What is Air Pollution?

Air pollutants may be defined as any substance in the air causing damage to human health and welfare and the environment. Harmful substances in the air originate from various sources, are related to various environment and health problems, and interact with each other. Because of these interactions the resultant problems are also interconnected and inevitably policies to mitigate these problems impact each other as well. Environmental problems generally associated with air pollutants are acidification—with adverse effects on ecosystems, crops, and infrastructure; eutrophication (overfertilization)—affecting mainly ecosystems; and ground-level ozone formation (smog), causing respiratory and cardiovascular problems for human health and damage to crops. Substances playing a role in these problems are sulfur dioxide (SO₂), nitrogen oxides (NO_x), carbon monoxide (CO), methane (CH₄), non-methane volatile organic compounds (NMVOCs), as well as particulate matter (PM), including soot and dust. NO_x, CO, CH₄, and NMVOCs are all ozone precursors. Depending on the composition of PM2.5 particles health risks can vary. Typical anthropogenic sources are transport, energy, industry, and agriculture (Fowler et al., 2020).

Early Air Pollution

Harmful substances originating from human activity have been in the air ever since humankind started to burn wood to prepare food and to keep warm (Makra & Brimblecombe, 2004). Traces of copper and lead in earth layers and in Greenland ice show that heavy metals have been transported through the air as a byproduct of mining and smelting activities in Europe and China as long as several thousand years ago (Hong et al., 1996). That living in the cities could be a choking experience and was bad for your health has been noted by authors through antiquity and the middle ages in various places in the world (Brimblecombe, 2011; Heidorn, 1978). In his treatise On the Regimen of Health written for one of the sons of Sultan Saladin the Great, the12th-century philosopher and doctor Maimonides (1138–1204) recommended residing at the outskirts of the city to avoid the “stagnant, thick, turbid, misty and foggy” air caused by activities of its inhabitants (Bos, 2019, pp. 112–113). In 1273, the English Parliament passed the Smoke Abatement Act, prohibiting the burning of soft coal because it was “prejudicial to health” (Heidorn, 1978). Thus, the idea that health problems, “bad air,” and human activities are connected and that something could be done about it has been around for quite a while. However, scientific exploration of the problem remained limited to isolated experiments for a long time, and it was not until 19th century that monitoring and measuring would lead to the first scientific insights in the substances and mechanisms playing a role (Fowler et al., 2020; Reis, 2017).

Early Measurements and Concern

Around the 1850s, English agriculturalists exploring fertilizing characteristics in rain (nitrogen) undertook systematic measurements near Rothamsted, a series which is still of interest in the 21st century (Miller, 1905; Perryman et al., 2018; Smith, 1872). Famously, the chemist Robert A. Smith measured various substances in the air around Manchester and London, the results and deductions of which he respectively published in a report in 1852 and his book Air and Rain: The Beginnings of a Chemical Climatology in 1872. Though in the decades after this publication there was no real follow-up to his findings. Smith’s work is seen as the start of how scientists in the 21st century understood, and still do, the issue of air pollution in general and acidification in particular (Gorham, 1982). At the same time, with increasing levels of population growth and industrial activity in North America and Europe, the increasing effects of emissions from these activities on human health, crops, buildings, and the environment could hardly be overlooked. Especially the burning of coal made the densely populated industrial areas an evidently unhealthy place to live and were cause for attempts to take action. In the United Kingdom, for example, Robert A. Smith was appointed the first chief inspector of the Alkali Inspectorate after the British government passed the Alkali Act in 1863. The Alkali Act aimed to curb the acid emissions of certain heavy industries. Inspector Smith advised industries on the best practices to do so and pushed for further regulation (Reed, 2012). In the United States, Cleveland established the Smoke Department, which between 1880 and 1918 issued various ordinances and promoted educational campaigns to reduce the amount of smoke emissions from burning coal. Likewise, in 1892 a group of businessmen in Chicago founded the Society for the Prevention of Smoke, hoping to encourage the city’s industry to install equipment to reduce smoke emissions. Unfortunately, these activities and ordinances proved only of little effectiveness (Rosen, 1995; see also Cleveland Department of Public Health, 2021). Also, in less densely populated areas damaging effects were acknowledged, like in the case of the toxic smoke from lead and zinc smelters in the United States and Canada, causing damage to crops in nearby farms. In 1924, the mining and smelter company in Trail in British Columbia, Canada, for example had to pay fines to compensate farmers for losses of crops due to deposition of sulfur and metals (Wirth, 2000). Still, measures to reduce impact of emissions mainly were limited to building higher smoke stacks, a mitigation approach already known by the ancient Romans (see, e.g., Brimblecombe, 2011), thereby turning the local air pollution issue into a regional one. In the case of the Trail smelter this also meant an international problem, because now farmers in the United States at the other side of the border were affected (Wirth, 2000). It was a long road to national and international policies.

With regard to effects on ecosystems, around 1900 Norwegian observations showed that populations of fish in lakes were diminishing. However, it was not until the 1920s that the fish expert Knut Dahl could present hypotheses about the relationship between declining trout and the acidity of water (Dahl,1927). Along a similar line, surveys in the 1920s enabled the geologist K. M. Strøm to explain acidity of Norwegian lakes from the interaction between certain soils’ characteristics in the catchment area and acid rainfall, but it was not till the 1950s that the relationships among dying fish, acidity of surface water, and acidifying compounds in precipitation could be shown (Dannevig, 1959; Strøm, 1925).

Concern and Regulation: Air Pollution as a Local Health Issue

In the meantime, a series of major smog events in the United States (in 1948 in Donora, Pennsylvania, with 20 deaths) and most notably in the United Kingdom (Great London Fog in 1952 with more than 10,000 deaths) changed attitudes toward air pollution regulation (Jacobs et al., 2018). These events and ones like them, in the case of the United States recurring smog events in Los Angeles and New York, are viewed to have increased public and political awareness of the health effects of air pollution, triggering dedicated work toward the 1956 U.K. Clean Air Act and the 1963 U.S. Clean Air Act and their successors (Rothschild, 2019).

Concern and Regulation: Air Pollution as a Transboundary Ecosystem Problem

At the end of the 1960s it was another event, albeit of a different character, that had a major impact on the way the public and policymakers perceived the transboundary character of the air pollution problem as well as the necessity for regulation. In Europe, especially Scandinavia, interest in the chemistry of the air had been slowly increasing. In 1947 systematic measuring and monitoring of the chemistry of precipitation started in Sweden, first, similar to the 19th-century measurements in Rothamsted, in the context of a network of scientists interested in the characteristics of atmospheric deposition for their fertilizing effects on crops (Grennfelt et al., 2020; Wilkening, 2004). In the 1950s this network expanded into the European Air Chemistry Network (also called IMI network after its home institution, the International Meteorologic Institute in Stockholm), which in the course of time would include more than 100 stations in various European countries, from 1957 on including Poland and the Soviet-Union (Jäger et al., 2001; Wilkening, 2004). It was based on data from this network that the Swedish scientist Svante Odén would develop his famous hypotheses that linked the increasing acidity of rainwater and surface waters to the large and increasing emissions of sulfur dioxide in Europe (Odén, 1968). Odén’s observations attracted a huge amount of public attention, not in the least because in a (his first) newspaper article of 1967 he used a map of Europe that visualized the problem for the readers. His findings, initially triggering Swedish and Norwegian public and policy interest in transboundary effects of air pollution, would become instrumental to forming a starting point for international scientific and policy cooperation in Europe (Grennfelt et al., 2020).

North America

From the beginning, Canadian and U.S. researchers have been active participants in the scientific network that was gradually built up in Europe and North America in the years preceding the signing of the Air Convention. Findings about the situation in Europe gathered in the 1977 OECD project were used to communicate the seriousness of the acidification problem and the need for long-term programs for measuring the composition of atmospheric deposition (e.g., Likens et al., 1979). In 1978 the U.S. National Atmospheric Deposition Program (NADP) was initiated. The Canadians had a similar program running from 1976, the Canadian Network for Sampling Precipitation (CAN SAP) (Likens et al., 1979). Political action or a bilateral agreement to reduce emissions, however, took more time to materialize. In the beginning, it was mainly Canada that was keen on bilateral negotiations with the United States in order to reduce the effects of acidification. About half of Canada’s acid deposition comes from sources in the United States, while only about 15% of U.S. deposition comes from Canadian sources. Also, a much higher percentage of Canadian territory is vulnerable to acid deposition (Carroll, 1982). Though negotiations started in 1978, it was only in 1991 that Canada and the United States signed the Air Quality Agreement, which aimed at reducing SO₂ and NO_x (Kakebeeke et al., 2004). Like in Europe, negotiations and the underlying science turned out to be highly sensitive and controversial (Alm, 1997; Carroll, 1983). While Canada took the lead in the so-called 30% Club of parties under the Air Convention, paving the way for the 1985 Sulphur Protocol, the United States did not sign the protocol. The United States had always been reluctant to sign any international binding agreement it was unsure it could keep, and it was no coincidence that the Air Quality Agreement was signed only after the adoption of the Clean Air Act Amendment of 1990. This amendment set a 40% reduction target on SO₂ emissions compared to 1980 and a cap on major emitters that would result in reduction of acid deposition not only in the Northeast United States but also in Canada (Cass, 2014). Also for Canada the Air Quality Agreement meant a confirmation of domestic policies. Targeting major smelting facilities and coal-burning power plants in 1985–1987, federal-provincial agreements were signed to reduce eastern Canadian SO₂ by half in 1994 compared to 1980 levels as the starting point (Kakebeeke et al., 2004). As part of Air Quality Agreement the United States and Canada furthermore agreed to sign and adopt the 1988 NOx Protocol of the Convention. In 1999 both countries signed the Gothenburg Protocol. In addition to the Air Quality Agreement, the United States and Canada implemented the provisions of the Air Convention through the Canada-U.S. Border Air Quality Strategy (since 2013) and (with Mexico) under the Environmental Cooperation Agreement (ECA; entered into force 2020).

Gathering Knowledge: Data Collection

An outstanding feature of the Air Convention, also generally seen as a crucial factor in its success, is the infrastructure in place to collect data to support the international negotiations. Negotiations on various emission reduction agreements (protocols) have coincided with a huge international effort to collect, validate, and streamline scientific information and integrate data on various aspects connected to air pollution. This data collection ranges from data on soil conditions, weather patterns, and deposition and concentrations of pollutants to estimates of emissions, control cost of those emissions, and the identification and reduction potential of available technologies. The data from this scientific assessment process serve as input for deliberation in the Working Group on Strategies, in which civil servants participate, representing governments of the countries party to the convention. Illustrative of this data-driven character of the Air Convention is the fact that its first protocol in 1983, about EMEP, was not a protocol about the reduction of a specific pollutant but about ensuring finances for information collection. Later, a Working Group on Effects and a Task Force on Integrated Assessment Modeling were formed to enable the development of effect-oriented and cost-effective policy strategies.

For the assessment of atmospheric dispersion, EMEP established meteorological synthesizing centers at the Norwegian Meteorological Institute (MSC-West) and the Hydrometeorological Service in Moscow (MSC-East), as well as a chemical coordinating center (CCC) at the Norwegian Institute for Air Research (NILU), responsible for the coordination of the chemical measurement and analysis. These centers use official emission inventories provided by national governments and evaluated by the Task Force on Emission Inventories and the Centre on Emission Inventories and Projections (CEIP) hosted by the Austrian Environmental Agency. In this way, the EMEP models have been able to relate deposition to emissions in each European country (Lövblad et al., 2004; Schneider & Schneider, 2004; Tørseth et al., 2012; see also EMEP website).

Sensitive Knowledge: Using Data

That the first protocol under the convention was one about scientific cooperation and not about emission reductions (and took four years to negotiate) was no coincidence and tells a tale about the hesitancy of policymakers to proceed to action. At the end of the 1970s and beginning of the 1980s many countries were still reluctant to discuss reduction obligations and transboundary air pollution was still viewed as mainly a problem for the Scandinavian countries. Countries like Germany and the United Kingdom were adverse to the high costs associated with reductions of emissions and the consequences for their industries and mining coal. The United States considered air pollution to be an interstate problem and not as a transboundary issue (though Canada, as net importer of pollution, had a different view). For these and other reasons, until the early 1980s the convention mainly functioned as a forum for scientific cooperation and exchange (Kakebeeke et al., 2004).

Even this exchange was not self-evident, as the context of the Cold War period made the exchange of data, for example, on emissions of pollutants or activity data of certain industries, a sensitive political issue. The parties had to not only build trust in each other’s data, but they also had to present and share information that could be strategically sensitive, like the location or specifications of power plants (Darst, 2001; Kakebeeke et al., 2004; Nordberg et al., 2004; Rindzevičiūtė, 2016; Tuinstra et al., 2006). Thus, the setting of the discussion on scientific issues was highly political.

Change came when “Waldsterben” (German for “dying woods”) became an issue of public concern in 1980s and German negotiators changed their mind. Now negotiations gained momentum, and in 1985 the first “30%” sulfur protocol was signed (Kakebeeke et al., 2004). Like the subsequent early protocols under the convention (for reducing NO_x in 1988 and VOC in 1991), it used simple “flat rate” concepts for determining the international distribution of reduction obligations, that is, all countries agreed to reduce their emissions by the same percentage or (in the case of NO_x protocol) stabilize emissions relative to a base year. The data collection efforts by EMEP helped convince negotiators that air pollution was a transboundary problem for which an international mitigation effort made sense. That said, in this early phase, diplomats were still reluctant to use the information on source-receptor relationships provided by EMEP. These were tables used and produced by the EMEP models showing the extent to which regions were sources or receptors of pollutants. Thus, the matrices show to what extent countries are responsible for pollution in other countries. The use of those matrices was received very hesitantly in the policy community because they were perceived as limiting space for maneuvering in the negotiations (Schneider & Schneider, 2004). In terms of “boundary work” between science and policy, policymakers felt that the existing division of labor between science and policy was being contested (Tuinstra et al., 2006). Therefore the EMEP steering body introduced them at first only to enable a general idea of the air quality situation in Europe. Slowly EMEP proceeded in building credibility and legitimacy in a continuous interaction with policymakers, negotiating and establishing areas of responsibility for the policy domain and the science domain and establishing the position of the source-receptor matrices (Schneider & Schneider, 2004; Tuinstra et al., 2006).

Differentiating Knowledge: Linking Data on Effects to Emission Reductions

The Second Sulphur Protocol (the Oslo Protocol), signed in 1994, was the first protocol based on a quantitative relation between agreed reductions in environmental impacts and an optimization in emission control. As an outcome, the most cost-effective distribution of efforts resulted in differences in reduction obligations among countries. Reductions obligations in 2005 compared to 1980 levels ranged from 3% for Greece to 87% for Germany (Second Sulphur Protocol, Annex II).

Also, in the multipollutant/multieffect Gothenburg Protocol (1999), reduction obligations differentiated among the countries. Such a differentiation in reduction obligations was possible because, starting from 1990, not only atmospheric dispersion of pollutants modeled under EMEP could be used but also maps of ecosystem sensitivity in various countries became available and were accepted for use in the negotiations. Again, this was facilitated by a huge data-collection effort by the participating countries, in this case organized by the Working Group on Effects (Bull et al., 2001); Hettelingh et al., 2015). The creation of an overarching integrative concept for the scientific assessment, the so-called critical loads, and not in the least the political acceptance of this simplified way to quantify ecosystem risks in negotiations played important roles in the uptake of differentiated reduction obligations (Bäckstrand, 2001, 2017; de Vries et al., 2015; Tuinstra, 2019). Furthermore, the development and use of computer-modeling tools (integrated assessment models) in order to combine and analyze all information consistently and efficiently has been instrumental. Like in the case of the political acceptance of the source-receptor matrices, one sees here the co-production of science and policy within the Air Convention at work.

Integrating Knowledge: Critical Loads and Integrated Assessment Modeling

The collection and mapping of data on sensitivity of soils to deposition of air pollutants, “critical loads,” was (and still is) coordinated by the Coordinating Centre for Effects (CCE). From 1990 to 2017 the CCE operated from the National Institute for Public Health and Environment (RIVM) in the Netherlands, and since 2018 it has been operating from the German Environment Agency (UBA). In all countries party to the convention, national focal centers are established to provide the data, and a unified methodology was developed to ensure international harmonization. Over time, various initiatives have been taken to further harmonization, including the development of a mapping manual, workshops with specialists from all over Europe, and expert training sessions. Thus, the mapping work has been and is an iterative process, and the maps are regularly updated (Hettelingh et al., 2015). Moreover, work on critical loads for ecosystems has been extended with critical level maps for NH₃ and O₃, referring to direct ecosystem and crop damage. For health, uniform exposure-response relationships for PM, NO₂, and O₃ are being used.

Integrated assessment models (IAMs) connecting the EMEP atmospheric dispersion models, the critical load maps, and databases with estimated costs of possible mitigation measures, as well as projections on activity levels and emissions, made it possible to identify cost-effective solutions. The collection of the information through the IAMs was organized through the Task Force on Integrated Assessment Modelling (TFIAM) of the Air Convention. This task force was established in 1986, before the actual negotiations on the Oslo Protocol began (Tuinstra et al., 2006). Its task was “to explore the possibilities to develop an analytical framework for a regional cost-benefit and cost-effectiveness analysis of concerted policies to control air pollution” (UN-ECE, 1986). At that moment, several models were under development at various research institutes in Europe. TFIAM discussed and compared various model approaches and made suggestions for further development. The final calculations that served as a starting point for the negotiations of the Oslo Protocol, as well as for the Gothenburg Protocol, were done by RAINS, the model of the International Institute for Applied Systems Analysis (IIASA) at Laxenburg, Austria (Alcamo et al., 1990; Amann et al., 1996; Tuinstra et al., 2006). Currently, IIASA functions as the conventions’ Centre for Integrated Assessment Modelling (CIAM) and centrally uses the GAIN -model, an extended version of the RAINS model (Amann et al., 2011). CIAM organizes bilateral consultation meetings with all parties to agree upon national input data and to gain credibility of the model output. In the past years, the modeling work has been extended, among others to include the health risks of air pollution on the local level (particulate matter) and to incorporate the relation between air pollution and climate change. Other changes have also been made, including the availability of an online version of GAINS. These extensions and changes, just like the early expansions with economic modules and the possibility to compare scenarios in the 1980s, have all been done in anticipation of, or in response to, needs of negotiators and other policymakers (Tuinstra, 2019).

Integrating Knowledge, Deliberating Solutions

While model calculations gave civil servants and negotiators insight into ranges of options and the order of magnitude of effects, this does not mean that negotiators blindly took over the modeling results in the protocols. At the same time, model output did not merely serve as a starting point for negotiations. Deliberations about possible solutions in the negotiations took place in parallel to model developments, and the development of indicators was based on the information needs of the negotiators. The development of the critical loads concept is an example of such an iterative process as well as the example of the “gap-closure” approach that was adopted in the preparations of the Oslo Protocol. At a certain moment in the negotiations, it had proved difficult to agree on which principles to base further negotiations on. It was clear that reaching critical loads everywhere would not be feasible and that, for example, a uniform reduction percentage for all countries would be far from cost-effective. At a meeting of the Working Group on Strategies in 1992, the Norwegian delegation proposed a compromise, implying a closure of the gap between current deposition and critical loads. This approach was appealing because it formed a direct link to critical loads in each EMEP grid cell and implied a kind of equity because the percentage for closing the gap would be the same everywhere. It would serve as an important point of departure to reach consensus. The modelers were asked to explore this approach, and in 1993 the negotiators choose a “60% gap closure” scenario to base further negotiations on (Maas et al., 2004; Tuinstra et al., 2006).

Gathering Knowledge: Country Participation

The credibility and legitimacy of the process of joint science and policy development within the Air Convention largely depends on the participation of the parties to the convention. The involvement of the countries in the collection of data is crucial. An important exponent of this involvement is that it simultaneously facilitates the development and maintenance of expertise in the countries themselves as well as the support of national policies (Grennfelt et al., 2020). Equally important is the enhancement of the credibility and legitimacy of the results of the international assessment effort because the countries themselves are responsible for the input (Gough et al., 1998; Kakebeeke et al., 2004; Reis et al., 2012; Tuinstra, 2019).

It should be noted, however, that over the course of time not all parties have been involved in the science and policy development under the Air Convention in the same way. Feeling that their ecosystems were evidently being affected, in the beginning it was the Nordic countries who took the lead in setting up infrastructure for data collection. They also invested in financing additional research and therefore could play an important role in the formulation of the issues. Germany followed suit only in a later stage, under pressure of public opinion, when it became more and more evident that they were affected as well and when feasible technical solutions (e.g., catalysts, flue gas denitrification, and desulfurization) came within reach (Kakebeeke et al., 2004). Southern European countries, like Spain and Italy, experienced quite different problems and had less reasons to play a pioneering role, especially in the early days. Consequently, for experts in those countries it was much harder to get support and acknowledgment from their governments for their participation in the international networks (Castells & Nijkamp, 1998; VanDeveer, 2004). Later, this changed because awareness of the dependence on international cooperation increased. For Eastern European countries, especially those that did not become a member of the European Union (EU) after the political and economic changes of the 1990s, it was difficult to keep scientific infrastructure and networks up and running, which hampered active involvement in international scientific networks (Rindzevičiūtė, 2016; VanDeveer, 2004, 2006). For this reason, the Air Convention has invested in capacity building to transfer knowledge in air quality monitoring, emission inventories, and projections and abatement strategies, particularly in the Caucasus and Central Asia. The idea behind this investment is that when emissions inventories improve, awareness of effective strategies will also increase, thus enabling the countries to meet the objectives of and obligations under the protocols, which will benefit all parties to the convention (UN-ECE, 2018).

Broadening Frames

In the course of time, the data collection and modeling work adapted to broadening frames of air pollution and links to other problems. In this way, the focus was extended from acidification to eutrophication and ozone formation to local health impacts and climate change. While this sounds self-evident, in practice it is not so straightforward. Assessment of effects of Long-range transboundary fluxes of air pollutants asks for a different kind of modeling and a different kind of data collection than assessment of, for example, effects of air pollutants on human health in cities (see, e.g., Cuvelier et al., 2007; Thunis et al., 2007). On the one hand, it means that alignment of various approaches is needed, implying both technical compromises and decisions about the relative importance of certain aspects in the negotiations. On the other hand, perhaps even more important, it means involving different communities of experts with different traditions of working. Also other levels of policymaking become relevant. In short, with the reframing of the air pollution issue, the scenes of science, policymaking, and negotiation also change. This is a phenomenon that can be observed with other environmental issues as well and asks for considerable flexibility in the structures for data collection, assessment, and negotiation in place (Halffman, 2019; Hulme, 2017, 2019). In the case of the Air Convention, the need for coordination with EU policies on health effects of air pollution stimulated the opening up of the framing of the issue and the adaptation of models, indicators, and data-collecting procedures. The orientation to health effects helped increase the relevance of the air pollution issue also in the perspective of those countries for whom the transboundary component and the aspect of degradation of ecosystems had been of less relevance before. Also, the fact that the EU had a much more powerful system for compliance in place played a role (Selin & VanDeveer, 2003; Tuinstra, 2008; Wettestad, 2002). With regard to climate change, already in quite an early stage the experts in the scientific network of the Air Convention had recognized the importance of linking emissions of air pollutants and greenhouse gases in the modelling work. CO₂, SO₂, NO_x, and PM for example all occur from the same combustion processes, and measures targeted at these processes would reduce all them simultaneously. Showing this link enabled experts to demonstrate to negotiators the cost-effectiveness and co-benefits of simultaneously tackling climate change and air pollution (Amann et al., 2011; Reis et al., 2012). A further important step in broadening the frame of the Air Convention was taken by specifically including black carbon (or soot) as a component of PM2.5 in the amended Gothenburg Protocol (2012) (UN-ECE, 2013a). Black carbon is also a short-lived climate pollutant and has a more radiative forcing power than CO₂.