Air Pollution as a Climate Forcing: A Workshop


Summary C. Aerosols

How do aerosols affect climate? Human activities, especially fuel combustion for energy production and land clearing, emit particles to the atmosphere. These particles are similar to carbon dioxide and other greenhouse gases because their atmospheric concentrations have increased greatly since the Industrial Revolution, and because they impose changes on the energy balance of the planet. Unlike greenhouse gases, these particles have environmental effects beyond altering climate: they directly affect regional air quality and the health of people and ecosystems in source regions.

All particles, especially those with diameters below 1 µm, interact with the solar radiation entering the earth-atmosphere system. Reflective aerosols reduce the incoming radiation, which cools the climate; of these, sulfates have received the most attention, and "organic" carbon (OC) particles are also important. Recent work (see ten Brink presentation) indicates that nitrates may also contribute significantly. Absorbing aerosols are so called because they absorb sunlight and transfer it to the atmosphere as heat, so that the earth-atmosphere system experiences both a net warming and a redistribution of energy. These absorbers heat the surrounding air, but also prevent energy from reaching the underlying surface, which is cooled. Black carbon (BC), or soot, is thought to dominate aerosol light absorption in many regions. Other contributors are desert dust and some organic carbon species; absorption by the latter is weaker, especially for green and red light, than that of black carbon (see Novakov presentation). The balance between reflection (scattering) and absorption determines whether the direct radiative effects of aerosols results in net warming or net cooling of the climate.

Emission of the climatically-relevant trace species — greenhouse gases and aerosols — is dominated by different sources, so quantifying the global budget of each presents a separate challenge. For example, industrialized nations emit about half of CO2 emissions, and are responsible for most of the accumulated anthropogenic CO2, but most of the emitted BC and OC particles come from Asia, Latin America, and Africa (see Streets presentation). About 80% of SO2 — from which sulfate aerosols are formed — results from fossil-fuel burning, dominated by coal combustion for power generation and industrial use, whereas only about 30% of BC comes from fossil-fuel burning, mostly from coal combustion for heating and cooking.

What properties are needed to assess climate effects? A few measurable aerosol properties, summarized in Table 1, are especially relevant to understanding climatic effects. Because aerosols are not well mixed in the atmosphere, and their climatic and other environmental effects depend on location, horizontally- and vertically-resolved distributions are needed. Measurements of these properties are needed for evaluating the present state of the atmosphere and for confirming model results; models require representations of these properties for current, past and future assessments.

Table 1: Climatically-relevant aerosol properties.
Characteristic Physical/chemical property Optical property Governing factors
Abundance Concentration Optical depth º Emissions, atmospheric formation, transport, removal
Absorption Content of black carbon or other light absorbing material Single-scatter albedo ª Relative fraction of light-absorbing aerosol, absorption per unit mass
Size Fine fraction, size distribution Angstrom exponent Source type (small particles from gases or combustion; large particles from abrasion or resuspension)
Number of particles Particle number -- Emissions of particles and precursor gases, particle concentration, meteorology
Interaction with water vapor Growth when humidified, cloud condensation nuclei -- Chemical composition
º Optical depth measures the total amount of light attenuated by the aerosol in an air parcel.
ª Single-scatter albedo is the fraction of light attenuation caused by scattering (reflection): 1 for purely reflective particles and 0 for purely absorbing particles.

While physical and chemical properties (concentration, composition, particle size) govern the transport and fate of aerosols in the atmosphere, optical properties (optical depth, single-scatter albedo) control the interaction with sunlight. Although there is correspondence between the two types of properties, considerable uncertainty is introduced when one type of measurements is used to derive the other. Optical properties also depend on the wavelength of light, and measurements at a single wavelength cannot usually represent properties across the solar spectrum.

What measurement assessments are available? Satellite-borne instruments offer broad spatial coverage as well as temporal resolution, and global coverage has been increasing during recent years. Satellite-based measurements such as AVHRR and TOMS have provided global, long-term records of aerosol optical depth over the ocean, which has led to greater understanding of seasonal behavior, source regions, and evolution of major events such as dust storms (see Kaufman presentation). Recently-launched satellites are directly targeted at providing aerosol information, including aerosol properties over land; optical depth can be sensed to within about 5%. A remote-sensing network of ground-based sun photometers (AERONET), also expanding to provide global coverage, provides long-term, time-resolved values of optical properties and inferred size distributions (see Holben presentation). For example, measurements of single-scatter albedo in urban areas identifies where more-absorbing aerosol heats the atmosphere (Mexico City) or less-absorbing aerosol cools it (Maryland).

Some aerosol properties must be measured by collecting and analyzing the particles (in-situ measurements). The longest history is available from this type of measurements, from stations operated by Global Atmosphere Watch (see )Baltensperger presentation) and by the U.S. National Oceanic and Atmospheric Administration. However, spatial coverage by these stations is currently insufficient to confirm trends in all regions of the globe. Intensive field campaigns involving dozens of scientists focus on specific regions for short periods. These studies provide very detailed knowledge about aerosol properties, test the understanding of the relationship between the chemical and radiative properties, and supply detailed data for model verification. Regions examined have included those near dense populations such as the Indian Ocean and the east coasts of Asia and the United States, those dominated by open biomass burning such as Brazil and African savannas, and the remote ocean. Chemical measurements of particles, especially those of carbonaceous aerosols, need to be better standardized for comparability. Furthermore, little information is available on the temporal evolution of BC. Ice cores from the European Alps have been shown to provide such data, but need to be complemented by data from other regions (see )Baltensperger presentation).

How well can models assess direct climate forcing by aerosols? Models of chemistry and transport predict both horizontally- and vertically-distributed concentrations of trace species and of radiative effects. They also provide the only opportunities for estimating past and future climate forcing. Models are used to produce the IPCC's estimates of current direct forcing, including -0.4 W m-2 for sulfates, +0.2 W m-2 for black carbon from fossil fuel burning, -0.1 W m-2 for fossil-fuel organic carbon, and -0.2 W m-2 from biomass burning (the latter is the net effect of positive and negative forcing by black and organic carbon, respectively). The contribution of nitrates was not estimated. Recent models (see Jacobson presentation, see Seinfeld presentation) have estimated forcing by black carbon as around +0.55 W m-2. Because this forcing includes BC from both fossil-fuel and biomass burning, it is expected to be much higher than the IPCC estimate. These model results are similar to estimated forcing derived from measurements of single-scatter albedo (Hansen et al., 1997) and AERONET observations (see Sato et al. poster).

The validity of model results is limited by the quality of the inputs and by the accuracy of parameterizing physical processes. The magnitude of climate forcing depends on the location of the particles, and on the physical and chemical properties listed in Table 1. The mechanisms controlling these processes are not understood well enough to constrain the models, especially for BC. While modeled sulfate forcing usually depends on the predicted atmospheric concentration, models disagree on the amount of BC forcing even when they predict similar concentrations.

The emission source strength is an important model input and a large source of uncertainty. Emission factors (the quantity of particles emitted per activity) are not well known for some sources, such as BC from small combustion devices, or OC from biogenic sources. The uncertainty in the global average source strength of BC is approximately a factor of three, and greater in some regions such as Asia (see Bond presentation).

Models that yield BC forcing estimates in the +0.5-+0.6 W m-2 range begin with published source strengths (Cooke et al., 1999) that have since been reduced by approximately a factor of two based on new measurements and more extensive literature review of emission factors (Bond et al., 2002; Andreae and Merlet, 2001). However, regional aerosol models using these latest estimates often underpredict measured atmospheric concentrations in the Indian Ocean and Asian regions (see Carmichael presentation). Existing emission-factor measurements may be dominated by lower-emitting combustion, so that both emission factors and total emissions could be underestimated. Some activities may also be uncounted or undercounted, such as military aircraft, waste burning, or biofuel combustion. Finally, predicted atmospheric concentrations could be too low if the modeled removal rate of particles is too high, which is possible because some common removal mechanisms, such as precipitation scavenging, are crudely represented in models.

While observations can be used to constrain current forcing by aerosols without knowledge of source strengths or major source sectors, targeting of mitigation efforts and projection of future aerosol trends cannot. Source apportionment techniques, using tracers or profiles that identify certain combustion types, can target major sources of aerosol in a region that need better characterization. This approach can distinguish the fractional contribution of major sources, such as transportation, within 10% when good source profiles are available; in the absence of that information, the apportionment may be incorrect by a factor of five (see Schauer presentation).

How large are regional effects of aerosols? As observed during recent field studies, the radiative effects of aerosols on specific regions can be more than a factor of ten greater than their global average effects. In one field study just before the monsoon season, forcing in the Indian Ocean region was measured as -5 to -7 W m-2 at the top of the atmosphere and -13 to -27 W m-2 at the surface. The difference between the two measurements is likely caused by light absorption by aerosols.

Measurements in China between 1960 and 1990 show a rapid increase in aerosols in heavily populated regions. The 25% increase in aerosol optical depth was accompanied by a 35% reduction in visibility, a 20% decrease in direct solar radiation, and a decrease in temperature that matched the spatial pattern of the optical-depth increase. Estimates of the forcing in the most highly affected region, the Yangtze River delta, range from -10 to -30 W m-2 depending on the season.

What are the "indirect" climatic effects of aerosols? So-called "indirect" climate effects are related to cloud behavior or properties. These effects depend greatly on the number of particles emitted in the 0.1-1 µm size range (the accumulation mode); according to models, this number concentration is dominated by primary particle emissions over North America, Eurasia and Africa, and by sulfate and sea salt over most of the oceans. Models suggest that the number of accumulation-mode particles has increased by nearly 300% since pre-industrial times (see Raes presentation). The first effect to be identified was cloud brightening, where elevated concentrations of particles result in smaller cloud droplets and therefore negative forcing. This mechanism has been extensively confirmed by observations, although its magnitude is uncertain. A second hypothesis states that these smaller droplets remain in the atmosphere longer, decreasing precipitation; the longer cloud lifetimes also result in negative forcing. Although it is still unclear whether the changes in either radiative forcing or precipitation are climatically significant, their potential magnitudes could surpass those resulting from increasing greenhouse gas concentrations.

What is the response of the climate system to aerosol forcing? It has been known for several years that inclusion of the direct effect of sulfate aerosol in global climate simulations gives more realistic patterns of temperature change compared to simulations forced only by increases in greenhouse gases. More recently, shifts of circulation and rainfall due to increased aerosol forcing have been postulated; although these are more difficult to quantify and confirm, recent model results are qualitatively similar to observations. For example, mean rainfall has decreased by 20-49% in the Sahel between the periods 1931-1960 and 1968-1997, and this observation is consistent with precipitation shifts due to hemispheric differences in the indirect effects of sulfate aerosols (Rotstayn and Lohmann, 2002). The observed southward shift of rainfall in China is consistent with modeled results when light-absorbing aerosols are introduced (see Menon presentation).

Can we assess past changes in climate-critical aerosols? Since "climate forcing" is generally defined as the change in radiative balance since pre-industrial times, estimating its magnitude requires assessing both pre-industrial and current conditions. Additionally, it is desirable to compare a time series of forcing with the measured temperature record, to deduce how the climate system responds to forcing. Relevant time series have been developed for greenhouse gases using ice core measurements. However, unlike greenhouse gases, aerosols are highly concentrated around source regions, and their deposition depends on atmospheric processes. For that reason, ice core records have not yielded quantitatively reconstructed atmospheric aerosol concentrations, although they do qualitatively confirm forcing trends. For example, sulfate trends in an ice core record from Switzerland match estimates of Central European emission trends, including the expected declining concentrations as sulfur controls became prevalent. Data from the same ice core matches the expected trend of black carbon emissions since about 1900. Both sulfate and nitrate trends in Greenland ice cores are less pronounced than the data from Europe, as would be expected for a record far from source regions. Further analysis of ice or lake-sediment records from a variety of regions may be critical to elucidate the histories of climate forcing agents.

What future changes are expected in aerosol concentration? Future climate forcing can be assessed only through models, and the predictions are highly sensitive to scenarios of changes in both energy consumption and technology. Sulfate emissions are predicted to remain constant over the next century, but nitrates may increase by a factor of 10 (see Seinfeld presentation). A scenario assuming business-as-usual energy use and emission factors resulted in a tenfold increase in BC emissions over the next century, while a more optimistic scenario predicted a twentyfold decrease over the same period (see Liousse presentation). These uncertainties obviously have large effects on the prediction of global climate.

What benefits accrue from reducing aerosol emissions? While there is evidence that some types of particles have greater deleterious health effects than others, it is generally agreed that reductions in fine particle emissions will have positive effects on health. Reducing emissions of aerosols or aerosol precursors will improve visibility in source regions. Because black carbon is preferentially emitted by poor combustion and small sources, actions taken to reduce emissions could improve combustion efficiency. Reduced particulate emissions will also ameliorate regional climate impacts, including effects on surface radiation, cloudiness, and precipitation.

Decreasing emissions of light-absorbing (warming) particles alone will reduce average positive climate forcing, even though the climatic effects of black carbon are not exactly comparable to those of greenhouse gases because of differences in their properties, distributions, and atmospheric behavior. Following observations that direct forcing by BC might exceed +0.5 W m-2 (Jacobson, 2000), Hansen et al. (2000) proposed that black carbon reductions could be used to slow global warming.

Considerations of human health are currently driving reductions in reflective aerosols, especially sulfates, so the present trend is to decrease emissions that provide climate cooling. Actions to reduce black carbon emissions through particulate controls, combustion improvements, or fuel switching will also be accompanied by decreases in reflective particles. For example, reduced coal use in China has decreased both black carbon and sulfate emissions over the past five years, and the estimated effect is a positive climate forcing (Streets et al., 2001). This trend may seem undesirable from a global-climate perspective, but it does have climate benefits by reducing regional perturbations. In order to compare mitigation measures, climate-change metrics that go beyond global averages are needed.

Finally, it should be noted that many health and environmental effects are not linearly related to emission source strength, so that the result of emission reductions may not be easily predictable.

How feasible are reductions of black carbon? Technology is available to reduce the emissions of black carbon from some of the largest contributors. Emissions from residential coal and biofuels (30% of total estimated emissions) can be reduced by combustion improvements or fuel-switching, both of which may result in substantial benefits in efficiency and indoor air quality. Effective exhaust treatments such as particle traps are available for diesel cars (10% of the total) and well-established particulate controls are available for stationary sources (5% of the total). Open burning of fields and forests contributes about 50% of the total; reduction of emissions in this sector will be more challenging, but experience in the United States shows that changes in practice are possible. As previously discussed, the global effect of emission reductions could be net warming or cooling depending on the fuel and the burning practice. However, these reductions will have desirable effects on regional scales.

BC reductions may also be achieved through public education and institutional changes. Because black carbon is associated with poor combustion, lowering emissions involves altering not only the combustion process, but also fuel quality, equipment maintenance, and fire-tending practices, all of which have strong behavioral elements. However, neither technology nor education will suffice to reduce emissions in the absence of institutional incentives to do so.

What benefits can be gained from cross-disciplinary dialogue? The properties of atmospheric aerosols listed in Table 1 are of interest not only to climate scientists, but also to researchers concerned about regional air quality, human health, and ecosystems. Common benefits can result from sharing information on chemical composition, physical properties, and emission rates, and from considering the information needs of these other communities when planning projects. In addition, organizations that are not traditionally scientific partners — such as programs in education, monitoring, and energy-efficiency — could provide information on source properties and receive benefits from strengthened interaction with the scientific community.


Citations above without a date refer to presentations at the workshop

  1. Andreae, M.O., and P. Merlet, Emissions of trace gases and aerosols from biomass burning, Global Biogeochemical Cycles 15, 955-966, 2001.
  2. Bond, T. C., D. G. Streets, S. M. Fernandes, S. M. Nelson, J.-H. Woo, and Z. Klimont, manuscript in preparation for J. Geophys. Res. 2002.
  3. Cooke W.F., C.Liousse, H. Cachier and J. Feichter, J. Geophys. Res. 104, 22137-22162, 1999.
  4. Hansen, J., M. Sato and R. Ruedy, J. Geophys Res. 102, 6831-6864, 1997.
  5. Hansen, J., M. Sato, R. Ruedy, A. Lacis and V. Oinas, Proc. Natl. Acad. Sci. 97, 9875-9880, 2000.
  6. Jacobson, M. Z. Nature 409, 695-697, 2001.
  7. Rotstayn, L.D. and U. Lohmann, J. Climate 15, 2103-2116, 2002.
  8. Streets, D.G., et al., Science 294, 1835-1836, 2001.

Workshop Homepage * Background
Summaries: Overview, Gases, Aerosols, Tech., Health, Agri./Eco.
Abstracts: Day 1, Day 2, Day 3, Day 4, Day 5 * Participants