Large-Scale Recirculation of Air over Southern Africa

View More View Less
  • 1 Climatology Research Group, University of the Witwatersand, South Africa
  • | 2 Department of Environmental Sciences, University of Virginia, Charlottesville, Virginia
© Get Permissions
Full access

Abstract

Kinematic air parcel trajectory analysis is used to determine patterns of horizontal air transport in 2000 km × 2000 km areas over southern Africa. From these, composite zonal and meridional transport fields are derived for the subcontinent to estimate the extent to which recirculation of air and aerosols may take place in the lower troposphere between the surface and 500 hPa. The nature and degree of recirculation beneath the persistent 500-hPa absolutely stable layer is demonstrated, and transport by recirculation in discrete streams is shown to constitute 44% of the total transport over the region.

From a determination of air volume fluxes and estimates of aerosol concentrations, the total mass flux of aerosols by direct transport and by recirculation in conditions during which semipermanent, subtropical, continental anticyclones prevail is estimated to be about 51 Mton yr−1 in the surface-to-hPa layer. Recirculation comprises approximately 22 Mton yr−1 of this amount. Of the recirculated transport, about 5 Mton yr−1 is recirculated to the west in easterly transport and 17 Mton yr−1 to the east in westerly transport.

Abstract

Kinematic air parcel trajectory analysis is used to determine patterns of horizontal air transport in 2000 km × 2000 km areas over southern Africa. From these, composite zonal and meridional transport fields are derived for the subcontinent to estimate the extent to which recirculation of air and aerosols may take place in the lower troposphere between the surface and 500 hPa. The nature and degree of recirculation beneath the persistent 500-hPa absolutely stable layer is demonstrated, and transport by recirculation in discrete streams is shown to constitute 44% of the total transport over the region.

From a determination of air volume fluxes and estimates of aerosol concentrations, the total mass flux of aerosols by direct transport and by recirculation in conditions during which semipermanent, subtropical, continental anticyclones prevail is estimated to be about 51 Mton yr−1 in the surface-to-hPa layer. Recirculation comprises approximately 22 Mton yr−1 of this amount. Of the recirculated transport, about 5 Mton yr−1 is recirculated to the west in easterly transport and 17 Mton yr−1 to the east in westerly transport.

2218JOURNAL OF APPLIED METEOROLOGYVo~.tnvm 35Large-Scale Recirculation of Air over Southern Africa P. D. TYSONClimatology Research Group, University of the Witwatersrand, South Africa M. GARSTANG AND R. SWAP*Department of Environmental Sciences, University of Virginia, Charlottesville, Virginia(Manuscript received 16 February 1996, in final form 28 May 1996) ABSTRACT Kinematic air parcel trajectory analysis is used to determine patterns of horizontal air transport in 2000 kmX 2000 lan areas over southern Africa. From these, composite zonal and mefidional transport fields are derivedfor the subcontinent to estimate the extent to which recirculation of air and aerosols may take place in the lowertroposphere between the surface and 500 hPa. The nature and degree of recirculation beneath the persistent 500hPa absolutely stable layer is demonstrated, and transport by re, circulation in discrete streams is shown to constitute 44% of the total transport over the region. From a determination of air volume fluxes and estimates of aerosol concentrations, the total mass flux ofaerosols by direct transport and by recirculation in conditions during which semipermanent, subtropical, continental anticyclones prevail is estimated to be about 51 Mton yr-t in the surface-to-500-hPa layer. Recirculationcomprises approximately 22 Mton yr-~ of this amount. Of the re, circulated transport, about 5 Mton yr-~ isrecirculated to the west in easterly transport and 17 Mton yr-~ to the east in westerly transport.1. Introduction Transport of aerosols and trace gases by the largescale circulation fields of the atmosphere has receivedincreasing attention with the growing need to identifysources and sinks of natural products and those generated by human activity. In midlafitudes such transportis usually manifested by elongated eastward-streamingplumes of material under the influence of the circumpolar westerly winds (Fishman 1991; Picketing et al.1994; Krishnamurti et al. 1993; Benkovitz et al. 1994;Garstang et al. 1996a,b,c; Tyson et al. 1996a). Moodyet al. (1996) have pointed out frequent, if not pervasive, synoptic-scale influences, which may couple thetroposphere and stratosphere within the. general westerly stream at latitudes poleward of the Tropics. Bycontrast, in subtropical latitudes, which are dominatedby the descending limbs of the Hadley cells of the general circulation and by the occurrence of large-surface,semipermanent, high pressure systems, the situation issignificantly different. On the equatorward and pole * Current affiliation: Climatology Research Group, University ofthe Witwatersrand, South Africa. Corresponding author address: Prof. P. D. Tyson, ClimatologyResearch Group, University of the Witwatersrand, PO Wits 2050,South Africa.E-mail: pdt @crg.bpb.wits.ae.zaward extremities of these systems, elongated zonalplumes do occur (Garstang and Tyson 1996). Withinthe large high pressure systems, however, re, circulationmay take place to a significant extent. In the case of southern Africa, re, circulation occurson subcontinental and regional scales (Tyson et al.1996a,b). The stable anticyclonic environment inhibitsvertical exchanges in the atmospheric column, stratifying the troposphere into persistent layers in whichresidence times are prolonged over days to tens of days.Under these conditions and over this period of time, airin a given volume may return to its original point oforigin or complete more than one cycle of recirculafion.For long periods of the year, atmospheric states undersuch anticyclonic conditions are typically free, ornearly free, of clouds, maximizing daytime insolation.Likewise, clear nights result in maximum nocturnalcooling at the surface and the consequent generation oflocal and mesoscale thermotopographic wind systems(Tyson 1967; Tyson and Preston-Whyte 1972; Tysonet al. 1988; Preston-Whyte and Tyson 1989; Annegarnet al. 1993; Held et al. 1994; Piketh 1995). Current anticyclonic mean surface circulation fieldsover southern Africa coincide with the orientation offossil dune fields tens of thousands of years old inSouth Africa, Namibia, Botswana, and southern partsof countries to the north (Lancaster 1979, 1981, 1988 ).The presence of these dunes suggests that anticyclonicrecirculation and aeolian transport have operated oversouthern Africa for a very long time (Garstang et al.c 1996 American Meteorological SocietyDECEMBER 1996 TYSON ET AL. 22191996b). Since the circulation system producing thistransport occurs on a subcontinental scale, it is probablethat a link to ecosystems exists. Not only is it importantto establish the validity of any such linkage, but it isessential to establish whether human intervention hasdisrupted a balance that may depend on the transportand deposition of critical trace elements within the system. First, however, it is necessary to examine the atmospheric recirculation process in detail. In this paper the physical processes and evidence forlarge-scale recirculation and retransport of air oversouthern Africa will be examined in detail, using kinematic trajectory analysis and an ensemble method ofanalyzing forward trajectories to produce air, aerosol,and trace gas transport fields. The extent and magnitudeof the recirculation will be established to provide afoundation for subsequent evaluation of the effects ofatmospheric transport of aerosols and trace gases onthe ecosystems of the subcontinent.2. Regional circulation Southern Africa is situated in the region of the general circulation of the atmosphere of the SouthernHemisphere that is dominated for much of the year bysubsidence in the descending limb of the Hadley cell(Newell et al. 1972). Semipermanent subtropical anticyclones consequently dominate the lower tropospheric circulation of the region (Garstang et al. 1996c;Tyson et al. 1996a) (Table 1 ). Other important circulation types include barotropic quasi-stationary easterlywaves, transient ridging anticyclones originating in themidlatitude westerlies, and baroclinic disturbances inthe westerlies. Easterly waves are a summer phenomenon, and ridging anticyclones occur regularlythroughout the year. Westerly disturbances exhibit asemiannual cycle, peaking in spring and autumn. Continental anticyclones are the single most frequently occurring systems in any month from March to September. From April to August they occur more frequentlythan all the other systems combined. An important consequence of the subcontinent's location in the region of the subtropical semipermanenthigh pressure cells of the Southern Hemisphere is thatthe thermodynamic structure of the atmosphere overthe region is characterized by a high degree of stability,particularly in winter (Taljaard 1955; Tyson et al.1976; Preston-Whyte et al. 1977; Tyson et al. 1988;Harrison 1993; Cosijn and Tyson 1996). A way ofcharacterizing the thermodynamic stability structure ofthe atmosphere that is less restrictive than specifyinginversions of temperature is to determine absolutelystable layers in which the observed environmental lapserate is less than the saturated adiabatic lapse rate. Theselayers prevent vertical diffusion and transport of aerosols and trace gases and effectively confine horizontaltransport to between the layers. Over the year as awhole, four such layers are observed over southern Africa, more frequently and persistently in the winter halfof the year than in summer. Over the interior plateau,three layers are observed (Fig. 1 ). The first occurs ator near the top of the midday mixing layer at about 700hPa and is broken every 6-7 days by passing westerlydisturbances (Preston-Whyte and Tyson 1973). Thesecond occurs at 500 hPa, is produced and maintainedby large-scale subsidence, and is remarkably persistent.For example, during the Southern African Fire-Atmosphere Research Initiative (SAFARI) the' layer waspresent as it oscillated in position from about 600 to450 hPa from 25 September to 3 November 1992that is, without interruption for 40 days. From 11 September 1992 onward it was observed for 53 days withthe exception of 1 day. The third layer occurs at aboutthe 200-hPa level and has consequences for the uppertropospheric horizontal and vertical transport of ozonerather than aerosols (Tyson et al. 1996b). Over thenarrow coastal margins, between the plateau and theoceans, a fourth layer is observed at approximately 800hPa. On days when the stable layers are observed,dense haze layers are also present. Over the plateau,the top of the first occurs at 700 hPa and of the secondat 500 hPa. The major discontinuity between the hazy,polluted lower-tropospheric air and clear air above isobservable almost always at around the 500-hPa level(specific examples are to be found in Figs. 2 and 3).On an annual basis the 500-hPa layer occurs with a78% frequency, having a mean base height of 540 hPa,with a 95% confidence limit of 65 hPa and a meandepth of 64 hPa (Cosijn and Tyson 1996). No studyof tropospheric aerosol or trace gas transport or recirculation over southern Africa can ignore the dominanteffect exerted for most of the year by the occurrenceof stable layers in the troposphere, particularly in themiddle troposphere.3. Data and methods South African Weather Bureau 0000 and 1200 UTCsynoptic charts at 850 and 500 hPa over the 7-yr pe TABLE 1. Monthly percentage frequencies of circulation types over subtropical southern Africa, 1986-92 (Tyson et al. 1996). January February March April May June July August September October November DecemberContinental highs 15 21 42 55 62 56 70 48 41 21 25 18Ridging highs 7 10 14 12 10 14 10 15 18 20 17 14Westerly waves 21 18 25 22 24 27 22 33 38 4l 29 27Easterly waves 55 45 15 9 4 3 I I 4 16 26 362220 JOURNAL OF APPLIED METEOROLOGY VOLUME35a)b)4~~0~07O0~0~010~PI PR BE BL UP SP CT PE DB4OO I I ~~~ ' I ~H ~H ~ ~ 50O~ 6ooc) 300 86% ~ 85% 400 500 1~78% 82% ~ 76% 1~0 ~-I YEAR MIu- SUMMER WINTERd)CH = Continental highRH = Ridging highEW = Easterly waveWW = Westerly wave (pre.frontal) PA. /~ UP. .PE / FIG. 1. The occurrence of absolutely stable layers over South Africa by circulation type and time of year.Absolutely stable layers are indicated by block shading, showing base heights (with 95% confidence limits)and depths (horizontal dimension is arbitrary) (a) for spatial distribution across South Africa, (b) by circulation type, and (c) by time of year. (d) Locations of stations. The results are based on the analysis of a totalof 2925 radiosonde ascents taken over the period 1986-92 (modified after Cosijn and Tyson 1996).riod 1986-92 are used to identify semipermanent highpressure circulation types over southern Africa. Sincemost of southern Africa consists of an elevated plateauvarying in height from about 1000 to over 1800 m, ithas been South African Weather Bureau practice formany years to define the surface flow field over theplateau as that at the 850-hPa level (Taljaard 1953;Schulze 1965). The European Centre for MediumRange Weather Forecasts (ECMWF) operationalanalyses of the 6-h, three-dimensional wind field atthe 1000-, 900-, 850-, 800-, 700-, and 500-hPa geopotential height levels provide the basis for derivingair transport fields from forward trajectories of air pareels originating from sites on the interior plateau at20-S, 25-E; 20-S, 25-E; 25-S, 30-E; and 30-S, 25-E,using Kfillberg's (1984) method of determining traDECEMBER 1996 TYS ON ET AL. 2221 6 October 1992 10 L640~3 -2 4 10- 20* 30* 40- 50*40- relative aerosor scaltedng (XIO00, [~ abs01utely stable layers at 12:00 UT. ~ >50 ~ 30-50 ~ 20-30 ~ 10-20~] 4-1011 October 1992 3O 20*40- ~ absolutel,/ stsble layers at 12:00 UT. [~ >50 --~ 30-50 ;~ 20-30 -[~ 10.20'r'~'~ 4-10 F[G. 2. 'Vertical sections of relative aerosol concen~ation determined from Browel]'s 0993) NASA UVD[^L data obtained duhng TR^CE-A flights along transects from Johannesburg to northern Zambia on 6October ]992 and from Johannesburg to 'Victoria Fails, Etosha National Park, and Windhoek~ South Afhca,on ! ] October ! 992. Absolutely stable layers arc vertically shaded. The relationship of aerosol concentrationsto absolutely stable layers and 700- and 850-hPa back and forward trajectohes are shown. Days of traveland geopotential heights are given for selected points on the trajectories. Cape Town is denoted by CT,PE--Port Elizabeth, S--Springbok, B--Bloemfontein/Beira, D--Durban/Dar-Es-Salaam, J--Johannesburg, P--Pretoria, M--Maputo, W--Windhoek, E--Etosha National Park, VF--Victoria Falls, KNP-Kruger National Park, L--Luanda/Lusaka, and H--Harare.jectories and the ensemble method of preparing transport fields outlined in Tyson et al. (1996a) and Garstang et al. (1996c). The method consists of followingindividual forward trajectories emanating from particular heights from specific points of origin through vertical planes erected on lines of longitude (interceptingzonal flow) and latitude (intercepting meridionalflow). In this study the vertical planes were erected at2222 JOURNAL OF APPLIED METEOROLOGY VOLUME35Low-and middle-level airflow 3 October 1992 850 hPa trajectory 800 hPa trajectory ,aK ~o / 10' 2o' 30' 40' 30' 40' 700 hPa trajectory 500 hPa trajectory ~ ~l~ ~ ~ at 12:~ ~. ~ >ffk ~ ~ ~ ~ ~ 10-~ ~ 4-10F~. 3. Back and fo~ard trajectories passing ~rough the Pretofia-Witwatersrand-Vereeniging aerosol layer at 850, 800, 700, and 500 hPa over Johannesburg on 3 October ~992. Symbols and layers are as defined in Fig. 1.2- intervals. The height, latitude, and longitude of thepenetration points are then located on the verticalplanes along successive meridians and parallels. Botheastward and westward penetrations are recorded.Penetrations of the meridional and latitudinal 'walls arerecorded as direct when they arrive directly from thesource and as recirculated when they return indirectlyin the opposite direction from the initial motion fromthe source. By recording direct and indirect strikes onsuccessive walls spaced at 2- intervals of latitude andlongitude, a clear depiction of the regional direct andrecirculated flow is obtained.After a large number of trajectories have passedthrough the plane erected on a meridian or parallel, itis possible to contour the field of strikes on that planeto give ensemble cross sections of percentages of penetration within trajectory dispersion fields (plumes).From the results on individual meridional and latitudinal planes, the transport fields, either in the horizontalor vertical, may be determined to enclose a percentagefrequency of occurrence of trajectories within theplume and within the spatial domain of interest. Botheastward and westward, as well as northward andsouthward, intersections on meridional and latitudinalplanes by individual trajectories are recorded to allowdirect (single crossing) and. recirculating (multiplecrossing) transport to be determined. In this case, directand recirculated zonal and meridional transports arecalculated through meridional and latitudinal planeswithin overlapping 2000 km x 2000 km squares (seeDECEMBER 1996 TYSON ET AL. 2223Fig. 4). For each of the four points of origin, 189 forward trajectories have been computed. From these,easteriy and westerly components of zonal transportand northerly and southerly components of meridionaltransport were defined in plumes bounded by 95% or98% frequency of occurrence contours. The uncertainties inherent in the ECMWF modeled velocity fields,which are contained in the calculated trajectories, arein part reflected by the dispersion of the penetrationpoints across the entire vertical plane. By accepting adistribution of points enclosed within the 95% or 98%frequency of occurrence contours, a measure of thatuncertainty is provided. In the analysis as a whole, over750 individual trajectory pathways were determined.From each station, maximum frequency pathways, aswell as the levels of maximum frequency transport andmean times taken for air parcels to reach specific meridians and latitudes, were computed. Given the stability layers that are observed over southern Africa, vertical integration of transport has been effected in thelayers of surface (850 hPa) to 800 hPa, 700 to 500 hPa,and surface to 500 hPa. Since the analysis in this paper is based on forwardtrajectories emanating from points over southern Africa, not back trajectories, to determine the origin ofrecirculating air, and in order to limit the scope of thestudy to manageable proportions, only trajectories originating over southern Africa are considered. The volume flux (m3 s-I) of air transported through a givenmeridional or latitudinal wall is the product of the area(m2) bounded by a given frequency contour (e.g.,95%) and the mean speed (m s-l) of transit throughthat wall. The mean speed of transit is obtained fromthe mean distance traveled by air parcels embedded inthe trajectories making up the central core transportbetween successive trajectory calculations. Mass transports per unit time of aerosols may then be estimatedfrom the product of the volume flux of air transportedand background aerosol concentrations. Vertical distributions of aerosols are taken from Browell's (1993)Ultra-Violet Differential Absorption Lidar (UVDIAL) measurements made from the National Aeronautics and Space Administration (NASA) DC-8aircraft during the SAFARI field experiment period ofAugust-October 1992 and from a variety of aerosolconcentration measurements cited later.4. Recirculation and individual trajectory pathways Simple anticyclonic recirculation within a high pressure system is evident from the 700-hPa back trajectoryto, and the forward trajectory from, a point on the aerosol profile, determined from NASA's UV-DIAL data,along a transect between Johannesburg, South Africa,and northern Zambia on 6 October 1992 (Fig. 2, upperpanel). In this instance, the air recirculated over southern Africa for 8 days. Recirculation was likewise evident on 11 October 1992, when air over Mozambique 1o0o1011I I10- 20-, / I 30- FIG. 4. The 2000 km x 2000 km areas bounding the central pointsof origin at 25-S, 30-E; 30-S, 25-E; and 25-S, 20-E from which 950forward trajectories were computed to determine recirculation fieldsbetween the surface and 500 hPa. In addition, 384 trajectories wereused to determine transport fields in a similar area around Skukuza(-25-S, 32-E) in the Kruger National Park (KNP). The boundingarea for the KNP transport field is not shown. Places are as denotedin Fig. 2.took 14 days to recirculate back to where it had originated (Fig. 2, lower panel). More prolonged recirculation of aerosol-laden andpolluted urban-industrial air from the Pretoria-Witwatersrand-Vereeniging area of South Africa occurredon 3 October 1992. The atmosphere on this occasionshowed a double decoupling between the lower boundary layer, the air in the middle to lower troposphere ataround 700 hPa, and that in the middle troposphere ataround 500 hPa and above. The near-surface 850- and800-hPa trajectories are similar (Fig. 3, top left, right).The passage time from entering the air space over thecontinent from the Atlantic Ocean to exiting to the Indian Ocean is about 13 days; the time the polluted airtook to move from the Johannesburg area to over theocean is 7 days. At the 700-hPa level the recirculationpattern is quite different from that of air recirculatingbelow at 800 hPa and from that circulating above at500 hPa (Fig. 3, lower left). At the 700-hPa level airrecirculated for 20 days over the continent without exiting to the oceans at all. Within the major recirculationpathway, minor closed recirculations occurred. Polluted air over Johannesburg on 3 October was transported to the northeast over southern Zimbabwe initially, before moving west to Namibia, then south backdown the west coast of South Africa, before finallytracking back to the Witwatersrand region from thewest to reach Johannesburg again 10 days later. Bycontrast, in the middle troposphere, at the 500-hPalevel, the recirculation was uncomplicated (Fig. 3,lower right). Nonetheless, the air still had a residencetime of 9 days over southern Africa.2224 JOURNAL OF APPLIED METEOROLOGY VOLUME355. Composite SAFARI trajectory fields In order to move beyond case studies, overlapping2000 km x 2000 km boxes covering southern Africaare considered (Fig. 4). Analyzing 384 forward trajectories from Kruger National Park (-25-S, 32-E) during the SAFARI field observation period from Augustto October 1992, it is clear that for vertically integratedsurface-to-800-hPa transport, the major direct zonaltransport stream is to the southeast, with subsidiarystreams to the west (Fig. 5). Recirculation from botheast and west occurs 400-700 km from the point oforigin of the air parcels. The meridional components oftransport, though much smaller, clearly show the anticyclonic nature of the overall air transport pattern fromthe particular locality. The maximum frequency pathway for air recirculating back from the west occursabout 400 km to the south of the point of origin at amean geopotential height of 750 hPa. The recirculationperiod for the air to reach this locality is 3.1 days. Themaximum frequency pathway for air recirculatin.g backfrom the east occurs about 700 km to the north of theorigin at 850 hPa, with a recirculation time of 3.4 days.The overall recirculation index (defined as the sum ofthe percentage of recirculating trajectories from the eastand west at the approximate longitude of the point oforigin, together with the percentage number of cases ofstagnation, i.e., calm conditions) for this locality is75%. Given that it is taking about 3.1 days for air initially being transported to the west to be recirculatedback in a westerly direction to the same longitude andthat it is taking 3.4 days for air initially being transported to the east to be recirculated back to the longitude of origin in an easterly direction, the mean overallrecirculation time will be at least 6.5 days, a figure thatcompares with the recirculation times quoted earlier forthe case studies. The vertically integrated 700-500-hPa transportfields show similar patterns (Fig. 6). In this case therecirculation index is 67%, the radius of the recirculation vortex is 400-500 km, the mean height of recirculation at the longitude of the origin is 750-6.50 hPa,and the mean overall recirculation time is 7.2 days. Forboth the lower and upper layer it would appear thatrecirculation of air does not take place uniformly withina vortex, but in discrete streams far removed t-om theorigin.6. Recirculation in continental anticyclones Analysis of the dominant synoptic-scale systems occurring over southern Africa over a 7-yr period, 198692 (Tyson et al. 1996a) shows that anticyclonic conditions (continental highs and ridging highs) occur70%-80% of the time in the midwinter months of Juneand July, declining to 22%-32% of the tirne in themidsummer months of December and January (Table1). The above analysis may thus be extended beyonda single observation period, such as that for SAFARI,within which different synoptic types occurred at different times, to focus only on anticyclonic circulations,which are the dominant type for most of the year andwhich are responsible for the most pronounced modesof recirculation. Conditions that are typical for continental anticyclonic circulations can then be extrapolated from periods of days to annual estimates usingthe frequency distributions given in Table 1. Four points of origin for air transport and recirculation under anticyclonic conditions over southern Africaare examined. The first is at 25-S, 30-E and is situatedat an altitude of approximately 1500 m on the inlandhighveld plateau of South Africa (approximately in thevicinity of Belfast-Middelburg) to the west of the Kruger National Park. It is representative of an area of coalmining, major power generation, and industrializationthat produces much air pollution (Tyson et al. 1988).An analysis of 189 forward trajectory runs reveals thatthe direct zonal component of vertically integrated surface-to-800-hPa transport from the site is both to theeast and west (Fig. 7, upper panel). Zonal recirculationlikewise occurs from both directions in two mainstreams, with negligible transport between them. Meridional recirculation streams are evident as well (Fig.7, lower panel). The maximum frequency pathway forair recirculating zonally back from the west occursabout 900 km to the south of the point of origin at amean height of 750 hPa. The recirculation period forthe air to reach this locality is about 4 days. The maximum frequency pathway for air recirculating backfrom the east occurs about 700 km to the north of theorigin at 800 hPa, with a recirculation time back to thesame longitude of 4 days. A significant amount of recirculation (29% at 850 hPa after 3 days) takes placeback to the site from the west on almost the direct lineof initial transport. The overall recirculation index forair originating over the eastern highveld plateau is 71%and the overall recirculation time 8 days. If transport and recirculation from a locality furtherto the west and south, in the vicinity of KimberleyBloemfontein in central South Africa (30-S, 25-E), isconsidered, a similar situation is evident with low-leveltransport (Fig. 8). Despite most direct transport beingto the southeast, a distinctive recirculation vortex is apparent in both the zon.al and meridional transport fields,in addition to substantial (up to 24-7o) localized recirculation about the point of origin. The latter occurs tothe north and south within a distance of 100 km of thesource, whereas the major recirculation vortex, as inthe previous case, has a radius of about 600-800 km,with the maximum frequency pathway of recirculatingair occurring at a mean height of 750-650 hPa and witha mean overall recirculation time of nearly 10 days(compared to 6-7 days for the localized recirculation).In this case the recirculation index is 58%. The vertically integrated 700-500-hPa zonal transport patterns from both the 25-S, 30-E and 30-S, 25-IEDECEMBER 1996 T Y S O N E T A L. 2225 / _-I./ / = 192 / X. ~,~ ~ / initial tran,port E: 39% W: 28% 1000 800 600 400 200 L 200 400 600 800 1000 ~__ ' ' ' ' / ' ' ' ' 00/ ~IN / ~- ~outherly/ 200 irect transport n0rlherly~ L/ initial transport ~ , ,, , ,'"'--dr, ;;J N: 20,/, s: 70/0 1000 800 600 400 200 0 200 4~0 600 800 ~'t000 F[o. 5. Vertically integrated surface-to-800-hPa transport from Sku~uza, ]O'uger National Park,during SAFARI. Initial transport is defined as transport in the first hour. Isolines indicate thepercentage frequency of trajectory pathways enclosed by the contour. Heavy lines denote maximum frequency pathways. Bold figures give the integrated percentage of transport along meridiansfor zonal transport and along latitudes for meridional transport. Light figures along the maximumfrequency pathways give geopotential heights and times taken for air parcels to reach those pointson the pathways.sources tend to be more complicated than their lowerlevel counterparts, notwithstanding the smaller recirculation indexes (Fig. 9). Recirculation vortices aremuch the same size. Mean overall recirculation timesare 6.4 days for the eastern plateau region and 7-10days for the central region. Semipermanent subtropical high pressure systemsare not the only anticyclones occurring in southern Africa, Transient ridging highs in the westerlies also affect the southern areas of the subcontinent. Their centers are usually located to the southeast of the SouthAfrican coastline from Durban to Port Elizabeth. A12226 JOURNAL OF APPLIED METEOROLOGY VOLLV~E 35 / ' ' ' ~"'/--. '~-" ' ' ~ ~ ~ ~ J ~ ~-- - easterly / ~ ~_ '~'~" '~" '~ '~/ .~"~ '~.~1~ 7 ~b ~~t _~~ ~~ ~rculationln~x / ~' ~'~ E: 4% W: 86% 1~ ~ ~ ~ ~ 0 200 ~ ~/'/ kX '~ /~'~ / N~ / ~,-~ t~di~t trans~A ~t ~ / / "~ ~ '"l -~ l hotter, y ~ ~ J r~i.ulat~ N: 2% S: 6% 1~ 8~ ~ 4~ ~ 0 200 4~ km~G. 6. Ve~ieally integrated ?00-500-hPa tr~spo~ &om Skukuza, ~ger National P~k, dutng SAFA~. Definitions ~e as in Figs. 1 and 4.though they may bring rain to coastal and adjacent inland areas, they are predominantly stable, rainless systems that produce transport patterns characteristic of alltypes of anticyclones over the subcontinent (Tyson etal. 1996a). Both systems produce recirculation vorticesin which the main stream of recirculating air parcels isseveral hundred kilometers from the source. Overallrecirculation times are shorter by about 2-3 days inridging highs. Together the systems account for 80%of all circulation types observed in midwinter (July)and 22% of those in midsummer (January).7. Combined recirculation fields At each of the points of origin examined so far, avortex of recirculating air with a radius of several hundred kilometers is found. Most of the recirculating airis contained in streams in an annular-like ring some1996 T Y S O N E T A L. 2227 800~ ~-~ \ 181 / / r 'l\ I%_ ) ~d,. / / ./ ' ~-~. ....:D~,, _Z ~oot - ., /~,.~~ ~t ~ 4/~.,o~~ ' ' W:g 0~ ) norgedy irculat~ /N , ~ ~ ~ ~ ~ ~.~ / / N:16%S:8% 1~ 8~ ~ 4~ 2~ 0 ~ 400 600 8~ 10~ kmF[o. 7. Vc~ica]ly integrated su~ace-to-800-hPa transport from 25-S, 30-E when semipermanentsub~opical continental high pressure systems prevail. Definitions ~e as in Figs. 1 and 4.hundreds of kilometers from the center. Little transportappears to occur in the interior of the ring. The picturebecomes progressively more complicated as the variousrecirculation fields are superimposed on one another.Integration into a uniform mass of recirculating air willtake place as the number of sources of transport increases. The complex picture may be approximated bycombining the vertically integrated surface-to-800-hParecirculation fields from the four site localities considered in this paper. This has been done to show the recirculation through 25-E, a central meridian roughlybisecting southern Africa (Fig. 10, upper panel). Eachtubelike transport conduit representing recirculating airin Fig. 10 encloses 95% of the trajectories recirculatingfrom a particular point of origin and is drawn in approximate proportion to the percentage of recirculationoccurring. It is clear that most of the air recirculatingin the semipermanent anticyclone over southern Africain the lowest layers eventually exits the subcontinentto the southeast over the Indian Ocean. In the vicinity2228 JOURNAL OF APPLIED METEOROLOGY VOLUME35 ~0~ ' -- easter~ --~-~S.~'~ ~7N.X ~ I / .t / ~ ~ ~ / ~_ ~ E:0% W:67%1~ 8~ ~ 4~ 2~ 0 ~ 4~ 6~ 8~ 1000 km ,oo; / .,,Yoe /~ -~ / / ~1 o/-' . ' ,p, ,-, r~i rculated initial transport ~hefiy N: 15% S: 5% I~ ~ '~ 4~ 2~' 0 200 400 6~ 8~ I0~ kmFIG, 8, Ve~icaIly integrated suffacc-to-800-hPa traaspo~ f~om 30-S, 25-E when semipe~aacntsub~opieal continental high pressure s~stems pmv~L Definitions ~e as in Figs, I and 4,of the central bisecting meridian shown in Fig. 10, 'themean easterly recirculated transport--that is, to thewest on the equatorward side of the system--occurs ata mean geopotential height of 790 hPa after a meanrecirculation through half the vortex of 4.9 days backfrom the east to the central meridian. The mean westerly recirculation--that is, to the east on the polewardside--takes place at a mean height of 775 hPa, with arecirculation time through half the vortex of 3.4 days.The overall mean recirculation time of lower-tropospheric air is 8.3 days. The recirculation is capped bythe 700-hPa absolutely stable layer over the subcontinent. The vertically integrated 700-500-hPa recirculationis capped by the 500-hPa stable layer (Fig. 10, lowerpanel). Mean easterly recirculation toward the AtlanticOcean on the northern margins of the system occurs ata height of 600 hPa after a recirculation time of 3.0days back to the central meridian. On the poleward sideof the system, westerly recirculation in the direction ofDECEMBER 1996 T Y S O N E T A L. 2229odgin: 30-E,25-S 8o0 = 128 4c~L ,~11"25' / ~t /~\ ~ 10. ~, _4 recirculationindex // Initial I ~ ~~L ~' . . . E: 29% W: 1~ 800 ~ 4~ ~0 0 ~ 400 6~ 8~ 10~ kmodgin: 25-E.30-S ' --~-7--"~'~ ........ 400 ej I ' ' ] ~- ~ S ~ ~ ~imulated / xx k / ,.,ti., t. nspor, I / . t / t [ ~ I ~ ~ ~ E:12%W:69% km ~m. 9. VeAica]]y integrated 700-500-hPa transport from 30-S, 25-E and 30-S, 25-~ whensemipc~cnt sub~op~ca] comi~c~] high pressure systems prcv~]. Pcfi~ifio.s ~~ as in Figs. and 4.the Indian Ocean takes place at a mean height of 610hPa after 3.0 days. The mean overall recirculation timeis 6.0 days. The climatology indicates that average recirculationtimes of air parcels recirculating in high pressure systems over southern Africa are on the order of a week.On particular occasions, as observations duringTRACE-A (Transport and Atmospheric Chemistrynear the Equator--Atlantic experiment) showed, individual recirculation times exceeding 2 weeks may beobserved. In one case, at the 700-hPa level, air wasobserved to recirculate over the land area of the subcontinent for 3 weeks. Once emitted into the atmosphere, the recirculation of aerosols within the recirculating air will be dependent on the residence timesof the particles suspended in the atmosphere.8. Aerosol residence times Direct transport of air below 500 hPa in anticycloniccirculations over the subcontinent takes 7-8 days onaverage to reach 10-E beyond the west coast at around20-S and 4-5 days to reach 35-E at 31-S (Tyson et al.1996a). By comparison, the mean overall recirculationtime over the subcontinental landmass, within the surface-to-500-hPa layer, in continental anticyclones is onthe order of 8-9 days.2230 JOURNAL OF APPLIED METEOROLOGY VOLUME35 i surface ! 800 ~ 20- I t ,.10-W 0- 10- 20-30- 40-50- 60- 70-E lO-W 0- 10- 20- 30- 40- 50- 60- 70-E FIG. 10. Upper panel shows superimposed surface-to-800-hPa recirculation under continental anticyclonic conditionsfrom all localities of origin. The cross-sectional area of each transport tube is proportional to the 95th percentile contourin the horizontal and vertical in the vicinity of the central meridian. The percentage of recirculation is given on eachtube and indicates the value in the vicinity of the central meridian unless otherwise indicated. The envelope withinwhich the 700-hPa absolutely stable layer occurs is shown by shading. Lower panel shows combined 700-500-hParecirculation from all localities of origin analyzed. The envelope within which the 500-hPa absolutely stable layeroccurs is shown by shading. The time spent in the atmosphere by aerosol particlesis a complicated function of their physical and chemicalproperties, together with their time and place and heightof release. Aerosol recirculation times are dependenton the size fraction of particles in the aerosol andwhether dry deposition occurs by fallout from the atmosphere and direct uptake at the surface, or onwhether wet deposition occurs in precipitation. Duringthe SAFARI field observation period of SeptemberOctober 1992, 99.5% of aerosols within the first 4 kmof the troposphere fell into the size range 0.1-0.4/~m,with a maximum frequency geometric mean volumediameter of about 0.275/~m (LeCanut et al. 1996). Atthat time, slightly larger particles tended to occur at thetop of the layer at around 4 km (which was the aircraftceiling height and not an atmospheric barrier). Pruppacher and Klett (1980) have reported thatglobal-mean residence times for aerosols are 2-7 d.aysin the lower troposphere, 1-2 weeks in the upper troposphere, and 3 weeks-1 month at tropopause level.In midlatitude winters, when precipitation is common,the time taken to remove anthropogenic sulphate fromDECEMBER 1996 TYSON ET AL. 2231the atmosphere is 4-5 days for North America and Europe (Benkovitz et al. 1994). For northern midlatitudes, winter zonal-mean aerosol turnover time is typically 4-6 days, whereas in the drier summers it is 615 days (Balkanski et al. 1993). Residence times of5-10 days have been inferred from measurements ofnuclear bomb-test radioactive species (Junge 1963;Chamberlain 1991 ) and of 7-9 days from radio nuclides after the Chernobyl accident (Cambray et al.1987). Saharan dust is transported in 4-7 and 6-11 days tothe Caribbean and South America ~from Africa, distances on the order of 5000 km (Carlson and Prospero1972; Talbot et al. 1986; D'Almeida 1987; Talbot etal. 1990; Swap et al. 1992). Particulate matter transported into, or formed in, the upper troposphere typically has lifetimes of weeks to months (Balkanski etal. 1993). Mean residence times of aerosols of about 7 days athigher latitudes appear to be of the same order as therecirculation period of air over southern Africa. Thesignificant difference is that the high-latitude aerosolsare transported in vigorous zonal airstreams of lowthermal stability, whereas the lower-latitude southernAfrican aerosols recirculate over the source region inhighly stable and comparatively sluggish airstreams.High-latitude transport over 7 days may remove aerosols from source by as much as 6000 km (with a 10m s-1 zonal flow). In contrast, over the South Africanplateau, an important fraction of total airflow by volume recirculates back to the meridian of origin in muchthe same time. Most of southern Africa is characterized by a summer rainfall regime. Little rainfall is experienced inwinter from May to September. At the same time, thelower and middle troposphere is characterized by theoccurrence of persistent atmospheric stability. Undersuch conditions, aerosol residence times are likely tobe on the order of weeks or longer. During the summer,rainy season residence times will be much more characteristic of their midlatitude counterparts and will beon the scale of days. Recirculation of aerosols at alltimes of the year, but particularly in winter, is likely tobe a major factor influencing long residence times inthe regional atmosphere over southern Africa.9. The annual cycle of recirculation Given the frequencies of occurrence of semipermanent continental anticyclones throughout the year overcentral South Africa (Table 1), recirculation fromthese systems alone may be expected to occur on 50%of all days from May to October and 30% of all daysfrom November to April. The highest frequency of occurrence is 70% in July, and the lowest is 15% in January. The average frequency of occurrence throughoutthe year is 40%. If the frequency of transient ridging anticyclones isadded to that of the semipermanent highs, then the frequency of all anticyclonic systems in midwinter risesto 80% in June and July and to an average of 27% inDecember and January. Using techniques developed previously (Tyson et al.1996a; Garstang et al. 1996c), it is possible to estimatemonthly and annual volume transports of air and masstransports of aerosols recirculating over southern Africa.10. Aerosol concentrations The mean daily surface particulate concentration deposited from the atmosphere in a periurban area site atMakalu, South Africa, between the urban-industrialcities of Johannesburg and Vereeniging over the period1989-93 was 20/~m m-z, with peak values exceeding120/~g m-3 (Rorich and Turner 1994). Over the sameperiod at Elandsfontein, South Africa, near Bethal, inthe industrial and electricity-producing area of the Eastern Transvaal Highveld, the mean was approximately15 /zg m-3. Peak daily values exceeded 140/~g m-~.Near Belfast, at the rural site of Palmer, South Africa,120 km to the northeast on the eastern limit of the plateau, the mean was about 10/~g m-z, with peak valuesreaching 60/~g m-3. At Verkykkop, South Africa, onthe escarpment near Volksmst 130 km to the southeastof Elandsfontein and located on the top of a 300-m hill,peak values of the particulate sulfate and nitrate components of the total aerosol loading have been shownto exceed 50 and 30/~g m-z, respectively, in elevatedlayers of pollution (Wells et al. 1987). The differencesbetween Palmer to the northeast and Verkykkop to thesoutheast are to be expected, since the latter site is situated on the northern edge of the main air transportcorridor from the urban-industrial heartland of SouthAfrica to the Indian Ocean (Tyson et al. 1996a). Some idea of seasonal differences in ambient aerosolloadings at remote rural sites may be inferred from thePalmer and Verkykkop data. Over the winter-earlysummer (June-December) period in 1991, 2-weekmaxima of 15/~g m-3 sulfates were observed at Palmer,in comparison to 39 ~g m-3 at Verkykkop. For latesummer (January-May) 1992, equivalents were. 53%and 36% of those observed in the June-December period, respectively (Taljaard and Zunckel 1992). During the 1992 SAFARI field observation period,median daily concentrations of inorganic particulatematerial less than 10 /~m in size (PM-10) at EtoshaNational Park and Victoria Falls were 28 and 34/zg m-3, respectively (Maenhaut et al. 1993). MedianPM-10 concentrations at Kruger National Park rangedfrom 25 to 33/zg m-3 (Maenhaut et al. 1996). Recentanalysis of surface particulates at Etosha National Parkhas yielded, for all size fractions of combined organicand inorganic aerosols, concentrations of 55 /~g m-3(Swap 1996). On specific occasions during SAFARI,high aerosol concentrations were observed in elevatedlayers above the surface. On 24 September 1992, for2232 JOURNAL OF APPLIED METEOROLOGY VOLUME35 TABLE 2a. Point transports, low level--surface to 800 hPa. Mass fluxes are calculated assuming a constant mean cohcentration of all sizefractions for the layer surface to 800 hPa of 50/~g m-3 for the anticyclonic conditions considered. Initial transport at the point of origin inthe easterlies (to the west) is indicated by E. Initial transport in the westerlies (to the east) is indicated by W. Direct and recirculated transportare denoted by subscripts d and r, respectively.Volume flux (m3 day-~ x 1014)Mass flux (ton day-~ x 103)Point of origin Ea Er Wa Wr Ea Er Wa Wr25-S, 30-E E 0.3 0.5 0.0 1.6 1.5 2.5 0.0 8.0 W 0.0 0.0 2.5 0.0 0.0 0.0 12.5 0.0200S, 25-E E 2.2 0.0 0.0 0.0 11.0 0.0 0.0 0.0 W 0.0 0.7 0.0 4.0 0.0 3.5 0.0 20.025-S, 20-E E 0.4 0.2 0.0 0.0 2.0 1.0 0.0 0.0 W 0.0 0.6 3.7 4.1 0.0 0.0 18.5 20.530-S, 25-E E 0.0 0.2 0.0 0.8 0.0 1.0 0.0 4.0 W 0.0 1.2 4.1 1.6 0.0 6.0 20.0 8.0Total 2.9 3.4 10.3 12.1 14.5 17.0 51.0 60.5instance, aircraft measurements revealed a maximumPM-10 concentration of 32 /xg m-3 at an altitude of2250 m over the Kruger National Park (LeCanut et al.1996) in air recirculating over the area (Piketh 1995).On this occasion, the total column loading to a heightof 3287 m was 42/~g m-3.11. Volume and mass fluxes in recirculating air Mass fluxes of aerosols out of southern Africa werepreviously estimated by determining the transport beyond 10-E into the Atlantic Ocean .and 35-E into theIndian Ocean (Tyson et al. 1996a). In the earlier investigation, recirculated transport was established byinference. In this study, volume and mass fluxes aredetermined explicitly for semipermanent corttinentalanticyclonic conditions by considering air recirculatingthrough meridional planes located 400 km to the westof each point of origin to establish the easterly component of transport to the west, and planes 400 km tothe east, to determine the westerly transport to the east.Volume fluxes are calculated by determining tlhe product of the plume area on the meridional plane and themean transport speed from the point of origin to theplane. Mass fluxes are determined from the product ofthe volume flux and the concentration in the l?lume byassuming a mean annual concentration of all aerosolsize fractions in the surface-to-800-hPa layer and the700-500-hPa layer to be 50 and 25 /~g m-3, respectively. Although likely to be of the correct order ofmagnitude, these figures probably are an underestimateof winter conditions and an overestimate of those insummer. But in the absence of more extensive free-airobservations than those quoted earlier, they must suffice for this exercise. Zonally direct and recirculated transports to the eastand west of the four points of origin located at 20-S,25-E; 25-S, 30-E; 30-S, 25-E; and 25-S, 30-E are givenin Table 2 for the surface-to-800-hPa and 700-500hPa layers. In the case of direct transport out of themean' 1000 km x 1000 km area centered on each pointof origin, both in the lower and upper layers, 22% ofthe mass transport is to the west in easterly flow and78% to the east in westerly flow. More pertinently forthis paper, 23% of recirculated mass transport is to thewest and 77% to the east. Up to 54% of the materialcompletes one cycle of recirculation before exiting thesubcontinent to the east into the Indian Ocean. By contrast, as much as 23% may be recirculated a secondtime over the land area. At lowest tropospheric levels (surface-800 hPa),111.5 x 103 ton day -~ are transported on the southern,poleward sides of continental anticyclones to the eastin westerly flow, both directly and by recirculation (Table 2a). Given that anticyclonic conditions prevail on40% of the days in a year (Table 1 ), this gives anannual westerly transport of 16.3 Mton. Similarly, thecombined direct and recirculated easterly transport tothe west on the northern, equatorward sides of the systems is 4.6 Mton each year. Within the integrated surface-to-500-hPa layer (Tables 2a and 2b), the combined direct and recirculated transport is 39.1 Mton tothe east and 11.5 Mton to the west. The westerly transport is almost the same as the 32 Mton yr-t reportedpreviously for anticyclonic transport to the east (Tysonet al. 1996a). Easterly transport toward the AtlanticOcean is considerably less than the 22 Mton previouslyestimated. Transport in the winter half of the year (heretaken to be May to October) by continental anticyclones is 10% greater and transport in the summer half10% less, reflecting the change in frequency of occurrence of these anticyclones in the two periods. Previously it was shown that total annual transport in semipermanent continental anticyclones constituted 73% oftransport in all synoptic situations. Scaling the anticyclonic transports quoted above in the same proportion,DECEMBER 1996 TYSON ET AL. 2233 TABLE 2b. Point transports, middle level--700-500 hPa. Mass fluxes are calculated assuming a constant mean concentration of all size fractions for the layer 700-500 hPa of 25/~m m-3 for the anticyclonic conditions considered. Definitions are as in Table 2a.Volume flux (m3 day-~ x 10]4)Mass flux (ton day-~ X 103)Point of origin Ea Er Wa W, Ea Er Wa Wr25-S, 30-E E 2.2 1.1 0.0 0.6 5.5 2.8 0.0 1.5 W 0.0 0.5 11.8 0.8 0.0 1.2 25,0 2.020-S, 25-E E 9.2 1.3 0.0 3.6 23.0 . 3.2 0.0 9.0 W 0.0 2.0 4.8 0.0 0.0 5.0 12.0 0.025-S, 20-E E 0.4 0.0 0.0 0.0 1.0 0.0 0.0 0.0 W 0.0 0.6 10.4 10,4 0.0 1.5 26.0 26.0 3.730-S, 25-E E 0.0 0.8 0.0 0.8 0.0 2.0 0,0 2.0 W 0.0 0.9 14,1 7.1 0.0 2.2 35.2 17.6 1.2Total 11.8 7.2 41.1 23.3 29.5 17.9 98.2 58,1the total direct and recirculated transport of aerosolsover southern Africa in all synoptic situations withinthe surface-to-500-hPa layer is 64.3 Mton yr-~. Combining the surface-to-800-hPa and 700-500hPa layers, anticyclonically recirculating transport inthe integrated surface-to-500-hPa layer constitutes 44%of the total aerosol load being transported over southernAfrica, and direct transport constitutes 56%. Considering only the recirculating fraction, it is apparent that17.3 Mton is recirculated annually on average to theeast over the central meridian of the plateau of southernAfrica in westerly transport and 5.1 Mton is recirculated to the west in easterly transport. It would appearthat 12.2 Mton recirculates once before exiting the subcontinent to the Indian Ocean and that up to 5.1 Mtonis available for inclusion in a second cycle of recirculation over the subcontinent.12. Conclusions Semipermanent subtropical anticyclones are the dominant circulation type affecting southern Africa. In themiddle of winter (July) they have 70% occurrence, andfor the six months May to October the frequency doesnot fall beneath 50%. Combined with transient ridginganticyclones that are associated with the westerlies, the / 10- recirculating flux: 44% of total / ///40-8 reci)~ulating flux: 44% of total/ 10-W 0- 10- 20- 30- 40- 50- 60- 70-E FiG. 11. Total and recirculating annual transport fluxes across the central meridian through southern Africa associatedwith continental high pressure systems. The total westward transport of 11.5 Mton yr-~ contains 5.1 Mton of recircu]ated material; the total eastward transport of 39.1 Mton contains 17.3 Mton of recirculated material in the layercapped by the 500-hPa absolutely stable layer. In both cases the recircu]ated transport is 44% of the total transport.2234 JOURNAL OF APPLIED METEOROLOGY VOLUME35combined frequency of anticyclonic circulation over thesubcontinent rises to 80% in June and July and equalsor exceeds 66% in the winter half of the year. For themost part, the anticyclonic wind fields over the interiorplateau in winter are associated with a highly stable stratification of the atmosphere, little or no cloud or rainfall,and maximum receipt of solar radiation. Different synoptic situations produce different aerosol and trace gas transport fields. Those associated withsubtropical anticyclones are distinctive, particularlywith respect to the degree to which recirculation takesplace. Transport is composed of two components: thedirect, in which material is advected directly with littledelay from the subcontinent to the oceans beyond, andthe recirculation, in which material recirculates backtoward the point of origin. Recirculation has beenshown to occur over distances that range from a fewtens to thousands of kilometers, and it exerts a controlling influence on the accumulation of aerosols andtrace gases within the greater southern African region. Analysis of forward trajectory fields within overlapping 2000 km x 2000 km areas reveals that, rather thanrecirculating in a quasi-uniform manner in the horizontal and vertical, air parcels emanating from particularpoints of origin tend to recirculate in anticyclonicallycurving streams having radii on the order of 500-700km. Between the streams and the origins, relatively little transport appears to take place. Combined fieldsfrom widely dispersed points over southern Africa reveal a large, slowly circulating, anticyclonic recirculation vortex over the subcontinent when anticyclonicsynoptic conditions prevail persistently. The recirculation vortex is evident from the surface to the frequently occurring absolutely stable layer at the 500-hPalevel over southern Africa, particularly in winter.Above that the increasing influence of the circumpolarwesterlies ensures that transport patterns become morezonal and that recirculation diminishes rapidly. Typical recirculation periods in the surface-to-500hPa layer are on the order of a week. On some occasions, up to 20 days of recirculation have been observed. Under subtropical anticyclonic conditions, recirculating transport constitutes 44% of all transportbetween May and October. Of the air parcels recirculating on the subcontinental scale, up to a quarter appear to recirculate a second time. On smaller scales,local and mesoscale recirculation in small vortices mayoccur along transport trajectories. By determining volume fluxes of air in both directtransport and recirculation streams and coupling theseto estimated concentrations of aerosols, aerosol masstransports and recirculation fluxes over southern Africahave been estimated. On the northern equatorwardflank of the continental anticyclones !t is estimated thatcombined direct and recirculating transport in the surface-to-800-hPa layer is 4.6 Mton yr -~; on the southernpoleward flank it is 16.3 Mton. The fact that the transport in the latter case is over three times greater thanin the former is a function of the mean location of thecenters of the high pressure systems over eastern SouthAfrica and the resulting asymetry of the systems overthe southern part of Africa. Within the lower troposphere to a height of 500 hPa, the combined direct andrecirculated transport is 11.5 Mton yr-~ to the west ineasterly flow and 39.1 Mton yr -~ to the east in westerlyflow. Considering recirculation alone, 5.1 Mton yr- ~ ofaerosols recirculate to the west in easterly transport and17.3 Mton yr-~ is recirculated to the east in westerlytransport (Fig. 11 ). On both the equatorward and poleward flanks of the recirculation vortex the recirculationflux is 44% of the total transport. The regional consequences of the recirculating aerosol flux may be considerable. Changes in the radiationbalance may produce ameliorative cooling to offset anticipated global warming. Such cooling is already beingpredicted by current general circulation models that incorporate the effects of sulphate aerosols alone (Mitchell et al. 1995a,b). Similarly, work done on closed-lakebasins (Kutzbach 1980; Benson 1981; Hastenrath andKutzbach 1983) shows that long-term changes in theenergy balance may lead to unforeseen changes in thehydrologic balance and precipitation. An abundance ofsmall aerosol particles in the regional atmosphere mayhave a negative effect on cloud microphysical processes, leading to a lowering of the median cloud dropsize and a consequent diminution in the efficiency ofthe coalescence process of rainfall generation (Mather1995). Long-term ecological changes may also dependon the transport and deposition of critical trace elements within the system and on the potential for humanintervention to disrupt linkages between ecosystemsdependent on such transport. Given the large-scale recirculation of aerosols and trace gases over southernAfrica, the phenomenon of recirculation and its consequences invite further investigation. Acknowledgments. It is with much appreciation thatthe authors thank Per Kfillberg for undertaking the analysis of forward trajectories and Matt Cobbett for completing the ensemble analyses. Philip Stickler andWendy Job prepared the figures. The research has beensupported by the Foundation for Research Development and the University of the Witwatersrand in SouthAfrica, and under Grants ATM92-07924 and ATM9424433 from the National Science Foundation to theUniversity of Virginia. Part of the study was initiatedwhile the first author was supported by the Universityof Virginia when on sabbatical leave at that institution.The work'is a contribution to South African globalchange research and to International Geosphere-Biosphere Programme, World Climate Research Programme, IHDP, and System for Analysis, Research,and Training regional research. REFERENCES Annegarn, H. J., M. A. Kneen, S. J. Piketh, A. J. Home, H. S. P. Hlapolosa, and G. A. Kirkman, 1993: Evidence for large-scaleDECEMBER 1996 TYSON ET AL. 2235 circulation of sulphur over South Africa. Proc. National Asso ciation for Clean Air Annual Conf., Dikhololo, South Africa, Nat. Assoc. for Clean Air.Balkanski, Y. J., D. J. Jacob, G. M. Gardiner, W. G. Gmustein, and K. K. Turekian, 1993: ~lYansport and residence times of tropo spheric aerosols inferred from a global three-dimensional sim ulation of 210 Pb. J. Geophys. Res., 98, 20 573-20 586.Benkovitz, C. M., C. M. Berkowitz, R. C. Easter, S. Nemesure, R. Wagenerand, and S. E. Schwartz, 1994: Sulfate over the North Atlantic and adjacent continental regions: Evaluation for Octo ber and November 1986 using a three-dimensional model driven by observation-derived meteorology. J. Geophys. Res., 99, 20 725-20 756.Benson, L. V., 1981: Paleoclimatic significance of lake level fluctu ations in the Lahontan Basin. Quat. Res., 16, 390-403.Browell, E. V., 1993: TRACE-A airborne DIAL ozone and aerosol data. NASA/Langley Research Center Rep.Cambray, R. S., P. A. Cawse, J. A. Garland, J. A. B. Gibson, P. Johnson, G. N. J. Lewis, D. Newton, L. Salmon, and B. O. Wade, 1987: Observations on radioactivity from the Chernobyl accident. Nucl. Energy, 26, 77-101.Carlson, T. N., and J. M. Prospero, 1972: The long-range movement of Saharan air outbreaks over the northern equatorial Atlantic. J. Appl. Meteor., 11, 283-297.Chamberlain, A. C., 1991: Radioactive Aerosols. Cambridge Uni versity Press, 255 pp.Cosijn, C., and P. D. Tyson, 1996: Stable discontinuities in the at mosphere over South Africa. S. Afri. J. Sci., in press.D'Almeida, G. A., 1987: Desert aerosol characteristics and effects on climate. Palaeoclimatology and Palaeometeorology: Mod ern and Past Patterns of Global Atmospheric Transport, M. Leinen and M. Sarnthein, Eds., Kluwer Academic Publishers, 311-338.Fishman, J., 1991: Probing planetary pollution from space. Environ.Sci. Technol., 25, 612-621. Garstang, M., and P. D. Tyson, 1996: Atmospheric circulation, ver tical structure and transport. Fire in Southern African Savannas: Ecological and Atmospheric Perspectives, B. W. van Wilgen, M. O. Andreae, J. G. Goldammer, and J. A. Lindesay, Eds., Witwatersrand University Press, in press. -, H. Cachier, and L. Radke, 1996a: Atmospheric transports ~f particulate and gaseous products by fire. Sediment Records of Biomass Burning and Global Change, J. Clark, Ed., Springer Verlag, in press. , E. Browell, and R. Swap, 1996b: Large scale transports 'of biogenic and biomass burning products. Biomass Burning and Global Change, J. Levine, Ed., MIT Press, in press. , R. Swap, M. Edwards, P. Kfillberg, and J. A. Lindesay, ~996c: Horizontal and vertical transport of air over southern Africa. J. Geophys. Res., in press.Harrison, M. S. J., 1993: Elevated inversions over southern Africa: Climatological properties, formation, processes and relation ships with rainfall. S. Aft. J. Sci., 75, 1-8.Hastenrath, S. L., and J. E. Kutzbach, 1984: Paleo-climate and water budget of East African lakes. Quat. Res., 19, 141-153.Held, G., H. Scheifinger, and G. M. Snyman, 1994: Recirculation of pollutants in the atmosphere of the South African highveld. S. Aft. J. Sci., 90, 91-97.Junge, C. E., 1963: Air Chemistry and Radioactivity. Academic Press, 382 pp.Kfillberg, P., 1984: Air parcel trajectories from analyzed or forecast windfields. Swedish Meteorological and Hydrological Institute Research and Development Note 37, 84 pp.Krishnamurti, T. N., H. E. Fuelberg, M. C. Sinha, D. Oosterhof, E. L. Bensman, and V. B. Kumai, 1993: The meteorological environ ment of the tropospheric ozone maximum over the tropical South Atlantic Ocean. J. Geophys. Res., 98, 10 621-10 641.Kutzbach, J. E., 1980: Estimate of past climate at paleolake Chad, North Africa, based on a hydrological and energy balance model. Quat. Res., 14, 210-223.Lancaster, I. N., 1979: Evidence for a widespread late Pleistocenehumid period in the Kalahari. Nature, 279, 145-148.--, 1981: Palaeoenvironmental implications of fixed dune systems in southern Africa. Palaeogeography, Palaeoclimatology and Palaeoecology, Vol. 33, 327-346.--, 1988: Development of linear dunes in the southwestern Kala haft. J. Arid Environ., 14, 233-244.LeCanut, P., M. O. Andreae, G. W. Harris, F. G. Wienhold, and T. Zenker, 1996: Aerosol optical properties over southern Africa during SAFARI-92. Biomass Burning and Global Change, J. Levine, Ed., M1T Press, in press.Maenhaut, W., I. Salma, M. Garstang, and F. Meixner, 1993: Size fractionated atmospheric aerosol composition and aerosol sources at Etosha, Namibia and Victoria Falls, Zimbabwe. Proc. AGU Fall Meeting (suppl.), San Francisco, CA, Amer. Geophys. Union, 104. , H. J. Annegarn, and M. O. Andreae, 1996: Regional atmospheric composition and sources in the eastern Transvaal, South Africa and the impact of biomass burning. J. Geophys. Res., in press.Mather, G. K., 1995: Some possible influences of aerosols on cloud microstructure in sub-Saharan Africa. Int. Symp. on Drought in Africa, Trieste, Italy, Int. Centre for Theor. Phys., 53.Mitchell, J. F. B., T. C. Johns, J. M. Gregory, and S. F. B. Tett, 1995a: Climate response to increasing levels of greenhouse gases and sulphate aerosols. Nature, 376, 501-504. , R. A. Davis, W. J. Ingrain, and C. A. Senior, 1995b: On surface temperature, greenhouse gases, and aerosols: Models and ob servations. J. Climate, 8, 2364-2386.Moody, J. L., and Coauthors, 1996: Meteorological mechanisms for transporting ozone over the WNAO (western North Atlantic Ocean)--A case study, August 24-August 1993. J. Geophys. Res., in press.Newell, R. E., I. W. Kidson, D. G. Vincent, and G. J. Boer, 1972: The General Circulation of the Tropical Atmosphere and Inter actions with Extratropical Latitudes. Vol. 1. MIT Press, 258 pp.Picketing, K. E., A. M. Thompson, D. P. McNamara, M. R. Schoe berl, L. R. Lair, P. A. Newman, C. O. Justice, and J. D. Kendall, 1994: A trajectory modeling investigation of the biomass burn ing tropical ozone relationship. Proc. Quad. Ozone Symp., NASA, 101-104.Piketh, S. J., 1995: Generation and transportation characteristics of suspended particles in the eastern Transvaal. M.S. dissertation, University of the Witwatersrand, 148 pp.Preston-Whyte, R. A., and P. D. Tyson, 1973: Note on pressure os cillations over South Africa. Mon. Wea. Rev., 101, 650-653. , and , 1989: The Atmosphere and Weather of Southern Africa. Oxford University Press, 374 pp. , R. D. Diab, and P. D. Tyson, 1977: Towards an inversion climatology of southern Africa: Part I1, Non-surface inversions in the lower atmosphere. S. Afr. Geogr. J., 59, 47-59.Pruppacher, H. R., and J. D. Klett, 1980: Microphysics of Clouds and Precipitation. D. Reidel, 714 pp.Rorich, R. P., and C. R. Turner, 1994: Ambient monitoring network annual data report for 1993 and regional trend analysis. Rep. TRR/S94/059/rw, TR1, ESKOM.Schulze, B. R., 1965: Climate of South Africa. Part 8: General survey. WB 28, South African Weather Bureau, Pretoria, South Africa, 330 pp.Swap, R. J., 1996: Transport and impact of southern African aerosols. Ph.D. dissertation, University of Virginia, 311 pp. , M. Garstang, S. Greco, R. Talbot, and P. K~llberg, 1992: Sa haran dust in the Amazon Basin. Tellus, 44, 133-149.Talbot, R. W., R. C. Harriss, E. V. Browell, G. L. Gregory, D. I. Sebacher, and S. M. Beck, 1986: Distribution and geochemistry of aerosols in the tropical North Atlantic troposphere: Relation to Saharan dust. J. Geophys. Res., 91, 5153-5182. , M. O. Andreae, H. Berreesheim, P. Artaxo, M. Garstang, R. C. Harriss, K. M. Beecher, and S. M. Li, 1990: Aerosol chemistry during the wet season in central Amazonia: The influence of long-range transport. J. Geophys. Res., 95, 16 955-16 969.2236 JOURNAL OF APPLIED METEOROLOGY VOLUME35Taljaard, J. J., 1953: The mean circulation in the lower troposphere over southern Africa. S. Afr. Geogr. J., 35, 33-45. ,1955: Stable stratification in the atmosphere over southern Af rica. Notos, 4, 217-230.--, and M. Zunckel, 1992: Comparative Study of Pollutants in the Eastern Transvaal. EMATEK, CSIR, 190 pp.Tyson, P. D., 1967: Some characteristics of the mountain wind over Pietermaritzburg. Proc. Jubilee Conf. South African Geographic Society, Durban, South Africa, S. Afr. Geogr. Sot., 103-128.~ , , and R. A. Preston-Whyte, 1972: Observations of regional to pographically induced wind systems in Natal J. Appl. Meteor., 11, 643-650. , --, and R. D. Diab, 1976: Towards an inversion climatology of southern Africa: Part 1, Surface inversions. S. Afr. Geogr. J., 58, 151-163. , F. J. Kruger, and C. W. Loi~w, 1988: Atmospheric pollution and its implications in the eastern Transvaal Highveld. S.A. Na tional Scientific Progress Rep. 150, Foundation for Research Development, Pretoria, South Africa, 114 pp. , M. Garstang, R. Swap, P. Kfillberg, and M. Edwards, 1996a: An air transport climatology for subtropical southern Africa. Int. J. Climatol., 16, 265-291. , E. V. Browell, R. D. Diab, and A. M. Thompson, ~996bi Transport and vertical structure of ozone and aerosol distributions over southern Africa. Biomass Burning and Global Change, J. S. Levine, Ed., MIT Press, in press.Wells, R. B., G. M. Snyman, G. Held, and A. Dos Santos, 1987: Air pollution on the eastern Transvaal Highveld. Foundation for Research Development Rep. CSIR Atmos/87/23, At mospheric Sciences Division, NPRL, CSIR, Pretoria, South Africa, 147 pp.

Save