1. Introduction
Wallace and Gutzler (1981) found organized disturbances characterized by recurring and persistent patterns of pressure and circulation that span vast geographic regions and called them teleconnection patterns. Later, Barnston and Livezey (1987) detected a series of teleconnection patterns in the Northern Hemisphere: the North Atlantic Oscillation (NAO) and the east Atlantic (EA), west Pacific (WP), east Pacific–North Pacific (EP/NP), Pacific–North American (PNA), east Atlantic–west Russian (EA/WR), Scandinavian (SCA), tropical–Northern Hemisphere (TNH), polar–Eurasian (POL), and Pacific transition (PT) patterns.
In the North Atlantic, the NAO is considered to be the primary mode of atmospheric variation (Barnston and Livezey 1987); it is characterized by a meridional dipole between the Azores high in the south and the Icelandic low in the north (Hurrell 1995). Numerous authors have described oceanic variations associated with the NAO (see the reviews of Marshall et al. 2001b and Hurrell et al. 2003): the tripole pattern of sea surface temperature (SST) (Wallace et al. 1990; Cayan 1992; Tanimoto and Xie 2002), a north–south shift of the Gulf Stream (Taylor and Stephens 1998; Coëtlogon et al. 2006), and the extent of sea ice cover in the Arctic Ocean (Maslanik et al. 1996; Hu et al. 2002).
The atmospheric teleconnection patterns are regarded as organized pressure patterns. From the viewpoint of oceanography, instead of the pressure field we want to know about the wind stress field, which is regarded as the driving source for the oceanic surface layer. However, the relationship between teleconnection patterns and wind stress fields at the sea surface is not self-evident. Recently, it has been reported that the primary wind stress curl (WSC) variation at the sea surface in the North Atlantic is associated with the NAO (Marshall et al. 2001a). However, except for the NAO, it remains unclear how other teleconnection patterns influence the WSC variation in the North Atlantic. It is necessary to clarify spatial and temporal behaviors of the WSC variations attributable to the teleconnection patterns to understand the oceanic variation.
The purpose of this study is to detect organized large-scale features of the wintertime WSC anomaly field in the North Atlantic and then to explore their relation to atmospheric teleconnection patterns. In the North Pacific, Ishi and Hanawa (2005) reported large-scale variabilities of the wintertime WSC anomaly field and discussed their relation to atmospheric teleconnection patterns. The present study is a counterpart for the North Atlantic to Ishi and Hanawa (2005).
The remainder of this paper is organized as follows: in section 2, the datasets and methods used for this study are outlined. Section 3 describes the extraction of dominant variations of WSC anomaly field using various statistical methods and the investigation of their relations to teleconnection patterns. Section 4 presents our summary and remarks.
2. Datasets and methodology
To identify large-scale atmospheric forcing fields robustly, we use two wind stress datasets: the National Centers for Environmental Prediction (NCEP)–National Center for Atmospheric Research (NCAR) reanalysis dataset (hereafter NRA; Kalnay et al. 1996) and the 40-yr European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-40) dataset (hereafter ERA; Uppala et al. 2005). The WSC is calculated using the spatial derivative of wind stress. This derivative operation tends to exaggerate small-scale features of the WSC field compared with the atmospheric pressure field. To highlight the large-scale variations, we use the WSC field smoothed by a Gaussian filter with a half power point of about 2000 km. We confirmed that the identical results were obtained even after changing the smoothing parameter. We also use geopotential height data at three levels from the datasets of NRA and ERA: the geopotential height at 700 hPa showing the lower troposphere, 500 hPa representing the midlevel, and 250 hPa showing the upper level (henceforth, Z700, Z500, and Z250). Although we used two reanalysis datasets, in the following sections we show only the results using the NRA because almost identical results were obtained using the ERA.
We use indices of teleconnection patterns constructed from leading modes of a rotated empirical orthogonal function (REOF) analysis for Z700 (Barnston and Livezey 1987; Bell and Halpert 1995). Indices of nine patterns detected in the Northern Hemisphere are used: NAO, EA, WP, EP/NP, PNA, EA/WR, SCA, TNH, and POL.
The present study is performed using the winter (December–February) mean dataset, since activities of teleconnection patterns dominate in winter at the mid- and high latitudes. The study area is the North Atlantic sector: 10°–80°N, 90°W–30°E. The analysis period is the 56-yr span 1950–2005. For a correlation analysis, a significance level is determined with the degrees of freedom calculated using the method of Davis (1976): the degrees of freedom are obtainable by the data length divided by the integral time scale in the combined autocorrelation functions of time series to be compared. A Student’s two-sided t test is applied to examine whether the correlation coefficient is statistically significant.
3. Results
a. Climatological features of the WSC anomaly field
In winter, both the Icelandic low and Azores high develop and strong westerlies blow along the zonal band centered around 50°N. Correspondingly, positive and negative WSCs are located respectively in the northern and southern parts of the study area (Fig. 1).
b. Dominant REOF modes of the WSC anomaly field and their relation to atmospheric teleconnection patterns
1) Comment on the results of EOF analysis
First, we apply an empirical orthogonal function (EOF) analysis using a correlation matrix method and obtain two dominant modes in the WSC anomaly field (not shown here). The first mode represents a dipole pattern having centers of action at 65°N, 20°W and 47.5°N, 25°W. The second mode is also a dipole pattern and its centers of actions are located out of phase with those of the first mode: 55°N, 25°W and 37.5°N, 40°W. We examine the robustness and significance of these two modes. The percentages of explained variances are almost identical (26.4% in the first mode and 22.7% in the second mode) and the difference of these values does not satisfy the criteria outlined in North et al. (1982). Therefore, we conclude that the EOF mode structures are not robust to sampling noise. Thus, the EOF analysis fails to produce statistically robust patterns of variability.
To extract meaningfully dominant variation patterns in the WSC anomaly field, a REOF analysis is applied. Here, we use a varimax rotation, which is widely accepted as the most accurate analytic algebraic orthogonal rotation (Kaiser 1958, 1959; Richman 1986). The first 10 modes, which account for more than 90% of the original variance, are used in rotation.
2) Dominant REOF modes of WSC anomaly field
Figure 2 shows time series and distributions of regression coefficients of WSC anomalies onto the time series for the four leading REOF modes. The four leading modes account for over 55% of the total variance.
The first mode (WSC 1; Fig. 2a) includes three significant areas (tripole pattern) at the eastern part of the sector. The centers of action are situated around 65°N, 15°W in the northern positive area, 45°N, 15°W in the central negative area, and 20°N, 15°W in the southern positive area. The time series with decadal-scale (about 10 years) variation shows a negative extreme in the 1960s and a positive extreme in the 1990s.
In the second mode (WSC 2; Fig. 2b), a north–south seesaw pattern, whose boundary lies along the latitude of about 40°N, is found at the eastern side of the basin. This pattern is dominant in the area off the British Isles and the northern African continent. The centers of action are located around 55°N, 15°W in the northern positive area and 30°N, 15°W in the southern negative area. The time series shows a positive peak in the late 1960s and a negative peak in the early 1990s.
The third mode (WSC 3; Fig. 2c) shows a dipole pattern in the central basin, whose boundary lies along the latitude of about 50°N. The centers of action are located around 60°N, 40°W in the northern positive area and 35°N, 40°W in the southern negative area. The time series shows long-term variation with a steplike change around 1970.
The fourth mode (WSC 4; Fig. 2d) is significant only in the area off the North American continent; the center of action in the western positive area is situated around 35°N, 75°W. No significant signal was found in any other area. The time series shows an interdecadal-scale (about 20 years) variation.
We confirmed that we got the identical results in using only the half length of the dataset. We therefore conclude that the extracted modes are robust without depending on the analysis period.
3) Relation to atmospheric teleconnection patterns
We explore a relation with teleconnection patterns regarded as the atmospheric prevailing organized disturbances. First, a correlation coefficient between the time series of WSC 1–WSC 4 and the indices of nine teleconnection patterns is calculated (Table 1). It is readily apparent that each mode has a one-to-one relation with a specific teleconnection pattern: NAO is characterized by the meridional dipole between the Azores high in the south and the Icelandic low in the north for WSC 1, EA has a north–south dipole of anomaly centers spanning the North Atlantic from east to west for WSC 2, TNH is distributed from the Gulf of Alaska to the Gulf of Mexico for WSC 3, and PNA is associated with a change in magnitude of the east Asian jet stream for WSC 4. Next, in order to confirm the above relation from the viewpoint of a spatial pattern, we perform a regression analysis between the WSC anomaly field and indices of four major teleconnection patterns: NAO, EA, TNH, and PNA (Fig. 3). As expected, these maps (Fig. 3) correspond well to those of WSC 1–WSC 4, respectively (Fig. 2).
In the Northern Hemisphere, it is known that the Arctic Oscillation (AO) is identified as a dominant EOF mode of sea level pressure (Thompson and Wallace 1998, 2000). Although the AO is considered not to be a teleconnection pattern, it might be expected that the AO forces the WSC variation in the North Atlantic. We examined the correlation coefficient between the time series of WSC 1–WSC 4 and the AO index prepared by Thompson and Wallace (1998, 2000). The AO had a significant correlation with only WSC 1 associated with the NAO (R = 0.66). This might reflect the fact that the AO resembles the NAO in many respects. Numerous authors have discussed whether or not the AO and NAO are mutually independent (e.g., Deser 2000; Ambaum et al. 2001; Itoh 2002; Wallace and Thompson 2002). Recently, Feldstein and Franzke (2006) reported that the NAO and AO events are indistinguishable using a statistical method. However, no conclusion or consensus has been reached at this time. Here, we only point out that the AO gives a forcing very similar to that of the NAO.
c. Dominant MCA modes of WSC anomaly field and their relation to teleconnection patterns
We attempt to provide further evidence that the WSC variations relate to pressure patterns associated with the teleconnection pattern by performing a maximum covariance analysis (MCA) that can extract covariability based on a covariance matrix constructed from the two data fields (Bretherton et al. 1992). We applied the MCA to the WSC anomaly field in the study area and various heights (Z700, Z500, and Z250) in the anomaly fields in the Northern Hemisphere north of 10°N. Figure 4 portrays results for the WSC and Z500 anomaly field. The four leading MCA modes account for about 90% of the squared covariance function (SCF), which is the ratio of the squared covariance to the total covariance; the contribution rates of the other higher modes are very limited. We use 1000 sets of MCA between the WSC anomaly field and a temporally shuffled Z500 anomaly field to investigate whether the four detected modes are significant. Based on those results, we found that they exceed a 5% significance level.
In the first through fourth modes, a spatial pattern for WSC is closely consistent with WSC 1–WSC 4, as extracted using REOF analysis (see Figs. 2a–d). The MCA structures resemble closely the teleconnection patterns displayed by Barnston and Livezey (1987): NAO for the first mode (MCA 1), EA for the second mode (MCA 2), TNH for the third mode (MCA 3), and PNA for the fourth mode (MCA 4). Actually, it is clearly established that the time series has a high correlation with the teleconnection pattern index (see Table 2).
An almost identical result is obtained using the Z700 anomaly field, which is used to define the teleconnection pattern index, and also the Z250 anomaly field (not shown): the teleconnection pattern signal is very similar throughout the troposphere, which implies that the pressure pattern associated with teleconnection pattern has the primary equivalent barotropic structure in vertical. These results strongly suggest that the pressure patterns attributable to four major teleconnection patterns directly relate to the WSC field at the sea surface.
4. Summary and remarks
We investigated the large-scale variations of wintertime WSC anomaly field in the North Atlantic, performing REOF and MCA analyses. In terms of temporal variation and spatial distribution, the first four leading modes show a one-to-one relation with four teleconnection patterns: the NAO, EA, TNH, and PNA, respectively. These four patterns forced WSC variations over different regions: the NAO and EA over the eastern side of the basin, the TNH over the central part of the basin, and the PNA over the western side of the basin. In general, no guarantee exists of a coincidence between pressure patterns in the troposphere and WSC patterns at the sea surface. Actually, we were unable to find good correspondence between them in other seasons, especially in summer.
Numerous authors have attempted to clarify the impact of NAO on the upper ocean field. In this study, it was newly found that other teleconnection patterns such as the EA may also be quite important for further understanding of the upper ocean circulation and its variation in the North Atlantic. We expect that the result of this study will attract attention to roles of the EA, TNH, and PNA teleconnection patterns in the upper oceanic variations in the North Atlantic.
Acknowledgments
The authors wish to express their sincere thanks to the members of Physical Oceanography Group at Tohoku University for their useful discussion. A number of valuable comments given by two reviewers are appreciated. The authors are financially supported by the Earth Science Global Center of Excellence (G-COE) program at Tohoku University.
REFERENCES
Ambaum, M. H. P., B. J. Hoskins, and D. B. Stephenson, 2001: Arctic Oscillation or North Atlantic Oscillation? J. Climate, 14 , 3495–3507.
Barnston, A. G., and R. E. Livezey, 1987: Classification, seasonality and persistence of low-frequency atmospheric circulation patterns. Mon. Wea. Rev., 115 , 1083–1126.
Bell, G. D., and M. S. Halpert, 1995: Atlas of intraseasonal and interannual variability, 1986–1993. NOAA Atlas 12, Climate Prediction Center, NOAA/NWS/MNC. [Available from the Climate Prediction Center, 5200 Auth Road, Camp Springs, MD 20746].
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.
Cayan, D. R., 1992: Latent and sensible heat flux anomalies over the northern oceans: Driving the sea surface temperature. J. Phys. Oceanogr., 22 , 859–881.
Coëtlogon, G. D., and Coauthors, 2006: Gulf Stream variability in five oceanic general circulation models. J. Phys. Oceanogr., 36 , 2119–2135.
Davis, R., 1976: Predictability of sea surface temperature and sea level pressure anomalies over the North Pacific Ocean. J. Phys. Oceanogr., 6 , 249–266.
Deser, C., 2000: On the teleconnectivity of the “Arctic Oscillation”. Geophys. Res. Lett., 27 , 779–782.
Feldstein, S. B., and C. Franzke, 2006: Are the North Atlantic Oscillation and the Northern Annular Mode distinguishable? J. Atmos. Sci., 63 , 2915–2930.
Hu, A., C. Rooth, R. Bleck, and C. Deser, 2002: NAO influence on sea ice extent in the Eurasian coastal region. Geophys. Res. Lett., 29 , 2053. doi:10.1029/2001GL014293.
Hurrell, J. W., 1995: Decadal trends in the North Atlantic Oscillation: Regional temperatures and precipitation. Science, 269 , 676–679.
Hurrell, J. W., Y. Kushnir, G. Ottersen, and M. Visbeck, 2003: An overview of the North Atlantic oscillation. The North Atlantic Oscillation: Climate Significance and Environmental Impact, Geophys. Monogr., Vol. 134, Amer. Geophys. Union, 1–35.
Ishi, Y., and K. Hanawa, 2005: Large-scale variabilities of wintertime wind stress curl field in the North Pacific and their relation to atmospheric teleconnection patterns. Geophys. Res. Lett., 32 , L10607. doi:10.1029/2004GL022330.
Itoh, H., 2002: True versus apparent arctic oscillation. Geophys. Res. Lett., 29 , 1268. doi:10.1029/2001GL013978.
Kaiser, H. F., 1958: The varimax criterion for analytic rotation in factor analysis. Psychometrika, 23 , 187–200.
Kaiser, H. F., 1959: Computer program for Varimax rotation in factor analysis. Educ. Psychol. Meas., 19 , 413–420.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77 , 437–471.
Marshall, J., H. Johnson, and J. Goodman, 2001a: A study of the interaction of the North Atlantic oscillation with ocean circulation. J. Climate, 14 , 1399–1421.
Marshall, J., and Coauthors, 2001b: North Atlantic climate variability: Phenomena, impacts, and mechanisms. Int. J. Climatol., 21 , 1863–1898.
Maslanik, J. A., M. C. Serreze, and R. G. Barry, 1996: Recent decreases in Arctic summer ice cover and linkages to atmospheric circulation anomalies. Geophys. Res. Lett., 23 , 1677–1680.
North, G. R., T. L. Bell, R. F. Cahalan, and F. J. Moeng, 1982: Sampling errors in the estimation of empirical orthogonal functions. Mon. Wea. Rev., 110 , 699–706.
Richman, M. B., 1986: Rotation of principal components. J. Climatol., 6 , 293–335.
Tanimoto, Y., and S-P. Xie, 2002: Inter-hemispheric decadal variations in SST, surface wind, heat flux and cloud cover over the Atlantic Ocean. J. Meteor. Soc. Japan, 80 , 1199–1219.
Taylor, A. H., and J. A. Stephens, 1998: The North Atlantic Oscillation and the latitude of the Gulf Stream. Tellus, 50A , 134–142.
Thompson, D. W. J., and J. M. Wallace, 1998: The Arctic Oscillation signature in the wintertime geopotential height and temperature fields. Geophys. Res. Lett., 25 , 1297–1300.
Thompson, D. W. J., and J. M. Wallace, 2000: Annular modes in the extratropical circulation. Part I: Month-to-month variability. J. Climate, 13 , 1000–1016.
Uppala, S., and Coauthors, 2005: The ERA-40 reanalysis. Quart. J. Roy. Meteor. Soc., 131 , 2961–3012.
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.
Wallace, J. M., and D. W. J. Thompson, 2002: The Pacific center of action of the Northern Hemisphere annular mode: Real or artifact? J. Climate, 15 , 1987–1991.
Wallace, J. M., C. Smith, and Q. Jiang, 1990: Spatial patterns of atmosphere–ocean interaction in the northern winter. J. Climate, 3 , 990–998.
Climatologies of winter wind stress curl (WSC; 10−6 kg m−2 s−2).
Citation: Journal of the Atmospheric Sciences 67, 5; 10.1175/2009JAS3339.1
REOF modes of winter WSC anomaly field: (a) WSC 1, (b) WSC 2, (c) WSC 3, and (d) WSC 4. (top) Time series. The thick line denotes the 5-yr running mean. (bottom) Distribution of regression coefficient of winter WSC. The contour interval is 0.01; the unit is 10−6 kg m−2 s−2. The shading denotes regions exceeding a 1% significance level. The percent of explained variance is shown in the upper right of each panel.
Citation: Journal of the Atmospheric Sciences 67, 5; 10.1175/2009JAS3339.1
(top) Time series of indices of four teleconnection patterns: (a) NAO, (b) EA, (c) TNH, and (d) PNA. The thick line represents the 5-yr running mean. (bottom) Distribution of regression coefficient of winter WSC. The contour interval is 0.01; the unit is 10−6 kg m−2 s−2. The shading denotes regions exceeding a 1% significance level.
Citation: Journal of the Atmospheric Sciences 67, 5; 10.1175/2009JAS3339.1
First four leading MCA modes between the winter WSC anomaly field in the study area and winter Z500 anomaly field in the NH north of 10°N: (a) MCA 1, (b) MCA 2, (c) MCA 3, and (d) MCA 4. (top) Time series of normalized expansion coefficient for WSC (solid line) and Z500 (dashed line). Percent represents SCF. The numeral in the upper right of the panel shows the correlation coefficient between time series of WSC and Z500. (center) Heterogeneous regression maps for WSC anomaly field. The contour interval is 0.01; the unit is 10−6 kg m−2 s−2. (bottom) Heterogeneous regression maps for the Z500 anomaly field. The contour interval is 10 m. The shading denotes regions exceeding a 1% significance level. The percentage of explained variance is shown in the upper left of each panel.
Citation: Journal of the Atmospheric Sciences 67, 5; 10.1175/2009JAS3339.1
Correlation coefficients of the time series of WSC 1–WSC 4 vs indices of nine teleconnection patterns. Bold typeface represents values exceeding a 1% significance level.
Correlation coefficients of the time series of MCA 1–MCA 4 vs the time series of WSC 1–WSC 4, NAO, EA, TNH, and PNA indices. Bold typeface numerals represent values exceeding a 1% significance level.
