Economically consistent long-term scenarios for air pollutant emissions
- First Online:
- Cite this article as:
- Smith, S.J., West, J.J. & Kyle, P. Climatic Change (2011) 108: 619. doi:10.1007/s10584-011-0219-1
- 282 Views
Pollutant emissions such as aerosols and tropospheric ozone precursors substantially influence climate. While future century-scale scenarios for these emissions have become more realistic through the inclusion of emission controls, they still potentially lack consistency between surface pollutant concentrations and regional levels of affluence. We find that the default method of scenario construction, whereby emissions factors converge to similar values in different regions, does not yield pollution concentrations consistent with historical experience. We demonstrate a methodology combining use of an integrated assessment model and a three-dimensional atmospheric chemical transport model, whereby a reference scenario is constructed by requiring consistent surface pollutant concentrations as a function of regional income over the 21st century. By adjusting air pollutant emission control parameters, we improve consistency between projected PM2.5 and economic income among world regions through time; consistency for ozone is also improved but is more difficult to achieve because of the strong influence of upwind world regions. Reference case pollutant emissions described here were used to construct the RCP4.5 Representative Concentration Pathway climate policy scenario.
Air pollutant emissions, in addition to their impacts on human health and ecosystems, impact the climate system through aerosol radiative effects, changes in cloud properties, the formation of tropospheric ozone, changes in methane lifetime, and changes in oxidant chemistry (Forster et al. 2007). Because of their health and ecosystem impacts, these emissions have come under emission controls in affluent countries and, increasingly, in developing countries. Consideration of future emissions and their impacts, therefore, depends on realistic scenarios that take into account these emission control trends.
The previous set of global scenarios from the Special Report on Emissions Scenarios (SRES; Nakicenovic and Swart 2001), while incorporating emissions controls for sulfur dioxide, were inconsistent with assumed future economic growth in terms of the global ozone levels, due to the lack of emissions controls for ozone precursors, including nitrogen oxides (NOx) and non-methane volatile organic compounds (VOCs) (Smith and Wigley 2006). Indeed, when these scenarios were later modeled using atmospheric chemical transport models, simulated ozone concentrations grew markedly, and perhaps unrealistically, around the world (Prather et al. 2003). Subsequent scenarios in the literature (van Vuuren et al. 2008) improved on these results by including pollution controls for a wider range of emissions. These scenarios have, in general, allowed emissions controls to increase with income (Smith and Wigley 2006), which can be described as a semi-Kuznets approach (Smith 2005). While these methodologies improved consistency between emissions and assumed income levels, regional pollutant levels will depend on many factors including the spatial and temporal distribution of emissions, geography, transport of pollutants from other regions, and local meteorological and climatic conditions. We would generally expect that the surface pollutant concentrations that future (more wealthy) societies would experience would be more closely related to income than emission control levels. A region that, due to geography or emissions density, has a higher pollution level would need to have stronger emission controls in order to reach the same pollutant standard as compared to a region with lower pollution levels. However, the integrated assessment models used to create century-scale, global emissions scenarios generally lack the ability to translate future air pollutant emissions into regional air pollutant concentrations.
This raises a fundamental research question: how can consistent long-term scenarios for pollutant emission be constructed, given that emissions have non-linear relationships with concentration as well as substantial cross boundary relationships (Dentener et al. 2010)? While previous scenarios (Smith et al. 2005; Smith 2005) appeared to be broadly consistent over time, it was not clear if surface pollutant levels were consistent with socio-economic assumptions. Here we develop and demonstrate a methodology to produce a set of air pollutant emissions scenarios that are consistent with projected greenhouse gas emissions and assumptions of economic growth over the 21st century. We will examine the air pollution consequences of emissions scenarios by using a global atmospheric chemical transport model. This activity was undertaken as part of the development of one of the new long-term Representative Concentration Pathway (RCP) scenarios (Thomson et al. 2011) developed for use in model inter-comparison and climate change assessment research (Moss et al. 2010).
The overall approach was to examine the surface pollutant concentrations that result from projected emissions, and iteratively change emission control levels so that concentrations are consistent with the assumed future income levels in various regions through time. The result of this procedure was a reference, no-climate policy scenario with pollutant and greenhouse gas emissions consistent with socio-economic assumptions. This consistency carries through to the RCP4.5 scenario, which was constructed by applying a climate policy to the reference scenario. The resulting reference and RCP4.5 greenhouse gas emissions are described elsewhere (Thomson et al. 2011), and we describe here the development of the pollutant emissions component of the reference scenario.
2 Pollutant emissions scenario development
Historical evidence indicates that within wealthy nations, industrialization initially led to poor air quality, which later improved as incomes increased. While there is not a unique relationship between income and pollutant levels (see reviews by Stern 2004; Carson 2010), Carson concludes, for example, that “a weak conceptual version of the EKC [Environmental Kuznets Curve] as an inverted U-shaped curve for a particular political jurisdiction” is “largely supported by the empirical evidence”. This effect can be illustrated by examining particulate concentrations across nations where it is apparent that high-income countries do not have pollutant concentrations above a certain threshold (see also Electronic Supplementary Material, ESM §S1). A scenario with high pollution levels in high-income regions is therefore inconsistent with historical experience. On this basis, we expect that developing nations that currently experience high pollution levels will reduce air pollutant concentrations once they have achieved a higher income level. While we do not enforce a single pollution vs income relationship over time, we aim to construct scenarios that are consistent with the overall trend of a declining upper threshold over time. Note that this is a somewhat conservative approach, given evidence that technological changes may allow pollution controls to be implemented at lower income levels over time (Stern 2005).
The scenario development process starts with the construction of a reference case scenario that embodies plausible socio-economic and technological developments, but no explicit climate policy. The scenario used here was developed using the Global Change Assessment Model (GCAM), a long-term integrated model incorporating energy supply and demand, agriculture, and climate (Kim et al. 2006). Details of the reference scenario are given in Thomson et al. (2011) (ESM §S3). While the reference case scenario does not include mitigation of greenhouse gas emissions, it does include consideration of pollutant emissions controls.
MOZART-2 simulations were conducted with two rounds of preliminary gridded emissions from the GCAM reference scenario (see also ESM §S5). Snapshot simulations were conducted for 2005, 2050, and 2095 using gridded anthropogenic emissions of NOx, VOCs, sulfur dioxide (SO2), carbon monoxide (CO), black carbon (BC), and organic carbon (OC). These emissions included forest and grassland fires, but not ships or aircraft (see below). The simulations used a horizontal resolution of about 2.8° and 34 vertical levels spanning the troposphere and stratosphere, with greater vertical resolution near the surface. This resolution is adequate for simulating global atmospheric pollution at the surface, but is too coarse to resolve peak concentrations in urban regions. The same meteorological inputs are used for all simulations, assuming that future concentrations are determined by changes in emissions and ignoring the possible effects of future climate change on air quality. PM2.5 is taken as the sum of sulfate, nitrate, ammonium, OC, and BC aerosols, assuming all of these species are in the PM2.5 size range, and omitting other species that are mainly influenced by natural emissions. MOZART-2 simulations with similar model configurations, meteorological inputs, and emissions were previously evaluated against observations (Horowitz et al. 2003; Horowitz 2006; Ginoux et al. 2006; West et al. 2009).
We find 2005 PM2.5 concentrations in high-income regions ranging from 5–10 μg/m3 and ozone concentrations ranging from 40–50 ppb (Fig. 1). We will define a consistent scenario as surface pollutant levels in developing countries falling within this range as incomes rise above roughly $15,000 per capita (ESM §S1). Because model results are available for only two future time points, consistency was judged by examining the graphs of pollutant concentrations plotted against income (e.g., Fig. 1, Figure ESM-6). We would expect the trend lines for each region in Fig. 1 to, approximately, go through the vertical line that indicates the current range of pollutant levels in high-income regions, as incomes rise above roughly $15,000 per capita, and use this criterion to define a scenario as having “consistency” between projected air pollutant concentrations and income.
Two iterations of this procedure were conducted. Maximum seasonal average fine PM2.5 concentrations in the first iteration fall to well below 10 μg/m3 in most world regions by 2050 (ESM §S6), due to transitions to cleaner fuels and the relative ease of controlling sulfur dioxide emissions (Smith et al. 2005). The first iteration revealed relatively high PM2.5 concentrations in East China, India, and South-East Asia, with concentration levels above the current range in high-income countries at equivalent income levels (Figure ESM-6). SO2 and OC were major contributors and emissions controls were, therefore, strengthened for OC globally and for SO2 in these specific regions. Both primary particulate and particulate precursor emissions have fallen substantially in most high-income regions since the 1970s (e.g., Smith et al. 2011; Granier et al. 2011).
Ozone concentrations in the first iteration for India and East China were higher than either Europe or the USA for an equivalent income level (ESM §S6). The assumed levels of future emissions controls were, therefore, increased for ozone precursors in India and China. Examination of the emissions data for India revealed that NOx emissions from the transport sector in India were very high in mid-century, a result of the high initial emissions factor implied by the base-year emissions inventory. A shorter transition time to implement emissions controls, relative to the default assumption, was applied to reduce these emissions more quickly (see ESM §S4). Emission control levels were increased in several other sectors for these regions to bring pollution levels down so that surface ozone concentrations would be closer to the overall trend (e.g., Fig. 1).
After this second iteration, emissions controls for industrial process emissions in most developing countries were further increased, particularly for aerosols in Asia, in order to bring pollutant levels closer in line with the historical experience. The final reference scenario used in the RCP process reflects these updates, and atmospheric simulations using these updated emissions are planned for future work.
We find that the initial GCAM scenario, with emissions factors converging to similar levels as a function of income, results in pollutant levels in Asia (China, India, South-East Asia) that are not consistent with projected incomes and historical experience. Refinement of these scenarios achieved broad consistency between regional average incomes and pollutant concentrations across all regions and through time. As seen in Fig. 1, there is a range of regional pollution levels today in high-income regions and we have aimed to keep developing country concentrations within a similar range as incomes increase there. Consistency with historical evidence is a conservative assumption given that pollutant controls may be stronger in newly developing countries due to greater knowledge of pollution impacts and diffusion of control technologies (Smith 2005). Countering this is the higher population and emissions density of many developing regions, which may make pollutant control more difficult.
Using the reference scenario analyzed here (after adjustment following the second iteration, and with no climate policy), the RCP4.5 scenario (Thomson et al. 2011) adds a climate policy such that total anthropogenic radiative forcing stabilizes at 4.5 W/m2 in 2100. The climate policy acts to shift toward technologies and practices that emit fewer greenhouse gases and sequester carbon in terrestrial ecosystems. As a result of these shifts, pollutant gas emissions are reduced below the reference case levels shown in this study, assuming the same level of pollutant emissions controls. RCP4.5 air pollutant emissions are 15–30% lower than in the reference scenario (ESM §S6). These lower air pollutant emissions represent a potential co-benefit of a climate policy in terms of reduced air pollution impacts (Syri et al. 2001; Cifuentes et al. 2001). It is also possible that pollutant emissions reductions that occur as a result of greenhouse gas mitigation could be used to achieve the same air pollution outcomes with a reduced level of explicit air pollutant controls, although that was not assumed in the RCP4.5 scenario. In such an alternative case, the co-benefit could be evaluated as the avoided cost of air pollutant emission controls (Burtraw et al. 2003).
For both PM2.5 and ozone, emissions from a variety of small sources can be significant contributors to total emissions. One third of total US NOx emissions in 2005 were estimated to be from non-road vehicles and other miscellaneous sources, which until recently have been largely unregulated (USEPA 2009). These sectors comprise a relatively small share of total greenhouse gas emissions, and are therefore not detailed in many integrated assessment models, even though they contribute substantially to air pollutant emissions. An improved understanding of these sectors may be necessary to improve both inventory estimates and future projections of emissions.
Similarly, shipping and aviation may have disproportionately large impacts on air pollution. The gridded version of GCAM projections for shipping and aviation were not available at the time of this analysis and were, therefore, not included. This is particularly important for NOx, as adding emissions from these two sectors would have increased global NOx emissions significantly in 2095 (ESM §S6). Emissions from shipping in 2095 in the final GCAM reference scenario are equal to 30% of all other surface NOx emissions despite an assumed 4-fold decrease in emissions factor over the 21st century. Impacts of NOx and SO2 from shipping emissions would be particularly important in highly populated coastal regions (Eyring et al. 2010), and the increase in aircraft emissions would also influence climate forcing (Lee et al. 2010). Emissions from these sectors in long-term scenarios should be further examined for their effects on air quality and climate.
Uncertainties result from the use of a coarse global atmospheric model that does not resolve urban concentrations well. Further investigation with higher resolution regional models is encouraged, noting that the methodology used for emissions downscaling (van Vuuren et al. 2010) may impact results at finer scales. A higher degree of temporal analysis, e.g. consideration of hourly peak concentration values, may also be useful, given that peak exposure is often also used in defining pollutant standards. We did not consider the impact of climate change on air pollutant levels (Jacob and Winner 2009), which could alter some of the results presented here. Any such effect would be smaller in the RCP4.5 scenario. Use of simplified models for deposition and impacts, such as the regional source-receptor relationships used in the RAINS and GAINS models (Schoepp et al. 1999; Amann et al. 2008; Cofala et al. 2010) may also be useful for long-term global scenario analysis in addition to their current use for short- to medium-term applications.
The work outlined here represents a methodological advance in terms of consistency between income and pollutant concentrations, and could potentially be used to examine the consistency of other pollutant emission scenarios. This approach needs to be applied in an appropriate context, in particular with regard to the new set of RCP scenarios being used for a number of model inter-comparison exercises. The RCP scenarios, such as the RCP4.5 scenario derived from the reference scenario examined here, were intended to span a range of climate forcing levels (Moss et al. 2010) and were not intended to be associated with any specific socio-economic pathway. Several socio-economic pathways could potentially be associated with a given RCP scenario, and a number of such additional pathways are under development (Kriegler et al. 2010). Air pollutant emissions, however, are not necessarily consistent with arbitrary socio-economic pathways. While high income scenarios, for example, could have either high or low greenhouse gas emissions, high-income scenarios with high pollutant emissions are unlikely. Approaches to creating consistency between greenhouse gas emissions, socio-economic conditions, and pollutant emissions (along with their climate forcing) will need to be further explored.
The authors would like to thank Jean-Francois Lamarque, Elaine Chapman, and several anonymous reviewers for helpful comments on the manuscript, and Larry Horowitz for assistance with MOZART-2 and its inputs. This work was supported by the Department of Energy Office of Science, and by an EPA STAR grant #834285.