Aerosol Recirculation and Rainfall Experiment (ARREX)

March 2004

Executive Summary


The Aerosol Recirculation and Rainfall Experiment (ARREX) was initiated in December 1997 to investigate the major transport pathways of atmospheric constituents over South Africa as well as to evaluate for the first time the effects that industrial aerosols, both primary and secondary, have on cloud microphysical processes. A total of seven intensive field campaigns have been undertaken over the past five years, four during the dry season and three during the wet season. Instrumented South African Weather Services aircraft (Aerocommanders 690A) have been utilised to investigate both the horizontal and vertical distribution of aerosols and cloud condensation nuclei. ARREX formed an important component of the SAFARI 2000 campaign. During the two last wet season campaigns particular attention was paid to cloud microphysical processes in convective clouds that have or have not been affected directly by industrial emissions. This report presents the data analysis and interpretation of the aircraft collected data, with particular emphasis on cloud condensation nuclei and regional-scale aerosol transportation in the atmosphere.


It is well established now that anthropogenic generated aerosols are transported in anticyclonic type circulation patterns and impact at remote sites all over the southern African subcontinent. In addition, recent research undertaken as part of the South African hygroscopic seeding experiment, has shown that modifying the cloud condensation nuclei (CCN) ingested by convective storms at cloud base with hygroscopic flares can have a marked influence on the precipitation formation processes. Increasing the availability of hygroscopic particles in the 0.5 µm size range suppresses the formation of numerous smaller CCN, broadening the size spectrum and hence reducing the cloud droplet concentration, resulting in faster droplet growth rates and thus higher possibilities for rain fallout. It has also been suggested by Mather (1995) that this mechanism is relevant to climate change studies. A positive feedback could, for example, exist between drought conditions (characterised by more abundant small CCN) and inefficient precipitation formation processes in convective storms. These feedback mechanisms would also occur with anthropogenic aerosols that act as CCN, although their chemical composition (especially hygroscopicity), size distributions and concentrations, relative to the natural CCN in an area will determine the sign of the effect. The relative contributions of natural to anthropogenic CCN and their effects on cloud formation have yet to be quantified. It is reasonable to expect that these effects will be best studied at a regional scale, especially where transport mechanisms ensure that both natural and anthropogenic aerosols impact atmospheric particulate compositions thousands of kilometres away from the source regions. Not only could this impact of aerosols on clouds influence the distribution of available ground and surface water, but also on incoming solar radiation, which ultimately affects the radiation budget of the globe. Anthropogenic induced changes in the rainfall, especially across national boundaries, has significant implications in terms of externality costs that will be incurred by the responsible industries in the future. Furthermore, the transport of aerosols over the subcontinent is important to understand if impacts on remote environments and transboundary impacts are to be understood. Transports over southern Africa are examined in the context of well-known vertical distributions in the atmosphere.


The objectives of ARREX are to determine the chemical and morphological characteristics of anthropogenic aerosols, to assess how these would influence cloud formation over southern Africa and to quantify the relative contributions of natural and anthropogenic aerosols to effective rain producing CCN. Additionally, atmospheric transport characteristics over the subcontinent are examined in the context of measured aerosol concentrations over South Africa.


Long and short flights were conducted over South Africa between the period 1997 - 2001 using either one or both Aerocommanders from the South African Weather Service, which were fitted with key aerosol, CCN and trace gas instruments, with the aim of answering ARREX-related questions. During the SAFARI 2000 campaign, flights were extended over a large portion of southern Africa to examine the atmospheric aerosol and trace gas loading exacerbated through the input of biomass burning products.

Measurements of cloud condensation nuclei have been taken in remote regions of southern Africa and over the industrialised Highveld region. Flights have been conducted both below and in clouds that are directly affected by industrial emissions and clouds that are upwind of the main source region. Measurements have been undertaken to estimate the relative importance of all sources on the subcontinent which may affect precipitation processes. This report covers all findings between 1997 and 2001.

Aerosol characterisation and transport

Aerosol characterisation

Examining the size distribution of the aerosol gives an indication of the aerosol's radiative properties and its interaction in the atmosphere in terms of residence time and transport. Accumulation mode sized particles, characteristic of combustion processes from industry and biomass burning, dominate over the Highveld and Lowveld during the biomass burning season.

It has been observed that the winter (May 1998) aerosols were of a coarser nature (>1mm) than during summer (December 1997) when fine aerosols outnumbered coarser aerosols by almost two orders of magnitude. Particles in the size range 0.2-0.4mm were observed to be enhanced at all levels of the lower- to mid-troposphere, especially over the Highveld. Aerosols over the marine region displayed the generation of coarser particles, as did aerosols measured over the Lowveld.

The thermodynamic structure of the atmosphere played a large role in the vertical distribution and concentrations of aerosols and trace gases, such as ozone. At the same time, the synoptic conditions prevailing at the time determined whether certain regions were ventilated and consequently total aerosol concentrations were lowered or larger particles deposited. Alternatively, stable conditions extended the residence time of the aerosols in the atmosphere.

Inhaca Island is influenced by various air masses of differing sources resulting in a high degree of optical variability between the periods April to December, the highest being between August and November. The aerosol optical thickness observed at Inhaca Island indicates high turbidity. In ~50% of the measurements, aerosol optical thickness values were above 0.2, with an overall mean of 0.26 ± 0.19. The Angström exponent parameter varied from 0.2 to 2, with a modal value of 1.6, indicative of a wide range in particle sizes and the dominance of fine mode aerosols at this site.

Data from Mongu, Zambia, Bethlehem, South Africa and Inhaca Island, Mozambique reveal seasonal variability, with a significant increase in aerosol content between August and October. This suggests a strong contribution of biomass burning characteristic of this time of year. A north to south gradient in aerosol optical thickness was confirmed. The highest aerosol content was observed over Mongu, while Bethlehem had the lowest.

It has been observed that there was a tendency for decreasing particle size as aerosol optical thickness increased.

Transport of aerosols over southern Africa
Preferred transport pathways are spatially organised both horizontally and vertically, with persistent elevated absolutely stable layers exerting a major controlling influence.
Almost half of all the air reaching the Highveld (43 %) is clean marine air advected within the westerly disturbances over the southern parts of South Africa. As much as 25 % of all transport to the Highveld is already aerosol laden from subtropical Africa.

The most important transport from the Highveld is directly to the Indian Ocean just below ~700 hPa. Over a third of the air from the industrial heartland of South Africa is recirculated on various scales over the subcontinent.

Neighbouring countries and remote sites of southern Africa are affected by industrial aerosols from the Highveld. Transport to Mozambique occur more than one third of the time, whilst roughly one out of three trajectories exiting the Highveld impacts Botswana. Swaziland is affected, on average, by one out of four trajectories. Direct southerly transport from the Highveld that is over Zimbabwe occurs 15 % of the time.

Elevated peak aerosol concentrations were detected over and exiting the Highveld industrial region during the ARREX flights. The layering of aerosols throughout the lower to mid-troposphere is very noticeable and relates closely to the elevated stable layer structure over South Africa. Horizontal and vertical mixing does not appear to be as pronounced as previously thought. Aerosols tend to accumulate in the atmosphere as fine plume filaments and are seemingly transported in this fashion over large distances with the overall decreases in concentrations being much lower than expected. As the thermodynamic profile over several days determines the vertical elevation of the plume filaments, so varying transport modes over several days will determine the position of these plume filaments over the subcontinent.

Improved estimates have been made of the volume and mass fluxes for westerly and easterly air transport out over the adjacent Indian and Atlantic Oceans at different levels of the troposphere. Based on a case study, the annualised aerosol flux across the coast of South Africa to the Indian Ocean was estimated to be around 40 Mt y-1. Long-term observations confirmed this figure.

Additionally, the deposition and iron fertilisation of the marine biota in the southwest Indian Ocean has been examined and it has been estimated that the mean daily deposition of iron into the sea in the central South Indian Ocean following a peak-concentration episode over eastern South Africa is around 0.99 mg.m-3. Such peak concentrations occur on around 33 days in a year; given that the average duration of an episode centred on the peak is 3 days, the number of days a year in which iron fertilisation may be significant appears to be around 100.

During several case studies the spatial variation in aerosol optical thickness between the interior sites of Pietersburg and Bethlehem and the Lowveld site of Skukuza clearly indicated the influence of biomass burning product transport from the north of South Africa.

Microphysical Processes in clouds

Measurements collected during the ARREX and SARAFI-2000 projects have shown that CCN concentrations over southern Africa are significantly elevated above background values. Seasonal differences have been observed in CCN concentration and characteristics.
CCN distribution over southern Africa

Anthropogenic aerosol emissions significantly enhance natural CCN concentrations over southern Africa

CCN distribution in the horizontal is influenced by the nature of aerosol sources, atmospheric transport and removal mechanisms.

Industries, power plants and urban areas clustered on the Highveld are a significant year-round source of anthropogenic aerosols. CCN concentrations are significantly elevated downwind of industrialised regions.

Biomass burning mainly occurs in the northern regions of the subcontinent in the late dry season (July-September). At that time, CCN concentrations reflect the north-south gradient in biomass burning emissions, with concentrations increasing towards the north.
Rainout and washout clean the atmosphere, and CCN concentrations may be four times lower after rain showers than after several days of no rain.

The advection of clean air over the subcontinent from the Indian or the Atlantic Ocean lowers CCN concentrations.

Most CCN are contained within the mixing layer, which typically extends to an altitude of between 3000 and 4000 m over the plateau and is capped by a temperature inversion at the ~700 hPa level.

Stratification of aerosols and CCN is common, due to the presence of absolutely stable layers and occasional convection in clouds.

Biomass burning aerosols from savanna fires in southern Africa are efficient CCN. They are composed mainly of soluble organic particles, often with the inclusion of potassium salts.

CCN concentrations are generally higher in the dry season than in the wet season due to the input of biomass burning products and the lack of precipitation scavenging in the dry season. Atmospheric stability and recirculation are also more prevalent in winter.

CCN activation spectra
The characteristics of aged smoke (>1 day) are relatively homogenous over the subcontinent and the relationship between the number of CCN NCCN (cm-3) and the supersaturation s (%) in aged biomass burning smoke can be summarized by the equation:

NCCN = 692 s 0.54

Newer smoke particles are more efficient CCN at lower supersaturations than aged smoke particles associated with the regional haze in the burning season. This is due to a change in the chemical composition of the particles as they are transported. Fresh emissions are composed of KCl particles, which are converted to K2SO4 and KNO3 particles within an hour. In the regional haze, ammonium sulphate particles dominate.

The CCN-aerosol relationship Twice as many aerosols are activated to form CCN in the dry season as opposed to in the wet season, since biomass burning products are more efficient CCN than industrial emissions. Soluble material is also preferentially removed by washout in the wet season.
There is a strong linear relationship between CCN and aerosol concentrations in the dry season because the atmospheric aerosol is fairly homogeneous throughout the region.
Cloud droplet size distributions

As has been found throughout the world, clouds forming in maritime environments have low droplet concentrations and a broad cloud droplet spectrum skewed towards larger droplets. Clouds forming in continental environments tend to have high cloud droplet number concentrations and a narrower cloud droplet spectrum skewed towards smaller droplet sizes.

Clouds influenced by emissions from industrial complexes and power stations have higher droplet concentrations and narrower spectra than those growing in background conditions.
Biomass burning emissions also have a profound impact on the droplet size distributions in clouds. Biomass smoke CCN overwhelm natural levels of CCN and dramatically increase the number of cloud droplets.

Radar-derived storm climatology
Most storms occurring over the Highveld only have short durations (~50% of storms do not last for longer than half and hour). The lifetime of storms is the major limiting factor to convective precipitation development.

There is a direct relationship between storm height and rain volume from the storm i.e. the higher the storm extends, the more rain it will produce.

Many storms have ineffective rainfall processes. Many storms with relatively low rain volumes grow to great heights, perhaps due to the slow initiation of rainfall.

The influence of anthropogenic aerosols on rainfall efficiency
Results from a cloud parcel model show that aerosol emissions from biomass and fossil fuel burning significantly inhibit the production of rainfall. Coalescence is retarded in clouds forming in polluted air masses, and almost no precipitation-size particles have formed after 30 minutes of growth. At least half the storms over the Highveld dissipate within half an hour, so in the majority of cases there is not sufficient time for rain to form. Giant CCN may have the potential to enhance coalescence, especially in highly polluted environments. They need to be further investigated before we can draw any conclusions about their role.

In-situ versus satellite data of clouds
Advanced Very High Resolution Radiometer (AVHRR) satellite data confirms that effective radii are higher in maritime clouds than in continental clouds. Different cloud processes are operating in the two environments: the 14 mm coalescence limit is not exceeded in the continental clouds.


For the first time it has been possible to characterise the nature and availability of CCN over South Africa. CCN concentrations have strong seasonal as well as spatial distribution patterns over the subcontinent. The industrialised Highveld acts as a major source of activated CCN. The numbers of CCN drop dramatically towards the coast. It was shown in a previous report that the nature and size of the aerosols also change along this east west gradient.

Long-range atmospheric transport and recirculation of aerosols and trace gases over southern Africa are expected to affect the radiative balance, photochemistry and biogeochemistry of the region. Improved knowledge with respect to the main industrial contribution to airborne material over southern Africa and its impact on other regions of the subcontinent is a starting point towards a comprehensive analysis of the air pollution status of the region.

ARREX; Cloud condensation nuclei; aerosols; aerosol transport, vertical and horizontal aerosol distributions; Southern Africa, industrial / biomass emissions; direct / indirect radiative forcing; radar; satellite retrievals.