• Anderson, J. R., and J. R. Gyakum, 1989: A diagnostic study of Pacific basin circulation regimes as determined from extratropical cyclone tracks. Mon. Wea. Rev.,117, 2672–2686.

  • Barnston, A. G., and R. E. Livezey, 1987: Classification, seasonality and persistence of low-frequency atmospheric circulation patterns. Mon. Wea. Rev.,115, 1083–1126.

  • Blackmon, M. L., 1976: A climatological spectral study of the 500-mb geopotential height of the Northern Hemisphere. J. Atmos. Sci.,33, 1607–1623.

  • ——, J. M. Wallace, N.-C. Lau, and S. L. Mullen, 1977: An observational study of the Northern Hemisphere wintertime circulation. J. Atmos. Sci.,34, 1040–1053.

  • ——, Y. Lee, and J. Wallace, 1984: Horizontal structure of 500-mb height fluctuations with long, intermediate, and short time scales. J. Atmos. Sci.,41, 961–979.

  • Bretherton, C. S., C. Smith, and J. M. Wallace, 1992: An intercomparison of methods for finding coupled patterns in climate data. J. Climate,5, 541–560.

  • Cai, M., and H. M. Van den Dool, 1991: Low-frequency waves and traveling storm tracks. Part I: Barotropic component. J. Atmos. Sci.,48, 1420–1444.

  • Dickson, R. R., and J. Namias, 1976: North American influences on the circulation and climate of the North Atlantic sector. Mon. Wea. Rev.,104, 1255–1265.

  • Hurrell, J. W., 1995: Decadal trends in the North Atlantic Oscillation regional temperatures and precipitation. Science,269, 676–679.

  • Jones, P. D., S. C. B. Raper, B. S. G. Cherry, C. M. Goodess, T. M. L. Wigley, B. Santer, P. M. Kelly, R. S. Bradley, and H. F. Diaz, 1991: An updated global grid point surface air temperature anomaly data set: 1851–1990. ORNL/CDIAC-37, 65 pp. [Available from National Technical Information System, U.S. Dept. of Commerce, 5285 Port Royal Road, Springfield, VA 22161.].

  • Lamb, P. J., and R. A. Peppler, 1987: North Atlantic Oscillation: Concept and an application. Bull. Amer. Meteor. Soc.,68, 1218–1225.

  • Lau, N.-C., 1988: Variability of the observed midlatitude storm tracks in relation to low-frequency changes in the circulation pattern. J. Atmos. Sci.,45, 2718–2743.

  • Lejenas, H., 1989: The severe winter in Europe 1941–42: The large-scale circulation, cut-off lows, and blocking. Bull. Amer. Meteor. Soc.,70, 271–281.

  • Makorgiannis, T. J., B. D. Giles, and A. A. Flocas, 1981: The problem of the extension of the Siberian Anticyclone towards southeast Europe, and its relation to atmospheric circulation anomalies over the Northern Hemisphere. Arch. Meteor. Geophys. Bioclimatol. Ser. A,30, 185–196.

  • Mann, K. H., and K. F. Drinkwater, 1994: Environmental influences on fish and shellfish production in the northwest Atlantic. Environ. Rev.,2, 16–32.

  • Moses, T., G. N. Kiladis, H. F. Diaz, and R. G. Barry, 1987: Characteristics and frequency of reversals in mean sea level pressure in the North Atlantic sector and their relationship to long-term temperature trends. J. Climatol.,7, 13–30.

  • Namias, J., 1964: Seasonal persistence and recurrence of European blocking during 1958–60. Tellus,16, 394–407.

  • ——, 1978: Multiple causes of the North American abnormal winter 1976–77. Mon. Wea. Rev.,106, 279–295.

  • Rex, D. F., 1951: The effect of Atlantic blocking action on European climate. Tellus,3, 100–112.

  • Rogers, J. C., 1984: The association between the North Atlantic Oscillation and the Southern Oscillation in the Northern Hemisphere. Mon. Wea. Rev.,112, 1999–2015.

  • ——, 1985: Atmospheric circulation changes associated with the warming over the Northern Atlantic in the 1920s. J. Climate Appl. Meteor.,24, 1303–1310.

  • ——, 1990: Patterns of low-frequency monthly sea level pressure variability (1899–1986) and associated wave cyclone frequencies. J. Climate,3, 1364–1379.

  • ——, and E. Mosley-Thompson, 1995: Atlantic Arctic cyclones and the mild Siberian winters of the 1980s. Geophys. Res. Lett.,22, 799–802.

  • Serreze, M. C., 1995: Climatological aspects of cyclone development and decay in the Arctic. Atmos.–Ocean,33, 1–23.

  • ——, J. C. Rogers, F. Carse, and R. G. Barry, 1997: Icelandic low cyclone activity: Climatological features, linkages with the NAO and relationships with recent changes in the Northern Hemisphere circulation. J. Climate,10, 453–464.

  • Shabbar, A., K. Higuchi, and J. L. Knox, 1990: Regional analysis of Northern Hemisphere 50 kPa geopotential heights from 1946 to 1985. J. Climate,3, 543–557.

  • Trenberth, K. E., and D. A. Paolino, 1980: The Northern Hemisphere sea level pressure data set: Trends, errors and discontinuities. Mon. Wea. Rev.,108, 855–872.

  • van Loon, H., and J. Williams, 1976: The connection between trends of mean temperature and circulation at the surface. Part I: Winter. Mon. Wea. Rev.,104, 365–380.

  • ——, and J. C. Rogers, 1978: The seesaw in winter temperatures between Greenland and northern Europe. Part I: General description. Mon. Wea. Rev.,106, 296–310.

  • Wallace, J. M., and D. S. Gutzler, 1981: Teleconnections in the geopotential height field during the Northern Hemisphere winter. Mon. Wea. Rev.,109, 784–812.

  • ——, G.-H. Lim, and M. L. Blackmon, 1988: On the relationship between cyclone tracks, anticyclone tracks and baroclinic waveguides. J. Atmos. Sci.,45, 439–462.

  • ——, C. Smith, and C. S. Bretherton, 1992: Singular value decomposition of wintertime sea surface temperature and 500-mb height anomalies. J. Climate,5, 561–576.

  • View in gallery

    A 120-winter-month (1951–90) average of the rms of high-pass filtered daily sea level pressures (in mb). Climatological data sites mentioned in the text are represented by dots in western Greenland, at Oslo, the Azores, and at gridpoint 65°N, 20°W in Iceland.

  • View in gallery

    Time series of scores associated with the first rotated principal component of the monthly rms fields of high-pass filtered sea level pressures. Monthly (thin solid line) and seasonal (thicker solid line) values are shown, including zero values for the missing data from December 1944 through 1945.

  • View in gallery

    Composites of monthly rms (in mb) for sets of months with extreme opposite modes of the first principal component of Atlantic area rms fields of high-pass filtered sea level pressures, 1900–92. The three diagrams include the (a) net mean rms differences (mb) between the (b) composite positive mode cases and (c) the composite negative cases.

  • View in gallery

    Composites of Atlantic mean sea level pressures (mb) for subset groups when the monthly scores (Fig. 2) of the first principal component monthly rms of high-pass filtered sea level pressures are (a) lower than −1.0; (b) between −1.0 and zero; (c) between zero and +1.0; (d) between +1.0 and +2.0; and (e) for cases higher than +2.0. Lighter and darker shading represent areas where the differences in pressure are statistically significant with 95% and 99% confidence between different combinations of maps. Shading in (b) through (e) represents significant differences with the preceding map while shading in (a) represents significant differences between (a) and (e).

  • View in gallery

    (Continued )

  • View in gallery

    Frequencies of once-daily pressures at 65°N, 20°W during winter months when the first principal component of monthly rms of high-pass filtered pressures has extreme (a) positive and (b) negative scores and frequencies of absolute day-to-day pressure changes at 65°N, 20°W during the same (c) positive and (d) negative months.

  • View in gallery

    The mean sea level mean pressure differences (mb) occurring between the months with extreme positive and negative scores of the first rotated pattern of the combined principal component analysis (CPCA) of monthly Atlantic root-mean-squares of high-pass filtered SLP and monthly mean sea level pressure, 1900–92. Lighter and darker shading represent areas where the differences in pressure are statistically significant with 95% and 99% confidence.

  • View in gallery

    Spatial distribution of coefficients of correlation between RPCA scores of the Atlantic storm track eigenvector and gridded winter mean air temperatures for land areas of the Northern Hemisphere, 1900–90 (from Jones et al. 1991). Correlation coefficients of r = ±0.32, r = ±0.44, and r = ±0.55 are significant at the 95%, 99%, and 99.9% confidence levels.

  • View in gallery

    Frequencies of RPCA storm track scores, at increments of 0.5, during individual winter months 1900–92 when the Greenland above (GA) and Greenland below (GB) modes occur because of the seesaw in winter air temperatures between Greenland and northern Europe.

  • View in gallery

    Mean sea level pressures (mb) when the Greenland below seesaw mode occurs and the RPCA storm track scores are (a) positive and (b) negative. (c) The net pressure differences, (b) minus (a), between those sets of cases. Mean sea level pressures are also shown (d) for the Greenland above seesaw cases that occur with negative RPCA storm track scores and for (e) the net pressure differences for the sets of cases, (d) minus (b). Lighter and darker shading in (c) and (e) represent areas where the differences in pressure are statistically significant with 95% and 99% confidence.

  • View in gallery

    (Continued )

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 440 440 29
PDF Downloads 312 312 26

North Atlantic Storm Track Variability and Its Association to the North Atlantic Oscillation and Climate Variability of Northern Europe

View More View Less
  • 1 Department of Geography, Ohio State University, Columbus, Ohio
© Get Permissions
Full access

Abstract

The primary mode of North Atlantic storm track variability is identified using rotated principal component analysis (RPCA) on monthly fields of root-mean-squares of daily high-pass filtered (2–8-day periods) sea level pressures (SLP) for winters (December–February) 1900–92. It is examined in terms of its association with 1) monthly mean SLP fields, 2) regional low-frequency teleconnections, and 3) the seesaw in winter temperatures between Greenland and northern Europe. The principal storm track component is characterized by high synoptic variability preferring one of two areas at any given time. The northeastern Atlantic center (identified by positive RPCA scores) is characterized by deep cyclones in the area extending from Iceland northeastward to the Norwegian and Barents Seas, whereas the Bay of Biscay center (negative scores) is linked to cyclone activity around that area and into the Mediterranean basin. Combined principal component analysis is used to link the high-frequency storm track pressure variability with that of lower frequencies (monthly mean pressures). In this, the primary storm track pattern is linked to large monthly mean SLP variations around the Bay of Biscay and near northern Scandinavia and the Barents Sea. This pattern does not suggest a strong storm track link to the North Atlantic Oscillation (NAO). Instead, the results presented indicate that the dominant mode of variability in the storm track is associated with low-frequency SLP anomalies in the extreme northeastern Atlantic. When the component scores reach their highest positive values, both the mean Atlantic subpolar low and subtropical high are unusually strong and displaced farther northeast than normal. The pressure field intensifies to the northeast and produces strong zonal flow extending into Europe, bringing abnormally high surface air temperatures as far east as Siberia and below normal temperatures over Greenland and northern Africa (the “Greenland below” seesaw mode, GB). Besides this eastward extension of the mean pressure field, anomalously high European winter temperatures can also be somewhat less frequently caused by mild return flow around the Siberian high, which is displaced farther west than normal. In this situation the Icelandic low is in its normal Denmark Strait location and cyclones move along the more southerly storm track toward the Mediterranean basin, contributing to the synoptic forcing that helps develop the westward extended high. The NAO appears to be only indirectly linked to the European component of the GB mode of the winter surface air temperature seesaw.

Corresponding author address: Dr. Jeffrey C. Rogers, Department of Geography, Ohio State University, Columbus, OH 43210-1361.

Email: jrogers@geography.ohio-state.edu

Abstract

The primary mode of North Atlantic storm track variability is identified using rotated principal component analysis (RPCA) on monthly fields of root-mean-squares of daily high-pass filtered (2–8-day periods) sea level pressures (SLP) for winters (December–February) 1900–92. It is examined in terms of its association with 1) monthly mean SLP fields, 2) regional low-frequency teleconnections, and 3) the seesaw in winter temperatures between Greenland and northern Europe. The principal storm track component is characterized by high synoptic variability preferring one of two areas at any given time. The northeastern Atlantic center (identified by positive RPCA scores) is characterized by deep cyclones in the area extending from Iceland northeastward to the Norwegian and Barents Seas, whereas the Bay of Biscay center (negative scores) is linked to cyclone activity around that area and into the Mediterranean basin. Combined principal component analysis is used to link the high-frequency storm track pressure variability with that of lower frequencies (monthly mean pressures). In this, the primary storm track pattern is linked to large monthly mean SLP variations around the Bay of Biscay and near northern Scandinavia and the Barents Sea. This pattern does not suggest a strong storm track link to the North Atlantic Oscillation (NAO). Instead, the results presented indicate that the dominant mode of variability in the storm track is associated with low-frequency SLP anomalies in the extreme northeastern Atlantic. When the component scores reach their highest positive values, both the mean Atlantic subpolar low and subtropical high are unusually strong and displaced farther northeast than normal. The pressure field intensifies to the northeast and produces strong zonal flow extending into Europe, bringing abnormally high surface air temperatures as far east as Siberia and below normal temperatures over Greenland and northern Africa (the “Greenland below” seesaw mode, GB). Besides this eastward extension of the mean pressure field, anomalously high European winter temperatures can also be somewhat less frequently caused by mild return flow around the Siberian high, which is displaced farther west than normal. In this situation the Icelandic low is in its normal Denmark Strait location and cyclones move along the more southerly storm track toward the Mediterranean basin, contributing to the synoptic forcing that helps develop the westward extended high. The NAO appears to be only indirectly linked to the European component of the GB mode of the winter surface air temperature seesaw.

Corresponding author address: Dr. Jeffrey C. Rogers, Department of Geography, Ohio State University, Columbus, OH 43210-1361.

Email: jrogers@geography.ohio-state.edu

1. Introduction

Climate variability associated with changes in intensity and location of storm tracks has not been studied to as great an extent as that associated with low-frequency components of the circulation such as standing waves (van Loon and Williams 1976; Shabbar et al. 1990), blocking flows (Rex 1951; Namias 1964, 1978; Dickson and Namias 1976; Lejenas 1989), and atmospheric teleconnection patterns. The latter constitute a set of slowly varying circulation features that retain their identities on monthly charts, each having specific spatial centers of action. In the lower troposphere over the North Atlantic, the North Atlantic Oscillation (NAO) is typically regarded as the primary regional teleconnection and is linked to observed climatological and oceanographic variability (van Loon and Rogers 1978; Lamb and Peppler 1987; Moses et al. 1987; Mann and Drinkwater 1994; Hurrell 1995). There are, however, other Atlantic regional teleconnections, including the west Atlantic and east Atlantic patterns, identified at the 500-mb level (Wallace and Gutzler 1981); the NAO; east Atlantic and Eurasian patterns at 700 mb (Barnston and Livezey 1987); and three additional sea level pressure patterns beside the NAO (Rogers 1990).

It is well established that synoptic-scale eddy activity is largest downstream of the major stationary wave troughs. For example, the amplitudes of the variance statistics of bandpass filtered (2.5–6-day periods) 500-mb heights are characterized by zonally elongated maxima extending across the western ocean basins from the east coasts of North America and Asia, representing areas of high temporal variability in geopotential heights and preferred trajectories of weather systems (Blackmon et al. 1977; Blackmon et al. 1984). Cai and Van den Dool (1991) demonstrate that storm tracks also occur ahead of troughs associated with traveling low-frequency waves. These waves are associated with monthly averaged circulation anomalies that are frequently associated with modes of specific teleconnection patterns. Lau (1988) showed that teleconnection patterns found in monthly time averages are linked to storm tracks. For example, he linked large latitudinal/meridional displacements in the easternmost or downstream portion of the Atlantic 500-mb storm track to the east Atlantic teleconnection pattern, while baroclinic activity in the westernmost or upstream portion of the storm track is linked to the west Atlantic pattern.

Relatively few papers have addressed how storm track variability is associated with regional variability of climate, as measured by monthly mean fields of surface air temperature, precipitation, and pressure. This may partly be due to difficulty in applying cyclone frequency and trajectory datasets, and interpreting results from them, and due to an emphasis on monthly mean circulation anomalies in studies of climate variability. The recent development of automated cyclone tracking algorithms (Serreze 1995) and the use of second-moment statistics of pressures filtered over synoptic timescales (Lau 1988) has helped focus interest on the role of storm tracks in climate variability. For example, Rogers and Mosley-Thompson (1995) link recent increases in storm activity in the Barents and Kara Seas to unusually mild winters of the 1980s in Siberia, and Serreze et al. (1997) examine the synoptic characteristics associated with the mean Icelandic low and recent atmospheric circulation changes.

The purpose of this paper is 1) to identify the primary mode of Atlantic storm track variability, 2) to show how it is related to low-frequency teleconnections, and 3) to relate it to regional climate variability in northern Europe and elsewhere. The paper argues for a new interpretation of how atmospheric circulation variability is linked to a well-known regional climatic phenomenon—the seesaw in winter air temperatures between Greenland and northern Europe. The air temperature seesaw was first qualitatively linked to the North Atlantic Oscillation by van Loon and Rogers (1978, hereafter vLR), and the assumed linkage has often been repeated. The Atlantic storm track and its long-term variability is examined with a historical sea level pressure dataset extending back to 1899. The analysis is performed (see section 2) so as to obtain both a spatial representation of the storm track pattern plus a long-term temporal index of the strength and polarity of the storm track. The storm tracks are also analyzed in conjunction with composites based on monthly mean pressure fields, in which the mean lows can generally be associated with the site of most frequent cyclone passages and even most frequent cyclogenesis and cyclolysis (Serreze et al. 1997).

Storm tracks can be identified by following low pressure centers on synoptic charts and plotting their trajectories on maps, thereby producing “cyclone tracks” in the pure sense. Serreze’s (1995) automated cyclone detection and cyclone tracking algorithm represents an improvement on trajectory methods and is used on twice-daily gridded National Meteorological Center (NMC, now known as the National Centers for Environmental Prediction) analyses in order to examine climatological characteristics of cyclones and their trajectories. Lau (1988) identifies storm track modes by applying empirical orthogonal function (EOF) analysis of the monthly root-mean-square statistics of bandpass (2.5–6 day) filtered twice-daily 500-mb geopotential heights. The unique patterns in the variance statistics represent a simple way to evaluate storm tracks, involving fewer arbitrary decisions than traditional trajectory (manual) methods (Wallace et al. 1988), but creating proxies of the cyclone tracks (Anderson and Gyakum 1989) that do not identify individual cyclone and anticyclone centers as might be required when diagnosing instantaneous weather conditions. Wallace et al. (1988) point out that both cyclones and anticyclones are associated with the maxima in the bandpass filtered pressure field variances. Although the usage of “storm tracks” is somewhat misleading in describing the product of variance methodologies, the isolated patterns have been found to closely match those found by manual methods. Hence, in this study, the phrase “storm track” refers to high-frequency fluctuations in the filtered pressure fields rather than to the trajectory of individual cyclones.

2. Data and methodology

Northern Hemisphere gridded daily and monthly mean sea level pressure (SLP) data are used. The data are available at every 5° of latitude and longitude from 20° to 85°N for the period November 1899–March 1992. Gridded maps are available once daily for either 1300Z (from 1899 to 1939) or 1200 UTC and are available twice daily (0000 and 1200 UTC) from 1955–56 through 1959–60 and for all winters starting with 1962–63. Daily maps are missing from December 1944–December 1945, but monthly charts are available for calendar year 1945. Other than this 13-month period, missing daily data were replaced by pressure averages of the day prior and the day after. The different SLP data sources are listed in Table 1 of Trenberth and Paolino (1980), who extensively examine the monthly SLP data for errors, inhomogeneities, trends, and discontinuities.

Monthly mean surface air temperature data on a 5° lat × 10° long grid (Jones et al. 1991) for 1899–1990 are used. Monthly surface air temperatures are also used for Oslo (59.9°N, 10.7°E), Norway, and Jakobshavn (69.2°N, 51.0°W) and Egedesminde (68.7°N, 52.8°W) in western Greenland. Temperature departures at the latter two stations are combined to make a complete western Greenland record, necessitated by a lack of data at Jakobshavn after 1970.

Once-daily gridded SLP data, spanning the period 27 November–4 March from 1899–1900 through 1991–92, were high-pass filtered using a binomial filter with weights −0.0625, −0.25, +0.625, −0.25, and −0.0625—the weights associated with the low-pass binomial filter of n = 4 (1-4-6-4-1). The filter has maximum response in the 2–8-day periodicity range, typically associated with passage of synoptic systems. The rms of the high-pass filtered data are then obtained for each winter month. A rotated principal component analysis (RPCA) is performed on Atlantic monthly rms values extending from 80°W to 20°E and from 30° to 70°N, following procedures outlined by Barnston and Livezey (1987) and Rogers (1990). The input data for the RPCA are not areally weighted.

RPCA was also performed on 1) monthly rms of 1948–92 once-daily SLP data in order to compare the 93-yr analysis to potentially more accurate recent daily data, and 2) on the monthly rms of twice-daily SLP data. The latter evaluation employs Blackmon’s (1976) 31-weight bandpass filter designed for twice-daily data that emphasizes 2.5–6-day periods. The first two component patterns of the additional RPCA analyses closely replicate spatial patterns and temporal score variations found in the 1899–1992 analysis. The first component pattern and its relation to regional climate variability is the subject of this paper. The second component pattern in each analysis exhibits a center of maximum pressure variability west of Greenland, similar to the second cyclone track pattern described by Serreze (1995). The time series scores of this component are not significantly correlated to monthly mean surface air temperatures elsewhere in the North Atlantic region, unlike those of the first component, and are not examined further here. Components beyond number 2 had fewer similarities among the three analyses. The first component pattern explained between 27% and 29% of the total dataset variance in all three analyses.

A combined principal component analysis (CPCA) was performed, with rotation, to examine the spatial interrelation between high-pass filtered SLP variability and that of monthly mean SLP fields. CPCA helps analyze the relationship within and among spatially and temporally large datasets (Bretherton et al. 1992). The methodology was found to extract coupled patterns between datasets somewhat more accurately than canonical correlation analysis and singular value decomposition (Bretherton et al. 1992), and Wallace et al. (1992) similarly find CPCA comparable to singular value decomposition as a methodological tool. Monthly mean SLPs used in the analysis are from Rogers (1990), except that they are updated to 1992 and exclude data for 1945, conforming with the rms dataset. Monthly pressures are on a 20° long × 5° lat, 13 × 11 grid extending from 160°E eastward to 40°E. The grid could not include Eurasia due to missing data before 1945. The CPCA is performed after standardizing the monthly data on both grid sets by creating departures from normal and then dividing by the monthly standard deviations.

3. The Atlantic storm track and SLP analyses

a. Climatology of the rms high-pass filtered SLP

Figure 1 shows the mean rms of high-pass filtered SLPs, averaged over 120 winter months from December 1950 through February 1990. The mean rms exceeds 6 mb over the northeastern United States, eastern Canada, and into the west-central Atlantic, reaching a maximum of 6.7 mb at grid point 45°N, 60°W near Nova Scotia. This maximum is very near that found by Blackmon et al. (1977, their Fig. 2b) at sea level and by Lau (1988, his Fig. 1) at 500 mb. The general configuration of rms values above 4 mb is similar to that found in these earlier studies although the area in Fig. 1 extends farther north and east toward the Barents Sea than Lau’s rms maxima. The 5-mb contour is oriented northwest–southeast over the Canadian Plains east of the Rocky Mountains, and in the Atlantic it extends to Iceland and eastward toward the United Kingdom. The mean rms values in the subtropical Atlantic are generally lower than 2 mb south of 30°N. The Pacific basin rms maxima reaches 5.8 mb at 45°N, 170°E and does not extend as far northeastward into high latitudes as in the Atlantic.

b. The storm track pattern scores

The scores of the first rotated principal component are a time series of the primary mode of Atlantic monthly rms fields of high-pass filtered SLPs. Monthly RPCA scores are standardized and have numerical values ranging from −2.1 to +3.3 (Fig. 2). Using an arbitrary cutoff score of ±1.25, the 32 highest positive and 27 lowest negative monthly scores are identified and used to illustrate spatial changes associated with extremes of the RPCA storm track pattern. Differences in monthly mean rms values (Fig. 3a), obtained by subtracting the mean rms distribution during negative cases from those of positive cases, have a spatial pattern very similar to that displayed in the rotated principal component loadings. The largest mean rms variations form a dipole (Fig. 3a) with centers in the extreme northeastern Atlantic and Norwegian Sea, where the net mean rms differences exceed 4 mb, and over the eastern Atlantic around Portugal. The mean rms differences between the two datasets are statistically significant at the 95% confidence interval, based on a two-tail t-test, over large areas of the northern Atlantic from 50°W to 50°E and from 55°N to 80°N, as well as over southern Siberia at 55°–60°N.

The composite mean rms for the 32 largest positive cases (Fig. 3b) exceeds 7 mb over Newfoundland and Labrador, and in the area around Iceland, with values over 5 mb from the East Greenland Sea northeastward to Novaya Zemlya. The rms maximum exceeds that of the long-term mean (Fig. 1) by over 2 mb in the northeastern Atlantic. The composite rms for the 27 negative cases (Fig. 3c) exhibits a maximum near Newfoundland, extending from Maine to nearly the southern tip of Greenland, and comparatively low values in the northeastern Atlantic, only reaching 3–4 mb. The main axis of the rms maxima is oriented toward the Bay of Biscay and the Mediterranean basin.

The rms values in the southern dipole center west of Portugal (Fig. 3a) range from 2 to 3.5 mb in the positive mode cases (Fig. 3b), but they are 3 to 4.5 mb in the negative mode cases (Fig. 3c). The net mean rms differences around the dipole center (Fig. 3a) are statistically significant with 95% confidence from 0° to 30°W and 30° to 45°N, due to small rms variability at these latitudes in the cases that make up the composites.

Composite means of raw monthly SLPs were obtained after stratifying the RPCA scores into five groups separated at points corresponding to numerical score values of −1.0, zero, +1.0, and +2.0 (Fig. 4). This data stratification illustrates changes occurring in mean intensity and spatial locations of Atlantic centers of action as score values change. The composite mean for the set of months with scores between zero and +1.0 (Fig. 4c) resembles closely the long-term mean Atlantic SLP field with a minimum of 996 mb over the Denmark Strait and a subtropical maximum near the Azores at 30°N, 30°W with a central pressure of just over 1024 mb.

In months with scores lower than −1.0 (Fig. 4a), the subtropical high and Icelandic low are weaker than normal and shifted to the south and west of their mean positions (Fig. 4c). The mean low (1001 mb) extends over a large area southeast of Greenland and relatively high pressure between 1010 and 1016 mb occurs across the northeastern Atlantic and the Barents and Kara Seas to Novaya Zemlya. The Azores high is relatively weak (1021 mb) and lies in the south-central North Atlantic near 25°N, 45°W (Fig. 4a). A trough of comparatively low pressure extends toward the Bay of Biscay and across southern Europe and the Mediterranean Sea. This case corresponds primarily to that of Fig. 3c with comparatively higher than normal mean rms values over the east-central Atlantic, suggesting an active storm track toward Portugal and the Mediterranean and a weak subtropical high.

In months with the highest positive RPCA scores (Fig. 4e), both the mean subpolar low and the subtropical high extend farther northeast of normal. The mean subpolar low is 994 mb at 70°N 10°E in the Norwegian–Barents Sea area, with pressure under 996 mb as far east as 50°E near Novaya Zemlya. The highest rms values of high-pass filtered pressures occur (Fig. 3b) over Iceland and farther northeast suggesting that cyclone activity proceeds northeastward into the Norwegian and Barents Seas and even farther east in these cases. The subtropical high extends northeast of normal, well over the Mediterranean Basin, with a maximum pressure of 1028 mb. The mean SLP is 1008–1012 mb around the Bay of Biscay in Fig. 4a but it is about 1026 mb in Fig. 4e. Shading in Fig. 4a shows that SLPs are significantly different between Figs. 4a and 4e in areas centered over southern Europe and northern Europe and the northeastern Atlantic.

It is apparent from Fig. 4 that with increasingly positive RPCA scores 1) the mean subpolar low intensifies as it shifts to the northeast, 2) the subtropical high intensifies and migrates northeastward of its mean position, 3) the pressure gradient between the centers of action intensifies as they shift northeastward, and 4) the storm track shifts from a northwest–southeast orientation (for low negative scores) to a southwest–northeast orientation, extending deep into the high Arctic. The eastward shift in the subpolar and subtropical SLP fields is apparent in the eastward shifts in areas of statistically significant pressure differences in Figs. 4b–d.

A further example of the nature of synoptic variability is presented in Fig. 5, showing the frequency distribution of daily raw pressures and absolute daily pressure differences at 65°N 20°W, a grid point over Iceland with large differences in monthly rms values (Fig. 3a) and mean pressures (Figs. 4a and 4e). During 32 months with extreme positive scores (Fig. 5a), daily pressures are most frequent between 995 and 1000 mb (998 mb mean), and they are most frequent between 1005 and 1010 mb (1009 mb mean) in months with negative scores (Fig. 5b). The range of daily pressures is roughly the same in both sets of extreme cases. However, in months with the lowest negative scores (Fig. 5b) there is a greater frequency of occurrence of pressures between 1015 and 1025 mb than in the positive cases (Fig. 5a) and far fewer pressures under 995 mb.

The day-to-day changes in SLP are examined during different categories of storminess as defined in the index. During the 27 months of lowest scores, 20 pairs of days show an SLP change that exceeds 20 mb from one day to the next (Fig. 5d), whereas daily pressure changes of less than 6 mb account for over half of the pairs of days. The daily absolute pressure change exceeds 20 mb on 203 pairs of days in positive months (Fig. 5c), about 21% of all cases. In summary, the daily absolute pressure changes at 65°N, 20°W tend to be higher (lower) when the mean monthly rms of high-pass filtered SLPs are higher (lower) and when the monthly mean SLP is lower (higher). The distributions of Figs. 5a and 5b, and 5c and 5d, were statistically different from each other with above 99% confidence, estimated using the chi-square test.

The differences in daily pressures between Figs. 5a and 5b, as well as those occurring between the extremes in monthly means in Figs. 4a and 4e, may be partly due to interannual changes in the background low-frequency (>8 days) circulation. The higher-frequency synoptic activity in a given month is superimposed upon a prevailing background low-frequency climatological mean monthly pressure field. Figure 5 suggests that, while there may be a contribution by the background climatological mean pressure field, the low mean pressures at 65°N, 20°W of the positive RPCA score months are accompanied by sizeable day-to-day pressure changes and many more daily pressures under 1000 mb. Conversely, low score cases have smaller daily pressure changes with higher mean pressures, suggestive of large, slow moving anticyclones. The results of Fig. 5 are consistent with the findings of Serreze et al. (1997) that a considerable increase occurs in the number of cyclones in the Denmark Strait area when the mean Icelandic low is unusually deep.

c. Association to low-frequency teleconnections

A combined principal component analysis (CPCA) was used to determine the spatial interrelationships between monthly high-pass filtered data and monthly mean SLPs, and to examine the degree to which the storm track information can be linked to known atmospheric teleconnections in monthly pressure fields. The CPCA produces a single time series of scores but eigenvector loadings for both datasets.

The correlation between scores of the monthly rms of high-pass filtered SLPs (Fig. 2) and the scores for the CPCA is r = +0.912 for n = 275 months (significance exceeding 99.9%). As such, the rms field loadings (not shown) for the CPCA are similar to those of Fig. 3a with only two slight differences: 1) the mean maximum rms difference is not as large (about 3.6 mb), and 2) it is shifted slightly eastward over the Norwegian Sea at 65°N, 5°E. The monthly mean SLP difference field in the CPCA (Fig. 6) exhibits a maxima over the Bay of Biscay and over northern Scandinavia and the Barents Sea, areas where the pressure variations between Figs. 4a and 4e are among the largest with high statistical significance (Fig. 4a). Note that the pressure anomaly pattern indicates that cold northerly flow would occur over western Greenland and milder maritime flow over northern Europe. The spatial dipole pattern of SLP in Fig. 6 does not resemble that typically associated with the NAO in which centers occur near the Azores and the Denmark Strait near Iceland. Instead the southern center in Fig. 6 is near the Bay of Biscay and the northern center is considerably northeast of Iceland. Areas of statistical significance in Figs. 4a and 6 are primarily located farther east of the Azores and Iceland, areas where three other teleconnections identified by Rogers (1990) have centers of action.

d. The seesaw in winter surface air temperatures

1) Stratification of seesaw events using the storm track scores

Winter means of the storm track scores (Fig. 2) were correlated to gridded hemispheric winter seasonal surface air temperatures spanning 1900–90 (Jones et al. 1991). Statistically significant coefficients of correlation (Fig. 7) have maximum positive values over Ireland, the United Kingdom, and southern Scandinavia extending eastward into north-central Asia between 55°–75°N and 70°–100°E. Positive correlations imply that an anomalously northeastward extension of the storm track (Fig. 3b) is associated with higher than normal winter surface air temperatures over Europe and Eurasia (Rogers and Mosley-Thompson 1995) and below normal air temperatures over western Greenland, Baffin Island, the Mediterranean Basin, and northern Africa. Conversely, Africa and the Mediterranean have above normal temperatures when the storm track scores are negative, occurring as storms migrate toward the Mediterranean basin and during which northern Europe and Eurasia have unusually cold winters.

The winter air temperature seesaw between western Greenland and northern Europe can be identified in Fig. 7. The seesaw is characterized by two temperature anomaly modes named “Greenland below” (GB) and “Greenland above” (GA), referring to the western Greenland temperature anomaly, and in which the northern Europe (represented by Oslo) temperature anomaly has the opposite sign. In keeping with vLR, monthly seesaw extreme GB and GA events occur if the sign of the western Greenland temperature anomaly is opposite that at Oslo, with an absolute temperature anomaly difference between them (Greenland minus Oslo) exceeding 4°C.

The storm track scores are negative in 55 of 67 GA winter months since 1899 (Fig. 8a), with scores most frequently falling between −0.5 and −1.5. Cyclone activity is concentrated near southern Greenland in these cases, and much higher mean pressures occur over northern Europe (Figs. 4a and 4b). On the other hand, the storm track index values have a wider distribution across the 63 GB events with 24 negative cases and only 39 positive (Fig. 8b). The two distributions differ significantly from each other with 95% confidence. Histograms such as these (Fig. 8) were constructed individually for temperature anomalies greater than absolute 4°C at Greenland and at Oslo, ignoring the seesaw criteria, and were then stratified by storm track scores. The tendency for a broader distribution of storm track scores across positive temperature anomalies is very predominant at Oslo (not shown), more so than is shown in Fig. 8b, but it occurs to a much lesser extent for the negative temperature anomalies at Greenland. The results suggest that, like GA cases, about 40% of GB cases have large synoptic activity along the southern dipole storm track.

2) SLP associated with GB and positive storm track anomalies

Mean SLP composites are created for the 30 GB cases when the storm track scores are greater than +0.5 and for 24 cases when the scores were negative, ignoring nine “overlap” cases with scores between zero and +0.5. The positive score months (Fig. 9a) are characterized by broad subpolar low pressure with centers west of Iceland and over the Norwegian Sea. Pressures under 1000 mb extend to Novaya Zemlya and the isobars across the double low extend zonally into Europe. The Atlantic subtropical high, as measured by the 1024-mb isobar, extends farther northeast than usual, and strong maritime westerly flow extends well into Europe and Asia. The entire pattern is typical of high RPCA scores in Figs. 4d and 4e.

A deep low also occurs near Iceland in GB cases with negative scores (Fig. 9b), but the isobars to the east generally lie parallel to the Scandinavian coast. Comparatively high pressure covers the Barents and Kara Seas. The Siberian anticyclone, as measured by its 1020-mb isobar, spreads much farther north and west in Fig. 9b than it did in Fig. 9a, while the Atlantic subtropical anticyclone is displaced west, over the midocean basin.

Mean SLP differences between these modes (Fig. 9c) consist of a dipole with centers over eastern Europe and southwest of Ireland with a strong pressure gradient between the two centers lying across much of Scandinavia and northern Europe. The eastern European dipole (Fig. 9c) is a center of anomalous high pressure in the GB–negative score months (Fig. 9b), and the flow around this anticyclonic anomaly produces an anomalous southeasterly flow over northern Europe (Fig. 9c) in conjunction with the above-normal surface air temperatures. This mild southeasterly return flow on the time-averaged charts, such as Fig. 9b, occurs during periods of westward extension of the Siberian anticyclone. The westward extension of the Siberian anticyclone is a well-known synoptic feature among meteorologists in southern Europe. Makorgiannis et al. (1981) obtained mean SLPs, 500-mb heights, and 1000–500-mb thicknesses for 20 winter cases when the Siberian high was displaced to the west and found 1) the westward extension develops due to negative vorticity advection aloft (it is not entirely due to radiational cooling), 2) cyclogenesis in the Bay of Biscay and Mediterranean basin often accompanies synoptic development leading to a westward extended Siberian high, and 3) much of western Europe, and particularly northwestern Europe, undergoes significant warm air advection during the westward extension of the anticyclone. Note that 2) is consistent with the synoptic development found here for negative score cases, whereas 3) is consistent with above-normal temperatures in northern Europe. The northern European dipole center in Fig. 9c is in the same location where positive correlation coefficients of Fig. 7 are somewhat lower than at other points between Ireland and eastern Siberia, suggesting that in this area another mechanism beside strong zonal flow (and positive RPCA scores) is linked to higher than normal winter air temperatures.

The dipole centers in Fig. 9c are not the standard centers of action of the NAO, having positions similar to those from the CPCA in Fig. 6. Pressure differences over Iceland and the Denmark Strait are not even statistically significant. In a climatic context, the results suggest that two separate synoptic settings and time-mean flow patterns are linked to mild winter months in the northern European segment of the winter temperature seesaw. The maritime flow producing mild conditions at Oslo in Fig. 9a is conditional on the extension of the low pressure into the Norwegian and Barents Seas and extending into northern Europe. In Fig. 9b, the flow is more southeasterly because of the westward extension of the Siberian high: the impact on abnormally high temperatures in Europe may primarily be due to the Siberian anticyclone extension. The Icelandic low, in the sense of the subpolar low over the Denmark Strait, seems to play little direct role in above-normal winter air temperatures over northern Europe.

There is virtually no difference in the pressure distribution (Fig. 9c) over the western Greenland seesaw center between the two cases. For the GB cases, the deep Icelandic low over the Denmark Strait brings northerly or northeasterly flow across western Greenland and the Davis Strait for both the positive (Fig. 9a) and negative (Fig. 9b) cases.

3) SLP associated with GA and negative storm track anomalies

Composite mean pressures are also obtained during the 55 Greenland above (GA) winter months, when the storm track scores are negative (Fig. 9d). The mean Icelandic low is weak and displaced south of Greenland with a trough of low pressure extending northwestward over the Davis Strait. The 1008-, 1012-, and 1016-mb isobars imply southeasterly flow and a trough over western Greenland, a situation often accompanying GA west-coastal above-normal winter air temperatures (Rogers 1985). A trough over the Norwegian Sea is very weak and high pressure extends westward over much of Europe.

The pressure differences obtained by subtracting GB (Fig. 9b) from GA (Fig. 9d), when the RPCA scores are negative, is shown in Fig. 9e. This pattern appears similar to the NAO, with centers near Iceland and the Azores, and areas of statistical significance over the Denmark Strait and central Atlantic. The pattern correlation between Fig. 9e and that of the winter NAO (Fig. 2a in Rogers 1990) is r = 0.865 across 76 grid points common to both figures. Pressure differences of 18 mb occur in the Denmark Strait and 8 mb near 35°N, 25°W. The elongated maximum of 6–8 mb extending northeastward of the Black Sea is the net result of the westward-extended Siberian anticyclone in Fig. 9b and its absence in Fig. 9d. Figure 9e suggests that the traditional NAO, with centers near Iceland and the Azores, is embedded in its entirety in the realm of negative and weakly positive (<+1.0) scores in the storm track index. The highest positive scores (>+1.0; Figs. 4d, 4e, and 9a) are instead cases when the storm track intensifies to the north and shifts farther east, having no single teleconnection clearly linked to it. The implied geostrophic flow variations around the Denmark Strait center indicate a stronger northerly (southerly) flow in GB (GA) over Greenland. The SLP anomalies illustrated in Fig. 9e represent the GA cases and cold flow over Europe would originate in the northeasterly flow across the Barents Sea and into Scandinavia.

4. Discussion

This study has shown the characteristics of sea level North Atlantic storm track variability, along with its association to teleconnection patterns found on monthly mean charts and to surface air temperature variability over northern Europe. The procedures used are similar to those of Lau’s (1988) 500-mb analysis that identified four Atlantic storm track modes. Lau’s first mode (“A1”) was a dipole pattern with centers located about 5° south and 10°–20° west of the dipole identified in this study (Fig. 3a), and represented “northward or southward migration of the storm tracks from their time mean position” (Lau 1988). In this study, the large latitudinal divergence in the storm tracks is especially noticeable over the central and eastern portion of the ocean basin and over Europe, with either a southwest–northeast oriented track to the Barents Sea or a northwest–southeast oriented track toward the Mediterranean basin. Lau also identified an “A2” pattern, representing an in situ strengthening or weakening of eddy activity over the Labrador region rms maximum in high-pass filtered pressures (Fig. 1). The changing strength of eddy activity is not identified in a separate storm track pattern in this study, appearing to be part of the apparent increases in eddy activity as the storm track becomes more active to the northeast (Figs. 4d and 4e).

This study shows that the GB mode of the winter air temperature seesaw can be explained by two separate sea level circulation patterns. Europe has mild winters 1) when the storm track and the mean subpolar low extend into the Norwegian and Barents Seas bringing strong maritime zonal flow far into northern Europe (Fig. 9a), and 2) when the storm track does not extend beyond Iceland, and the mean low lies over the Denmark Strait with isobars parallel to the Scandinavian coast (Fig. 9b). The latter case is the less common, but in these winters the NAO centers of action are near their normal ocean basin positions (Fig. 9b) and a westward-extended Siberian anticyclone assists in producing a strong southeasterly GB-event flow into northern Europe. The first case, with strong European zonal circulation, has long been considered (vLR and others) the NAO-based cause of GB events. This is indeed the more frequent mechanism for GB events (Fig. 8), but it is brought about by the strong northeastward-extended storm track with a deep trough in the Norwegian Sea occurring in conjunction with northeastward movement of the subtropical high. This case of maritime flow is arguably linked to a non-NAO eastward extension of Atlantic cyclone activity.

The question of whether the NAO has an atmospheric circulation and climatic imprint extending well into Europe hinges on whether the Atlantic subpolar low is really “Icelandic” when there is a Norwegian or Barents Sea pressure minima characterized by either a single mean low or a deep extension of the primary low near Iceland. On the long-term climatological charts, the wintertime Icelandic low almost always appears over the Denmark Strait with a trough extending to the northeast. The strength of the trough is at issue, reflecting the amount of eddy activity occurring in the extreme northeastern Atlantic. This study has linked high rms scores (>+1) with 1) high rms variability in the extreme northeastern Atlantic (Fig. 3b), 2) deep mean low pressure over the Norwegian and Barents Seas at the expense of a separate low over the Denmark Strait (Figs. 4d and 4e), and 3) strong zonal flow into Europe linked to above-normal surface air temperatures as far east as Siberia (Fig. 7; see also Rogers and Mosley-Thompson 1995).

Finally, comparison is made between the results of this study and the cyclone trajectories obtained in the extremes of other SLP low-frequency teleconnections described in Rogers (1990). The cyclone tracks in the extremes of the NAO (Rogers 1990; his Figs. 8a and 8b) are characterized by large latitudinal differences over the central Atlantic. The NAO positive mode has maximum cyclone frequency near the Denmark Strait with few cyclones occurring east of Iceland, and the cyclones have a path toward the Bay of Biscay in the NAO negative mode but they do not enter the Mediterranean basin. Each of the other three Atlantic sector sea level teleconnections (Rogers 1990; Figs. 8c–h) have one polarity mode characterized by a pronounced cyclone frequency maximum to the east or northeast of Iceland (along with a northeastward-extended mean subpolar low), while in the other phase there is a tendency for cyclones to penetrate into the Mediterranean. The results of this and the earlier study suggest that while the NAO is linked to the latitudinal variability in the central Atlantic, the bulk of the variability in the storm track is in the easternmost portion of the basin and is associated with low-frequency changes in SLP that are confined to that region. The NAO also seems confined to negative scores (GA cases; Figs. 4a and 9d) and those near zero or slightly positive (GB cases; Figs. 4c and 9b).

5. Conclusions

The primary mode of North Atlantic sea level storm track variability is identified using rotated principal component analysis on monthly rms fields of high-pass filtered (2–8 day periods) sea level pressures for the winters 1899–1900 through 1991–92. The primary component of variability is a dipole pattern with centers in the extreme northeastern Atlantic and west of Portugal indicating that the storm track in a given month is either very active in the extreme northeastern Atlantic Arctic, or it is oriented toward the Mediterranean basin. These storm tracks correspond respectively to positive and negative mean score values in the time series of the principal component. As the component scores become increasingly positive, there is a substantial northeastward extension of the storm track and the subpolar low intensifies in the monthly mean fields over the Norwegian and Barents Seas. The mean subtropical high simultaneously intensifies and moves northeastward, spreading over the Mediterranean basin. The westerlies strengthen in response to the intensification and movement of the centers of action and spread far into Europe, generally bringing above-normal temperatures. Months with high component scores are associated with a relatively high number of days with pressure under 1000 mb (Fig. 5), and with comparatively large interdiurnal pressure changes (over 20 mb in 24 h) and relatively low monthly mean pressures.

The storm track component scores (Fig. 2) vary somewhat randomly through the twentieth century, unlike the NAO index (Rogers 1984), and are significantly positively correlated to winter air temperatures over a region from Europe to Siberia. These correlations also show the seesaw in winter air temperatures between Greenland and northern Europe (Fig. 7). The Greenland above mode of the winter temperature seesaw primarily occurs in months with negative storm track scores (Fig. 8), indicating that the storm track is oriented toward the Mediterranean basin with high pressure over the northeastern Atlantic Arctic. The GB mode of the temperature seesaw typically occurs in months with positive storm track scores, but in 40% of the cases the monthly scores are negative. With positive scores (Fig. 9a), mild European winters (GB) are due to strong maritime flow associated with the northeastward-extended storm track and anomalous low pressure in the Norwegian Sea. With negative scores (Fig. 9b), the Icelandic low is confined to the Denmark Strait, and mild southerly or southeasterly flow occurs east of the low assisted predominantly by a westward extension of the mean Siberian anticyclone (Makorgiannis et al. 1981). The strong maritime flow in positive score cases has long been associated with the NAO but it is argued here (section 4) that this is not so. The differences in mean SLP between the sets of GB months with differing score polarities (Fig. 9c) are not significant near Iceland and the Denmark Strait, the key northern center of the NAO. Instead the strong zonal case (Fig. 9a) represents the situation in which the storm track extends far into the northeastern Atlantic with both the mean subpolar low and Azores high extending far to the northeast of their seasonal mean positions (Figs. 4d and 4e). It is suggested that the NAO may be closely linked to the latitudinal aspects of storm track variability in the central Atlantic, but the low-frequency teleconnections that are linked to the predominant mode of the storm track variability in the North Atlantic are in the far northeastern Atlantic.

Acknowledgments

This work is supported by the NOAA Office of Global Programs-Atlantic Climate Change Program under Grants NA36GP0236 and NA56GP0213. I thank Mark Serreze of CIRES and the anonymous reviewers for valuable comments that improved this paper. Chung-Chieh Wang, of the Atmospheric Science Program at The Ohio State University, assisted in producing the diagrams. This is Byrd Polar Research Center contribution no. 1002.

REFERENCES

  • Anderson, J. R., and J. R. Gyakum, 1989: A diagnostic study of Pacific basin circulation regimes as determined from extratropical cyclone tracks. Mon. Wea. Rev.,117, 2672–2686.

  • Barnston, A. G., and R. E. Livezey, 1987: Classification, seasonality and persistence of low-frequency atmospheric circulation patterns. Mon. Wea. Rev.,115, 1083–1126.

  • Blackmon, M. L., 1976: A climatological spectral study of the 500-mb geopotential height of the Northern Hemisphere. J. Atmos. Sci.,33, 1607–1623.

  • ——, J. M. Wallace, N.-C. Lau, and S. L. Mullen, 1977: An observational study of the Northern Hemisphere wintertime circulation. J. Atmos. Sci.,34, 1040–1053.

  • ——, Y. Lee, and J. Wallace, 1984: Horizontal structure of 500-mb height fluctuations with long, intermediate, and short time scales. J. Atmos. Sci.,41, 961–979.

  • Bretherton, C. S., C. Smith, and J. M. Wallace, 1992: An intercomparison of methods for finding coupled patterns in climate data. J. Climate,5, 541–560.

  • Cai, M., and H. M. Van den Dool, 1991: Low-frequency waves and traveling storm tracks. Part I: Barotropic component. J. Atmos. Sci.,48, 1420–1444.

  • Dickson, R. R., and J. Namias, 1976: North American influences on the circulation and climate of the North Atlantic sector. Mon. Wea. Rev.,104, 1255–1265.

  • Hurrell, J. W., 1995: Decadal trends in the North Atlantic Oscillation regional temperatures and precipitation. Science,269, 676–679.

  • Jones, P. D., S. C. B. Raper, B. S. G. Cherry, C. M. Goodess, T. M. L. Wigley, B. Santer, P. M. Kelly, R. S. Bradley, and H. F. Diaz, 1991: An updated global grid point surface air temperature anomaly data set: 1851–1990. ORNL/CDIAC-37, 65 pp. [Available from National Technical Information System, U.S. Dept. of Commerce, 5285 Port Royal Road, Springfield, VA 22161.].

  • Lamb, P. J., and R. A. Peppler, 1987: North Atlantic Oscillation: Concept and an application. Bull. Amer. Meteor. Soc.,68, 1218–1225.

  • Lau, N.-C., 1988: Variability of the observed midlatitude storm tracks in relation to low-frequency changes in the circulation pattern. J. Atmos. Sci.,45, 2718–2743.

  • Lejenas, H., 1989: The severe winter in Europe 1941–42: The large-scale circulation, cut-off lows, and blocking. Bull. Amer. Meteor. Soc.,70, 271–281.

  • Makorgiannis, T. J., B. D. Giles, and A. A. Flocas, 1981: The problem of the extension of the Siberian Anticyclone towards southeast Europe, and its relation to atmospheric circulation anomalies over the Northern Hemisphere. Arch. Meteor. Geophys. Bioclimatol. Ser. A,30, 185–196.

  • Mann, K. H., and K. F. Drinkwater, 1994: Environmental influences on fish and shellfish production in the northwest Atlantic. Environ. Rev.,2, 16–32.

  • Moses, T., G. N. Kiladis, H. F. Diaz, and R. G. Barry, 1987: Characteristics and frequency of reversals in mean sea level pressure in the North Atlantic sector and their relationship to long-term temperature trends. J. Climatol.,7, 13–30.

  • Namias, J., 1964: Seasonal persistence and recurrence of European blocking during 1958–60. Tellus,16, 394–407.

  • ——, 1978: Multiple causes of the North American abnormal winter 1976–77. Mon. Wea. Rev.,106, 279–295.

  • Rex, D. F., 1951: The effect of Atlantic blocking action on European climate. Tellus,3, 100–112.

  • Rogers, J. C., 1984: The association between the North Atlantic Oscillation and the Southern Oscillation in the Northern Hemisphere. Mon. Wea. Rev.,112, 1999–2015.

  • ——, 1985: Atmospheric circulation changes associated with the warming over the Northern Atlantic in the 1920s. J. Climate Appl. Meteor.,24, 1303–1310.

  • ——, 1990: Patterns of low-frequency monthly sea level pressure variability (1899–1986) and associated wave cyclone frequencies. J. Climate,3, 1364–1379.

  • ——, and E. Mosley-Thompson, 1995: Atlantic Arctic cyclones and the mild Siberian winters of the 1980s. Geophys. Res. Lett.,22, 799–802.

  • Serreze, M. C., 1995: Climatological aspects of cyclone development and decay in the Arctic. Atmos.–Ocean,33, 1–23.

  • ——, J. C. Rogers, F. Carse, and R. G. Barry, 1997: Icelandic low cyclone activity: Climatological features, linkages with the NAO and relationships with recent changes in the Northern Hemisphere circulation. J. Climate,10, 453–464.

  • Shabbar, A., K. Higuchi, and J. L. Knox, 1990: Regional analysis of Northern Hemisphere 50 kPa geopotential heights from 1946 to 1985. J. Climate,3, 543–557.

  • Trenberth, K. E., and D. A. Paolino, 1980: The Northern Hemisphere sea level pressure data set: Trends, errors and discontinuities. Mon. Wea. Rev.,108, 855–872.

  • van Loon, H., and J. Williams, 1976: The connection between trends of mean temperature and circulation at the surface. Part I: Winter. Mon. Wea. Rev.,104, 365–380.

  • ——, and J. C. Rogers, 1978: The seesaw in winter temperatures between Greenland and northern Europe. Part I: General description. Mon. Wea. Rev.,106, 296–310.

  • Wallace, J. M., and D. S. Gutzler, 1981: Teleconnections in the geopotential height field during the Northern Hemisphere winter. Mon. Wea. Rev.,109, 784–812.

  • ——, G.-H. Lim, and M. L. Blackmon, 1988: On the relationship between cyclone tracks, anticyclone tracks and baroclinic waveguides. J. Atmos. Sci.,45, 439–462.

  • ——, C. Smith, and C. S. Bretherton, 1992: Singular value decomposition of wintertime sea surface temperature and 500-mb height anomalies. J. Climate,5, 561–576.

Fig. 1.
Fig. 1.

A 120-winter-month (1951–90) average of the rms of high-pass filtered daily sea level pressures (in mb). Climatological data sites mentioned in the text are represented by dots in western Greenland, at Oslo, the Azores, and at gridpoint 65°N, 20°W in Iceland.

Citation: Journal of Climate 10, 7; 10.1175/1520-0442(1997)010<1635:NASTVA>2.0.CO;2

Fig. 2.
Fig. 2.

Time series of scores associated with the first rotated principal component of the monthly rms fields of high-pass filtered sea level pressures. Monthly (thin solid line) and seasonal (thicker solid line) values are shown, including zero values for the missing data from December 1944 through 1945.

Citation: Journal of Climate 10, 7; 10.1175/1520-0442(1997)010<1635:NASTVA>2.0.CO;2

Fig. 3.
Fig. 3.

Composites of monthly rms (in mb) for sets of months with extreme opposite modes of the first principal component of Atlantic area rms fields of high-pass filtered sea level pressures, 1900–92. The three diagrams include the (a) net mean rms differences (mb) between the (b) composite positive mode cases and (c) the composite negative cases.

Citation: Journal of Climate 10, 7; 10.1175/1520-0442(1997)010<1635:NASTVA>2.0.CO;2

Fig. 4.
Fig. 4.

Composites of Atlantic mean sea level pressures (mb) for subset groups when the monthly scores (Fig. 2) of the first principal component monthly rms of high-pass filtered sea level pressures are (a) lower than −1.0; (b) between −1.0 and zero; (c) between zero and +1.0; (d) between +1.0 and +2.0; and (e) for cases higher than +2.0. Lighter and darker shading represent areas where the differences in pressure are statistically significant with 95% and 99% confidence between different combinations of maps. Shading in (b) through (e) represents significant differences with the preceding map while shading in (a) represents significant differences between (a) and (e).

Citation: Journal of Climate 10, 7; 10.1175/1520-0442(1997)010<1635:NASTVA>2.0.CO;2

Fig. 5.
Fig. 5.

Frequencies of once-daily pressures at 65°N, 20°W during winter months when the first principal component of monthly rms of high-pass filtered pressures has extreme (a) positive and (b) negative scores and frequencies of absolute day-to-day pressure changes at 65°N, 20°W during the same (c) positive and (d) negative months.

Citation: Journal of Climate 10, 7; 10.1175/1520-0442(1997)010<1635:NASTVA>2.0.CO;2

Fig. 6.
Fig. 6.

The mean sea level mean pressure differences (mb) occurring between the months with extreme positive and negative scores of the first rotated pattern of the combined principal component analysis (CPCA) of monthly Atlantic root-mean-squares of high-pass filtered SLP and monthly mean sea level pressure, 1900–92. Lighter and darker shading represent areas where the differences in pressure are statistically significant with 95% and 99% confidence.

Citation: Journal of Climate 10, 7; 10.1175/1520-0442(1997)010<1635:NASTVA>2.0.CO;2

Fig. 7.
Fig. 7.

Spatial distribution of coefficients of correlation between RPCA scores of the Atlantic storm track eigenvector and gridded winter mean air temperatures for land areas of the Northern Hemisphere, 1900–90 (from Jones et al. 1991). Correlation coefficients of r = ±0.32, r = ±0.44, and r = ±0.55 are significant at the 95%, 99%, and 99.9% confidence levels.

Citation: Journal of Climate 10, 7; 10.1175/1520-0442(1997)010<1635:NASTVA>2.0.CO;2

Fig. 8.
Fig. 8.

Frequencies of RPCA storm track scores, at increments of 0.5, during individual winter months 1900–92 when the Greenland above (GA) and Greenland below (GB) modes occur because of the seesaw in winter air temperatures between Greenland and northern Europe.

Citation: Journal of Climate 10, 7; 10.1175/1520-0442(1997)010<1635:NASTVA>2.0.CO;2

Fig. 9.
Fig. 9.

Mean sea level pressures (mb) when the Greenland below seesaw mode occurs and the RPCA storm track scores are (a) positive and (b) negative. (c) The net pressure differences, (b) minus (a), between those sets of cases. Mean sea level pressures are also shown (d) for the Greenland above seesaw cases that occur with negative RPCA storm track scores and for (e) the net pressure differences for the sets of cases, (d) minus (b). Lighter and darker shading in (c) and (e) represent areas where the differences in pressure are statistically significant with 95% and 99% confidence.

Citation: Journal of Climate 10, 7; 10.1175/1520-0442(1997)010<1635:NASTVA>2.0.CO;2

Save