating the important roles of the synoptic-scale waves in the formation and maintenance of the Atlantic blocking circulation. For DBL, the large negative value of L(2,5-9) indicates that, during the occurrence of double blocking. n=2 transfers kinetic energy not only to n= l but also to the synoptic-scale waves through nonlinear wave-wave interaction. This supp()fts the argument by Kung et al. ( 1993) that the double blocking is baroclinic in nature. The dominant n=2 should be supported by the baroclinic energy source and may be come the barotropic kinetic energy source to other wavenumbers.

Blocking Episodes and Time Series Analysis
111e daily values of l ,.c and l,n are used to determine the blocking episodes in the Pacific and Atlantic sectors. Blocking episodes are selected as those stationary anticyclonic circula tions lasting for more than I 0 consecutive days. The following presentations and discussions are based on the composite means of 18 Pacific blocking episodes, 15 Atlantic blocking epi sodes. and 16 double blocking episodes. respectively although there exist episode-to-episode variations. The onset date and the duration for each episode are listed in Table 5. Figure 7 shows the composite mean flow patterns and anomaly fields of 500 mb height, which generally illustrate the same features as the climatological study. Austin (1980) re ported that n= I and 2 tend to interfere constructively in the Atlantic sector, and n=2 and 3 interfere in the Pacific sector. In the study of Kung et al. ( 1990), it was suggested that during the development of a Pacific blocking in January 1979, n=l and 2 came in phase at the location of the blocking, and later the interference of n=I and 2 in the Atlantic was identified with the development of a major Atlantic blocking. Figure 8 demonstrates the 500 mb trough-ridge diagram in the 54 ° • 70°N latitude band. Although the blocking indices are defined between 38 °-58 °N to account for the circulation patterns, the 54 ° -70 °N is used in this application because it is the latitudinal band in which the ridge is maximized (Kung er al., 1993). The "O" day marks the onset of blocking. It is noted that the planetary-scale waves n=l-3 interfere constructively in the longitudinal segments where blocking occurs. During the onset of Pacific blocking. traveling wave n= I come in phase with the quasi-stationary wave n=2. For Atlantic blocking, n= I ampl ifies at its normal position and reaches a peak 4-5 days after the onset of the blocking ri dge. In the double blocking situation, strong amplification of n=2 is found during the onset of blocking. As Madden (1983) indicated, the interference of stationary waves and traveling waves of the same longitudinal scale cause the time variations in the large-scale circulation.
Figures 9-1 1 are the composite time evolutions of K(n) and L(n,m). The x-axis denotes

87
indicati ng that n=2 has its source at n=3 in forming the Pacific blocking. The synoptic-scale waves also contribute to the ma intenance of n=2 after the onset. Subsequently, K( I) and K(3) .
In the case of double blocking, K(2) and K(3) increase simultaneously and reach their peaks at day 4, whereas K( I) decreases after the onset of blocking without adequate transfer of kinetic energy from n=2 (as seen in Figure 10). It is of interest to note that L(3, 5-9) reaches its peak I to 2 days before the peaks of K(3), which suggests that n=3 may be maintained by the synoptic-scale wave in the double blocking case. However, the kinetic energy of planetary scale waves (n= 1-3) for the Atlantic blocking shows no evidence of significant increase except that two days before the onset K( I) reaches its maximum. This may suggest that 1he blocking circulation in the Atlantic, in lack of adequate amplification of planetary-scale waves, might be forced by di fferent mechanisms. The kinetic energy of the synoptic-scale waves (n=4-9) for the Pacific blocking shows no significant changes during the 10 day period after the onset of blocking. For the Atlantic blocking. K(4-9) increases and maximizes at clay 5, and fluctuates  in a period of about 5-6 days. During the life cycle of double blocking, the K(4-9) reaches its peak at the onset day and fluctuates in a period of about S-10 days. The time series of L( 1,2), L( 1,3) and L(2.3) in Figure 10 also show large differences between different blocking circulations. For the Pacific blocking, L( 1,2) and L(2,3) increase during the course of blocking flow. A strong increase of L( 1,2) after the onset is well illus lrated for the Pacific blocking, indicating that n=l carr ies over a large amount of kinetic en ergy from n=2 after the blocking is fully developed. For the Atlantic blocking and double

10
. . . , . , . . blocking, L( 1,3), distinguished from Pacific blocking, increases after the onset of blocking and reaches its maximum at day 5. Although L( 1,3) has a small positive value at its peak, the net effect on the kinetic energy budget could be tremendous since n=3 normally receives a large amount energy from n= I (see Table 4). It seems likely that the forcing of the planetary. scale waves by the synoptic-scale waves through nonlinear wave-wave interaction is an im portant factor in the life cycle of blocking, particularly for the Pacific blockings.

S. CONCLUDING REMARKS
The characteristics of wave activities in the Northern Hemisphere winter for 1956-I 991 are examined using the space-time spectral analysis, and summarized in the wavenumber and L( 1,5-9), L(2,5-9), L(3,5-9). Unit: 10·5 m1s·3. 91 frequency domain. The results are in reasonable agreements with the results of previously available studies with shorter period datasets (e.g., Fraedrich and Bi:lttger, 1978;Speth and Madden, 1983;Hansen et al., 1989). As revealed by the spatial distributions of time spectra, the variability of the winter circulation in the Northern Hemisphere winter is attributed to the low frequency planetary-scale waves. The most prominent phenomenon in the winter circulation is the quasi-stationary anticy clonic ridges or blockings in the Pacific and the Atlantic. When they occur, they exert domi nant influences on the local weather. Classification of the seasonal circulation in the Northern Hemisphere winter on the basis of blocking activities, as used in this study, seems to ad equately identify four categories of winter circulation patterns: PAC (Pacific blocking), ATL (A1lan1ic blocking), DBL (double blocking), and NBL (no blocking). Char.icterislics among these categories are distinguished in terms of power spectra in the wavenumber and frequency domain, kinetic energy, and nonlinear energy transfer.
The differences of the power spectra among PAC, ATL and DBL are clearly seen in the standing variance and traveling variances. II is shown in this study 1ha1 Pacific blocking is associated wilh lhe strengthening of the low frequency planetary-scale waves and the Atlantic blocking is associated with lhe enhancement of the high frequency synoptic-scale waves. For lhe double blocking, standing variance shows peaks al n=2-4 with a larger than 10 day period.
Different circulation patterns may involve di fferent physical mechanisms. For the Pacific block ing, the barotropic energy transfer from n=2 to n=I plays an important role in forming and maintaining the blocking circulation; whereas for lhe Allanlic blocking the source of kinetic energy 10 sustain the blocking circulation is lhe baroclinic conversion from potential energy 10 kinetic energy. Time series analysis indicates lhat synoptic-scale waves are important in the development and maintenance of Atlantic and double blocking, and lhat planetary-scale waves play a crucial role in lhe Pacific blocking.
Although it would be difficult 10 draw definite conclusions from this study becaus e of1he limitations of lhe datasets, ii is demonstrated that nonlinear transfer of kinetic energy plays a n important role in the formation and maintenance of dominant circulation pauerns associated with winter blocking.