1. Introduction
Most of the subtropical anticyclones reside over the eastern portions of the subtropical ocean basins throughout the year. In the Northern Hemisphere (NH), they show apparent seasonal variability in their strength and configuration. Each of them strengthens and develops into a basin-wide cell-type configuration in summer. The surface wind stress curl associated with these anticyclones contributes to the maintenance of subtropical oceanic gyres, with warm poleward western boundary currents and cool equatorward currents off the west coasts of the continents. The strong equatorward alongshore winds to the east of the anticyclones also contribute to the maintenance of underlying cool sea surface temperatures (SSTs) by enhancing surface evaporation, coastal upwelling, and entrainment at the bottom of the oceanic mixed layer. In the Southern Hemisphere (SH), as well, the summertime surface subtropical anticyclones are in a cell-type configuration (as shown below), suggesting that their dynamics may be understood within essentially the same framework as is applicable to their NH counterpart.
As argued by Hoskins (1996) and Rodwell and Hoskins (2001), the theory of a zonally symmetric Hadley circulation predicts much weaker subtropical subsidence in the summer hemisphere than in the winter hemisphere (Lindzen and Hou 1988). In relation to the formation of zonally asymmetric circulation in the subtropics as a manifestation of the climatological summertime planetary waves, the surface NH subtropical anticyclones have been argued to be a steady response to monsoonal convective heating to their east (Hoskins 1996; Rodwell and Hoskins 2001) or west (Chen et al. 2001) or to shallow heating–cooling couplets across the west coasts of the subtropical continents (Liu et al. 2004; Nakamura and Miyasaka 2004; Miyasaka and Nakamura 2005, hereafter MN05).
Specifically, the argument by Hoskins (1996) is based on the “monsoon–desert mechanism” (Rodwell and Hoskins 1996). Rodwell and Hoskins (2001) concluded that the combined effects of topography and monsoonal heating to the east are of primary importance for the generation of a surface subtropical anticyclone and subsidence aloft (e.g., the Indian, Mexican, and South American monsoons for the subtropical anticyclones in the North Atlantic, North Pacific, and South Pacific, respectively), with the reinforcement by local cooling over the eastern oceans. However, Wu and Liu (2003) suggested the potential importance of the combined effects of surface sensible heating observed over the dry western portions of subtropical continents and lower-tropospheric radiative cooling off their west coasts in the formation of the subtropical anticyclones.
In fact, stationary responses to prescribed diabatic heating in the numerical experiments by Liu et al. (2004), Nakamura and Miyasaka (2004), and MN05 have shown the importance of the shallow continental heating and maritime cooling in the formation of surface NH subtropical anticyclones. Through coupled general circulation model (GCM) experiments, Seager et al. (2003) also suggested the potential importance of the local air–sea interaction for the subtropical anticyclones in attaining their full strength, although they argued that the monsoonal influence is particularly important. The presence of cool SSTs and a midtropospheric subsidence associated with the anticyclone favors the local development of marine stratus in the planetary boundary layer (PBL; Klein and Hartmann 1993). The high albedo of the clouds also acts to maintain the underlying cool SSTs (Hartmann et al. 1992), which may imply the involvement of a local air–sea feedback system.
The present study is an extension of Nakamura and Miyasaka (2004) and MN05 on the summertime NH subtropical anticyclones into their SH counterparts. Similarly to the NH, marine stratus and stratocumulus develop off the west coasts of the SH subtropical continents (Klein and Hartmann 1993), while surface sensible heat flux is strongly upward over the western portions of those continents (Wu and Liu 2003; Liu et al. 2004). In their modeling study, Rodwell and Hoskins (2001) investigated the dynamics of the South Pacific subtropical anticyclone. In the thermal forcing in their model, contributions from monsoonal convective heating and surface sensible heating over the South American continent were not well separated, because they assigned diabatic heating to all the model tropospheric levels. Seager et al. (2003) showed that the monsoon–desert mechanism exerts a weaker impact in the SH than in the NH. They concluded that oceanic processes, which may be driven by the wind stress associated with the subtropical anticyclones, are more important for the maintenance of low SSTs on the eastern portions of the South Pacific and South Atlantic and thus for the formation of the subtropical anticyclones. Through experiments with their regional model in which marine stratiform clouds are reproduced realistically, Wang et al. (2005) showed that radiative cooling associated with those clouds in the eastern South Pacific can reinforce the subtropical anticyclone in austral winter and spring. The aforementioned findings seem consistent with the hypothesis that the subtropical anticyclones in the SH summer can be formed through the essentially the same dynamics as in the NH summer. Nevertheless, the three-dimensional structures of each of the SH anticyclones and the relative importance between the low-level, land–sea thermal contrast and the monsoonal convective heating in the formation of each of those anticyclones have not been fully understood.
In the present study, the three-dimensional structure and mechanisms of the three SH subtropical anticyclones are investigated through data analysis and numerical modeling as in MN05. Although the monsoon–desert mechanism argued by Hoskins (1996) and Rodwell and Hoskins (1996, 2001) may be important in maintaining the dryness of the western portions of the continents and thereby setting fundamental zonal asymmetries in the diabatic heating pattern, the present study suggests that the summertime climatological SH subtropical anticyclones can essentially be interpreted within the same framework of local atmosphere–ocean–land interactions as their NH counterpart.
2. Data and atmospheric model
a. Data and analysis methods
As in MN05, the potential vorticity (PV) inversion technique (Hoskins et al. 1985) is used to assess the impacts of a zonally asymmetric distribution of surface temperature upon the tropospheric circulation. It is based on the invertibility principle for PV (q), which indicates that the surface θ* can act as PV anomalies (q*). Specifically, warm and cold anomalies at the surface are equivalent to cyclonic and anticyclonic PV anomalies, respectively. In practice, the observed 1000-hPa θ* is assigned as the lower-boundary condition with no q* assigned in the free atmosphere and at the upper boundary. The combination of the PV inversion and the omega equation diagnosis [(2)] enables us to evaluate midtropospheric vertical motion that would be induced in the presence of a given θ* distribution at the surface.
b. Atmospheric model
As in MN05, a global primitive equation model with simplified physical processes is used to investigate the atmospheric response to a given zonally asymmetric distribution of the tropospheric diabatic heating, Q*. The model, originally developed at the University of Reading (Hoskins and Simmons 1975), has been modified for the study of summertime planetary waves as used by Rodwell and Hoskins (2001) and MN05. The model details can be found in Hoskins and Rodwell (1995). The nonlinear spectral model with a triangular spectral truncation at wavenumber 42 (T42) has 30 vertical levels in the σ coordinate. To resolve shallow Q* confined mainly into the PBL, the model vertical resolution has been set twice as high as in Rodwell and Hoskins (2001). Throughout each of the model integrations, the zonal-mean state was fixed to the climatological-mean state taken from the NCEP–DOE reanalyses. In the model it took 10∼15 days for a surface subtropical anticyclone to develop after Q* had been switched on. We therefore examine 10-day mean fields from the 16th to the 25th day of the integrations. In experiments with orography, the surface elevation had been raised gradually over a 5-day period before Q* was imposed.
In our experiments, Q* measured as the rate of a local temperature change was taken from the NCEP–DOE reanalysis data at the 28 σ levels of the reanalysis system and then interpolated onto the 30 equally spaced model levels from σ = 0.983 to σ = 0.017. In some of our experiments, Q* was assigned with its full intensity only within a given rectangular domain or a latitudinal band. Outside of the domain, the intensity of Q* was reduced linearly in such a manner that Q* vanishes within 10° in latitude or longitude away from the domain boundaries. Strictly speaking, the heating outside of the domains does not necessarily equal zero locally, since the zonal average of Q* has to remain zero in our experimental design.
3. Observations
a. Three-dimensional structure
Figure 1 shows the climatological-mean January sea level pressure (SLP) field for the SH. As in the summertime NH, a cell-type subtropical anticyclone is observed over each of the three SH ocean basins (i.e., the South Pacific, South Atlantic, and South Indian Oceans). The anticyclonic center positioned at ∼30°S is displaced zonally to the eastern portion of the basin with isobars almost parallel to the west coast of the nearby continent. Strong alongshore equatorward winds are thus observed in the eastern flank of the anticyclone. The vorticity balance near the surface is achieved mainly by the cyclonic planetary vorticity advection due to the alongshore winds and vorticity-tube shrinking associated with midtropospheric subsidence (Fig. 1b), in agreement with Rodwell and Hoskins (2001). In the upper troposphere, anticyclonic and cyclonic vorticity centers located above the poleward and equatorward portions of a particular surface anticyclone (Fig. 1d) can both induce anticyclonic vorticity tendencies to their immediate east by the mean zonal advection and by anomalous planetary-vorticity advection, respectively (not shown). These tendencies tend to be in balance with the offsetting effect due to the local subsidence. The upper-level meridional vorticity dipole is a manifestation of the barotropic and baroclinic structures in the poleward and equatorward portions, respectively, of a given surface anticyclone. These features are essentially the same as in the summertime NH as observed by White (1982), Wallace (1983), and MN05.
Reflecting the tendency that upper-level vorticity dipoles in the SH are weaker than in the NH, the double-jet structures in upper-level westerlies are less clear above the SH subtropical anticyclones (Fig. 1c), especially over the South Atlantic anticyclone. Nevertheless, in a manner consistent with the presence of the meridional vorticity dipoles, the upper-level westerly wind speed (ug) increases eastward above the equatorward portions of those subtropical anticyclones, including the South Atlantic anticyclone (see the curvature of the isotachs in Fig. 1c). The primary momentum balance, ug(∂ug/∂x) ≈ fυa, requires poleward ageostrophic motion (υa < 0) above the anticyclones, where ∂ug/∂x > 0 (Fig. 1e). The poleward ageostrophic motion appears to be the primary contributor to the upper-level convergence and the associated midtropospheric subsidence over the SH surface anticyclones (Fig. 1b). In addition, equatorward ageostrophic motion (υa > 0) from the midlatitudes also contributes to the convergence over the southeastern Indian Ocean (Fig. 1e). This motion is required in the momentum balance in the downstream regions of the polar-front jet cores (Fig. 1c; Nakamura and Shimpo 2004). Thus, the summertime surface subtropical anticyclones in the SH are similar to their NH counterparts with respect not only to their three-dimensional structure but also to their dynamical coupling with the upper-level vorticity dipoles. Though somewhat weaker than in the reanalysis data, the midtropospheric vertical motion, including the subsidence over the subtropical anticyclones, can be reproduced through a diagnosis based on the omega equation [(2)] (Fig. 2a), which is linearized about the climatological zonal-mean state for the SH summer and forced with the heat and vorticity transport associated with the observed planetary waves (Fig. 2a). This result indicates that those subtropical anticyclones are a fundamental element of the climatological planetary wave in the SH summer.
b. Wave-activity propagation
The structural relationship between the surface subtropical anticyclones and upper-level planetary waves has been investigated in relation to the structure of diabatic heating (Wu and Liu 2003; Liu et al. 2004) and by using Plumb’s (1985) wave-activity flux (MN05). We conduct the same wave-activity flux diagnosis as in MN05 to gain some insight into the forcing region of the upper-level planetary waves in focusing particularly on its vertical component in the midtroposphere. Over the summertime SH, the horizontal component of the wave-activity flux (W) associated with the upper-tropospheric planetary waves shown in Fig. 2b indicates no apparent incoming wave train into any of the core regions of the subtropical anticyclones in the South Pacific and South Indian Oceans, where eastward W diverges out of the vorticity dipoles with its upward component in the midtroposphere above the surface anticyclones (Fig. 2c). While a weak influx of wave activity is observed over the South Atlantic, the midtropospheric wave-activity flux is slightly upward to the east of the surface anticyclone. This upward wave-activity flux suggests a source of the SH summertime planetary waves that is located in the lower troposphere and associated with the surface anticyclones, as observed in the NH summer (Nakamura and Miyasaka 2004; MN05). In fact, in the subtropical SH (20°∼40°S), the zonal wavenumber-3 component seems dominant in the upper-level planetary wave field (Fig. 1d), showing some correspondence with the three high pressure cells at the surface. This feature is in sharp contrast with the dominance of the zonal wavenumber-1 component poleward of 45°S. Interestingly, poleward wave-activity propagation is evident from the subtropical anticyclone over the southeastern Indian Ocean. This propagation appears to add a signature of higher zonal wavenumbers to the midlatitude planetary wave field to the south of Australia and over the western South Pacific. In the SH summer, equatorward wave-activity dispersion over the subtropical anticyclones tends to be stronger than in the NH summer. This is probably because upper-tropospheric zonal-mean westerlies extend more equatorward in the summertime SH than in the summertime NH (see the vertical arrows along the right axis of Fig. 2b). In fact, at the 250-hPa level the critical latitude for stationary Rossby wave over the summertime SH is displaced equatorward by more than 10° in latitude relative to its NH counterpart (13°S versus 25°N). In fact, the wave-activity flux is evident from each of the vorticity dipoles to a cyclonic anomaly located to its northeast (Fig. 2b). The northeastward propagation is particularly strong over the eastern South Pacific.
c. Diabatic heating and land–sea thermal contrasts
As in the summertime NH, the summertime SH subtropics are characterized by strong land–sea heating–cooling contrasts across the west coasts of the continents (Fig. 3). As pointed out by Wu and Liu (2003), continental heating near the west coast is associated with near-surface sensible heating over the dry land surface heated by strong insolation. In fact, over the South African continent and Australia, the diabatic heating Q* is maximized near the surface (Fig. 3b). Over the South American continent, however, both the near-surface sensible heating and deep convective heating above the eastern slope of the Andes contribute to the continental heating Q*. In the reanalysis data, the maritime cooling, which peaks in the lower troposphere at the level of σ = 0.85, is associated mainly with the radiative cooling due to marine stratus and/or stratocumulus.
To confirm the potential impacts of the pronounced summertime land–sea thermal contrasts in the SH subtropics upon the maritime anticyclones, continental lows, and strong alongshore winds at the surface, the PV inversion technique was applied to the observed field of the zonally asymmetric component of the 1000-hPa temperature θ* (Fig. 4a). As in the NH case (MN05), the induced surface winds are the strongest along the west coasts of the subtropical continents, across which the zonal θ* contrasts are most pronounced. The induced alongshore equatorward winds advect cooler air and negative (i.e., cyclonic) planetary vorticity, which can be balanced if the midtropospheric subsidence and associated near-surface divergence are induced, respectively. In fact, subsidence is evident just to the east of each of the surface subtropical anticyclonic centers, in the field of vertical motion (Fig. 4b) as diagnosed by substituting the wind and temperature fields obtained through the PV inversion (Fig. 4a) into the omega equation [(2)] that has been linearized about the climatological zonal-mean state for January. The diagnosed subsidence is weaker than in the observations, since the PV anomalies were assigned only at the surface. The diagnosed descent is shifted slightly to the east of its observational counterpart (Fig. 1b), because the induced surface wind over the land surface is much stronger than in the real atmosphere due to the lack of surface friction in the PV inversion. Nevertheless, the PV inversion and the subsequent diagnosis of the induced vertical motion have confirmed the potential importance of the surface thermal contrast across the west coast of a subtropical continent in the formation of a SH summer subtropical anticyclone, as has been revealed for the NH summer (MN05).
4. Model simulations
a. Remote versus local influences
As in MN05, the primitive-equation planetary wave model as described in section 2b was used to examine the reproducibility of the zonally asymmetric component of the climatological-mean SH summertime tropospheric circulation, including the surface subtropical anticyclones. Compared with their observational counterparts (Fig. 5a), the maritime subtropical anticyclones and continental cyclones are well reproduced in the model SLP response (Fig. 5b) to the zonally asymmetric diabatic heating (Q*; cf. Fig. 3a) assigned globally at all the vertical levels in the model with global topography (control experiment). This result is in good agreement with Rodwell and Hoskins (2001). Over the South Atlantic and Pacific, the surface anticyclones are reproduced with their full intensities. In another experiment in which the prescribed Q* is limited to a latitudinal band of 40°∼10°S (Fig. 5c), no substantial reduction is found in the magnitude of any of the subtropical anticyclones. Rather, the anticyclone in the South Indian Ocean (i.e., Mascarene high) is enhanced to attain its observed intensity. A comparison between Figs. 5b and 5c reveals that remote influences either from the tropical convection or from higher latitudes do not act as the primary direct forcing for any of the SH summertime subtropical anticyclones. The same is the case for their NH counterparts (MN05).
b. Model response to a low-level heating–cooling couplet in each of the ocean basins
To substantiate the importance of the local land–sea thermal contrasts in the formation of the SH subtropical anticyclones in January, additional numerical experiments were performed with the same model but driven by Q* prescribed only in the lower troposphere (below the σ = 0.667 level) within the regional domains, as indicated with heavy solid lines in Figs. 3b and 3c. In the first model experiment of this kind, the atmospheric response to the low-level Q* prescribed within the southeastern Pacific–South American domain (Fig. 6a) is investigated. The continental heating is associated mainly with sensible heat flux over the (semi-) desert region in the subtropics and midlatitudes and convective heating in the tropics (Fig. 3). Though maximized at the σ = 0.4 level, the convective heating extends into the lower troposphere (Fig. 3b). The offshore cooling confined to the lower troposphere (Fig. 3b), which is associated with marine stratus–stratocumulus, extends more equatorward than in the summertime NH. In this particular experiment, both the maritime cooling and continental heating were assigned together, because the PV thinking predicts that the strength of the alongshore winds should be related to the zonal contrast of the surface temperature, but not solely to either the heating or cooling. This is also because the zonally integrated Q* is much weaker for the heating–cooling couplet than for either the heating or cooling alone and thus more suited for our planetary wave model experiment. To isolate the response to Q*, no topography was imposed in the model, although we have confirmed that the model surface response is insensitive to the inclusion of topography (not shown).
As a stationary response to the imposed Q* (Fig. 6), a surface temperature contrast across the west coast of the South American continent, though somewhat overestimated, is reproduced reasonably well in the model (Figs. 2d and 6b). The subtropical anticyclone and alongshore equatorward winds are also reproduced reasonably well in the surface response (Fig. 6c), although the simulated anticyclonic center is shifted equatorward by ∼10° in latitude compared with the observations (Fig. 5a). We have confirmed that the surface anticyclonic response is reproduced well solely with the maritime cooling (not shown). The intensity of the surface anticyclone (Fig. 6c) reaches as much as ∼80% of its counterparts in the observations (Fig. 5a) and in the control experiment described in the preceding subsection (Fig. 5b). Correspondingly, subsidence is also induced above the surface anticyclone as observed (Fig. 6d). Unlike in the observations (Fig. 1b), however, the ascent is simulated noticeably poleward of the surface anticyclone (Fig. 6d). In the presence of this dipolar pattern in vertical motion, a meridional vorticity dipole is generated above the simulated surface anticyclone (Fig. 6e). The anticyclonic response that constitutes the dipole is overestimated, probably due to the effects of the unrealistically strong midtropospheric ascent and the associated upper-level divergence simulated upstream. This unrealistic equatorward shift of the upper-level dipolar vorticity pattern is consistent with the corresponding equatorward shift of the surface anticyclone in the model response. The node line of this meridional dipole above the surface anticyclone displays a northwest–southeastward tilt as in the observations (Fig. 1d), consistently with the northeastward wave-activity flux (Fig. 6f) as observed (Fig. 2b). The model simulation therefore suggests that the land–sea heating–cooling contrast can contribute to the formation of the surface subtropical anticyclone and the upper-level planetary wave pattern over the southeastern Pacific and South America, although the anticyclonic portion of the upper-level meridional dipole was reproduced in the region of cyclonic vorticity in the observations. As in the NH summer, the contribution from the land–sea heating–cooling contrast to the upper-level planetary wave pattern is comparable or even weaker than that from the response to the mid- and upper-tropospheric diabatic heating, as suggested by another experiment (not shown). Of course, for a particular thermal forcing with a given intensity, the upper-level response should be stronger if the forcing is placed at the upper level than in the case where the forcing is placed near the surface. Another factor that can result in the weakness of the upper-level response to the near-surface heating–cooling contrast may be the presence of the zonal-mean easterlies in the region of low-level heating–cooling, especially in the SH summer. Unlike in the case of zonal-mean westerlies, a near-resonant stationary Rossby wave response cannot be realized in the presence of the easterlies.
A similar numerical experiment was performed with the lower-tropospheric heating–cooling contrast between the African continent and the South Atlantic (Figs. 3b, 3c, and 7a). In the model integration carried out without topography, the near-surface land–sea thermal contrast is reproduced reasonably well in intensity (Fig. 7b). In Fig. 7c, the strength of the surface subtropical anticyclone reaches ∼80% of its observational counterpart (Fig. 5a) and ∼67% of that in the control experiment (Fig. 5b). The simulated anticyclone is, however, displaced equatorward compared with the observations (Fig. 5a), as in our simulation for the South Pacific (Fig. 6c). Midtropospheric subsidence is simulated to the southeast of the surface anticyclone (Fig. 7d), but in reality it is observed over the eastern portions of the anticyclone (Fig. 1b). As in the South Pacific case, unrealistically strong ascent simulated upstream of the subsidence (Fig. 7d) leads to the overestimation of the upper-level anticyclonic anomaly that constitutes a meridional dipole with a cyclonic anomaly to the north (Fig. 7e). Though shifted equatorward in correspondence with the subsidence, the meridional dipole above the surface anticyclone is reproduced upstream of the subsidence (Fig. 7e). A node line of the simulated dipole is tilted northwest–southeastward (Fig. 7e) as observed (Fig. 1d), consistent with the northeastward wave-activity flux in the model (Fig. 7f) and in the observations (Fig. 2b). The enhanced upper-level anticyclonic anomalies accompany well-defined poleward wave-activity propagation that is much stronger than in the observation. Our wave-activity flux analysis indicates that the low-level land–sea heating–cooling couplet can act as a local source of planetary waves in the subtropics over the South Atlantic and Africa (Fig. 7f), although the upward flux is somewhat overestimated in the model.
The numerical experiment was repeated with the lower-tropospheric heating–cooling contrast between the southeastern Indian Ocean and the Australian continent (Figs. 3b, 3c, and 8a). Again, no surface elevation was imposed. Compared with its observational counterpart (Fig. 2d), the land–sea thermal contrast across the Australian coast is well reproduced in intensity (Fig. 8b). The surface subtropical anticyclone is also reproduced well (Fig. 8c). Although its center is slightly equatorward relative to the observation, the strength of the surface anticyclone in Fig. 8c reaches ∼100% of its counterpart in the observations (Fig. 5a) and ∼125% of that in the control run (Fig. 5b). Though shifted somewhat eastward relative to the observations (Fig. 2a), the associated midtropospheric subsidence is also reproduced (Fig. 8d). However, the model generates unrealistic updraft in the mid- and upper troposphere above the surface anticyclone. Although the upper-level vorticity dipole (Fig. 8e) is shifted northeastward relative to the observations due to the unrealistic updraft simulated, two Rossby wave trains are reproduced in a realistic manner (Fig. 8e): one propagating northeastward directly into the tropics and the other propagating southeastward and then refracted northeastward into the western South Pacific. The latter wave train is also evident in the observed SLP field (Fig. 5a) and in the model SLP response (Fig. 5f). These wave trains are confirmed through a wave-activity flux diagnosis applied to the model response (Fig. 8f). The aforementioned numerical results suggest that the land–sea heating–cooling couplet can act as one of the local sources of the SH summertime planetary waves with the dominant zonal wavenumber-3 component as observed, while contributing to the formation of the local wave trains as observed in the Australian–South Pacific sector (Figs. 2b and 2c).
5. Summary and discussion
In the present study, we have clarified the three-dimensional structure of the climatological-mean summertime subtropical anticyclones observed in the SH. As in the NH summer (MN05), the SH subtropical anticyclones are accompanied locally by pronounced zonal contrasts in surface air temperature and near-surface diabatic heating across the west coasts of the subtropical continents. Using the same nonlinear atmospheric primitive-equation model as was used in MN05 for the NH summer driven by a given zonally asymmetric diabatic heating pattern, we have shown that the model SLP response to a localized near-surface heating–cooling contrast as observed in January can account for more than 80% of the observed strengths of the SH summertime subtropical anticyclones as a component of the climatological planetary waves, although the upper-level vorticity balance in the model response appears to be distorted to a certain degree. Thus, the formation of the summertime subtropical anticyclones and associated upper-level vorticity dipoles in the two hemispheres can be interpreted within the same framework where an anticyclone forms as an atmospheric response to a local thermal forcing. The model response is understandable from a PV thinking viewpoint, since cool and warm anomalies at the ocean and land surfaces act as anticyclonic and cyclonic PV anomalies, respectively, to yield strong equatorward alongshore winds in between. In a steady response, vorticity and heat transport by the alongshore winds require descent above the eastern portion of the subtropical anticyclone, as observed, to maintain the vorticity and heat balance (Hoskins 1996). This descent acts to maintain the surface anticyclone against the friction and to force upper-level planetary waves. In fact, our diagnosis has revealed the upward component of a wave-activity flux of stationary Rossby waves and its upper-level divergence in the vicinities of the surface subtropical anticyclones observed and simulated over the South Pacific, South Atlantic, and South Indian Oceans. The upward wave-activity flux is a manifestation of poleward heat transport associated with the cool equatorward alongshore winds. The wave-activity flux diagnosis reveals no substantial influx of Rossby wave activity from upstream into any of the subtropical anticyclones. Our results indicate that the pronounced near-surface land–sea thermal contrasts accompanied by those anticyclones can act as local sources of the planetary waves observed in the summertime subtropical SH, contributing to the formation of the wavenumber-3 pattern that prevails in the subtropics. We have shown that one of the wave trains emanating from the subtropical anticyclone over the South Indian Ocean reaches into the midlatitudes, modifying the midlatitude planetary wave pattern in the Australian–southwestern Pacific sector.
Each of the observed SH surface subtropical anticyclones accompanies a meridionally oriented vorticity dipole in the upper troposphere, as was found for their counterparts in the NH summer (see MN05; the signs of vorticity related to the trough and ridge are opposite between the SH and NH). The summertime subtropical anticyclones in the two hemispheres commonly exhibit barotropic structures in their poleward portions and baroclinic structures in their equatorward portions where the low-level zonal-mean flows are easterly. In view of the vorticity and zonal momentum balance, this structure is favorable for inducing descent effectively over a surface anticyclone, particularly over its eastern portion.
The land–sea thermal contrast involved in the formation of each of the subtropical anticyclones is mainly a manifestation of the thermal effect by relatively low SSTs over the eastern ocean with radiative cooling by marine stratus and by heating through a sensible heating flux over a dry, heated landmass to the east. The importance of low SSTs has been suggested through coupled GCM experiments by Seager et al. (2003), in which the subtropical anticyclones in the SH summer are not well reproduced because the model SSTs are not as low as observed over the eastern portions of the ocean basins. A regional model experiment by Wang et al. (2005) showed that radiative cooling off the Peruvian coast in austral spring reinforces a surface subtropical anticyclone off the coast. The anticyclonic SLP response in their model reaches ∼1 hPa in intensity when the maritime cooling around 20°S is enhanced by ∼1 K day−1. This relationship is consistent with our result that a ∼4 hPa SLP response is induced by the cooling of ∼3 K day−1. The temperature inversion forms in association with midtropospheric descent, low SSTs, and cold advection by the alongshore equatorward winds, all of which can be augmented as the nearby subtropical anticyclone intensifies seasonally into summer. Strong radiative cooling associated with marine stratus and stratocumulus has been shown in the present study to possibly force the subtropical anticyclone. The cooling is thus, in turn, maintained in the presence of the anticyclone itself, suggestive of a local feedback loop in the atmosphere–ocean–land system in the summertime subtropics (Ma et al. 1994; Hoskins 1996; Seager et al. 2003; Nakamura and Miyasaka 2004; MN05). Reinforcement of continental heating through the suppression of both moist convection and cloud formation by descent can also be an additional factor for the feedback loop. In fact, Grotjahn and Osman (2007) showed that intraseasonal anomalies in surface air temperature over the western portion of the United States tend to lag the development of the North Pacific subtropical anticyclone.
Although the present study and MN05 suggest that the surface summertime subtropical anticyclones in the two hemispheres form mainly through local feedback processes, one may nevertheless wonder why the summertime subtropical anticyclones observed in the SH are weaker than in the NH. Figure 5a indicates that their strength measured in terms of zonally asymmetric SLP is 4 ∼ 5 hPa, which is only half of their counterpart in the NH (MN05). One of the reasons may be that the SH continents are zonally narrower than their counterparts in the NH. Since the zonally asymmetric surface temperature distribution acts as anomalous PV, the induced circulation should be weaker if the zonal scale of the thermal anomalies is smaller (Hoskins et al. 1985). Another reason may be that climatologically the lower-tropospheric zonal-mean westerlies in summer are more strongly confined meridionally into the midlatitudes over the SH than over the NH. To a given forcing, the near-resonant stationary response is possible in the presence of the zonal-mean westerlies but not in the presence of easterlies. Another interhemispheric difference is that the upper-tropospheric wave-activity flux that diverges out of the regions around the surface subtropical anticyclones tends to extend equatorward farther in the SH, which is probably because the upper-tropospheric zonal-mean westerlies extend into the lower latitudes over the SH (13°S versus 25°N). These interhemispheric differences in the climatological westerlies are climatological manifestations of a stronger meridional thermal gradient in the summertime subtropics over the SH than over the NH, where the Asian monsoon dominates.
The importance of the near-surface heating–cooling patterns for the formation of the SH surface subtropical anticyclones has also been confirmed through our other model simulations with deep convective heating imposed as forcing in any of the rectangular domains in Fig. 3a, which cannot induce any noticeable anticyclonic response at the surface in the SH subtropics (not shown). This tendency has been confirmed through a similar numerical experiment but with the mid- and upper-tropospheric diabatic heating taken from the Japanese 25-yr reanalysis (JRA-25) dataset (Onogi et al. 2007), as a product of the cooperative research project of the JRA-25 long-term reanalysis by the Japan Meteorological Agency and the Central Research Institute of Electric Power Industry. The wave-activity flux diagnosed from the 500-hPa temperature, wind, and geopotential height fields based on the JRA-25 data is downward, as opposed to the corresponding flux diagnosis based on other reanalysis data, including the 40-yr European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-40; Uppala et al. 2005). The discrepancy is probably due to some uncertainties in the reanalysis data currently available for the SH. It should be pointed out that the corresponding numerical experiments, as in Figs. 5 –8 but with low-level diabatic heating taken from the JRA-25 data, can reproduce the results shown in the upward flux of Rossby wave activity over the surface subtropical anticyclones. Reexamination with a more reliable dataset is required in the future in order to substantiate the wave-activity propagation in the observed field and its relative importance in the formation of upper-level planetary waves between the near-surface diabatic heating and the mid- and upper-tropospheric diabatic heating.
In our model simulation with only local near-surface diabatic heating (Fig. 5d), the strength of the surface summertime subtropical anticyclone over the South Pacific is somewhat underestimated. One of the reasons for this underestimation is a possible influence of midlatitude planetary waves characterized by the zonal wavenumber-1 component (Fig. 1d). In association with this structure, there is some enhancement of the upper-level polar-front jet (Fig. 1c) to the south of that surface subtropical anticyclone. In the jet-entrance region, transient eddy activity is relatively weak (Fig. 9a). A planetary wave model simulation with the zonally asymmetric components of the heat and vorticity fluxes associated with transient eddies prescribed as forcing (Fig. 9b) suggests that the eddy forcing acts to suppress the poleward extension of the surface subtropical anticyclone over the South Pacific. In contrast, the eddy forcing contributes positively to the poleward extension of the surface subtropical anticyclones in the South Atlantic and South Indian Oceans (Fig. 9b), where the core region of the SH storm track is located along a prominent oceanic frontal zone (Nakamura and Shimpo 2004). Especially over the South Indian Ocean, upper-level vorticity flux convergence associated with transient eddies directly forces the upper-level meridional vorticity dipole (not shown). The role of the transient eddy feedback for the seasonality of the subtropical anticyclones will be examined in a future study. Also to be examined in a future study is what mechanisms are involved in the interannual variability of the surface summertime subtropical anticyclones, including the one over the South Pacific (Grotjahn 2004).
Acknowledgments
We are grateful to Prof. B. J. Hoskins for kindly providing the code of the Reading spectral model and giving encouragement and advice. We also thank Dr. T. Enomoto for his technical assistance with the model and his valuable comments. We thank Prof. R. Grotjahn for his useful comments on an earlier version of the paper. We are grateful to Dr. K. Takaya and Profs. S.-P. Xie, I. Hirota, Y. Takayabu, T. Yamagata, M. Kimoto, H. Niino, and M. Takahashi for their valuable comments and encouragement. Sound criticism by the two anonymous referees and the editor (Dr. M. Alexander) has led to the substantial improvement of the paper. This work is supported, in part, by a grant-in-aid for scientific research (18204044) of the Japan Society for the Promotion of Science and by the Global Environment Research Fund (S-5) of the Ministry of the Environment, Japan.
REFERENCES
Chen, P., M. P. Hoerling, and R. M. Dole, 2001: The origin of the subtropical anticyclones. J. Atmos. Sci., 58 , 1827–1835.
Grotjahn, R., 2004: Remote weather associated with South Pacific subtropical sea-level high properties. Int. J. Climatol., 24 , 823–839.
Grotjahn, R., and M. Osman, 2007: Remote weather associated with North Pacific subtropical sea-level high properties. Int. J. Climatol., 27 , 587–602.
Hartmann, D. L., M. E. Ockert-Bell, and M. L. Michelsen, 1992: The effect of cloud type on Earth’s energy balance: Global analysis. J. Climate, 5 , 1281–1304.
Hoskins, B. J., 1996: On the existence and strength of the summer subtropical anticyclones. Bull. Amer. Meteor. Soc., 77 , 1287–1292.
Hoskins, B. J., and A. J. Simmons, 1975: A multi-layer spectral model and the semi-implicit method. Quart. J. Roy. Meteor. Soc., 101 , 637–655.
Hoskins, B. J., and M. J. Rodwell, 1995: A model of the Asian summer monsoon. Part I: The global scale. J. Atmos. Sci., 52 , 1329–1340.
Hoskins, B. J., M. E. McIntyre, and A. W. Robertson, 1985: On the use and significance of isentropic potential vorticity maps. Quart. J. Roy. Meteor. Soc., 111 , 877–946.
Kanamitsu, M., W. Ebisuzaki, J. Woollen, S-K. Yang, J. J. Hnilo, M. Fiorino, and G. L. Potter, 2002: NCEP–DOE AMIP-II Reanalysis (R-2). Bull. Amer. Meteor. Soc., 83 , 1631–1643.
Klein, S. A., and D. L. Hartmann, 1993: The seasonal cycle of low stratiform clouds. J. Climate, 6 , 1587–1606.
Lindzen, R. S., and A. Y. Hou, 1988: Hadley circulations for zonally averaged heating centered off the equator. J. Atmos. Sci., 45 , 2416–2427.
Liu, Y. M., G. X. Wu, and R. Ren, 2004: Relationship between the subtropical anticyclone and diabatic heating. J. Climate, 17 , 682–698.
Ma, C-C., C. R. Mechoso, A. Arakawa, and J. D. Farrara, 1994: Sensitivity of a coupled ocean–atmosphere model to physical parameterizations. J. Climate, 7 , 1883–1896.
Miyasaka, T., and H. Nakamura, 2005: Structure and formation mechanisms of the Northern Hemisphere summertime subtropical highs. J. Climate, 18 , 5046–5065.
Nakamura, H., and J. M. Wallace, 1990: Observed changes in baroclinic wave activity during the life cycles of low-frequency circulation anomalies. J. Atmos. Sci., 47 , 1100–1116.
Nakamura, H., and T. Miyasaka, 2004: Formation of summertime subtropical highs. Bull. Amer. Meteor. Soc., 85 , 1062–1064.
Nakamura, H., and A. Shimpo, 2004: Seasonal variations the Southern Hemisphere storm tracks and jet streams as revealed in a reanalysis dataset. J. Climate, 17 , 1828–1844.
Onogi, K., and Coauthors, 2007: The JRA-25 Reanalysis. J. Meteor. Soc. Japan, 85 , 369–432.
Plumb, R. A., 1985: On the three-dimensional propagation of stationary waves. J. Atmos. Sci., 42 , 217–229.
Rodwell, M. J., and B. J. Hoskins, 1996: Monsoons and the dynamics of deserts. Quart. J. Roy. Meteor. Soc., 122 , 1385–1404.
Rodwell, M. J., and B. J. Hoskins, 2001: Subtropical anticyclones and summer monsoons. J. Climate, 14 , 3192–3211.
Seager, R., R. Murtugudde, N. Naik, A. Clement, N. Gordon, and J. Miller, 2003: Air–sea interaction and the seasonal cycle of the subtropical anticyclones. J. Climate, 16 , 1948–1966.
Takaya, K., and H. Nakamura, 2001: A formulation of a phase-independent wave-activity flux for stationary and migratory quasigeostrophic eddies on a zonally varying basic flow. J. Atmos. Sci., 58 , 608–627.
Uppala, S. M., and Coauthors, 2005: The ERA-40 Re-Analysis. Quart. J. Roy. Meteor. Soc., 131 , 2961–3012.
Wallace, J. M., 1983: The climatological mean stationary waves: Observational evidence. Large-Scale Dynamical Processes in the Atmosphere, B. J. Hoskins and P. R. Pearce, Eds., Academic Press, 27–53.
Wang, Y., S-P. Xie, B. Wang, and H. Xu, 2005: Large-scale atmospheric forcing by southeast Pacific boundary-layer clouds: A regional model study. J. Climate, 18 , 934–951.
White, G. H., 1982: An observational study of the Northern Hemisphere extratropics summertime general circulation. J. Atmos. Sci., 39 , 24–40.
Wu, G. X., and Y. M. Liu, 2003: Summertime quadruplet heating pattern in the subtropics and the associated atmospheric circulation. Geophys. Res. Lett., 30 , 1201. doi:10.1029/2002GL016209.
Climatological-mean January distribution of (a) SLP [contour interval (CI) 4 hPa] and 1000-hPa horizontal wind velocity (arrows), (b) 500-hPa p velocity (CI 0.02 Pa s−1; solid lines for subsiding and dashed lines for rising), (c) 250-hPa zonal wind speed (CI 5 m s−1), (d) zonally asymmetric component of 250-hPa relative vorticity (unit, 10−5 s−1; ±0.2, 0.6, and 1.0; solid and dashed lines for anticyclonic and cyclonic vorticity, respectively), and (e) horizontal divergence at the 250-hPa surface (CI 5 × 10−5 s−1) and 250-hPa ageostrophic wind component (arrows) based on the NCEP–DOE reanalyses. In (a) and (e), the scaling for the arrows is given above the panels. In (a)–(e), shading is applied where SLP exceeds 1016 hPa. In (c) and (d), local 250-hPa westerly jet axes are indicated with dots. In (b) and (e), zero lines are omitted.
Citation: Journal of Climate 23, 8; 10.1175/2009JCLI3008.1
Climatological-mean planetary waves in January, based on zonally asymmetric fields of the NCEP–DOE reanalyses. (a) Wave-associated 500-hPa p velocity diagnosed with the omega equation [(2)] (CI ±0.005, 0.015, and 0.025 Pa s−1; solid and dashed lines for subsiding and rising, respectively). (b) Horizontal component of Plumb’s (1985) wave-activity flux (arrows) at the 250-hPa level, superimposed onto the 250-hPa relative vorticity associated with planetary waves (contoured as in Fig. 1d). Scaling (m2 s−2) for arrows is given below this panel. Arrows are plotted where the zonal mean westerly is stronger than 2 m s−1 and the horizontal flux is stronger than 1 m2 s−2. (c) Vertical component of the wave-activity flux (m2 s−2) at the 500-hPa level. Solid lines signify the upward flux. (d) The 1000-hPa horizontal wind velocity (arrows; unit, m s−1; with scaling given below this panel; plotted where wind speed >1 m s−1) and temperature (CI 2 K; dashed for negative values), both associated with planetary waves. In (a)–(d), shading is applied where the total SLP exceeds 1016 hPa, as in Fig. 1, and zero lines are omitted. In (b), the solid arrow along the right axis indicates the latitudinal extent of the zonal-mean westerly in the SH for January, while the dashed arrow indicates its NH counterpart for July.
Citation: Journal of Climate 23, 8; 10.1175/2009JCLI3008.1
Zonally asymmetric component of climatological-mean diabatic heating (Q*) for January obtained from the NCEP–DOE reanalyses. (a) Column-integrated heating (CI 50 W m−2). (b) Longitude–vertical section of Q* as an average between 40° and 10°S (CI 1 K day−1). (c) Vertically integrated heating below the level σ = 0.667 (CI 30 W m−2). Solid and dashed arrows along the right axis indicate latitudinal extents of the zonal-mean westerlies at the 1000- and 250-hPa levels, respectively. In (a)–(c), dashed lines denote cooling and zero lines are omitted. In (b) and (c), rectangles signify forcing domains for model simulations, as discussed in section 4b. In (a), long-dashed rectangles signify forcing domains for model simulations for deep convective heating, as discussed in section 5.
Citation: Journal of Climate 23, 8; 10.1175/2009JCLI3008.1
(a) Quasigeostrophic PV anomalies [ f0θ*/(∂Θ/∂z); CI 0.01 m s−1; solid and dashed lines for anticyclonic and cyclonic PV, respectively] associated with the zonally asymmetric component of the climatological-mean temperature for January at the lower boundary that is assumed to be placed at the 1000-hPa level. The associated 1000-hPa geostrophic wind diagnosed through the PV inversion is indicated with arrows whose scaling (m s−1) is given below the panel. (b) The 500-hPa vertical p velocity (CI 0.003 Pa s−1) obtained through substituting the wind field obtained through the PV inversion in (a) into the linearized omega equation [(2)]. Solid and dashed lines denote subsiding and rising, respectively. In (a) and (b), zero lines are omitted, and shading is applied where the observed total SLP exceeds 1016 hPa. The 1000-hPa temperature of the zonal-mean state for (2) has been taken from the NCEP–DOE reanalysis data.
Citation: Journal of Climate 23, 8; 10.1175/2009JCLI3008.1
(a) Zonally asymmetric component of the climatological-mean January SLP based on the NCEP–DOE reanalysis data (CI 1 hPa; dashed lines for negative values). (b) The zonally asymmetric SLP response in the atmospheric model with global topography forced with the zonally asymmetric component of the climatological-mean three-dimensional diabatic heating Q* over the globe for January, whose horizontal distribution is shown in Fig. 3a. The response was averaged for 10 days. The heating was taken from the NCEP–DOE reanalyses. CI 1 hPa. Dashed lines used for negative values. Zero contours are omitted. (c) As in (b) but for the model SLP response for the diabatic heating prescribed only in a latitudinal band of 40°∼10°S (as indicated with boldface lines). (d) As in (b) but for the model SLP response for the diabatic heating prescribed only below the level of σ = 0.667 in rectangular domains over a South Pacific domain (40°∼10°S, 120°∼60°W) as indicated. Flat topography is included. (e) As in (d) but for the heating in a South Atlantic domain (40°∼10°S, 30°W∼40°E) only. (f) As in (d) but for the heating in an Indian Ocean domain (40°∼10°S, 80°∼160°E) only. Note that the threshold for the shading is 3 hPa for the South Pacific and Atlantic Oceans, and 2 hPa for the Indian Ocean.
Citation: Journal of Climate 23, 8; 10.1175/2009JCLI3008.1
Model simulation of the South Pacific subtropical anticyclone with the lower-tropospheric diabatic heating prescribed only within a rectangular domain (40°∼10°S, 120°∼60°W) as indicated in (a). Response was averaged for 10 days. The zonally asymmetric climatological-mean heating Q* for January, as shown in Fig. 3, was taken from the NCEP–DOE reanalyses and assigned only in the lower troposphere (layers for 0.667 ≤ σ ≤ 1). No surface elevation is applied to the model. (a) Column-integrated diabatic heating (CI 30 W m−2; dashed for cooling). (b) Potential temperature response (CI 2 K; dashed for negative values) at the lowest model level (σ = 0.983). (c) The zonally asymmetric SLP response (CI 1 hPa; dashed for negative values). (d) Model response in 500-hPa vertical p velocity (CI ±0.005, 0.015, 0.025, … Pa s−1; solid and dashed lines for subsiding and uprising, respectively). (e) The 250-hPa relative vorticity response (10−5 s−1; contoured for ±0.2, 0.6, 1.0, …; solid and dashed lines are for anticyclonic and cyclonic vorticity, respectively). The observed climatological-mean westerly jet axes are superimposed with dots. (f) Plumb’s (1985) wave-activity flux evaluated for the model response. Its horizontal component at the 250-hPa level is plotted with arrows whose scaling (m2 s−2) is given below the panel. Arrows are plotted where the zonal-mean westerly is stronger than 2 m s−1 and the horizontal flux is stronger than 0.5 m2 s−2. The vertical component of the flux at the 500-hPa level is superimposed with solid lines (CI 0.01 m2 s−2; solid for upward). In (a)–(f), shading is applied where the SLP response exceeds 3 hPa.
Citation: Journal of Climate 23, 8; 10.1175/2009JCLI3008.1
As in Fig. 6 but for a model simulation of the South Atlantic subtropical anticyclone with the lower-tropospheric diabatic heating prescribed only within a rectangular domain (40°∼10°S, 30°W∼40°E) as indicated in (a).
Citation: Journal of Climate 23, 8; 10.1175/2009JCLI3008.1
Same as Fig. 6, but for a model simulation of the Indian Ocean subtropical anticyclone with the lower-tropospheric diabatic heating prescribed only within a rectangular domain (40°∼10°S, 80°∼160°E) as indicated in (a). Shading is applied where the SLP response exceeds 2 hPa.
Citation: Journal of Climate 23, 8; 10.1175/2009JCLI3008.1
(a) Climatological-mean January distribution of an “envelope function” of the 250-hPa geopotential height (CI 30 m), defined as the square root of the squared time series of the 8-day high-pass-filtered height fluctuations, as in Nakamura and Wallace (1990). (b) Zonally asymmetric SLP response (CI 1 hPa) to heat and vorticity flux convergence associated with transients defined as 8-day high-pass-filtered fluctuations. In (b) shading is applied where the zonally asymmetric component of the SLP based on the reanalysis exceeds 3 (2) hPa for the South Pacific and Atlantic (Indian) Oceans. Zero lines are omitted.
Citation: Journal of Climate 23, 8; 10.1175/2009JCLI3008.1
