• Bengtsson, L., 1991: Advances and prospects in numerical weather prediction. Quart. J. Roy. Meteor. Soc.,117, 855–902.

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

  • Cheng, X., G. Nitsche, and J. M. Wallace, 1995: Robustness of low-frequency circulation patterns derived from EOF and rotated EOF analyses. J. Climate,8, 1709–1713.

  • Craddock, J. M., and C. R. Flood, 1969: Eigenvectors for representing the 500-mb geopotential surface over the Northern Hemisphere. Quart. J. Roy. Meteor. Soc.,95, 576–593.

  • Farrara, J. D., M. Ghil, C. R. Mechoso, and K. C. Mo, 1989: Empirical orthogonal functions and multiple flow regimes in the Southern Hemisphere winter. J. Atmos. Sci.,46, 3219–3223.

  • Gordon, N. D., 1986: The Southern Oscillation and New Zealand weather. Mon. Wea. Rev.,114, 371–387.

  • Horel, J. D., and J. M. Wallace, 1981: Planetary-scale phenomena associated with the Southern Oscillation. Mon. Wea. Rev.,109, 813–829.

  • Jolliffe, I. T., 1986: Principal Component Analysis. Springer-Verlag, 271 pp.

  • Jones, D. A., and I. Simmonds, 1993: A climatology of Southern Hemisphere extratropical cyclones. Climate Dyn.,9, 131–145.

  • —— and ——, 1994: A climatology of Southern Hemisphere anticyclones. Climate Dyn.,10, 333–348.

  • Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-year reanalysis project. Bull. Amer. Meteor. Soc.,77, 437–471.

  • Karoly, D. J., 1989: Southern Hemisphere circulation features associated with El Niño–Southern Oscillation events. J. Climate,2, 1239–1252.

  • ——, 1990: The role of transient eddies in low-frequency zonal variations of the Southern Hemisphere circulation. Tellus,42A, 41–50.

  • Kidson, J. W., 1975: Eigenvector analysis of monthly mean surface data. Mon. Wea. Rev.,103, 177–186.

  • ——, 1988a: Indices of the Southern Hemisphere zonal wind. J. Climate,1, 183–193.

  • ——, 1988b: Interannual variations in the Southern Hemisphere circulation. J. Climate,1, 1177–1198.

  • ——, and M. R. Sinclair, 1995: The influence of persistent anomalies on Southern Hemisphere storm tracks. J. Climate,8, 1938–1950.

  • Kousky, V. E., M. T. Kagano, and I. F. A. Cavalcanti, 1984: A review of the Southern Oscillation: Oceanic–atmospheric circulation changes and related rainfall anomalies. Tellus,36A, 490–502.

  • 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.

  • ——, and M. J. Nath, 1991: Variability of the baroclinic and barotropic transient eddy forcing associated with monthly changes in the midlatitude storm tracks. J. Atmos. Sci.,48, 2589–2613.

  • Mo, K. C., and G. H. White, 1985: Teleconnections in the Southern Hemisphere. Mon. Wea. Rev.,113, 22–37.

  • ——, and M. Ghil, 1987: Statistics and dynamics of persistent anomalies. J. Atmos. Sci.,44, 877–901.

  • Murray, R. J., and I. Simmonds, 1991: A numerical scheme for tracking cyclone centres from digital data. Part I: Development and operation of the scheme. Aust. Meteor. Mag.,39, 155–166.

  • 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.

  • O’Lenic, E. A., and R. E. Livezey, 1989: Relationships between systematic errors in medium-range numerical forecasts and some of the principal modes of low-frequency variability of the Northern Hemisphere 700-mb circulation. Mon. Wea. Rev.,117, 1262–1280.

  • Pittock, A. B., 1980: Patterns of climatic variation in Argentina and Chile—I. Precipitation, 1931–60. Mon. Wea. Rev.,108, 1347–1361.

  • Quinn, W. H., and V. T. Neal, 1982: Long-term variations in the Southern Oscillation, El Niño, and the Chilean subtropical rainfall. Fish. Bull.,81, 363–374.

  • Richman, M. B., 1986: Rotation of principal components. J. Climatol.,6, 293–335.

  • Rogers, J. C., and H. van Loon, 1982: Spatial variability of sea level pressure and 500-mb height anomalies over the Southern Hemisphere. Mon. Wea., Rev.,110, 1375–1392.

  • Ropelewski, C. F., and M. S. Halpert, 1996: Quantifying Southern Oscillation–precipitation relationships. J. Climate,9, 1043–1059.

  • Sinclair, M. R., 1994: An objective cyclone climatology for the Southern Hemisphere. Mon. Wea. Rev.,122, 2239–2256.

  • ——, 1995: A climatology of cyclogenesis for the Southern Hemisphere. Mon. Wea. Rev.,123, 1601–1619.

  • ——, 1996a: A climatology of anticyclones and blocking for the Southern Hemisphere. Mon. Wea. Rev.,124, 245–263.

  • ——, 1996b: An extended climatology of extratropical cyclones over the Southern Hemisphere. Wea. Climate,15, 21–32.

  • Taljaard, J. J., 1967: Development, distribution and movement of cyclones and anticyclones in the Southern Hemisphere during the IGY. J. Appl. Meteor.,6, 973–987.

  • Taylor, K. E., 1986: An analysis of biases in traditional cyclone frequency maps. Mon. Wea. Rev.,114, 1481–1490.

  • Trenberth, K. E., 1981: Interannual variability of the Southern Hemisphere 500-mb flow: Regional characteristics. Mon. Wea. Rev.,109, 127–136.

  • ——, 1991: Storm tracks in the Southern Hemisphere. J. Atmos. Sci.,48, 2159–2178.

  • ——, and K. C. Mo, 1985: Blocking in the Southern Hemisphere. Mon. Wea., Rev.,113, 3–21.

  • ——, and J. G. Olson, 1988: An evaluation and intercomparison of global analyses from the National Meteorological Center and the European Centre for Medium-Range Weather Forecasts. Bull. Amer. Meteor. Soc.,69, 1047–1057.

  • van Loon, H., and R. A. Madden, 1981: The Southern Oscillation. Part I: Global associations with pressure and temperature in northern winter. Mon. Wea. Rev.,109, 1150–1162.

  • Walker, G. T., 1924: Correlation in seasonal variation of weather. Part IX: A further study of world weather. Mem. India Meteor. Dept.,24, 275–332.

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

  • View in gallery

    Domain-average cyclone numbers (per 10° latitude radius circle per month) during 1980–94 with the mean and standard deviation indicated for (a) raw data, (b) slow trend removed, and (c) annual cycle removed. See text for more details.

  • View in gallery

    (a) Mean monthly cyclone track density for 1980–94, drawn every 10 cyclones per 10° latitude radius circle. (b) As in (a) except for anticyclones, every one cyclone per 10° circle. (c) Mean MSL pressure for 1980–94, every 5 hPa. (d) Standard deviation of monthly cyclone track density anomalies, drawn every one cyclone per 10° circle, with values greater than 9 hatched. (e) As in (d) except for anticyclone anomalies, every 0.2 anticyclones, with values greater than 1.2 hatched and the 1.1 contour added (dashed) in the Pacific. (e) Standard deviation of monthly MSL pressure anomalies, every 0.5 hPa, with values greater than 5 hatched.

  • View in gallery

    Eigenvectors 1–3 of monthly cyclone track density anomalies, with positive (negative) values solid (dashed).

  • View in gallery

    Eigenvectors 1–5 of monthly anticyclone track density anomalies, with positive (negative) values solid (dashed).

  • View in gallery

    Difference in (a) mean cyclone track density for 1987–94 minus 1980–86, every one cyclone per 10° latitude radius circle, with positive (negative) values solid (dashed). Regions where these differences are statistically significant at better than the 99% level are shaded. (b) As in (a) except for MSL pressure, drawn every 1 hPa.

  • View in gallery

    Eigenvectors 1–3 of monthly MSL pressure anomalies, with positive (negative) values solid (dashed).

  • View in gallery

    (a)–(c) Distribution of the correlation coefficient between cyclone track density anomalies and the three leading pressure PCs. (d)–(f) As in (a)–(c) but for anticyclones.

  • View in gallery

    High (a) and low (b) cyclone track density anomaly composites for months where the first pressure PC exceeds ±1 SD. Values are expressed as a percentage of the 1980–94 mean track density and are drawn every 5%. (c) and (d) The cyclone track density composites corresponding to (a) and (b), drawn every 10 cyclones per 10° circle per month.

  • View in gallery

    (a)–(c) Time series for pressure PCs 1–3, normalized so that the top and bottom of each plot corresponds to ±3 SDs. (d) and (e) Number of days of anomalously high pressure in regions SE-NZ and SE-PAC, expressed as a departure from normal. (f) and (g) Blocking days for (f) SE-NZ, and (g) SE-PAC. See text for more details.

  • View in gallery

    Composite MSL pressure anomalies, drawn every 1 hPa, for the months in which more than five blocking days occurred in the region (a) SE-NZ and (b) SE-PAC. See text for more details.

  • View in gallery

    Correlation, drawn every 0.1, between the SOI and (a) and (b) MSL pressure anomalies during (a) winter (May–October) and (b) summer (November–April), and (c) and (d) winter cyclone track density anomaly composites, expressed as a percentage of mean track density, for months having (c) SOI < −1.5 and (d) SOI > 0.5.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 164 78 4
PDF Downloads 131 61 3

Low-Frequency Variability of Southern Hemisphere Sea Level Pressure and Weather System Activity

View More View Less
  • 1 National Institute of Water and Atmospheric Research, Ltd., Wellington, New Zealand
Full access

Abstract

This study examines the month-to-month variations in the tracks of Southern Hemisphere weather systems and their relation to low-frequency circulation variability. Cyclones and anticyclones are identified and tracked from ECMWF analyses during 1980–94 via an automated method and the principal patterns of variation identified by EOF analysis of monthly track density anomaly fields. Only the first three EOFs of cyclone track density involving about one-third of the total track variance were distinguishable from noise. Spatial patterns derived from both unrotated and rotated EOF analysis were not reproducible on subsets of the data, pointing to secular changes in the variance structure of the cyclone dataset. An increase in cyclone numbers over the Southern Ocean during the 1980s suggested that detection of small-scale cyclones is sensitive to changes in data coverage and analysis procedure, as associated changes in the mean circulation were small during this period. EOFs of anticyclone track data were found to be more robust, indicating a variety of seesaw patterns across the hemisphere.

The leading modes of sea level pressure variability are found to be associated with regional variations in cyclone and anticyclone activity. The first pressure EOF, the so-called high-latitude mode, modulates cyclone activity between middle and high latitudes, with increased (decreased) westerlies near 55°–56°S accompanied by more (fewer) cyclones in the circumpolar regions and fewer (more) in middle latitudes. The second and third EOFs have centers of action near 60°S, 120°W and 55°S, 165°W, respectively, and are linked with blocking activity in these two regions. Blocks in the New Zealand sector occur in conjunction with a zonal wavenumber 3 pattern, while southeast Pacific blocks have no significant correlations outside the Pacific. A coherent cyclone response to ENSO was also found. During El Niño winters, increased cyclone activity occurs in a band spiraling southeastward from the subtropical Pacific toward South America, while fewer cyclones are found across the subtropical Indian Ocean, Australasia, and the southwest Pacific. During La Niñas, these patterns are almost exactly reversed, suggesting a predominantly linear cyclone response to ENSO.

Corresponding author address: Dr. Mark R. Sinclair, NIWA, P.O. Box 14-901, Kilbirnie, Wellington, New Zealand.

Email: msinclair@niwa.cri.nz

Abstract

This study examines the month-to-month variations in the tracks of Southern Hemisphere weather systems and their relation to low-frequency circulation variability. Cyclones and anticyclones are identified and tracked from ECMWF analyses during 1980–94 via an automated method and the principal patterns of variation identified by EOF analysis of monthly track density anomaly fields. Only the first three EOFs of cyclone track density involving about one-third of the total track variance were distinguishable from noise. Spatial patterns derived from both unrotated and rotated EOF analysis were not reproducible on subsets of the data, pointing to secular changes in the variance structure of the cyclone dataset. An increase in cyclone numbers over the Southern Ocean during the 1980s suggested that detection of small-scale cyclones is sensitive to changes in data coverage and analysis procedure, as associated changes in the mean circulation were small during this period. EOFs of anticyclone track data were found to be more robust, indicating a variety of seesaw patterns across the hemisphere.

The leading modes of sea level pressure variability are found to be associated with regional variations in cyclone and anticyclone activity. The first pressure EOF, the so-called high-latitude mode, modulates cyclone activity between middle and high latitudes, with increased (decreased) westerlies near 55°–56°S accompanied by more (fewer) cyclones in the circumpolar regions and fewer (more) in middle latitudes. The second and third EOFs have centers of action near 60°S, 120°W and 55°S, 165°W, respectively, and are linked with blocking activity in these two regions. Blocks in the New Zealand sector occur in conjunction with a zonal wavenumber 3 pattern, while southeast Pacific blocks have no significant correlations outside the Pacific. A coherent cyclone response to ENSO was also found. During El Niño winters, increased cyclone activity occurs in a band spiraling southeastward from the subtropical Pacific toward South America, while fewer cyclones are found across the subtropical Indian Ocean, Australasia, and the southwest Pacific. During La Niñas, these patterns are almost exactly reversed, suggesting a predominantly linear cyclone response to ENSO.

Corresponding author address: Dr. Mark R. Sinclair, NIWA, P.O. Box 14-901, Kilbirnie, Wellington, New Zealand.

Email: msinclair@niwa.cri.nz

1. Introduction

The regular passage of anticyclones, cyclones, and their associated fronts accounts for most of the shorter-term “weather” variations seen in middle latitudes. Weather systems develop in preferred regions where atmospheric and physiographic features combine to provide conditions favoring their growth. The time-averaged behavior of weather systems has been extensively documented. In the Southern Hemisphere (SH), cyclones tend to form and intensify in middle latitudes and near the principal upper-tropospheric jet streams, and migrate eastward and poleward as they mature and decay (Taljaard 1967; Jones and Simmonds 1993; Sinclair 1994, 1995). On the other hand, anticyclones have a slight equatorward component of motion and are generally concentrated in the subtropics between 25° and 40°S (Taljaard 1967; Jones and Simmonds 1994; Sinclair 1996a).

As with other aspects of the atmospheric circulation, weather system behavior can vary considerably from the time-averaged picture. Marked variations in the location and strength of storm tracks can occur within individual seasons (e.g., Lau 1988). Anomalous local weather patterns can occur when weather systems depart from their usual paths. For example, during the 1982 winter when very few cyclones crossed New Zealand (NZ), rainfall was well below normal around the country. In contrast, a preference for cyclones to track across NZ during the winters of 1990–92 produced unusually wet and stormy conditions (Sinclair 1996b).

Previous studies of storm track variability have focused on variations in transient eddy activity and its interaction with the mean flow (Blackmon et al. 1984; Lau 1988; Wallace et al. 1988; Karoly 1990; Trenberth 1991; Lau and Nath 1991). In such studies, a “storm track” is defined in terms of variance statistics for geopotential height fluctuations with periods shorter than about a week. Unfortunately, the exact nature of the relation between eddy statistics and the tracks of individual weather systems is not always clear (Wallace et al. 1988; Jones and Simmonds 1993). Moreover, both positive and negative geopotential perturbations are considered together in variance-based analyses, despite the fact that the preferred tracks of cyclones and anticyclones differ. While there is reasonable correspondence between the regions of maximum high-frequency variability in the SH (e.g., Trenberth 1991) and the preferred paths that cyclones take (e.g., Sinclair 1994), anticyclones tend to occupy subtropical regions well equatorward of the storm track (Sinclair 1996a). The positive geopotential perturbations that contribute most to the storm track arise from mobile ridges embedded within the baroclinic westerlies. Persistent, slow-moving features like blocks that interrupt or divert the normal flow of transient eddies would be expected to contribute somewhat less to fluctuations on periods shorter than about a week.

In this study, we address the issue of weather system variability by determining tracks of SH cyclones and anticyclones directly from a 15-yr series of twice-daily analyses obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF) and examining how they vary from month to month. Empirical orthogonal function (EOF) analysis is used to determine the characteristic patterns of cyclone and anticyclone variance. However, we also examine how monthly cyclone and anticyclone tracks vary in response to characteristic large-scale circulation anomalies by relating the temporal EOF coefficients (PCs) for monthly sea level pressure anomalies to weather system activity. The low-frequency variability of the SH circulation has been documented in a number of studies (e.g., Trenberth 1981; Kidson 1988a,b; Karoly 1990). The approach for this part of the study is similar to Kidson and Sinclair (1995), who explored the links between SH storm tracks and leading modes of 500-hPa variability. Here, we extend the analysis to anticyclones and blocking, and make use of a longer (15-yr) series of ECMWF analyses. We also briefly consider the cyclone response to El Niño–Southern Oscillation (ENSO) variability.

Section 2 describes how monthly track density anomaly fields are obtained by identifying cyclone and anticyclone centers and tracking them via an automated method (Sinclair 1994). Section 3 outlines climatological features of these fields and their variance, while in section 4 the principal patterns of cyclone, anticyclone, and surface pressure variations are identified by EOF analysis. Section 5 examines the relationship between weather systems and low-frequency variability of the large-scale circulation. We first relate the leading EOFs of monthly mean sea level pressure anomalies to variations in cyclone and anticyclone activity and identify connections with South Pacific blocking. We also explore the cyclone response to ENSO before summarizing results in section 6.

2. Data and processing

a. Cyclone and anticyclone track data

The cyclone finding and tracking is based on Sinclair (1994). Cyclones are identified from the twice-daily ECMWF mean sea level (MSL) pressure analyses on a 2.5° × 2.5° latitude–longitude grid, as local maxima of gradient wind cyclonic vorticity. The use of vorticity captures many mobile centers for which a pressure minimum does not exist. A constant radius Cressman-type spatial smoother is used on the pressure fields prior to the vorticity calculation to avoid bias caused by varying grid spacing over the computational domain. Anticyclones are identified as local maxima of MSL pressure, as described in Sinclair (1996a). Centers are then tracked, and the date, position, pressure, and vorticity saved for each track point. Tracking uses an algorithm first developed by Murray and Simmonds (1991). For more details, see Sinclair (1994).

The measure of weather system activity used here is track density, which is a count of the number of tracks passing within a fixed distance (1111 km or 10° latitude) of each point, with each tracked weather system permitted just one count per grid point. The simpler method of counting centers that fall within 10° of each grid point (system density) does not require centers to be formed into tracks, but produces noisier fields, with high counts for slow-moving systems (Sinclair 1994). The use of fixed-area circular geometry eliminates bias that arises when latitude–longitude boxes are used to accumulate statistics (Taylor 1986). To avoid including weaker transient systems, only systems persisting through four or more analyses (2 or more days) as tracked were used.

b. Preprocessing

Each dataset was interpolated onto a polar stereographic projection of resolution 600 km at 40°S. An annular subregion of this domain comprising 444 grid points between 20° and 70°S was used. Poleward of 70°S, cyclone and MSL pressure data are based on the artificial reduction of pressure to sea level over Antarctica, while equatorward of 20°S, the use of gradient wind vorticity to identify cyclones becomes unreliable. The 10° radius circles used to accumulate cyclone and anticyclone statistics ensure complete coverage between grid points and result in a fourfold increase in counts over the 5° cells used by Sinclair (1994). This improves the signal-to-noise ratio of month to month variations at the expense of resolving spatial detail. For the MSL pressure analyses, a constant radius Cressman-type spatial smoother was used prior to interpolation to the coarse analysis grid. This avoids aliasing by only admitting spatial variations resolvable by the 600-km grid.

The raw track density data averaged over the 444 grid points (Fig. 1a) reveals an increase in the overall number of cyclones detected during the early 1980s, particularly during 1983. This increase is possibly due to changes to the ECMWF model and the data it used, as discussed at the end of section 4. This trend was removed (Fig. 1b) by determining a 12-month running mean for the domain average track density and normalizing (dividing) the value at each grid point by this running mean, assumed to apply at the sixth month. This procedure reduced the number of months used from 180 to 169.

The final step was to obtain an anomaly field by removing the annual cycle, which contributed about half the variance of the raw data. The monthly mean was computed at each grid point from the 15 years of data and subtracted from the gridpoint data for each month. Figure 1c shows the resulting domain average with the annual cycle removed (but with the multiyear mean added). The resulting sets of 169 monthly anomalies at 444 grid points composed the basic datasets for this study.

3. Mean fields and the distribution of variance

In this section, we briefly review the synoptic climatology of the cyclone and anticyclone track density fields as obtained from the 15-yr ECMWF dataset and determine the geographic distribution of the month-to-month variance.

The climatological locations for SH cyclone tracks are seen in Fig. 2a. The storm track maximum near 55°S in the south Indian Ocean extends westward into the Atlantic Ocean and eastward into the southwest Pacific. The location of the storm track is close to that obtained by Sinclair (1994, cf. his Fig. 14) and is also consistent with bandpassed variance statistics derived by Karoly (1990), Trenberth (1991), and others. Further discussion and interpretation of climatological cyclone behavior and its seasonality is found in Sinclair (1994, 1995).

The standard deviation (SD) of the monthly cyclone track anomalies (Fig. 2d) maximizes in the South Pacific near 60°S and near the track density maximum in the Indian Ocean (Fig. 2a). The SD amounts to around 15%–20% of the total track numbers within the storm track and 25%–60% in subtropical regions where cyclone numbers are smaller. This variance peaks in winter (domain-average SD for May–October is 6.2 compared with 4.9 for November–April). This means that results are biased slightly toward winter. However, the data series used here is too short to analyze this variability by season.

Anticyclones (Fig. 2b) are found in largest numbers in the 25°–40°S band, similar to the results of Jones and Simmonds (1994) and Sinclair (1996a). However, the localized maxima obtained in those studies are not seen here because systems are counted just once per grid point per track, reducing counts for slow-moving anticyclones. Very few highs are found in the south Indian Ocean, the region most frequently occupied by cyclones (Fig. 2a).

Anticyclone variability (Fig. 2e) generally maximizes in regions having the highest track density. However, regions of the South Pacific east of NZ also exhibit heightened variability, especially in relation to overall anticyclone numbers. This is consistent with Sinclair (1996a), who found enhanced interannual variability in this region, with winter highs perferring a path south of 55°S in some years while remaining north of 40°S in others. Sinclair also found this sector to be home to the highest frequency of blocking anticyclones, which occur most commonly southeast of NZ and west of South America. In section 4, we will determine how blocking frequency in these two regions of the South Pacific is related to the leading modes of atmospheric circulation variability.

The long-term mean of MSL pressure (Fig. 2c) is zonally symmetric except for broad troughs in the Pacific and Indian Ocean sectors. The cyclone track density maximum (Fig. 2a) occurs between the circumpolar trough and the strongest westerlies, consistent with the distinctive relation between the SH storm track and the zonal wind maximum (Trenberth 1991). The intermonthly SD of pressure anomalies (Fig. 2f) is largest at high latitudes of the South Pacific and in two regions north and south of the Indian Ocean storm track. Other lobes of enhanced variability are found just east of NZ and in the Atlantic. These patterns are similar to those obtained by Kidson (1988b) for low-pass-filtered 500-hPa geopotential.

4. Eigenvector analysis

In this section, EOF analysis (e.g., Jolliffe 1986) is used to determine the characteristic spatial patterns of weather system variance. The EOFs were computed from the temporal covariance matrices constructed from the 169 monthly anomaly fields of MSL pressure, cyclone, and anticyclone track density. Results based on the correlation matrix (not shown) were found to be noisy because of the tendency to highlight variability in regions of low track counts.

Table 1 shows the proportion of variance associated with the first 10 EOFs of pressure, cyclone, and anticyclone track density anomalies, contributing 80.0%, 69.6%, and 63.1% of the total variance, respectively. The criterion of North et al. (1982) yields a fractional uncertainty for each eigenvalue, δλi/λl, of ±0.15 (expressed as absolute errors in Table 1), based on 84 independent observations. The number of independent observations was estimated by computing the absolute autocorrelation of the anomaly time series at each grid point and averaging over all grid points for a range of lags. At lag ±1 month, the mean autocorrelation was 0.09 for cyclones, 0.11 for anticyclones, and 0.21 for pressure, with near-zero values at longer lags. The near absence of autocorrelation suggests that the monthly track density fields are almost independent, but pressure fields less so. If, as in Mo and White (1985), we make the conservative assumption that observations 2 months apart are independent, we obtain a total of 84 independent observations for the 169-month period.

A straight-line fit to log(λl) versus EOF number i explained between 97.7% and 99.3% of the variance for EOFs 11–20, indicating these to be noise (Craddock and Flood 1969). When this exponential fit was extrapolated back to the lower-numbered EOFs (result shown in parentheses in Table 1), only EOFs 1–3 are distinguishable from noise for MSL pressure and cyclones, with five anticyclone EOFs above the noise level. Some overlap exists between pressure and anticyclone eigenvalues numbered 2 and above in Table 1, and between all the cyclone eigenvalues, implying that some mixing is likely between the associated eigenvectors.

The cyclone eigenvectors are shown in Fig. 3. The first, explaining 13.7% of the variance, is monopolar, with maximum amplitude in the Indian Ocean. The second eigenvector (Fig. 3b) involves a seesaw of cyclone activity between middle and high latitudes, analogous to the high-latitude mode of circulation variability, which involves out-of-phase zonally symmetric geopotential height anomalies between middle and high latitudes. The third eigenvector (Fig. 3c) depicts cyclonic activity varying in antiphase between the Atlantic and South Pacific, and between the low and high latitudes of the Pacific sector. Together, these three patterns explain 35% of the total variance.

For anticyclones (Fig. 4), about five EOFs can be distinguished from noise (see Table 1). As with cyclones, the leading anticyclone eigenvector (Fig. 4a) is nearly monopolar, with maximum amplitude spanning the Atlantic and Indian Oceans near 40°S. Eigenvectors 2 and 3 depict wavenumber 1 patterns comprising out-of-phase fluctuations between highs near South Africa and the South Pacific (Fig. 4b), and between south Australia and South America (Fig. 4c). Eigenvector 4 (Fig. 4d) has a wavenumber 2 pattern, with centers of action east of Africa and west of South America out of phase with centers near southwest Australia and east of South America. Eigenvector 5 (Fig. 4e) exhibited the most zonally symmetric pattern: a seesaw between highs in the 40°–50°S band and those north of 30°S, akin to the high-latitude mode.

Following EOF analysis, a varimax rotation (Richman 1986) was applied to the leading 15 cyclone EOFs, as rotated patterns are thought to be more robust and physically meaningful than their unrotated counterparts (Richman 1986; Cheng et al. 1995). The leading rotated modes (not shown) were similar in form to the unrotated patterns, but with increased amplitude near the strongest centers of action.

To test the temporal stability of the EOF patterns, the datasets were split into two halves, 1980–86 and 1987–94, and the EOFs computed separately for each half (not shown). The anticyclone and pressure EOFs appeared to be stable. However, for cyclones, it was found that both the “raw” and rotated EOFs varied substantially in form between the two halves of the data, suggesting not only sampling uncertainty due to mixing, but also the presence of some sort of secular change in the variance structure of the cyclone dataset.

Changes in the variance structure of the cyclone dataset may be due to modifications in analysis procedure at the ECMWF and/or changes in data coverage over the period (Trenberth and Olson 1988; Bengtsson 1991). We have already documented a slow increase in the overall number of cyclones detected prior to 1984 (Fig. 1a). Sinclair (1994) also observed a jump in cyclone numbers in 1983 that coincided with an increase in model resolution. Figure 5 shows a significant 10%–15% increase in the number of cyclones detected during the analysis period. Increases occurred primarily over expanses of the Southern Ocean devoid of conventional data. Detection of smaller cyclones in these locations may be sensitive to changes in remote sensing methodology and introduction of new data types (e.g., drifting buoys). The accompanying circulation changes (Fig. 5b) are less significant statistically and appear insufficient to account for the increase in cyclone frequency. On the other hand, stability of the anticyclone patterns may result from their almost unambiguous detection by virtue of their large size and persistence characteristics. Highs also tend to occur near populated midlatitude and subtropical regions that have more observations.

5. Low-frequency variability and weather system tracks

In this section, we analyse the impact of variations in low-frequency circulation on weather systems by relating the EOFs of sea level pressure variability to changes in cyclone and anticyclone activity. The approach is similar to that of Kidson and Sinclair (1995), who used EOF analysis of 500-hPa heights to define the circulation anomalies. They found increased (decreased) counts of surface cyclone centers below and to the east of persistent negative (positive) 500-hPa height anomalies, with preferred regions of cyclone occurrence lying to poleward of 50-hPa jet axes in a dynamically consistent fashion.

a. MSL pressure eigenvectors

The leading MSL pressure eigenvectors are shown in Fig. 6. The first eigenvector (explaining 19.9% of the variance) is the high-latitude mode representing a zonally symmetric seesaw between middle and high latitudes, and corresponds to the first EOF of low-pass-filtered 500-hPa data found in the work of Mo and Ghil (1987), Kidson (1988b), Kidson and Sinclair (1995), and others. The positive (negative) phase of this mode is associated with decreased (increased) westerlies in the 50°–65° latitude band (not shown). Here, the phase reversal occurs around 55°–60°S (Fig. 6a), close to results of Kidson (1988b) and Kidson and Sinclair (1995) but somewhat poleward of earlier results (Kidson 1975; Rogers and van Loon 1982). We can have more confidence in the stability of the present results, as they are based on 15 years of recent state-of-the-art numerical analyses.

The second pressure EOF (Fig. 6b) shows a center of action in the southeast Pacific near 60°S, 120°W that is out of phase with variations in the subtropical Pacific on one side and Antarctica on the other. While the southeast Pacific center is common to all studies, some studies (e.g., Kidson 1988b) show additional wavelike structure for this mode involving extrema near NZ and the Antarctic peninsula. Eigenvector 3 (Fig. 6c) describes a wave train extending from the Pacific across South America, with the dominant center of action southeast of NZ. The wave train form of this eigenvector is similar to the third eigenvectors described by Mo and Ghil (1987), Farrara et al. (1989), and Kidson (1988b). Eigenvectors for the two halves of the data were similar to those in Fig. 6, but with reversed order in the early years, suggesting stable patterns but of differing importance in terms of explained variance. Rotated eigenvectors retained the dominant centers of action in Fig. 6 and were also almost identical in the two subsets.

b. Weather system relationships

To determine how the modes of circulation variability in Fig. 6 affect cyclone and anticyclone activity, we computed correlation coefficients between the PCs for each of the three leading pressure eigenvectors, and the time series of cyclone (Figs. 7a–c) and anticyclone (Figs. 7d–f) anomalies at each grid point. Only statistically significant values (<−0.2 or >0.2) are contoured. The statistical significance of correlation coefficients was assessed by assuming that the statistic t = r(N − 2)1/2(1 − r2)−1/2 has Student’s distribution with N = 84 independent samples (82 degrees of freedom). This yielded threshold correlation values r of 0.2 and 0.3 at the 95% and 99% confidence levels, respectively.

The first pressure PC is correlated with above-normal cyclone activity (Fig. 7a) just east of NZ and in an elongated zone spanning the Indian and Atlantic Oceans near 40°S and below normal activity within the circumpolar trough near Antarctica. Thus, stronger (weaker) westerlies near 55°S imply more (fewer) cyclones in the circumpolar regions and fewer (more) in middle latitudes. The relation between PC 1 and anticyclones (Fig. 7d) is less coherent, the main response being a correlation with subtropical highs to the north of NZ and in the Indian Ocean. The time series for pressure PC 1 is also correlated with cyclone PC 2 (r = −0.60, Fig. 3b) and anticyclone PC 5 (r = −0.49, Fig. 4e). However, as the cyclone EOFs are not well defined, these correlations yield little physical insight beyond representing a general association between low (high) pressure anomalies and (anti-) cyclones.

The second MSL pressure eigenvector (Fig. 6b) is associated with a meridional dipole of decreased cyclone activity (Fig. 7b) near 60°S, 120°W and increased cyclones near 30°S, 110°W. A weaker dipole of reversed polarity occurs in the Atlantic. Anticyclones (Fig. 7e) are highly correlated (r > 0.6) with pressure PC 2 near the center of action in the southeast Pacific. Sinclair (1996a) showed that anticyclones at these higher latitudes often have central pressures more than 20 hPa above climatology and involve an anomalous breakdown of the westerlies. Comparison of Figs. 7b and 7e suggests that the positive phase of this mode is often associated with a blocking signature comprising an anticyclone near 60°S, 120°W and a cyclone near 30°S, 110°W. Pressure PC 2 is also correlated with cyclone PC 3 (r = −0.51, Fig. 3c).

Similarly, the third pressure PC (Fig. 6c) correlates with a blocking signature comprising highs near 55°S, 180° (Fig. 7f) and lows to the north of NZ (Fig. 7c). However, the increased anticyclone frequency in the central Indian Ocean in Fig. 7f occurs at lower latitudes (45°S) and without cyclones to the north, suggesting instead a small poleward shift in the position of the subtropical anticyclone rather than an anomalous breakdown of the westerlies.

Because the high-latitude mode (Fig. 6a) is the leading mode of low-frequency circulation variability in the SH (e.g., Kidson 1988a,b; Karoly 1990; Kidson and Sinclair 1995), we examine its cyclone response in more detail. Cyclone track density and track density anomaly composites were prepared (Fig. 8) for opposite polarities of the high-latitude mode. Composites were obtained for the 54 months where the time coefficient exceeded ±1 SD. The cyclone anomaly composites (Figs. 8a,b) show 15%–20% more (fewer) cyclones near NZ and in the Indian Ocean near 40°S and fewer (more) south of 55°S for high (low) composites. The symmetry between Figs. 8a and 8b suggests a linear cyclone response to the high-latitude mode (O’Lenic and Livezey 1989). The track density maximum is slightly broader and extends into lower latitudes during positive excursions (Fig. 8c), whereas cyclones are more confined to higher latitudes for the opposite polarity (Fig. 8d). Figures 8c,d are very similar to the corresponding band-passed 500-hPa variance composites obtained by Kidson and Sinclair (1995, cf. their Figs. 4b,e). Their cyclone system density composites (their Figs. 4c,f) were noisy in comparison because of the artificially high counts given to slower-moving features. As a measure of cyclone activity, track density is more consistent with band-passed variance (Sinclair 1994) and yields more coherent results because each cyclone is limited to a single count per grid point.

c. South Pacific blocking

Sinclair (1996a) found that persistent SH blocks occurred with highest frequency southeast of NZ and in the southeast Pacific near 60°S, 120°W. These two South Pacific locations coincide with the centers of action of EOFs 2 and 3 (Figs. 6b,c) and with the occurrence of anticyclones in Figs. 7e,f. To extend Sinclair’s analysis, we derived monthly statistics for high pressure and blocking during 1980–94 in these two regions (Fig. 9). Episodes of high pressure were identified as instances where daily MSL pressure in at least two consecutive 12-h analyses exceeded the monthly mean by more than 15 hPa at any point within 1000 km of 55°S, 120°W (SE-PAC) and 55°S, 165°W (SE-NZ). These regions correspond approximately1 to the centers of action for EOFs 2 and 3, respectively (Figs. 6b,c). To be included, a high pressure episode also had to exceed the climatological mean by more than 6 hPa when averaged over the whole circular region, to eliminate transient high pressure ridges affecting only a small fraction of each area. High pressure episodes were most frequent during winter in both regions but occurred with sufficient frequency throughout the year to enable the calculation of a meaningful annual cycle. Anomalies from this are shown in Figs. 9d,e. Blocking events were identified as high pressure episodes persisting for at least 5 days (Figs. 9f,g). The time series for pressure PCs 1–3 are also included (Figs. 9a–c).

The frequency of high pressure episodes in SE-PAC (Fig. 9e) is correlated with PC 2 (Fig. 9b, r = 0.74) and weakly with PC-1 (Fig. 9a, r = 0.29), while those in SE-NZ (Fig. 9d) are correlated with PC-3 (Fig. 9c, r = 0.65). On the other hand, high pressure anomalies in the two regions are only weakly related to each other (r = 0.29), suggesting that episodes of high pressure in both regions in the same month are comparatively rare.

Blocking episodes (Figs. 9f,g) almost always occur in months having a higher than normal number of high pressure episodes (positive swings in Figs. 9d,e), with SE-PAC (SE-NZ) blocks mostly occurring during positive excursions of PC 2 (PC 3). Of course, not all such positive swings led to blocks. More blocking days were found in the SE-PAC region than in SE-NZ, especially late in the period. This is consistent with Sinclair (1996a), who found around 30% more blocks in SE-PAC during 1980–89. Here, there were 16 blocking months in SE-NZ and 27 in SE-PAC, including 7 months in which blocking occured in both regions. Thus, while most blocks occur in isolation, multiple blocking events are sometimes seen, a feature also noted by Trenberth and Mo (1985).

To shed some light on the SH circulation accompanying blocks, pressure anomaly composites were prepared for the 16 blocking months in SE-NZ (Fig. 10a) and the 27 SE-PAC blocking months (Fig. 10b). As expected, these indicate positive anomalies in the two blocking regions, with negative anomalies to the north. Cyclone anomaly composites (not shown) indicate an enhanced storm track (10%–20% more cyclones) to the north of each region. Figure 10a bears a striking resemblance to pressure eigenvector 3 (Fig. 6c) and suggests that blocking southeast of NZ is also associated with somewhat higher than average pressure in the Indian and Atlantic Oceans. This is in agreement with Trenberth and Mo (1985), who found that blocking SE of NZ was sometimes associated with a zonal wavenumber 3 pattern. One-point pressure correlation maps between pressure anomalies at 55°S, 165°W (not shown) yielded a wavenumber 3 pattern similar to Fig. 10a, with statistically significant positive correlations with Indian and Atlantic Ocean pressure. On the other hand, remote links with SE-PAC blocking appear more tenuous (Fig. 10b), with the corresponding one-point correlations (not shown) yielding no significant values outside of the Pacific.

d. ENSO relationships

The relationship between the Southern Oscillation and global circulation has been explored since early this century, providing ample evidence of large-scale ENSO-related variability in a number of meteorological fields (e.g., Walker 1924; van Loon and Madden 1981; Gordon 1986; Karoly 1989). This is illustrated in Figs. 11a,b, which show the correlation field between the Southern Oscillation index (SOI), obtained as the normalized difference of monthly MSL pressure anomalies at Tahiti (18°S, 150°W) minus Darwin (12°S, 131°E) (Gordon 1986), and monthly MSL pressure anomalies for winter (May–October) and summer (November–April). In winter, positive correlations with the SOI extend from a maximum over the tropical Pacific (Tahiti) across South America into the South Atlantic, while negative correlations are found throughout the Indian Ocean and Australasia region. In summer, the negative correlations are more confined to the Australian region, with positive values south of NZ, across the tropical Pacific and south of Africa. These patterns imply stronger (weaker) westerlies over the South Pacific and decreased (increased) winter westerlies over the Indian Ocean for positive (negative) swings of the SOI.

In view of the above relationships between the SOI and hemispheric pressure anomalies, it is not unrealistic to expect ENSO-related changes in cyclone and anticyclone activity. The strongest response was found in winter, when statistically significant positive correlations with the SOI are found for cyclone track density anomalies over Australasia, with negative correlations in the tropical east Pacific and in the high latitudes of the Indian Ocean (not shown). These relations are illustrated by cyclone anomaly composites for opposite polarities of the SOI (Figs. 11c,d). Because the period 1980–94 was dominated by negative SOI, months having SOI < −1.5 (El Niño) and SOI > 0.5 (La Niña) were composited, with each subset comprising 35 months. The cyclone composite for El Niño winters (Fig. 11c) shows 10%–20% fewer cyclones occurring in the subtropical Indian Ocean, across Australasia, and southeastward into the South Pacific, with more cyclones in the subtropical eastern Pacific and across South America, and over the Southern Ocean south of Australia. Winter cyclone anomalies during La Niñas (Fig. 11d) are almost the reverse of those for El Niño, with more cyclones across the Indian Ocean and Australasia and fewer over the east Pacific and South America. This symmetry between opposite phases of the SOI in winter suggests a linear cyclone response to ENSO (O’Lenic and Livezey 1989). In summer (not shown), the main response to swings in the SOI is an east–west phase reversal in cyclone frequency across the tropical Pacific.

These ENSO-related variations in weather system activity are in harmony with precipitation variations. For example, during El Niño conditions, above normal precipitation occurs over southern Brazil, Paraguay, and northern Argentina (Kousky et al. 1984); central Chile (Pitock 1980; Quinn and Neal 1982); and the tropical central Pacific (e.g., Horel and Wallace 1981), consistent with the higher cyclone frequency across those regions. On the other hand, southeast Australia, Fiji, and New Caledonia have above (below) normal rainfall during positive (negative) SOI episodes (Ropelewski and Halpert 1996).

Anticyclone–SOI relations (not shown) are more tenuous. A combination of highs to the south of NZ and lows to the north occurs in summer with highest frequency during positive SOI episodes, while a similar blocking pattern occurs in the southeast Pacific near 120°W during negative episodes. Thus, summer blocking in these regions may be partially modulated by the Southern Oscillation, as also suggested by correlations of −0.2 and 0.35 with the SOI for pressure PCs 2 and 3, respectively (Figs. 9b,c).

6. Summary and concluding remarks

Month-to-month fluctuations in weather system activity during 1980–94 have been analyzed via EOF analysis of monthly track density anomalies, generated via an automated algorithm. For cyclones, around a third of the total variance (the first three EOFs) was distinguishable from noise. There was an increase in the number of cyclones identified during the period, particularly over the oceans surrounding Antarctica, where cyclone detection appeared to be sensitive to changes in analysis procedure and data coverage. As a consequence of these trends, the rotated cyclone EOFs were not stable across subsets of the data. Anticyclone EOFs yielded more stable patterns, with the five modes discernable from noise depicting a variety of seesaw patterns across the hemisphere. The increased stability of the anticyclone eigenvectors reflects the large size and longevity of high pressure systems and their location near relatively data-rich subtropical regions. This makes their detection almost unambiguous.

The leading modes of MSL pressure variability were then related to regional variations in cyclone and anticyclone activity. The leading mode of circulation variability, the so-called high-latitude mode, was found to be correlated with a similar seesaw in cyclone activity between middle and high latitudes, with increased westerlies near 55°–60°S associated with more cyclones in the circumpolar regions and fewer in middle to low latitudes. The second and third eigenmodes of the pressure field have centers of action in the southeast Pacific (EOF 2) and southeast of New Zealand (EOF 3) and are correlated with blocking frequency in these respective regions. Simultaneous blocks in both regions are comparatively rare. Blocks southeast of NZ occur in conjunction with a wavenumber 3 pattern of high pressure anomalies in the the Indian and Atlantic sectors, while southeast Pacific blocks have no associated remote links.

Finally, we documented links between ENSO and weather system tracks. A coherent cyclone response to ENSO was found, especially in winter. During El Niño winters, over 20% more cyclones occur in a broad band extending southeastward from the subtropical Pacific toward South America, while 10%–20% fewer cyclones are found across the subtropical Indian Ocean, Australasia, and the southwest Pacific. During La Niñas, these patterns are almost exactly reversed, suggesting a linear component to the cyclone response to ENSO. Summer blocking anticyclones south and southeast of NZ are also more frequent during La Niñas.

We plan to extend this analysis by using a longer series of data, obtained as a result of the reanalysis efforts (Kalnay et al. 1996) at the National Centers for Environmental Prediction. Although changes in data coverage may still impact results, the reanalyses are based on a consistent data assimilation procedure. The present use of a discrete weather system count probably introduces additional noise in regions of low occurrence. Accumulation of track statistics over longer periods (say, a 3-month season) or use of a continuous variable based on transient eddy variance (e.g., Lau 1988) should reduce noise. However, explicit tracking of weather systems offers unique advantages by allowing separate consideration of highs and lows and/or stratification according to intensity or stage of development (e.g., Sinclair 1995).

Acknowledgments

Gridded atmospheric analyses were provided by ECMWF. This work was supported by the New Zealand Foundation for Research, Science and Technology under Contract CO1522.

REFERENCES

  • Bengtsson, L., 1991: Advances and prospects in numerical weather prediction. Quart. J. Roy. Meteor. Soc.,117, 855–902.

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

  • Cheng, X., G. Nitsche, and J. M. Wallace, 1995: Robustness of low-frequency circulation patterns derived from EOF and rotated EOF analyses. J. Climate,8, 1709–1713.

  • Craddock, J. M., and C. R. Flood, 1969: Eigenvectors for representing the 500-mb geopotential surface over the Northern Hemisphere. Quart. J. Roy. Meteor. Soc.,95, 576–593.

  • Farrara, J. D., M. Ghil, C. R. Mechoso, and K. C. Mo, 1989: Empirical orthogonal functions and multiple flow regimes in the Southern Hemisphere winter. J. Atmos. Sci.,46, 3219–3223.

  • Gordon, N. D., 1986: The Southern Oscillation and New Zealand weather. Mon. Wea. Rev.,114, 371–387.

  • Horel, J. D., and J. M. Wallace, 1981: Planetary-scale phenomena associated with the Southern Oscillation. Mon. Wea. Rev.,109, 813–829.

  • Jolliffe, I. T., 1986: Principal Component Analysis. Springer-Verlag, 271 pp.

  • Jones, D. A., and I. Simmonds, 1993: A climatology of Southern Hemisphere extratropical cyclones. Climate Dyn.,9, 131–145.

  • —— and ——, 1994: A climatology of Southern Hemisphere anticyclones. Climate Dyn.,10, 333–348.

  • Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-year reanalysis project. Bull. Amer. Meteor. Soc.,77, 437–471.

  • Karoly, D. J., 1989: Southern Hemisphere circulation features associated with El Niño–Southern Oscillation events. J. Climate,2, 1239–1252.

  • ——, 1990: The role of transient eddies in low-frequency zonal variations of the Southern Hemisphere circulation. Tellus,42A, 41–50.

  • Kidson, J. W., 1975: Eigenvector analysis of monthly mean surface data. Mon. Wea. Rev.,103, 177–186.

  • ——, 1988a: Indices of the Southern Hemisphere zonal wind. J. Climate,1, 183–193.

  • ——, 1988b: Interannual variations in the Southern Hemisphere circulation. J. Climate,1, 1177–1198.

  • ——, and M. R. Sinclair, 1995: The influence of persistent anomalies on Southern Hemisphere storm tracks. J. Climate,8, 1938–1950.

  • Kousky, V. E., M. T. Kagano, and I. F. A. Cavalcanti, 1984: A review of the Southern Oscillation: Oceanic–atmospheric circulation changes and related rainfall anomalies. Tellus,36A, 490–502.

  • 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.

  • ——, and M. J. Nath, 1991: Variability of the baroclinic and barotropic transient eddy forcing associated with monthly changes in the midlatitude storm tracks. J. Atmos. Sci.,48, 2589–2613.

  • Mo, K. C., and G. H. White, 1985: Teleconnections in the Southern Hemisphere. Mon. Wea. Rev.,113, 22–37.

  • ——, and M. Ghil, 1987: Statistics and dynamics of persistent anomalies. J. Atmos. Sci.,44, 877–901.

  • Murray, R. J., and I. Simmonds, 1991: A numerical scheme for tracking cyclone centres from digital data. Part I: Development and operation of the scheme. Aust. Meteor. Mag.,39, 155–166.

  • 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.

  • O’Lenic, E. A., and R. E. Livezey, 1989: Relationships between systematic errors in medium-range numerical forecasts and some of the principal modes of low-frequency variability of the Northern Hemisphere 700-mb circulation. Mon. Wea. Rev.,117, 1262–1280.

  • Pittock, A. B., 1980: Patterns of climatic variation in Argentina and Chile—I. Precipitation, 1931–60. Mon. Wea. Rev.,108, 1347–1361.

  • Quinn, W. H., and V. T. Neal, 1982: Long-term variations in the Southern Oscillation, El Niño, and the Chilean subtropical rainfall. Fish. Bull.,81, 363–374.

  • Richman, M. B., 1986: Rotation of principal components. J. Climatol.,6, 293–335.

  • Rogers, J. C., and H. van Loon, 1982: Spatial variability of sea level pressure and 500-mb height anomalies over the Southern Hemisphere. Mon. Wea., Rev.,110, 1375–1392.

  • Ropelewski, C. F., and M. S. Halpert, 1996: Quantifying Southern Oscillation–precipitation relationships. J. Climate,9, 1043–1059.

  • Sinclair, M. R., 1994: An objective cyclone climatology for the Southern Hemisphere. Mon. Wea. Rev.,122, 2239–2256.

  • ——, 1995: A climatology of cyclogenesis for the Southern Hemisphere. Mon. Wea. Rev.,123, 1601–1619.

  • ——, 1996a: A climatology of anticyclones and blocking for the Southern Hemisphere. Mon. Wea. Rev.,124, 245–263.

  • ——, 1996b: An extended climatology of extratropical cyclones over the Southern Hemisphere. Wea. Climate,15, 21–32.

  • Taljaard, J. J., 1967: Development, distribution and movement of cyclones and anticyclones in the Southern Hemisphere during the IGY. J. Appl. Meteor.,6, 973–987.

  • Taylor, K. E., 1986: An analysis of biases in traditional cyclone frequency maps. Mon. Wea. Rev.,114, 1481–1490.

  • Trenberth, K. E., 1981: Interannual variability of the Southern Hemisphere 500-mb flow: Regional characteristics. Mon. Wea. Rev.,109, 127–136.

  • ——, 1991: Storm tracks in the Southern Hemisphere. J. Atmos. Sci.,48, 2159–2178.

  • ——, and K. C. Mo, 1985: Blocking in the Southern Hemisphere. Mon. Wea., Rev.,113, 3–21.

  • ——, and J. G. Olson, 1988: An evaluation and intercomparison of global analyses from the National Meteorological Center and the European Centre for Medium-Range Weather Forecasts. Bull. Amer. Meteor. Soc.,69, 1047–1057.

  • van Loon, H., and R. A. Madden, 1981: The Southern Oscillation. Part I: Global associations with pressure and temperature in northern winter. Mon. Wea. Rev.,109, 1150–1162.

  • Walker, G. T., 1924: Correlation in seasonal variation of weather. Part IX: A further study of world weather. Mem. India Meteor. Dept.,24, 275–332.

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

Fig. 1.
Fig. 1.

Domain-average cyclone numbers (per 10° latitude radius circle per month) during 1980–94 with the mean and standard deviation indicated for (a) raw data, (b) slow trend removed, and (c) annual cycle removed. See text for more details.

Citation: Monthly Weather Review 125, 10; 10.1175/1520-0493(1997)125<2531:LFVOSH>2.0.CO;2

Fig. 2.
Fig. 2.

(a) Mean monthly cyclone track density for 1980–94, drawn every 10 cyclones per 10° latitude radius circle. (b) As in (a) except for anticyclones, every one cyclone per 10° circle. (c) Mean MSL pressure for 1980–94, every 5 hPa. (d) Standard deviation of monthly cyclone track density anomalies, drawn every one cyclone per 10° circle, with values greater than 9 hatched. (e) As in (d) except for anticyclone anomalies, every 0.2 anticyclones, with values greater than 1.2 hatched and the 1.1 contour added (dashed) in the Pacific. (e) Standard deviation of monthly MSL pressure anomalies, every 0.5 hPa, with values greater than 5 hatched.

Citation: Monthly Weather Review 125, 10; 10.1175/1520-0493(1997)125<2531:LFVOSH>2.0.CO;2

Fig. 3.
Fig. 3.

Eigenvectors 1–3 of monthly cyclone track density anomalies, with positive (negative) values solid (dashed).

Citation: Monthly Weather Review 125, 10; 10.1175/1520-0493(1997)125<2531:LFVOSH>2.0.CO;2

Fig. 4.
Fig. 4.

Eigenvectors 1–5 of monthly anticyclone track density anomalies, with positive (negative) values solid (dashed).

Citation: Monthly Weather Review 125, 10; 10.1175/1520-0493(1997)125<2531:LFVOSH>2.0.CO;2

Fig. 5.
Fig. 5.

Difference in (a) mean cyclone track density for 1987–94 minus 1980–86, every one cyclone per 10° latitude radius circle, with positive (negative) values solid (dashed). Regions where these differences are statistically significant at better than the 99% level are shaded. (b) As in (a) except for MSL pressure, drawn every 1 hPa.

Citation: Monthly Weather Review 125, 10; 10.1175/1520-0493(1997)125<2531:LFVOSH>2.0.CO;2

Fig. 6.
Fig. 6.

Eigenvectors 1–3 of monthly MSL pressure anomalies, with positive (negative) values solid (dashed).

Citation: Monthly Weather Review 125, 10; 10.1175/1520-0493(1997)125<2531:LFVOSH>2.0.CO;2

Fig. 7.
Fig. 7.

(a)–(c) Distribution of the correlation coefficient between cyclone track density anomalies and the three leading pressure PCs. (d)–(f) As in (a)–(c) but for anticyclones.

Citation: Monthly Weather Review 125, 10; 10.1175/1520-0493(1997)125<2531:LFVOSH>2.0.CO;2

Fig. 8.
Fig. 8.

High (a) and low (b) cyclone track density anomaly composites for months where the first pressure PC exceeds ±1 SD. Values are expressed as a percentage of the 1980–94 mean track density and are drawn every 5%. (c) and (d) The cyclone track density composites corresponding to (a) and (b), drawn every 10 cyclones per 10° circle per month.

Citation: Monthly Weather Review 125, 10; 10.1175/1520-0493(1997)125<2531:LFVOSH>2.0.CO;2

Fig. 9.
Fig. 9.

(a)–(c) Time series for pressure PCs 1–3, normalized so that the top and bottom of each plot corresponds to ±3 SDs. (d) and (e) Number of days of anomalously high pressure in regions SE-NZ and SE-PAC, expressed as a departure from normal. (f) and (g) Blocking days for (f) SE-NZ, and (g) SE-PAC. See text for more details.

Citation: Monthly Weather Review 125, 10; 10.1175/1520-0493(1997)125<2531:LFVOSH>2.0.CO;2

Fig. 10.
Fig. 10.

Composite MSL pressure anomalies, drawn every 1 hPa, for the months in which more than five blocking days occurred in the region (a) SE-NZ and (b) SE-PAC. See text for more details.

Citation: Monthly Weather Review 125, 10; 10.1175/1520-0493(1997)125<2531:LFVOSH>2.0.CO;2

Fig. 11.
Fig. 11.

Correlation, drawn every 0.1, between the SOI and (a) and (b) MSL pressure anomalies during (a) winter (May–October) and (b) summer (November–April), and (c) and (d) winter cyclone track density anomaly composites, expressed as a percentage of mean track density, for months having (c) SOI < −1.5 and (d) SOI > 0.5.

Citation: Monthly Weather Review 125, 10; 10.1175/1520-0493(1997)125<2531:LFVOSH>2.0.CO;2

Table 1.

Percentage of variance associated with the first 10 EOFs of pressure, cyclone, and anticyclone track density anomalies, with the cumulative fraction of the total variance from these EOFs at the bottom of each column. The absolute uncertainty obtained from the North et al. (1982) test is included. The values in parentheses are obtained by extrapolating exponential regression fits to the contributions (not shown) of EOFs 11–20.

Table 1.

1

Actually, these regions are centered slightly to the north to avoid the data cutoff south of 65°S.

Save