Volume 11, Issue 6 e2022EF003185
Research Article
Open Access

Shift in Peaks of PAH-Associated Health Risks From East Asia to South Asia and Africa in the Future

Sijia Lou

Corresponding Author

Sijia Lou

School of Atmospheric Sciences, Joint International Research Laboratory of Atmospheric and Earth System Sciences, Jiangsu Provincial Collaborative Innovation Center of Climate Change, Nanjing University, Nanjing, China

Frontiers Science Center for Critical Earth Material Cycling, Nanjing University, Nanjing, China

Correspondence to:

S. Lou and M. Shrivastava,

[email protected];

[email protected]

Contribution: Conceptualization, Formal analysis, Methodology, Writing - original draft, Writing - review & editing

Search for more papers by this author
Manish Shrivastava

Corresponding Author

Manish Shrivastava

Pacific Northwest National Laboratory, Richland, WA, USA

Correspondence to:

S. Lou and M. Shrivastava,

[email protected];

[email protected]

Contribution: Conceptualization, Formal analysis, Methodology, Writing - original draft, Writing - review & editing

Search for more papers by this author
Aijun Ding

Aijun Ding

School of Atmospheric Sciences, Joint International Research Laboratory of Atmospheric and Earth System Sciences, Jiangsu Provincial Collaborative Innovation Center of Climate Change, Nanjing University, Nanjing, China

Contribution: Writing - review & editing

Search for more papers by this author
Richard C. Easter

Richard C. Easter

Pacific Northwest National Laboratory, Richland, WA, USA

Contribution: Writing - review & editing

Search for more papers by this author
Jerome D. Fast

Jerome D. Fast

Pacific Northwest National Laboratory, Richland, WA, USA

Contribution: Writing - review & editing

Search for more papers by this author
Philip J. Rasch

Philip J. Rasch

Pacific Northwest National Laboratory, Richland, WA, USA

Contribution: Writing - review & editing

Search for more papers by this author
Huizhong Shen

Huizhong Shen

School of Environmental Science and Engineering, Southern University of Science and Technology, Shenzhen, China

Contribution: Methodology, Writing - review & editing

Search for more papers by this author
Staci L. Massey Simonich

Staci L. Massey Simonich

Department of Chemistry and Department of Environmental and Molecular Toxicology, Oregon State University, Corvallis, OR, USA

Contribution: Writing - review & editing

Search for more papers by this author
Steven J. Smith

Steven J. Smith

Pacific Northwest National Laboratory, Joint Global Change Research Institute, College Park, MD, USA

Contribution: Methodology, Writing - review & editing

Search for more papers by this author
Shu Tao

Shu Tao

Laboratory for Earth Surface Processes, College of Urban and Environmental Sciences, Peking University, Beijing, China

Contribution: Writing - review & editing

Search for more papers by this author
Alla Zelenyuk

Alla Zelenyuk

Pacific Northwest National Laboratory, Richland, WA, USA

Contribution: Writing - review & editing

Search for more papers by this author
First published: 31 May 2023
Citations: 1
This article was corrected on 3 JUL 2023. See the end of the full text for details.

Abstract

Lung cancer risk from exposure to ambient polycyclic aromatic hydrocarbons (PAHs) is expected to change significantly by 2050 compared to 2008 due to changes in climate and emissions. Integrating a global atmospheric chemistry model, a lung cancer risk model, and plausible future emissions trajectories of PAHs, we assess how global PAHs and their associated lung cancer risk will likely change in the future. Benzo(a)pyrene (BaP) is used as an indicator of cancer risk from PAH mixtures. From 2008 to 2050, the population-weighted global average BaP concentrations under all RCPs consistently exceeded the WHO-recommended limits, primarily attributed to residential biofuel use. Peaks in PAH-associated incremental lifetime cancer risk shift from East Asia (4 × 10−5) in 2008 to South Asia (mostly India, 2–4 × 10−5) and Africa (1–2 × 10−5) by 2050. In the developing regions of Africa and South Asia, PAH-associated lung cancer risk increased by 30–64% from 2008 to 2050, due to increasing residential energy demand in households for cooking, heating, and lighting, the continued use of traditional biomass use, increases in agricultural waste burning, and forest fires, accompanied by rapid population growth in these regions. Due to more stringent air quality policies in developed countries, their PAH lung cancer risk substantially decreased by ∼80% from 2008 to 2050. Climate change is likely to have minor effects on PAH lung cancer risk compared to the impact of emissions. Future policies, therefore, need to consider efficient combustion technologies that reduce air pollutant emissions, including incomplete combustion products such as PAH.

Key Points

  • Population-weighted global average BaP concentrations under all RCPs consistently exceeded the WHO-recommended limits from 2008 to 2050

  • Peaks in PAH-associated ILCR shift from East Asia in 2008 to South Asia and Africa by 2050 mainly due to changes in traditional biofuel use

  • Policies that encourage using clean energy and complete combustion technologies will help mitigate health risks from PAHs

Plain Language Summary

Polycyclic aromatic hydrocarbons (PAHs) are inevitable by-products in the combustion processes of organic matter, and are contaminants of global concern because they increase the risk of lung cancer and are detrimental to human health and the ecosystem. While high concentrations of PAHs were already measured in 2008, future changes in energy use, land use, and climate policy may alter the PAHs concentrations. In this work, we estimated how future changes in emissions and climate would affect PAH distribution and human health. We found that the peaks of PAH-associated lung cancer risks will likely shift from East Asia in 2008 to South Asia and Africa by 2050, due to increasing traditional solid biofuel use with rapid population growth in these regions. Our work implies that developing efficient combustion technologies and reducing traditional biomass fuels in the future are needed in South Asia and Africa to avoid the projected deterioration of air quality and human health.

1 Introduction

Polycyclic aromatic hydrocarbons (PAHs) are unavoidable by-products from combustion processes involving organic matter. They are contaminants of global concern, and several PAHs are persistent organic pollutants in the atmosphere that increase the risk of lung cancer in humans (Boffetta et al., 1997; Chen & Liao, 2006; Muir et al., 2019; Perera, 1997). As one of the most carcinogenic PAHs, benzo(a)pyrene (BaP) is commonly used as an indicator of cancer risk from PAH mixtures (Boruvkova, 2015; Delgado-Saborit et al., 2011; IARC, 2021). High concentrations of BaP (>1 ng m−3) have been measured in several megacities across Asia, Africa, Europe, and North America (Bostrom et al., 2002EMEPIADN; Mu et al., 2018; Shen et al., 2014), with biofuel combustion dominating the global BaP emission budget (Shen et al., 2013; Xu et al., 2006). Approximately 60% of total atmospheric BaP emissions are from residential indoor biomass burning, while deforestation and wildfires contribute another 14% (Shen et al., 2013). In addition, Shen et al. (2013) suggested that the remaining 26% of the total atmospheric BaP emissions come from the use of fossil fuels, including industry (12%), residential fossil fuels (10%), and transportation (4%).

To mitigate human health risks from human exposure to PAHs, it is essential to understand how these pollutant concentrations will change in the future. Projecting air pollutant emissions into the future is more difficult than making greenhouse gas projections. While changes in energy and land use affect both greenhouse gases and air pollutants, additional dimensions complicate air pollutant projections. For example, cleaner combustion technologies substantially decrease air pollutant concerns, although their effects on greenhouse gases such as CO2 may be minimal and they may exacerbate emissions of some greenhouse gases such as CH4. Specific sectors such as residential biomass burning contribute substantially to BaP emissions, while fossil fuel use broadly across the economy is the major contributor to greenhouse gas emissions.

In 2010, biofuel use accounted for 10% of the total primary energy demand, while more than 60% was traditional solid biomass use (O'Neill et al., 2014). Traditional biofuel use is concentrated in developing regions of Africa, South Asia, and China, primarily in households for cooking, heating, and lighting (Bauer et al., 2017; Hailu & Kumsa, 2021; Tao et al., 2018). As the demand for biomass energy is projected to increase to meet increasing energy demand, and perhaps also to satisfy the Paris Agreement (Rogelj et al., 2018), BaP emissions from biofuel use are challenging to reduce by the 2050s, especially in developing regions. Therefore, regulatory control policies aimed at improving air quality, human health, and socioeconomic development could greatly change sectoral profiles and spatial distributions of BaP emissions and PAH-associated lung cancer risks.

Two different multimodel projection scenarios, the Representative Concentration Pathway (RCP) and the shared socioeconomic pathways (SSP) are often used to estimate future pollution concentrations. RCP scenarios are multimodel global scenarios of greenhouse gases and air pollutants, which were used in the Coupled Model Intercomparison Project (CMIP5) (Taylor et al., 2012) to span a range of future climate forcing levels. More recently, the SSPs have offered a broader range of future air pollution developments in different regions over the 21st century (Feng et al., 2020; Gidden et al., 2019; Kriegler et al., 2012; Rao et al., 2017). Therefore, we use RCP scenarios for assessing future emissions changes and discuss our results within the context of SSPs. Primary PAHs (including BaP), while not included in either RCP or SSP, are emitted from incomplete combustion of organic matter and are expected to covary with directly emitted organic carbon (OC) emissions in space and time. We derive future regional, sectoral and temporal changes in BaP emissions using RCP OC scenarios and discuss our results within the context of SSPs (Section 2.5). Globally, PAH emissions are expected to decrease in the future ultimately, but varied regional trends are possible (Friedman, Zhang, & Selin, 2014; Shen et al., 2013).

Additionally, future climate changes may alter the atmospheric transport and lifetime of PAHs. For example, a previous study predicted climate change would reduce BaP concentrations from 2000 to 2050 in developing regions of Africa, South Asia, and China, but increase in Western Europe and the United States (Friedman, Zhang, & Selin, 2014). In their study, adsorptive partitioning of most PAHs to black carbon (BC) was needed to reproduce the observed gas-particle phase distribution (Friedman, Pierce, & Selin, 2014; Friedman, Zhang, & Selin, 2014). However, using advanced model formulations of PAHs and SOA within a global model, we found that >90% of the particle-bound BaP is absorbed within organic matter (Shrivastava et al., 2017). In addition, recent laboratory studies have shown that the frequent presence of viscous secondary organic aerosol (SOA) coatings can shield BaP from rapid ozone oxidation (Berkemeier et al., 2016; Ringuet et al., 2012; Shiraiwa et al., 2017; Zhou et al., 2012). Due to variations in the viscosity of SOA associated with temperature and humidity, the shielding effectiveness of SOA coatings varies in the atmosphere. Warm and humid tropical regions with liquid-like SOA coatings will be less effective in shielding PAHs from ozone than cooler and drier high-latitude regions. Previously, we implemented a new modeling approach in a three-dimensional global atmospheric chemical transport model, which accounts for the shielding of BaP by viscous SOA coatings, and showed that this shielding increased BaP lifetime from ∼2 hr to 5 days (Shrivastava et al., 2017). Simulations using the new modeling approach yielded significantly better agreement with measured BaP concentrations than treatments that neglected to shield by SOA coatings. Our previous study focused on the present-day simulation of BaP concentrations. Since the effectivity of SOA coatings is sensitive to changes in temperature and humidity (Shrivastava et al., 2017; Zelenyuk et al., 2012), it is unclear how climate change will affect the atmospheric degradation of BaP.

Although modeling studies have investigated the relative importance of future emissions and future climate influence on air pollutants, they mainly focus on the aerosols and ozone concentrations (Fenech et al., 2021; Nolte et al., 2018; Rogelj et al., 2014; Silva et al., 20162017; Zhang et al., 2018; Zhao et al., 2021). Using the novel treatment that shields PAHs with viscous SOA coatings, we examine how BaP concentrations and PAH-associated health risks could change in the future due to emissions/climate change scenarios represented by various RCPs. In addition, this study helps to understand how future changes in climate policy, energy structure, and land use will affect future health risks related to PAHs.

2 Materials and Methods

2.1 Emissions

PAHs are incomplete combustion byproducts produced by burning organic matter from fossil fuel and biofuel/biomass burning sources. BaP emissions in the model are taken from the 2008 PKU-PAH global emission inventory (available from http://inventory.pku.edu.cn), which includes 69 major sources of fuel consumption and −40% to +60% uncertainty (Shen et al., 2013). The global and regional ratios of PKU-BaP to PKU-OC vary widely based on different types of fuel consumption (Table S1 in Supporting Information S1). The PKU-BaP emissions in 2008 were assigned to various emission sectors, including residential biofuel, residential fossil fuel, industry, transportation, agricultural waste burning (AWB), and open-fire biomass burning using regional ratios from Shen et al. (20132014) (Table S2 in Supporting Information S1). Since PAH emissions are a part of primary OC emissions, we scale the 2008 BaP emissions with temporally/vertically varying profiles of OC emissions based on different sectors and regions. Monthly variations in BaP emissions from residential (including both biofuel and fossil fuel), industrial, and traffic sectors follow OC emissions from the Hemispheric Transport of Air Pollution (HTAP_v2.2) 2008 emission inventory (Janssens-Maenhout et al., 2015). Daily/monthly variations in BaP emissions from open-fire biomass burning and agricultural waste burning to follow those in OC emissions from Global Fire Emissions Database (GFED) version 3 (van der Werf et al., 2010) and the Emissions Database for Global Atmospheric Research (EDGAR) v4.3 emissions inventory (Crippa et al., 2016), respectively. The 2008 emissions of OC and BC (black carbon), aerosols, and ozone precursors are from the HTAP_v2.2 2008 emission inventory (near-surface fossil fuel and biofuel emissions), EDGAR v4.3 (AWB) and GFED3.0 (open biomass burning). The details of the emission assumptions are followed by previous studies (Shrivastava et al., 20152017).

Future BaP emissions were generated assuming the same spatial and temporal trends as OC emission changes projected by different RCP scenarios (Masui et al., 2011; Riahi et al., 2011; Thomson et al., 2011; van Vuuren et al., 20102011) from 2008 to 2020 and 2050. For each grid, we derived PAH emissions for each of the six sectors from the PKU-BaP emission inventory in 2020 and 2050 by scaling these sector-specific emissions from the 2008 inventory based on ratios of different RCP 2020/2050 OC emissions to those of 2008 (Figure S1 in Supporting Information S1). Source-specific BaP scaling based on OC is performed within different source categories, including residential (both biofuel and fossil fuel), industry, transportation, AWB, and open-fire biomass burning. In this study, we assumed that the ratio of BaP to OC, which varies by sector and region, will be constant from 2008 to 2050 (Table S1 in Supporting Information S1). However, we acknowledge that variations in the ratio of BaP to OC are uncertain since BaP is a minor component of OC. Therefore, based on the changes in OC emissions, we calculated the variations in BaP emissions from 1960 to 2014, and compared them with the realistic interannual BaP emissions from the PKU-BaP emissions inventory in the same period. The modified normalized mean biases (MNMB) of −3.4% globally or −1.9% to −11.8% regionally (Figure S2 in Supporting Information S1) indicate that the BaP to OC ratio did not change substantially in the past half-century, giving us more confidence in projecting future BaP emissions.

Since RCPs do not provide 2008 emissions, we estimated 2008 emissions assuming a linear change between 2005 and 2010 (RCPs provide emissions every 5–10 years after 2000). BaP emissions vary across the different RCPs, ranging from 4.2 Gg yr−1 in RCP6 to 4.6 Gg yr−1 in RCP4.5 in 2020, and from 3.0 Gg yr−1 in RCP4.5–4.4 Gg yr−1 in RCP6 in 2050 (Figure S1 in Supporting Information S1). Residential biofuel and industry sources are major contributors to BaP emissions under RCP4.5 in 2020 due to the projected increase in biomass burning and industry emissions early in the 21st century (Bellouin et al., 2011; Thomson et al., 2011). In comparison, RCP6 projects an increase in biomass burning emissions due to an increase in the use of cropland, traditional biomass use, and coal consumption, peaking in the mid-21st century, which leads to large BaP emissions in 2050 (Masui et al., 2011).

Note that the emission budgets differ significantly between the RCP projections and HTAP_v2.2. For example, HTAP estimates global annual BC and OC emissions at the 2010 level to be 5.5 and 12.3 Tg yr−1, respectively (Janssens-Maenhout et al., 2015). In contrast, the RCPs project much higher BC and OC emissions than actual emissions, ∼8 Tg yr−1 for BC and 35 Tg yr−1 for OC at the 2010 level (IPCC, 2013). Therefore, Emissions of BC, OC, secondary aerosol precursors including SO2, NOx, NH3, emissions of semivolatile and intermediate volatility (SIVOC) organic compounds from biomass burning and fossil fuel sources, and ozone precursor emissions are also scaled from HTAP_v2.2 2008 inventory to 2020 and to 2050 based on RCP scenarios. We converted OC to primary organic aerosol emissions (OM) using a factor of 1.4, which is a conservative estimate given evidence that the ratio of OM/OC may be closer to 1.8 for biomass fuels (Klimont et al., 2017). SOA formation is treated explicitly in the global model from multigenerational aging of SIVOC organics emitted by fossil fuel and biomass burning, and biogenic VOCs, including isoprene and terpenes. The SOA formulations used in this work are the same as those in Shrivastava et al. (2015) and were shown to yield good agreement with global organic aerosol measurements.

2.2 Model Overview and Simulation Design

We used the Global Community Atmosphere Model, version 5.2 (CAM5), with a new PAH representation (Shrivastava et al., 2017) to simulate the global distribution of BaP. Gas-phase chemistry was represented by the MOZART (Model for Ozone and Related Chemical Tracers) chemical mechanism (Emmons et al., 2010). The properties and processes of aerosol species for mineral dust, black carbon, primary organic aerosols, secondary organic aerosols, sea salt, and sulfate are included in the Modal Aerosol Module (MAM3) (Liu et al., 2012), with changes to both SOAs and their precursor gases (Shrivastava et al., 2015). The model includes gas-phase reactions of BaP with hydroxyl radicals (OH) and heterogeneous reactions of particle phase BaP with ozone and OH radicals (Shrivastava et al., 2017). Gas-particle partitioning of BaP is calculated by the poly-parameter linear free energy relationship (pp-LFER) model, which includes both absorption into organic aerosols and adsorption onto soot surfaces (Shahpoury et al., 2016). Viscous SOA can significantly slow the heterogeneous oxidation of PAHs by shielding them from ozone oxidation, but this shielding is less effective in warm and/or humid locations (Mu et al., 2018; Shrivastava et al., 2017). Wet and dry deposition of particle BaP (including oxidized PAH) is treated similarly to those of other aerosol species in CAM5 (Liu et al., 2012).

To investigate the relative effects of changes in each emission sector on near-surface BaP concentrations, we performed the following model simulations (Table 1):
  1. 2008_CTRL. The standard BaP simulation for 2008 is conducted with BaP, aerosols, and ozone precursor emissions at the 2008 level. Winds and temperature are nudged toward the European Center for Medium-Range Weather Forecasts Reanalysis-Interim (ERA-Interim) reanalysis data (Dee et al., 2011).

  2. 2008_sector. Simulations to quantify individual source contributions, that is, with BaP emissions from different source sectors turned on in the model, one at a time (residential biofuel, residential fossil fuel, industry, transportation, AWB, and open-fire biomass burning) for each simulation. The aerosols and ozone precursor emissions, as well as the dynamic fields, are the same as in 2008_CTRL.

  3. RCPs_emis. Simulations to examine future BaP concentrations in 2020 and 2050 under four different RCP emissions scenarios. The emissions (e.g., aerosols, ozone precursors, etc.) in these simulations are projected to be at the 2020 and 2050 levels under four RCP scenarios. The dynamic fields are the same as those in 2008_CTRL.

  4. RCPs_sector. Simulations to quantify the individual source contributions in 2020 and 2050 under different RCP scenarios. BaP emissions from different source sectors are turned on in the model, one at a time (residential biofuel, residential fossil fuel, industry, transportation, agriculture waste burning, and open biomass burning) for each simulation. The aerosols and ozone precursor emissions are taken from four RCP scenarios in 2020 and 2050, respectively. The dynamic fields are the same as those in 2008_CTRL.

Table 1. Description of Model Simulations
Simulation Emissions Dynamic field
2008_CTRL 2008 2007–2008a
2008_sector 2008 (turn on one sector at a time) 2007–2008a
RCPs_emis 2020/2050 under four RCP scenarios 2007–2008a
RCPs_sector 2020/2050 under four RCP scenarios (turn on one sector at a time) 2007–2008a
RCP8.5_2008 2008 2007–2010b
RCP8.5_2050_Clim 2008 2047–2050b
RCP8.5_2050 2050 2047–2050b
  • a Meteorological fields from ERA-Interim (Dee et al., 2011).
  • b Meteorological fields under RCP8.5 from NCAR (NCAR, 2011).
To investigate the relative effects of changes in emissions and climate on BaP concentrations and associated health risks, we performed simulations under the RCP8.5 scenario. While RCP8.5 describes the scenario with the highest CO2 emissions from fossil fuels, it is not the worst scenario for PAHs. RCP8.5 represents the most severe future global warming scenario. More recently, the sixth phase of the Coupled Model Intercomparison Project (CMIP6) provided a series of future climate predictions based on the combined pathways of SSP and RCP (CMIP6, 2021). A previous study reported that the temperature and humidity changes in RCP8.5 were very close to those in SSP5-8.5 in CMIP6 (Zhu et al., 2021). Therefore, we chose the RCP8.5 scenario to explore the maximum impact of climate change on future BaP concentrations. The RCP8.5 simulations include the following (Table 1):
  1. RCP8.5_2008. Standard simulation to examine the impact of climate change on BaP concentrations under the RCP8.5 climate scenario. Emissions are the same as RCP8.5_2008 but with the winds and temperature nudged toward 2007–2010 fields from a CCSM4 (CAM4) RCP8.5-scenario simulation made for CMIP5 (NCAR, 2011).

  2. RCP8.5_2050_Clim. Simulation to examine BaP concentrations in 2050 under the RCP8.5 climate scenario. Emissions are the same as RCP8.5_2008 but with the winds and temperature nudged toward 2047–2050 fields from a CCSM4 (CAM4) RCP8.5-scenario simulation made for CMIP5 (NCAR, 2011).

  3. RCP8.5_2050. Simulation to examine BaP concentrations in 2050 under the RCP8.5 climate/emission scenario. BaP, aerosol, and ozone precursor emissions correspond to the 2050 projections under RCP8.5, and the winds and temperature are nudged toward the RCP8.5-scenario from 2047 to 2050 (NCAR, 2011).

All simulations for 2008 (1–4) were performed for 2 years and used ERA-Interim meteorology from January 2007 to December 2008 for nudging (Dee et al., 2011). The RCP8.5 simulations (5–7) were performed for 4 years, using CCSM4-simulated meteorology nudged from January 2007 to December 2010, and from January 2047 to December 2050 (NCAR, 2011), representing past (2008) and future (2050) climatic conditions, respectively. The first year in all simulations is for model spin-up, and only results from the second year (or second to fourth years) are used in our analyses. All simulations fixed biogenic emissions at the 2008 level.

By comparing model-simulated results from 2008_sector and 2008_CTRL, we derive contributions from each BaP emission sector, while a comparison of simulations for RCP_sector and RCP_emis represent emission sector contributions in 2050. Differences between 2008_CTRL and RCP_emis simulations yield effects of future emissions on BaP concentrations. Furthermore, the difference between RCP8.5_2050 and RCP8.5_2008 represents the combined effect of future climate and emission changes on BaP concentrations, where the impact of climate change is reflected by comparing RCP8.5_2050_Clim and RCP8.5_2008 simulations.

2.3 Global Model Downscaling Formulation

To rectify the bias due to the coarse grid resolution of the global model (2.5° × 1.9°), we downscaled the model-calculated near-surface BaP concentrations to a much higher resolution (0.1° × 0.1°), as described by a previous study (Shen et al., 2014). A weighting factor (urn:x-wiley:23284277:media:eft21326:eft21326-math-0001) for the ith 0.1° × 0.1° receiving grid was the sum of the contributions of the emissions from all 0.1° × 0.1° emission grids with nine 2.5° × 1.9° model grids (one covering the 0.1° × 0.1° receiving grid plus the eight surrounding grids), and is formulated as follows:
urn:x-wiley:23284277:media:eft21326:eft21326-math-0002
where Qj (ng/s), fj (dimensionless), and uj (m/s) are emissions of the density of the jth emission grid, wind frequency (0–1), and wind speed in directions from 1 to 16 in the jth emission grid derived from the ERA-interim reanalysis wind field, respectively; rd (s−1) is the BaP degradation rate in the gas phase by OH, and the particle phase by heterogeneous oxidation in the receiving grid from the simulation output. tji (s) and xji (m) is the distance and transport time from the jth emission grid to the ith receiving grid, respectively; σ (m) is the vertical standard deviation of the concentrations. Finally, Wi is used as a proxy to disaggregate the model-calculated concentration of each 2.5° × 1.9° model grid cell to a 0.1° × 0.1° grid. The downscaling BaP in the same simulation from Shrivastava et al. (2017) successfully reproduced the distribution and magnitude of the observed BaP with a modified normalized mean bias of −0.39.

2.4 Incremental Lifetime Cancer Risk (ILCR)

The ILCR induced by exposure to PAHs in ambient air is calculated with the cancer slope factor (CSF), lifetime average daily dose (LADD), and a factor SUS describing individual susceptibility, respectively, depending on age, gender, ethnicity and geographic region as follows (Shen et al., 2014)
urn:x-wiley:23284277:media:eft21326:eft21326-math-0003
LADD is calculated from BaP exposure concentration (C, mg/m3), inhalation rate (IR, m3/day), exposure duration (y, year), body weight (BW), and average life expectancy of the global population (LE, 70 years). ILCR in this study is a population-weighted average and represents the maximum likelihood estimate; the unit for ILCR is one death per 100,000 persons.

2.5 A Comparison Between BaP Emissions in RCPs and SSP Projections

Four RCPs were developed as part of the IPCC AR5 activities to represent emissions of future greenhouse gas and air pollutants in climate model simulations, with different assumptions about future changes in social and economic development, technological change, energy, and land use. As a result, RCPs differ in projecting biomass burning trends, energy use, and technology development (Table S3 in Supporting Information S1). Figure S3 in Supporting Information S1 plots the time evolution of BaP emissions calculated at three discrete points in time (2008, 2020, and 2050) under different RCP scenarios. Globally, PAH emissions are expected to ultimately decrease in the future, but varied regional trends are possible (Friedman, Zhang, & Selin, 2014; Shen et al., 2013). However, future PAH emissions projections are uncertain, because they depend on socioeconomic development trajectories, in particular levels of modern energy access and air pollutant abatement policies. For example, the projections of future PAH emissions used in a previous study (Friedman, Zhang, & Selin, 2014) are not consistent with estimated emission trends to date, which find a faster decrease in South Asia than in East Asia, so this study uses the SSPs (Gidden et al., 2019; Kriegler et al., 2012; Rao et al., 2017) and 2008–2014 emissions trends determined from observations and surveys (Shen et al., 2013) to provide a broader context for our results (Figure 1).

Details are in the caption following the image

Comparison of BaP emission trajectories under Representative Concentration Pathways (RCPs) with shared socioeconomic pathway (SSP) projections. Global Change Assessment Model (GCAM) SSP scenarios span a range of assumed air pollution control policies: SSP1/SSP5—strong control (green shaded area), SSP2—medium control (cyan shaded area), and SSP3/SSP4—weak control (magenta shaded area) (Calvin et al., 2017; Fujimori et al., 2017; Gidden et al., 2019; Kriegler et al., 2017; Rogelj et al., 2018). Symbols represent Benzo(a)pyrene (BaP) emissions in 2020 and 2050 scale from RCPs as used in this work. Bounding black lines represent a range that combines GCAM SSP and Coupled Model Intercomparison Project (CMIP6) harmonized data for SSPs. Colored shaded areas represent the overall range in GCAM model SSP results across the full range of climate policies from baseline (no climate policy) to the SSP-1.9 policy with large greenhouse gas emission reductions. BaP emissions from additional scenarios, for example, AIM/CGE (Asia-Pacific Integrated Model/Computable General Equilibrium, olive) and REMIND-MAgPIE (Regional Model of Investments and Development-Model of Agricultural Production and its Impact on the Environment, red) as used in CMIP6 harmonized data are shown as dashed lines. In comparison, the black line with circle markers represents BaP emissions from the PKU PAH inventory from 2008 to 2014.

The SSPs provide five possible future development trajectories that are designed to address air pollutant emissions with strong, medium, and weak pollution control goals over the 21st century (Gidden et al., 2019; Rao et al., 2017). We find that RCP-based BaP projections used in this work generally represent the middle to upper range of the (newer) SSP projections (Figure 1a). Globally integrated BaP emissions in 2050 under all RCPs are higher than those under the medium air pollution control scenarios (SSP2) (Figure 1a), which follow the current trends of emission control (Shen et al., 2013). Among all RCPs, RCP4.5 projects the lowest global BaP emissions by 2050, while RCP6 projects the highest emissions in 2050 (Figure S1a in Supporting Information S1). In contrast, RCP8.5 projections represent weak air pollution control scenarios (SSP3/4, Figures 1b and 1c) in developing regions of Africa and South Asia, and thereby project high emissions in these regions (Figures S1h–S1i in Supporting Information S1). However, in developed regions of the United States and Europe, RCP8.5 estimates the lowest regional BaP emissions (Figure S1b-c). In comparison, the recent global BaP emission trends (PKU emission inventory of 2008–2014) are closer to the most stringent air pollution control assumptions under the SSP1/5 projections, while the emission trends in developing regions, including South Asia and Africa are closer to the medium and weak air pollution control projections (SSP2/3/4) (Figure 1).

3 Results

3.1 Variation in BaP Concentrations Due To Changes in Emissions

In this study, we estimated urban population exposure to BaP by downscaling global model BaP concentration estimates from ∼200 km horizontal grid spacing to a higher resolution of ∼10 km, to resolve strong gradients and high BaP concentrations near urban areas. The World Health Organization (WHO) suggests that lifetime exposure to 0.1 ng m−3 of BaP would theoretically lead to one extra lung cancer death in 100,000 exposed individuals (Bostrom et al., 2002). Modeled BaP concentrations in 2008 over large portions of East Asia, South Asia, Southeast Asia, Europe, and parts of Russia, Africa, and North and South America exceeded the WHO guidelines (Figure 2a). The simulated global population-weighted exposure of 1.28 ng m−3 (Figure 2a) also greatly exceeds the WHO-recommended limit. Model predictions were evaluated at 69 background/remote sites (before downscaling) and 294 urban sites (after downscaling) around the world (Figure S3 in Supporting Information S1). The current model (with BaP shielded by viscous SOA coatings) agrees with field measurements, with normalized mean biases of +24.7% and +15.9% at the background and urban sites, respectively.

Details are in the caption following the image

Spatial distribution of near-surface Benzo(a)pyrene (BaP) concentration (ng m−3) in (a) 2008 and (b–e) 2050 under Representative Concentration Pathway (RCP2.6, RCP4.5, RCP6, and RCP8.5), respectively. Here, we present simulated BaP concentrations after downscaling to 0.1° × 0.1° grid spacing. White areas are grid cells with BaP concentrations <10−5 ng m−3 “PWGA” (at the top of each plot) is population-weighted global average near-surface concentrations.

Simulations using meteorological characteristics of 2008, but with changing emissions predict that the global population-weighted average BaP concentrations will exceed the 0.1 ng m−3 between 2008 and 2050 (Figures 2b–2e), although global BaP concentrations will decrease by 50–100% in 2050 compared to 2008 (except RCP6). The decrease in BaP concentrations coincides with strong OC emission reductions projected by three of the RCPs in many regions of the world, especially in some developed and moderately developed countries (such as Europe, Russia, China, and the United States) (Masui et al., 2011; Riahi et al., 2011; Thomson et al., 2011; van Vuuren et al., 20102011). However, in rapidly developing regions of the world, including South Asia and Africa, BaP emissions are projected to increase by 2050 due to local increases in cropland and pasture related to more AWB and deforestation fires, and more primary energy consumption (Liousse et al., 2014). As a result, high levels of BaP exposure are likely to persist from 2008 to 2050 under the four RCPs (Figures 2b–2e) in East Asia, South Asia, and Africa.

Figure 3 shows the spatial distribution of BaP concentrations at three points in time in each RCP simulation (2008, 2020, and 2050). Relative to 2008, global population-weighted average concentrations of BaP are estimated to decrease by ∼9% in 2020 and 41% in 2050 (Figure 3a) under the RCP4.5 and RCP8.5 scenarios. However, BaP concentrations are projected to increase in Africa (Figure 3i) due to increasing biofuel use, changes in land use (Masui et al., 2011; Riahi et al., 2011; Thomson et al., 2011; van Vuuren et al., 2010), and rapid industrialization throughout the 21st century (Liousse et al., 2014). With the fastest population growth in the world, the increasing traditional biomass use in households for cooking and lighting will largely offset the reduction in emission intensity (Meng et al., 2019; Hailu & Kumsa, 2021), consequently enhancing BaP emissions in Africa. Moreover, shifts in land use increase both AWB and forest fires in Africa, increasing BaP emissions. For example, under RCP8.5, increasing deforestation is projected from 2008 to 2050 to meet food demands with agriculture (Riahi et al., 2011). On the other hand, RCP2.6 projects major changes in land use from the forest and agricultural land in 2008 to land clearing for the cultivation of bioenergy crops in 2050 (van Vuuren et al., 2010) that will increase deforestation and abandoned agricultural land (Fargione et al., 2008; Searchinger et al., 2015; Searchinger & Heimlich, 2015). Similarly, many developing countries in South Asia, including India, could significantly increase BaP emissions, at least from 2008 to 2020 (Figure 3h), due to increases in AWB and residential primary energy consumption (van Vuuren et al., 2011). In portions of the United States, Europe, Russia, South America, East Asia, and Southeast Asia, RCPs project strong decreases in BaP concentrations from 2008 to 2020 and 2050 (Figures 3b–3g). In these regions, emission control regulations yield the adoption of better combustion technologies, such as improved cooking stoves, and a shift from coal and oil to renewable energy sources, such as wind, solar, and nuclear energy. In the RCP8.5 scenario, decreases in domestic solid biomass fuels and AWB emissions, along with broader air pollution controls (Riahi et al., 2011), are the main drivers for reducing BaP emissions in developed and moderately developed countries (Figures S4b–S4g in Supporting Information S1).

Details are in the caption following the image

Time series of population-weighted average Benzo(a)pyrene (BaP) concentrations in different regions. Line charts show the Representative Concentration Pathway (RCP)-projected, population-weighted, downscaled BaP (ng m−3) for individual regions.

In general, socioeconomic developments promote cleaner fuels and improvements in combustion technologies. Globally, air pollution controls are expected to become more stringent with rising income, but trends differ for specific regions and particular times. For example, in 2008, 55% of global BaP emissions were from the residential use of firewood and crop residues. BaP emissions from such biofuel use could remain similar to 2050 levels if energy access gains are insufficient to outpace population growth, or decrease dramatically if the modern energy transition accelerates, for example, replacing traditional stoves with improved stoves burn more efficiently or shifting to modern fuels.

3.2 Variation in PAH-Associated ILCR Due To Changes in Emissions

We use BaP as an indicator of lung cancer risk caused by exposure to all PAH mixtures (not just BaP), using a method based on epidemiological data (Shen et al., 2014). On a global population-weighted basis, ILCR is projected to exceed the WHO-acceptable guideline limit (1 death per 100,000 persons) in 2050 under all RCP scenarios (Figure 4a). The PAH-associated ILCR is projected to increase the most in Africa by 37–64% under all RCP scenarios except RCP6, followed by ∼30% in South Asia under RCP8.5 (Figures 4b and 4c). The increase in cropland (high biomass burning), residential biomass use, as well as industrialization throughout the 21st century (Glotfelty & Zhang, 2017; Liousse et al., 2014), will lead to an increase in BaP concentrations (Figures 3h and 3i), which will increase future ILCR in these developing regions.

Details are in the caption following the image

Polycyclic aromatic hydrocarbon (PAH)-associated Incremental lifetime cancer risk (ILCR) in 2008 and 2050 in different regions. Bar charts show the Representative Concentration Pathway (RCP)-projected, population-weighted, global and regional ILCR (deaths per 100,000 persons). Source contributions, including residential biofuel, residential fossil fuel, industry, transport, agricultural waste burning (AWB), and open-fire biomass burning, are shown in different colors.

Although the rate of increase of ILCR is projected to be the greatest in Africa (due to a rapid increase in human activities), the absolute ILCR in South Asia (∼3 deaths/100,000 persons) is projected to be significantly higher than in Africa (∼2 deaths/100,000 persons) even in 2050. Our results demonstrate that the greatest PAH-associated lung cancer risk in the 21st century will be localized over South Asia, Africa, and East Asia.

Residential biofuel use dominated the global lung cancer risk in 2008 (66%) (Figure 4a) due to the following factors: (a) lower combustion efficiency of biomass burning compared to other fuel types; (b) higher human exposure due to colocation of population and residential biofuel emissions. The contribution of residential biofuel use to global lung cancer risk is projected to increase in 2050 (∼70–80%). In developing regions of South Asia and Africa, increasing use of residential biofuels with low combustion efficiency co-occurs with rapidly growing populations, exacerbating human exposure to PAHs (Figure 4 and Figure S4 in Supporting Information S1).

In East Asia, PAH-associated ILCR is projected to decrease under all RCPs, except RCP6 (Figure 4d), due to the use of cleaner fuels that accompany socioeconomic development in China (Tao et al., 2018). In Russia, PAH-associated ILCR is projected to decrease substantially in 2050 (compared to 2008) due to the assumption that the implementation of air quality regulations will reduce industrial emissions (Rafaj et al., 2010; Riahi et al., 2011) (Figures S4d and S5c in Supporting Information S1). In other developed countries and regions, for example, the United States and Europe, PAH-associated ILCRs were already much lower in 2008 than in developing regions and are expected to decrease further by 2050, resulting from the decline in residential consumption (Figures S5a and S5b in Supporting Information S1). In contrast, the contribution from transportation and AWB will most likely increase.

3.3 Variation in PAH-Associated ILCR Associated With Climate and Emission

To investigate the effect of climate change on BaP concentrations, RCP8.5 was chosen as the reference scenario because it represents the most severe future global warming scenario among four RCP scenarios, significantly impacting air pollutants. In addition, RCP8.5 assumes a fragmented world that restricts international trade in energy and technology and describes an energy-intensive, fossil-based economy (Riahi et al., 2011). These assumptions are consistent with current global realities and fossil fuel production plans (https://productiongap.org/2021report/).

Due to the climate change from 2008 to 2050, BaP concentrations change substantially in different regions. In tropical areas of South Asia, Southeast Asia, and Central and South Africa, BaP concentrations are reduced by 7%–10% with increasing temperature (Figure 5a and Figure S6b in Supporting Information S1). The shielding effectiveness (from SOA coatings) declines with increasing temperature in these regions (Figure S6b in Supporting Information S1), leading to a much faster future BaP oxidation. Additional factors (including minor increases in precipitation and ozone concentrations) also contribute to a decrease in BaP concentrations (Figures S6f and S7b in Supporting Information S1) over Southeast Asia and Central and South Africa. In contrast to warm and moisture regions, BaP concentrations increase by 2–3% from 2008 to 2050 in East Asia and Europe, mainly due to a decrease in O3 concentrations and wind field convergence changes at 850 hPa (Figures S6h and S7b in Supporting Information S1). The shielding effectiveness is not sensitive to a 1–2°C increase in temperature under cold conditions, such as wintertime in East Asia and Europe. Therefore, our results indicate that although regional temperature and precipitation increases could reduce BaP concentrations, these effects appear minor compared to changes in O3 and wind field convergence anomalies (Figures S6 and S7 in Supporting Information S1).

Details are in the caption following the image

Impacts of changes in climate and emissions on (a–c) Benzo(a)pyrene (BaP) concentrations and (d–g) polycyclic aromatic hydrocarbons (PAHs)-associated incremental lifetime cancer risk (ILCR) (deaths per 100,000 persons) from 2008 to 2050 under Representative Concentration Pathway (RCP8.5). Simulated (downscaled to 0.1° × 0.1° grid spacing) spatial distribution of near-surface BaP concentration (ng m−3) in (a–c).

In this study, the effect of increasing temperature and moisture on BaP concentrations are likely upper-bound estimates (Shrivastava et al., 2017) since the heterogeneous reaction of particle-bound BaP is assumed to be completely shut off under cool and dry conditions in a manner similar to solid eicosane (a highly viscous organic coating) (Zhou et al., 2012). Nonetheless, the impact of climate change on BaP concentrations is still lower than that of changes in emissions. For example, although the maximum “climate benefit” reduces population-weighted average BaP concentrations by 10% (Figure 5a), BaP concentrations are predicted to increase significantly in most of South Asia from 2008 to 2050, especially in areas with high population densities (Figure 5c). Thus, “climate benefit” will partially offset increases in BaP concentrations in developing regions of South Asia and Africa, but emission changes will dominate future BaP trends.

We also estimate how PAH-associated health risks will change due to variations in climate and emissions (Figures 5d–5g). Overall, the global average PAH-associated ILCR is predicted to decrease by 37% due to future emissions reductions, while climate change contributes another 2% decline (Figure 5d). However, in developing regions of South Asia (mostly India) and Africa, the PAH-associated ILCR is projected to increase by ∼20% from 2008 to 2050 (Figures 5e and 5f). As mentioned in Section 3.2, incomplete combustion of fossil fuels and traditional biomass for cooking and burning associated with deforestation as cropland expands are projected to increase human health risks from these toxic components of fine particulate matter.

Compared to 2008, PAH-associated ILCR is also expected to decrease by 70–95% in developed and moderately developed regions such as the United States, Western Europe, Russia, Southeast Asia, and South America by 2050 (Figure S8 in Supporting Information S1). These considerable declines are attributed to strong regional emission reductions from the residential and industrial sectors (Figures S4 and S5 in Supporting Information S1). Similarly, the PAH-associated ILCR peaked in East Asia in 2008, twice as high as in South Asia, and will decline by 76% in 2050, primarily attributed to a decrease in residential biomass consumption in rural areas (Riahi et al., 2011; Shen et al., 2013).

4 Conclusions and Discussions

In this study, we integrate a state-of-the-art global atmospheric chemistry model and a lung cancer risk model to assess how global PAH concentrations and their associated lung cancer risks may change with respect to several plausible future emissions trajectories. We project that the global population-weighted exposure to BaP will significantly exceed the WHO-recommended limit from 2008 to 2050 under all RCP scenarios. PAH-associated lung cancer risks, which peaked in East Asia (mostly China, 4 deaths per 100,000 persons) in 2008, are likely to shift to South Asia (mostly India, ∼3 deaths per 100,000 persons) and Africa (∼2 deaths per 100,000 persons) in 2050. The increment in residential energy demand in households for cooking, heating, and lighting accompanied by rapid population growth in India and Africa, as well as the continued use of traditional biomass use, increases in agricultural waste burning, and forest fires, could lead to the increase in health risks from 2008 to 2050. Although future climate change may be beneficial for reducing PAH concentrations and their associated health risks, future PAH-associated ILCR will strongly depend on socioeconomic developments and air pollution policies.

Our analysis shows that residential biofuel combustion is connected with the greatest lung cancer risk in both the past (2008) and the future (2050). The largest use of residential biofuels occurs in currently developing regions, primarily Africa, East Asia, Southeast Asia and South Asia. Based on observed trends between 1992 and 2017, we find that East Asia will see stronger emission controls in the future (Tao et al., 2018; Zhang et al., 2019; Zheng et al., 2018), represented by SSP1/5 emissions projections (Figure 1d). For example, a recent study (Tao et al., 2018) reported a 5% per year reduction in residential biomass fuel use in rural China from 2008 to 2012. Due to rising incomes and strict air pollution control policies, rural China is switching from traditional biomass energy use to electricity and liquefied petroleum gas (LPG) for cooking and heating. Therefore, if residential biofuel emissions in China continue to decrease at a 5% per year rate up to 2050, the projected BaP emissions in East Asia would be 11% and 28% lower than projections in RCP4.5 and RCP8.5 by 2050, respectively.

In contrast with East Asia, emission control regulations in Africa and South Asia are still weak (Liousse et al., 2014; Rao et al., 2017). A recent study estimated that African OC emissions are expected to increase significantly, comprising ∼50% of global OC emissions in 2030 in the absence of regulations (Liousse et al., 2014). Moreover, another study reported that OC emissions in India increased by 5% from 2006 to 2010 (Li et al., 2017). Recent inventories and the SSP projections are consistent with the RCP data for Africa and South Asia in the near term, although there is potential for lower emissions than used here by 2050 if emission controls are substantially strengthened (Figures 1b and 1c).

Note that this study's conclusion is highly dependent on BaP concentrations, which can be affected by several factors, such as the model horizontal resolution, the VBS treatment for the secondary organic aerosols, and the particle-bound PAH degradation scheme. For example, due to the horizontal resolution of the global model, we cannot represent the high concentrations near source regions. Therefore, we downscaled global model BaP concentrations from 200 km per grid to 10 km based on the emission density, wind frequency, wind speed, and the BaP degradation rate. However, there are still uncertainties in the downscaling method because it does not consider the impact of potential complex topography on the pollutant dispersion processes.

There is also uncertainty in particle-bound PAH heterogeneous reactions. Following Shrivastava et al. (2017), we assumed that particle-bound PAHs were encapsulated by sticky secondary organic aerosols, completely preventing particle-bound PAHs from oxidizing with ozone under cool/dry conditions. However, Mu et al. (2018) suggested that a one-step function that describes particle-bound BaP degradation cannot represent the complex multiphase reactions of BaP. Therefore, they developed a new kinetic scheme ROI-T (Reactive oxygen intermediates-temperature), including the effects of temperature and humidity on SOA phase state and BaP degradation chemical reaction rate (Mu et al., 2018). Figure S9 in Supporting Information S1 compares simulated BaP with observations at 69 background sites using two different degradation treatments. The simulated BaP using our shielded treatment are higher than those using the ROI-T treatment, but closer to the observed values.

Our analysis suggests a great range of possibilities for PAH emission changes in the future driven by variations in air quality policies across different regions. Since residential biofuel dominates the PAH-associated ILCR in India and Africa, more stringent controls on residential fuel use appear critical to avoid deterioration of air quality and human health. To achieve environmental targets, our study suggests that policies that encourage a shift from traditional solid biomass-based technologies to those using higher temperatures and more complete combustion will be important.

Acknowledgments

This research was supported by the National Natural Science Foundation of China (Grant 42075095), the Laboratory Directed Research and Development program at Pacific Northwest National Laboratory (PNNL), the Energy Exascale Earth System Model (E3SM) project, the Fundamental Research Funds for the Central Universities (Grant DLTD2107), and the U.S. Department of Energy (DOE) Office of Science, Office of Biological and Environmental Research's Early Career Research program. A. Z., J. F., and M. S. also acknowledge support from the U.S. Department of Energy (DOE) Atmospheric System Research (ASR) program via the Integrated Cloud, Land-surface, and Aerosol System Study (ICLASS) Science Focus Area at PNNL. PNNL is operated for the DOE by Battelle Memorial Institute under Contract DE-AC05-76RL01830.

    Data Availability Statement

    The ERA-Interim reanalysis data and NCAR CCSM4-simulated future meteorology data are available from https://www.ecmwf.int/en/forecasts/datasets/reanalysis-datasets/era-interim and https://doi.org/10.5065/D6TH8JP5, respectively. The PKU and RCPs emission inventories are available from http://inventory.pku.edu.cn and https://tntcat.iiasa.ac.at/RcpDb/dsd?Action=htmlpage&page=welcome, respectively. The global population density in 2008 is from the Gridded Population of the World fourth version (GPWv4), available at http://sedac.ciesin.columbia.edu/data/collection/gpw-v4. The population growths for individual regions are projected by the United States Census Bureau, available from https://www.census.gov/programs-surveys/popproj.html. Additional data related to the modeling results are available at figshare data publisher: https://figshare.com/articles/dataset/Surface_NV_LFER_EDGAR_BAP_2008_nc/21200935.

    Erratum

    In the originally published article, co-authors Manish Shrivastava, Aijun Ding, Richard C. Easter, Jerome D. Fast, Philip J. Rasch, Huizhong Shen, Staci L. Massey Simonich, Steven J. Smith, Shu Tao, and Alla Zelenyuk were omitted from the Author Contributions. The Author Contributions have been corrected to read as follows: Conceptualization: Sijia Lou, Manish Shrivastava; Formal analysis: Sijia Lou, Manish Shrivastava; Methodology: Sijia Lou, Manish Shrivastava, Huizhong Shen, Steven J. Smith; Writing – original draft: Sijia Lou, Manish Shrivastava; Writing – review & editing: Sijia Lou, Manish Shrivastava, Aijun Ding, Richard C. Easter, Jerome D. Fast, Philip J. Rasch, Huizhong Shen, Staci L. Massey Simonich, Steven J. Smith, Shu Tao, Alla Zelenyuk. This may be considered the authoritative version of record.