Australian winter anticyclonicity, 1850–2006



[1] The latitude, longitude, and longitudinal extent of Australian winter anticyclonicity are reconstructed for the period 1850–2006 using the gridded mean sea level pressure data set of the Met Office (HadSLP2r). The latitude of the system has varied about 29°S, with no long-term trend evident, while the longitudinal extent has expanded by some 17° and the center pressure increased by some 1.5 hPa.

1. Introduction

[2] The mean sea level pressure (MSLP) field over the Australian continent exhibits a large annual variation. In summer, intense heating over the continent results in the development of thermal heat lows, while the subtropical anticyclones follow a more poleward track to the south of the continent.

[3] In autumn the anticyclones begin to track farther north, and in May the land-sea temperature contrast reverses, with the continent becoming cooler than the adjacent oceans. In winter, defined here as June–August (JJA), the anticyclones tend to track more over the Australian continent at ∼30°S, becoming “blocking” systems ∼18% of the time [Browne, 1975]. This winter maximum in anticyclonicity over the continent, subsequently referred to as Australian winter anticyclonicity (AWA), significantly affects the atmospheric circulation over Australia and leads to an increased prevalence of cool southerly airflows over the Tasman Sea and New Zealand in the winter months.

[4] Temporal variations of the mean latitude of the anticyclones have been the subject of a number of studies, with a detailed review provided by Drosdowsky [2005]. These have typically used a transect of surface stations down the east coast of Australia (∼150°E), from which the monthly MSLP were used to find the latitude (L) of highest MSLP, i.e., the latitude of the subtropical ridge, or “L index” [Pittock, 1973]. Most recently, Drosdowsky [2005] used daily reanalysis data from the National Centers for Environmental Prediction-National Center for Atmospheric Research (NCEP-NCAR) and monthly MSLP means from the Australian National Climate Centre (NCC) to construct L indices, again for the east coast of Australia.

[5] The availability of the high-quality global gridded MSLP data set, HadSLP2r, now permits a continental-scale approach to be made. The purpose of this study is to use these data to construct a new, long, and easily updated index for the variability of the AWA, with the results of this presented here.

2. Data and Method

[6] The global gridded MSLP data set, HadSLP2r [Allan and Ansell, 2006] covering the period 1850–2006, was obtained from the Met Office. These data are gridded at 5° × 5° latitude/longitude resolution. The data for the months June, July, and August, for longitudes 105–180°E and latitudes 5–45°S were extracted (Figure 1).

Figure 1.

Location map, showing the JJA MSLP pattern (hPa) for the decade 1996–2005, the location of the grid points (filled circles) from the HadSLP2r data set, and the location of the Alice Springs station (filled square).

[7] The zonal means for the latitudes 5–45°S were calculated at 5° resolution, averaged over the longitudinal range 105–180°E. A natural cubic spline curve was fitted to these data for each winter, and from this the latitude of maximum MSLP (i.e., the latitude of the center of the AWA) was obtained. The same approach was used to determine the longitude of the center of the AWA, using the meridional means for 105–180°E, again at 5° resolution, averaged over the latitudinal range 5–45°S.

[8] Both land and marine observations are included in the HadSLP2r data, and in common with other meteorological series, the quality and quantity of the data available vary with time. Figure 2 shows the number of observations of MSLP in each winter (JJA) for the period 1850–2004 for the study region defined above. The 1850s have the least data, and coverage becomes better from about 1910 on. The AWA is a large-scale synoptic feature and might be expected to be resolved relatively well even with fewer observations. However, the results for data prior to ∼1910 must be taken with greater caution.

Figure 2.

Number of MSLP observations in the region 5–45°S, 105–180°E for the months June, July, and August. Note the log scale on the y axis.

3. Results

[9] The latitude of the AWA has remained centered at about 28.8°S over the period of record (Figure 3 and Table 1) with no long-term trend poleward or equatorward. This lack of any significant change in the latitude of the AWA is consistent with the results of Drosdowsky [2005].

Figure 3.

Latitude of the center of the Australian winter anticyclone (AWA), °S.

Table 1. Latitude of the Center of the AWAa
  • a

    Latitude is in °S determined from the maxima of a cubic spline fit to the zonal mean HadSLP2r data, 5–45°S, 105–180°E, 1850–2006. The negative sign indicates southern latitude.


[10] Figure 4 plots the longitude of the center of the AWA. This suggests that a westward shift occurred in the 1930s.

Figure 4.

Longitude of the center of the AWA, °E.

[11] A Hövmöller plot of the HadSLP2r MSLP at 30°S and over the longitudinal range 105–180°E (Figure 5) indicates an increase in the longitudinal extent of the AWA over time, also consistent with the earlier result of Allan and Haylock [1993]. This expansion initially began in the west of the AWA starting about the 1920s–1930s when the limit of the 1021 hPa contour (for example) was about 133°E and continuing to the present where it is now less than 125°E. The eastern margin of the AWA did not start to show an expansion until about the 1980s, until which time the 1021 hPa contour was around 143°E, before expanding to the present limit of more than 150°E. This represents an expansion of more than 15° in the longitudinal extent of the AWA since the 1920s–1930s. Figure 5 shows that the westward shift in the longitude of the center of the AWA (Figure 4) has resulted, at least in part, from the expansion of the AWA, beginning with the western margin. This is consistent with the observed reductions in rainfall seen over southwestern Australia [Allan and Haylock, 1993; Haylock and Nicholls, 2000] and the conclusions of Allan and Haylock [1993] that these were due to an increase in MSLP.

Figure 5.

Hövmöller plot of the longitudinal extent of the AWA (degrees) and MSLP (hPa) at 30°S.

[12] The decreasing trend in south Australian rainfall began earlier in the southwest of the continent, in the 1930s–1940s [Allan and Haylock, 1993], where it has also been more marked. Since about 1997, significantly drier conditions have been experienced in southeastern Australia. These observations are what might be expected given the expansion of the western margin of the AWA in the 1930s, increasing the MSLP over southwestern Australia. As the AWA subsequently continued to expand, higher MSLP also occurred over southeastern Australia, notably since the mid-1990s (Figure 5). This coincides with the decrease in rainfall observed there.

[13] Interestingly, the minimum extent of the AWA occurred in the 1890–1920 period, a time when hemispheric temperatures (in the period of the instrumental record) were coolest. If this period were chosen as the baseline, the longitudinal expansion of the AWA to its present size would be closer to 17°. Prior to this, in the 1880s, a period when hemispheric temperatures were warmer than those in the 1890–1920 period, the longitudinal extent of the AWA is again observed to have been greater.

[14] The MSLP at the center of the AWA was determined from synoptic maps produced for each winter from the HadSLP2r data. This is shown in Figure 6. No correction has been made for year-to-year changes in the latitude of the AWA, which increases the variance of the series. A slight increase (1.5 hPa) in MSLP is observed from 1021.1 hPa in the period 1890–1899 to 1022.6 hPa in the period 1997–2006.

Figure 6.

MSLP (hPa) at the center of the AWA (top curve) and at Alice Springs (bottom curve).

[15] In order to compare these results from the reconstructed HadSLP2r data with observational data, the long-term climate station of Alice Springs 23.7°S, 133.9°E was chosen (Figure 1). This is relatively close to the center of the AWA, and data for the two stations (Alice Springs Post Office (station 15540), 1878–1953, and Alice Springs Airport (station 15590), 1951 to present) were obtained from the Australian Bureau of Meteorology with the JJA MSLP plotted in Figure 6. For Alice Springs, the JJA MSLP for the decade 1890–1899 was 1019.7 hPa, while for 1997–2006 it was 1021.5 hPa, an increase of 1.8 hPa. The correlation between the Alice Springs MSLP and that of the center of the AWA is 86% (n = 122). This increase is consistent with the longitudinal expansion of the AWA, although as discussed in section 4, less than would be expected from dynamical considerations alone.

4. Discussion

[16] It is useful to compare the values for the latitude of the AWA obtained here with previous work. Figure 7 plots the published values of the L index of Pittock [1973] against the latitude of the AWA from this study. While the two indices are highly correlated (0.75, n = 31), the variance of the L index (σ = 3.07) is much greater than for that calculated here (σ = 0.92). This arises because the AWA is centered over middle Australia, some 2000 km away from the east coast stations used by Pittock [1973] to derive the L index. Subtle changes in the orientation of the AWA are observed from year to year. These result from changes in the location of the upper level Rossby wave pattern, which alters the mean track of the anticyclones across the continent. During some years then, the AWA is orientated slightly more WSW to ENE, while in others it is more WNW to ESE. This has little effect on the location of the center, as calculated here, but it introduces significant variability at the margins of the system and therefore in the L index for the winter months.

Figure 7.

Latitude (°S) of the AWA (solid curve, this study) and of the subtropical ridge L index (dashed line) of Pittock [1973].

[17] A similar effect is seen in the L index of Drosdowsky [2005] (Figure 8) where for the period 1890–2003, σ = 2.32 compared to σ = 0.88 for the method used here.

Figure 8.

Latitude (°S) of the AWA, this study (top curve) and of the subtropical ridge L index of Drosdowsky [2005] (bottom curve).

[18] The latitude of the AWA from this study is some 2–4° equatorward of that reported by Pittock [1973] and Drosdowsky [2005]. Again, this reflects the difference between measuring the latitude of the center of the AWA as opposed to measuring the L index on the eastern margin. The anticyclones do not track directly west to east but tend to track, on average, a little more southward as the eastern coast of Australia is approached, and this biases the L index to slightly higher latitudes.

[19] Having said this, the index for the latitude of the AWA obtained here is significantly correlated at the 99% level with the L index of Drosdowsky [2005] (0.80) and Pittock [1973] (0.75) for the period 1941–1971 when all three overlap. Over the period 1890–2003, the correlation between the AWA latitude (this study) and that of Drosdowsky [2005] is 0.62. This is still significant at the 99% level.

[20] In previous work, Larsen [1998a, 1998b] used a similar methodology to that described here for the gridded MSLP data set of Jones [1991]. The results from this study were again very similar to those reported here: no long-term change in the mean latitude of the AWA but an increase in the longitudinal extent of the system.

[21] The lack of any significant long-term trend in the latitude of the AWA appears to be a robust result, having been obtained from the following four data sets: (1) analysis of the Jones [1991] series by Larsen [1998a, 1998b], (2) the NCEP-NCAR reanalysis data, (3) the NCC data by Drosdowsky [2005], and (4) the HadSLP2r data from this study. This is interesting in light of the work of Chylek et al. [2001] and Lu et al. [2007], who, on theoretical and modeling grounds, predict that a poleward movement in the location of the subtropical ridge should occur with increasing global temperatures. Why this is not apparent over wintertime Australia is unclear.

[22] The center pressure elevation, ΔP (hPa), of an anticyclone is dynamically constrained:

equation image

where ρ is the air density (kg m−3), r is the radius of the system (km), and f is the Coriolis parameter (f = 2Ωsin ϕ), with Ω the rotation rate of the Earth and ϕ the latitude [McIlveen, 1992].

[23] As the long-term mean latitude of the AWA has remained constant over time (Figure 3), the observed increase in MSLP (Figure 6) cannot be due to an increase in the Coriolis parameter. Assuming that ρ has remained unchanged, or slightly decreased, as the continent has warmed, then the only dynamical explanation for the increase in MSLP is an increase in the radius of the AWA, which has occurred (Figure 5).

[24] For an anticyclonic system, equation (1) predicts that the observed increase in MSLP at the center of the AWA could be achieved with a modest expansion in the radius of the AWA of some 100 km. This is much less than the ∼1440–1640 km (15–17° longitude at 30°S) expansion measured here. Any decrease in air density from a warming of the atmosphere is also insufficient to explain this. The most likely explanation is a combination of the following factors: (1) the anticyclones have become slightly more intense and/or (2) have been moving more slowly (or with more frequent blocking) over the continent and/or (3) have been tracking across the continent over a progressively broader range of longitudes, leading to the observed mean pattern of a larger AWA. That MSLP has increased more at Alice Springs than at the center of the AWA, which lies south-eastward (Figure 1), is supportive of an increase in anticyclonic activity to the west, reflected also in the westward shift of the mean longitude of the AWA (Figure 4). In order to determine which of these factors are responsible requires an analysis of the daily synoptic charts.

5. Conclusions

[25] The use of the HadSLP2r gridded MSLP data set provides a new and readily updated means of calculating the location of the AWA. In winter, when the anticyclones are located some distance from the eastern coast, the approach described here has greater precision than indices based on a transect of stations along the Australian east coast.

[26] While the latitude of the AWA has remained essentially constant, the AWA has become longitudinally more extensive and intensified over the past century. The AWA was weakest around 1900–1920, a period when hemispheric temperatures were coolest, and appears to be more extensive/intense in periods of warmer climate (1875–1890) and especially post-1920.

[27] The slight longitudinal shift westward in the center of the AWA during the 1930s coincides with the onset of the drying trend seen over southwestern Australia. The shift of the AWA to the west would have delayed the increase in MSLP over eastern Australia, while with the continued expansion of the system, higher MSLP has subsequently affected eastern Australia.


[28] The author gratefully acknowledges the provision, in the public domain, of the HadSLP2r data set by the Met Office, Wasyl Drosdowsky for the provision of his L index and helpful comments, and those of the three anonymous reviewers.