1.2.1. Introduction

Although there are hundreds of sources of outdoor air pollution, the source categories that are the largest contributors to most air pollutants in many locations are: vehicle emissions; stationary power generation; other industrial and agricultural emissions; residential heating and cooking; re-emission from terrestrial and aquatic surfaces; the manufacturing, distribution, and use of chemicals; and natural processes (Unger et al., 2010). Given the large differences in the number and density of these sources as well as in their design, fuel source, and effectiveness of emission control technology, the relative contribution of these sources to air pollution concentrations and exposures varies considerably across locations.

Daily, weekly, and seasonal changes in source activity, as well as meteorological factors, can also lead to very large changes in the temporal trends in atmospheric pollutant concentrations and the relative contributions from different sources.

Sources of air pollutants can be divided into several types. These can be helpful in understanding the spatial and temporal distribution of source emissions, which has a large impact on exposures to emissions from different sources. Sources are commonly classified into three broad groups: primary, secondary, and re-emission sources. A primary source results from the direct emissions from an air pollution source. In contrast, a secondary source results from the formation of a pollutant in the atmosphere from the chemical reaction of precursors emitted from air pollution sources. Finally, a re-emission source results from primary or secondary pollutants depositing on the Earth’s terrestrial or aquatic surfaces, followed by re-emission to the atmosphere.

Not all pollutants fall exclusively into one group, but in many locations, the classification of a pollutant into these categories can provide insight into exposure gradients. Secondary and re-emission sources tend to have smaller temporal and spatial concentration gradients than primary sources, due to the physical processes controlling their emissions. Primary sources can be further subdivided into point sources, mobile sources, and area sources. Point sources’ emissions are from emissions stacks and tend to lead to very large spatial and temporal gradients in concentration. Mobile sources are associated with transportation and tend to have large spatial gradients close to roadways but tend to be more homogeneous away from roadways in urban areas. Area sources are sources with relatively dispersed emissions over large areas and lead to relatively constant source contributions over space but can have very large temporal changes in emissions. In addition, fugitive sources, including VOCs and dust, result from the leakage of gases from storage and handling facilities and the resuspension of dust, respectively. The nature of these source categories leads to source contributions and exposures that can be parameterized with physical and statistical models to represent pollutant concentrations, given knowledge of emission factors.

Estimates of the source contribution to pollutant concentrations in the atmosphere and to exposures can be obtained with transport models, receptor models, or hybrid models that integrate aspects of transport models and receptor models. Transport models use emissions inventories along with mathematical representations of wind speed and direction to estimate pollutant concentrations over time and space. Receptor models use measurements of pollutants at a given location or from personal exposure measurements to elucidate the sources of the pollutants (EPA, 2014; European Commission, 2014). Reasonable confidence in source apportionment models usually requires agreement between transport and receptor models, but this is not always achieved if the applied models are not adequately developed.

In locations or scenarios where transport and receptor models have not been developed, the use of emissions inventories and source-specific intake fractions can provide reasonable estimates of exposures and the sources of the exposures.

Table 1.3 provides a global anthropogenic emissions inventory of key global pollutants by sector in 2000. On a global average, the power and industry sectors were the two major anthropogenic sources of SO₂ emission. These two sectors together with biomass burning and on-road transportation also contributed greatly to NO_x emission. Biomass burning, household biofuel, on-road transportation, and industry were the most important sources of carbonaceous emissions, including CO, BC, OC, and VOCs (Unger et al., 2010). It is important to note that the relative source contribution and absolute source contribution to these pollutants vary considerably across different regions of the world, across urban areas, and across seasons.

Table 1.3

Global anthropogenic emissions inventory of air pollutants by sector in 2000.

1.2.2. Photochemical oxidants

Photochemical oxidants are secondary pollutants that are formed during photochemical reactions in the atmosphere. These oxidants have short lifetimes but are continuously formed and destroyed through chemical reactions, leading to pseudo-steady-state concentrations that are important for chemical processing and can be inhaled. These oxidants include ozone, hydrogen peroxide, acids, peroxyacetyl nitrate, and reactive radicals. The reactive radicals, which include hydroxyl radical, oxygen radical, hydrogen radical, and several other radicals, have very short lifetimes and are not commonly measured (Finlayson-Pitts & Pitts, 2000a). A large number of VOCs, SVOCs, and non-volatile organic compounds are also produced in photochemical smog, and some are oxidants (see Section 1.2.10). Ozone is often used as an indicator for these oxidant compounds.

Photochemical oxidants are formed in the presence of sunlight from the chemical reactions of VOCs and NO_x. A more detailed discussion of the sources of NO_x and VOCs is presented in Sections 1.2.6 and 1.2.10, respectively.

Given the nonlinear response of ozone production from the reaction of VOCs and NO_x, the relative source contributions to ozone cannot be directly scaled from the relative source contributions to VOCs and NO_x. Chemical transport models are needed to apportion the incremental ozone to sources (Cohan et al., 2005).

1.2.3. Particulate matter

The size of atmospheric particles can be related to their sources, due to the physical processes that form atmospheric particles and the atmospheric processes that control the fate and evolution of particle size distributions in the atmosphere.

Coarse PM (particles with aerodynamic diameters between 2.5 μm and 10 μm) is generated largely by physical processes, including resuspension of soil and road dust, sea spray, agricultural tilling, vehicular abrasion (i.e. tyre and brake wear), and fugitive dust emission from industrial sources.

Accumulation mode particles (particles with diameters between 0.2 μm and 2.5 μm) comprise predominantly the condensation of secondary inorganic and organic compounds and coagulated nuclei mode particles (particles with diameters < 0.2 μm). These particles comprise predominantly secondary sulfate and bisulfate ion, secondary nitrate ion, secondary ammonium ion, and carbonaceous PM from primary and secondary sources, but also include some crustal materials due to the fact that accumulation mode particles include supermicrometre particles.

Nuclei mode particles originate predominantly from combustion sources and atmospheric nucleation. They have relatively short atmospheric lifetimes before they either grow to become accumulation particles or coagulate to form accumulation particles. Nuclei mode particles tend to be enriched in carbonaceous aerosols and metals from the combustion of heavy oil and fuel as well as emissions from the high-temperature processing of metals.

Coarse PM comprises predominantly inorganic crustal materials, abrasion particles from mobile sources and industrial sources, and sea spray.

It should be noted that PM_2.5 includes nuclei mode particles and accumulation mode particles and PM₁₀ includes nuclei mode particles, accumulation mode particles, and coarse particles (Watson, 2002).

Source apportionment efforts for PM have typically been directed at source apportionment of particle mass; however, there are some studies that have been used to apportion the sources of components of PM (Querol et al., 2007; Heo et al., 2013).

Zhang et al. (2007) analysed the bulk composition of fine PM at more than 30 sites in the Northern Hemisphere, including urban, rural, and remote locations. They found that organic compounds accounted for 18–70% of the PM mass, sulfate ion accounted for 10–67%, nitrate ion accounted for a few percent to 28%, and ammonium ion accounted for 7–19% of the PM mass. EC and crustal materials are also important contributors to fine PM in the context of human exposure and health. Crustal material typically contributes 5–20% to PM_2.5 in most locations in Europe and the USA (Chow & Watson 2002; Belis et al., 2013), and EC usually contributes about 5–10% of the fine PM mass. Although PM_2.5 levels in China are much higher than those in cities in North America and Europe, the relative composition in megacities in China is similar (Chan & Yao, 2008; Cao et al., 2012). In addition, sea spray and road salt (used in cold climates to melt snow and ice on roadways) can account for up to 5–10% of fine PM mass (Chow & Watson 2002; Belis et al., 2013).

Sulfate ion in fine and ultrafine PM is predominantly from the oxidation of SO₂, which is largely from the combustion without emission controls of sulfur-containing fossil fuels. More information on the sources of SO₂ is provided in Section 1.2.4.

The contribution of nitrate ion and ammonium ion to fine PM is influenced by the fact that the two major forms of nitrate ion – nitric acid and ammonium nitrate – are semivolatile compounds, which can exist in both the gas phase and the particle phase. Atmospheric chemistry, temperature, and humidity control the rate of NO_x conversion to nitric acid. Further details are given in Section 1.2.6

The sources of carbonaceous fine PM have been a large area of research over the past decade, and the tools to understand the contribution of primary sources of carbonaceous PM and the split between primary and secondary organic aerosols are quite advanced and show reasonably good agreement (Docherty et al., 2008; Snyder et al., 2009a; Zhang et al., 2009a; Heo et al., 2013). In contrast, it is still difficult to quantify the specific sources of secondary organic aerosols at this time. The primary sources of fine particle organic aerosols are dominated by combustion sources, including gasoline-powered engines, diesel-powered engines, coal and residual oil combustion, biomass burning, and food cooking operations (Schauer et al., 1996; Bond et al., 2004). As previously noted, the distribution of sources and their fuels, operations, and degree of emission controls can have a very large impact on their relative contributions to primary organic aerosols, which can be dominated by mobile sources in cities such as Los Angeles (USA), Tel Aviv (Israel), Amman (Jordan), and Mexico City (Mexico) (Stone et al., 2008; von Schneidemesser et al., 2010; Heo et al., 2013), by biomass burning in locations such as Kathmandu (Nepal) and rural North Carolina (USA) (Sheesley et al., 2007; Stone et al., 2010), or by multiple combustion sources in locations such as Beijing (China) (Zheng et al., 2005).

EC emissions are mainly in the submicrometre range, and the contribution of EC to atmospheric PM is largely in the PM_2.5 fraction. EC is mainly from pyrolysis during combustion from sources including coal combustion, fuel oil combustion, diesel engines, poorly operating gasoline engines, and biomass burning. As PM controls are being placed on most stationary power generation sources, as well as diesel engines, in Europe, the USA, and Canada, the concentrations of EC in these locations continue to decrease. In regions of the world where diesel engine emissions are not being controlled and there are large primary emissions from residual fuel and solid fuel combustion, these sources dominate contributions to EC.

Source contributions to PM₁₀ can be represented as the sum of source contributions to fine PM plus source contributions to coarse PM. In the Los Angeles Basin (USA), coarse PM was found to have average contributions of about 50% from crustal material, 20% from secondary inorganic ions, 20% from OC, and 10% from sea spray (Cheung et al., 2011). Similar results were observed in the United Kingdom (Yin & Harrison, 2008). In locations affected by dust storms, the dust contributions to coarse PM and fine PM can be significantly larger in terms of concentrations and relative contribution.

Emissions inventory data can provide an assessment of sources of primary emissions of PM on a global or local scale (Bond et al., 2004; Corbett et al., 2007).

1.2.4. Sulfur dioxide

Natural sources of SO₂ include the atmospheric oxidation of sulfur compounds emitted from microbial activity in the ocean and from the anaerobic degradation of organic material in terrestrial environments. In some locations, such as Mexico City and parts of Japan, SO₂ emissions from volcanoes also affect urban areas and SO₂ exposures (de Foy et al., 2009; Kitayama et al., 2010).

However, in most locations in the world that are influenced by anthropogenic emissions, SO₂ emissions from natural sources are usually much lower than anthropogenic emissions. SO₂ in urban and industrialized areas is largely from the combustion without emission controls of sulfur-containing fuels and from uncontrolled metal processing facilities that roast sulfide ores to make metal oxides. Emissions inventories can provide a good understanding of the sources of SO₂, given the ability to accurately estimate sulfur contents of fuels (Bhanarkar et al., 2005; Smith et al., 2011; Ozkurt et al., 2013). Many countries have adopted regulations and technologies to reduce sulfur levels in gasoline and diesel fuels; however, there are still a large number of countries around the world that do not have good controls for SO₂ emissions and have not reduced sulfur levels in mobile-source fuels. Historically, there have been petroleum refining and coal liquefaction facilities that have removed sulfur during fuel processing and emitted it as SO₂ directly to the atmosphere. It is unclear whether such facilities are still operating, but they may be important sources in some local areas where adequate emission controls do not exist.

In addition, in some regions where coal is burned for residential heating and cooking, very high exposure to SO₂ can occur.

1.2.5. Carbon monoxide

The formation of CO is largely due to poor mixing of combustion air and combustion fuel, resulting in incomplete combustion. The dominant sources of outdoor concentrations of CO in urban areas are on-road transportation (gasoline- or diesel-powered engines) (IARC, 2013a), off-road engines, and biomass burning activity.

The use of catalytic convertors to convert emissions of CO to CO₂ for on-road gasoline-powered engines decreases CO emissions.

In rural areas and locations where biomass fuels are commonly used for residential cooking and heating, outdoor concentrations of CO are typically dominated by these biomass burning activities. Likewise, forest fires and controlled burns of vegetation can also be very large sources of CO.

Several global assessments of CO can be used to understand the regional distribution of CO sources using emissions inventory and chemical transport models (Holloway et al., 2000). On an urban scale, inverse models can be used to understand the local contributions of sources to CO (Bergamaschi et al., 2000).

1.2.6. Nitrogen oxides

Globally, the sources of NO_x are dominated by fossil fuel combustion, microbial activity in soils, and biomass burning, with smaller contributions from lightning and stratospheric oxidation of nitrous oxide (N₂O).

In urban areas, fossil fuel combustion is often the dominant source and includes stationary power generation, diesel-powered engines, and gasoline-powered engines. There has been some concern that diesel aftertreatment technologies aimed at reducing PM emissions will shift the distribution of NO_x emissions towards NO₂, which will lead to higher NO₂ exposures near roadways (Grice et al., 2009).

In rural areas where residential combustion of solid fuels is common, the residential combustion of solid fuels and microbial activity in soils are typically the dominant sources of NO_x.

Ammonia is a primary pollutant and on national scales is emitted largely as a result of agricultural practices, including direct emissions from livestock waste, emissions from spreading of manure, and emissions from the use of synthetic fertilizers (Battye et al., 2003). In urban areas, ammonia emissions are dominated by mobile-source emissions (Battye et al., 2003), which result from three-way catalytic converters over-reducing NO_x to ammonia (Fraser & Cass, 1998).

As part of the photochemical cycle, NO reacts with ozone to form NO₂, and NO₂ undergoes photolysis in the presence of sunlight to form NO. This photochemical cycle is a key component of ozone formation and the production of photochemical oxidants.

Chemical transport models that use emissions inventories have been very successful at modelling both near-roadway (Karner et al., 2010) and continental-scale NO_x concentrations (Stedman et al., 1997; Martin et al., 2003). Such models are an effective means of quantifying the sources of NO_x on different time scales for current and future scenarios.

1.2.7. Lead and other toxic metals

Non-volatile metals are components of atmospheric PM and can greatly influence its biological activity. Industrial sources can be very large sources of metals that can be found in atmospheric PM even though the metals are not major contributors to particle mass (Schauer et al., 2006; Snyder et al., 2009b). In the absence of industrial sources, roadway emissions and stationary power generation are typically the largest source of many toxic metals in the urban atmosphere. The braking systems of motor vehicles and underground public transportation emit metals that are potentially of concern for human exposure, including iron, copper, chromium, strontium, manganese, and antimony (Schauer et al., 2006; Kam et al., 2013). Stationary power generation that does not have suitable particle controls can have substantial impacts on metal concentrations and exposures. In locations where residual oils are used for heating and emission controls do not exist, very high concentration of nickel and vanadium can be found in atmospheric PM (Peltier et al., 2009). Likewise, coal fly ash can contain relatively high levels of arsenic, copper, chromium, zinc, antimony, selenium, and cadmium (Ratafia-Brown, 1994), and if the fly ash is not controlled with aftertreatment technologies, then emissions will contribute to an increased presence of toxic metals in the PM downwind of the facility. In developing countries, the uncontrolled emissions from brick kilns, waste incineration, and cement plants are important sources of metals to communities close to these facilities (Christian et al., 2010; Tian et al., 2012). There are very few comprehensive studies of the emissions inventory of fine particulate metals; Reff et al. (2009) provided an assessment of the spatially resolved emissions inventory for 10 metals classified as air toxics by the United States Environmental Protection Agency (US EPA) from 84 source categories.

1.2.8. Volatile metals, including mercury

Atmospheric mercury concentrations are largely dominated by gaseous elemental mercury (GEM), reactive gaseous mercury (RGM), and particulate mercury (Hg-P). RGM and Hg-P are formed in the atmosphere from the oxidation of GEM. Global anthropogenic emissions of mercury have been assessed by Pacyna et al. (2010). Globally, in 2005, burning of fossil fuel (mostly coal) was the largest single source of GEM emissions, accounting for about 45% of the anthropogenic emissions; artisanal/small-scale gold mining was responsible for about 18%, and industrial gold production accounted for 5–6%. Other mining and metal production activities and cement production were each responsible for about 10% of global anthropogenic releases to the atmosphere. The proportion of emissions from waste incineration and product-use sources is more difficult to estimate (Pacyna et al., 2010).

GEM is a global pollutant that has an atmospheric lifetime in the range of months to years. In most urban and rural outdoor locations, GEM levels are typically in the range of 2–10 ng/m³, and the concentrations of RGM and Hg-P are typically in the range of tens to hundreds of pictograms per cubic metre. Local sources of GEM, including anthropogenic sources and re-emissions from terrestrial and aquatic surfaces, can increase local concentrations to 5–10 ng/m³, or hundreds of nanograms per cubic metre near large mercury sources (Manolopoulos et al., 2007).

In addition to mercury, other volatile metals have been measured in the atmosphere, including alkyl-lead compounds (Wang et al., 1997), arsines and methyl arsines (Mestrot et al., 2009), and selenium compounds (Zhang et al., 2002).

1.2.9. Polycyclic aromatic hydrocarbons

Poor combustion conditions can lead to high emissions of PAHs and are often associated with liquid and solid fuel combustion. Benzo[a]pyrene (B[a]P) is a specific PAH formed mainly from the burning of organic material, such as wood, and from car exhaust fumes, especially from diesel vehicles. B[a]P pollution is predominantly a problem in countries where domestic coal and wood burning is common (EEA, 2013).

In 2007, it was estimated that the global total atmospheric emission of 16 PAHs came from residential/commercial biomass burning (60.5%), open-field biomass burning (agricultural waste burning, deforestation, and wildfire) (13.6%), and petroleum consumption by on-road motor vehicles (12.8%) (Shen et al., 2013).

1.2.10. Other organic compounds, including VOCs, SVOCs, and particulate organic matter

Thousands of organic compounds can be found in the atmosphere. They are components of fossil fuel, partially combusted components of fossil fuel, and pyrolysis products of fossil fuel; industrial chemical, food cooking, and biomass burning emissions; biogenic compounds emitted from plants; and organic compounds formed in the atmosphere (EEA, 2013; Oderbolz et al., 2013). These compounds include VOCs, non-volatile organic compounds that are present in atmospheric PM, and SVOCs that are present in both the gas phase and the particle phase. Many known or suspected carcinogens (Table 1.2) come from combustion sources; they include benzene, 1,3-butadiene, formaldehyde, acetaldehyde, acrolein, and naphthalene (EPA, 2006). Industrial facilities and consumer products are also important sources of aromatic VOCs, oxygenated VOCs, and halogenated VOCs. These chemicals include benzene, toluene, xylenes, ethylbenzene, methyl ethyl ketone, acetophenone, and trichloroethylene. In addition, some VOCs of potential concern are also formed in the atmosphere from photochemical reactions; these include formaldehyde, acetaldehyde, and nitrobenzene. There is also a group of persistent organic pollutants (POPs), which include many SVOCs such as polychlorinated biphenyls, polybrominated biphenyls, furans, and dioxins, and several pesticides and insecticides that can be directly emitted from air pollution sources or re-emitted from previous contamination through volatilization or resuspension of soil material (EPA, 2006; EEA, 2013).

The three major sources of VOCs in Asia are stationary combustion, solvent and paint use, and transportation; the proportion of each of these sources varies between 25% and 50%, depending on the region (Kurokawa et al., 2013). In Europe, solvent and product use was reported to contribute to about half of the total VOC emissions; the contributions of three other major sources of VOCs – commercial, institutional, and household energy use; road transportation; and energy production – were 10–20% each (EEA, 2013). In the USA, the relative source contribution reported in 2008 by the US EPA was 50% for transportation and 20% each for solvent use and industrial processes (EPA, 2013d).

In recent years, significant progress in the development of emissions inventories has been made, including the current and future emissions of dioxins (Quass et al., 2004). To assess the sources of organic compounds that both are formed in the atmosphere and react in the atmosphere, such as formaldehyde, chemical transport models are needed (Zheng et al., 2005). Several integrated assessments of emissions inventories of toxic organic compounds have been conducted and are used to provide an integrated risk from these sources by source and receptor (George et al., 2011; Luo & Hendryx, 2011).

1.2.11. Mineral dust and fibres

Resuspended dust from roadways, agricultural lands, industrial sources, construction sites, and deserts is a major source of PM in many regions of the world. Roadway dust also contains metals associated with motor vehicles (Schauer et al., 2006). Agricultural soils often contain metals that accumulate from fertilizer and animal waste, and the content of dusts from industrial sources and construction sites will depend on the specific process activities occurring at those facilities.

Although fibres, such as asbestos, are not commonly measured in the outdoor atmosphere, they can be part of the atmospheric pollution mixture. The use of asbestos has been restricted or banned in many countries. However, outdoor air pollution with asbestos may still arise in some areas from releases from asbestos-containing building materials, asbestos brakes used on vehicles, and asbestos mining activity (IARC, 2012a).

1.2.12. Bioaerosols

Bioaerosols are part of the atmospheric PM. The term “bioaerosol” refers to airborne biological particles, such as bacterial cells, fungal spores, viruses, and pollens, and their products, such as endotoxins (Stetzenbach et al., 2004). A wide range of these biological materials have been measured in the outdoor atmosphere, including moulds, spores, endotoxins, viruses, bacteria, proteins, and DNA (Yeo & Kim, 2002). Knowledge about the dynamics and sources of airborne microbial populations is still scanty. Bioaerosols are believed to be ubiquitous, and studies demonstrate the long-range transport of microorganisms and biological particles in the atmosphere (Gandolfi et al., 2013). Bioaerosols may derive from many sources, for example plants, suspension of soils containing biological materials, cooking, and burning of biological materials.