1. Introduction
In 2012, the World Health Organization (WHO) estimated that 7 million premature deaths per year were caused by exposure to air pollution [
1]. Exposure to air pollution causes adverse effects such as strokes, chronic obstructive pulmonary disease (COPD), acute lower respiratory infections, and lung cancer [
2]. In addition to the alarming numbers published in the WHO report, more than 43% of Americans live in areas with poor air quality, putting their lives at risk [
3]. Air pollution costs the global economy trillions of dollars annually due to healthcare expenditures and lost work and is one of the leading causes of morbidity and mortality [
4,
5].
Sources of air pollution are of biogenic and anthropogenic origins, such as emissions resulting from energy production by combustion, industrial processes, metallurgical industry, residential heating, waste treatment, and mobile sources. Mobile sources are divided into two categories: on-road, including trucks, passenger cars, buses, and motorcycles, and non-road, including commercial marine vessels, airplanes, railroads, agricultural tools, snow equipment, and recreational vehicles. On-road mobile sources emit more than 1162 compounds into the atmosphere via tailpipe and evaporative emissions [
6,
7].
On-road mobile sources, which are the focus of this paper, also emit non-exhaust emissions from the release of gasoline vapor in the fuel storage system [
8], fluid leaks (e.g., fuel, lubricants, and refrigerants), particulate matter from brake dust and tire wear, and re-entrained road dust [
9].
Numerous peer-reviewed studies have clearly demonstrated that mobile source emissions are the most significant contributor to air pollution in urban areas, and they represent a risk to human and ecological health [
10,
11,
12,
13,
14,
15].
In addition to the previously mentioned studies, the 2014 National Air Toxics Assessment (NATA), the EPA’s most recent large-scale and recurring risk assessment, showed that the United States national cancer risk from exposure to air toxics emitted from on-road sources was approximately “4-in-1 million” [
16]. This means that if one million people were exposed to the same concentration of MSATs continuously over 70 years, four people would likely contract cancer from this exposure. On-road sources contributed to 12.43% of the 2014 NATA national cancer risk by source groups (NCRSG) as the second leading contributor.
In the United States, more than 13% of the population live, attend school, or work within 100 m of a major road [
17]. Kingsley et al. [
18] showed that in 2005–2006, more than 6 million students attended schools within 250 m of a major road.
MSATs and criteria pollutant concentrations are higher near the roadway, with the maximum concentrations typically occurring within the first 100 to 150 m of a roadway [
19,
20]. For example, according to the Mobile Source Air Toxics report published by the Health Effects Institute [
6], near-road monitoring and air monitoring network data showed that the highest levels of ambient benzene and formaldehyde concentrations were measured at urban roadside monitors.
Depending on many factors, such as the air toxics chemical structure, daily meteorological conditions, vehicle fleet mix, presence of barriers, and the terrain surrounding the roadway, ambient concentrations of MSATs can be found within 600 m of a roadway [
19]. Additionally, the EPA’s transportation conformity and hot-spot guidance recommend placing air dispersion modeling calculation points (receptors) to estimate concentration impacts at distances as close as five meters from a roadway edge [
21].
Unlike criteria pollutants, air toxics lack effective ambient air quality standards or guidelines in Canada or the United States that are directly applicable to protecting public health or the surrounding environment [
22,
23]. Moreover, regulatory agencies around the world do not monitor the vast majority of air toxics but monitor only a small fraction, less than 70, of the list of thousands of air toxics being emitted from on-road sources [
24,
25].
This work employs the chronic and acute toxicity approach [
26] to estimate the multi-pathway cancer risk and noncancer hazard from exposure to MSATs. This approach, coupled with air dispersion modeling and estimation of population exposures, addresses the absence of ambient air quality standards for most of the MSATs and the lack of monitoring by regulatory agencies by determining through modeling the health impacts resulting from extended and acute exposure to MSATs.
Humans (also known as human receptors) may come into contact with air toxics via two main exposure routes or pathways: (1) direct via inhalation and (2) indirect via ingestion of contaminated water, produce, and animals. Examples of indirect pathways include deposition of air toxics onto the soil and produce such as lettuce and tomatoes. Additionally, biotic uptake and accumulation from contaminated soil or water can occur, such as irrigation water transporting soil deposits containing air toxics and resulting in uptake by fish and livestock, including cattle, swine, chickens, and sheep. In addition, the incidental ingestion of soil itself is considered an important indirect exposure pathway. Soil contamination is attributed to direct discharge to the soil, atmospheric deposition of air toxics, or transport from other media (e.g., through runoff). Consideration of indirect exposure pathways is critical for certain air toxics. For example, the indirect risk from dioxin and furans can be orders of magnitude higher than the direct risk [
27]. Heavy-duty diesel vehicles emit furans and dioxins [
28,
29,
30,
31]. Furans and dioxins are highly toxic compounds that contain the dibenzofuran nucleus and dibenzo-p-dioxin nucleus, respectively. 2,3,7,8-Tetrachlorodibenzodioxin (TCDD) is the most toxic and most comprehensively studied dioxin [
32]. Moreover, TCDD is highly lipophilic and tends to bioaccumulate in the food chain [
33], thereby presenting a potentially elevated risk through the consumption of food exposed to TCDD, further contributing to the risk already caused by inhalation exposure to MSATs.
This work presents a novel and practical (streamlined and integrated) methodology to assess the significant human health impacts associated with chronic and acute exposure to MSATs by calculating the total inhalation cancer risk and the total inhalation hazard quotient. Furthermore, this methodology computes the cumulative cancer risk and the air toxics hazard index due to exposure via indirect pathways such as ingestion. Finally, the methodology is verified using monitored data.
Literature Review
Previous work performed on assessing human health impacts from exposure to air toxics includes the National Air Toxics Assessment [
16], Regional Air Impact Modeling Initiative [
34,
35], and MPCA Statewide Cumulative Risk Study [
36], which are briefly described here.
The most recent large-scale and recurring risk assessment is the National Air Toxics Assessment (NATA). A NATA is conducted every three years, and the most current one is from 2017. NATAs provide a snapshot of the outdoor air quality and risks associated with exposure to air toxics. The main objective of the NATA is to identify air toxics that are of the highest potential concern in terms of contribution to population risk. NATAs only focus on risks resulting from direct inhalation of air toxics rather than evaluating other indirect pathways, such as ingestion. Cancer and noncancer health effects are estimated from breathing air toxics at a constant concentration over a lifetime (i.e., 70 years). Results from the NATA are used in prioritizing pollutants and emission sources. The 2014 NATA results reported that cancer risk due to inhalation of MSATs was 12.43% of the national cancer risk by source groups [
16], and this is equal to “4 in 1 million people”. This percentage does not include the cumulative cancer risk due to indirect exposure of MSATs, nor other noncancer health impacts such as damage to the central nervous system or cardiovascular disease. A NATA is not appropriate for identifying local-scale air toxics “hot spots”, nor is it appropriate for identifying localized risks or individual risks from air toxics from near-roadway-related exposures. Furthermore, NATAs do not include risks due to non-inhalation exposure pathways.
The Regional Air Impact Modeling Initiative (RAIMI) included development of a suite of integrated software tools and reports, developed under contract by Lakes Environmental Software for the EPA Region 6 Compliance Assurance and Enforcement Division [
34]. The RAIMI program and tools were developed to conduct community-wide cumulative air dispersion and human health risk assessment modeling for hundreds of air toxics from thousands of air pollution sources. Human health risk assessment modeling involves evaluating chronic cancer risks, noncancer hazards, and short-term acute exposure and is conducted following the methodologies published in the Human Health Risk Assessment Protocol (HHRAP) for Hazardous Waste Combustion Facilities [
26]. Exposure pathways include direct inhalation and multiple indirect pathways, such as water, plant, and animal tissue ingestion. The first RAIMI pilot study was conducted in Harris County, Texas [
35], including the Houston Ship Channel industrial corridor. The most significant obstacle to the implementation of the RAIMI program was source parameterization (i.e., characterizing physical source parameters required to support air dispersion modeling). After preliminary modeling was completed, it was determined that many sources driving adverse human health risk were in fact fugitive sources, which required the development of necessary physical source parameters to model as fugitive sources instead of point sources as originally reported in the Texas Commission on Environmental Quality (TCEQ) emissions inventory database. The RAIMI was the first regional study to integrate all the necessary tools and peer-reviewed guidance required to conduct large-scale cumulative human health risk assessments. The project outcome resulted in the identification of multiple risk hotspots, where the EPA was able to utilize its regulatory authority to enforce existing permit limits and, where necessary, modify permits to ensure that emissions control or process changes were made to protect cumulative health impacts to the communities affected.
The dispersion of emissions from on-road sources cannot be accurately estimated using a single point source. This paper employs non-point sources to estimate the dispersion of pollution from a road network. Another weakness of the RAIMI program is the emissions inventory inputs. The RAIMI emissions inventories tend to follow the EPA’s AP-42 emission factors, and these emission factors are not up to date with the current science, such as improved fuel formulation with less sulfur content or efficient internal combustion engines equipped with advanced emission control technologies. The on-road mobile emissions inventory in the proposed methodology is prepared using the EPA’s state-of-the-science maintained Vehicle Emission Simulator called MOVES.
The Indiana Department of Environmental Management’s RAIMI study [
12] demonstrated that the incremental cancer risk from exposure to MSATs was “18.6 in 1 million people”, and the cancer risk drivers were formaldehyde and benzene. Moreover, the highest noncancer risk was also due to MSAT exposure, with the noncancer risk driver being acrolein. Acrolein is a common air toxic, found in urban settings, that is emitted from motor vehicles [
37].
The Minnesota Pollution Control Agency (MPCA) conducted a statewide screening level human health risk assessment study of all inventoried emission sources, including on-road mobile sources, non-road mobile sources, area (non-point sources), and permitted point sources located in Minnesota [
36]. The MPCA conducted a cumulative risk study according to the methodologies and science in the HHRAP [
26]. The statewide screening level human health risk assessment was conducted using the Minnesota statewide risk screening (MNRISKs) tool [
36]. MNRISKs automates the process of compiling the emissions inventory by incorporating data for point, area, and mobile sources, air dispersion modeling, and risk assessment.
MNRISKS incorporates the AERMOD air dispersion model (American Meteorological Society (AMS)/United States Environmental Protection Agency (EPA) Regulatory Model) and risk assessment protocols outlined in the HHRAP to predict cancer risk and noncancer hazard indices.
The cumulative human risk study is updated for the entire state every three years, with a comprehensive emissions inventory of all source categories, including more than 12,271 industrial point source emissions processes, on-road and non-road mobile sources, and 18 subcategories of area sources. More than 250 pollutants from all source categories are included in the study.
Following the study by MPCA, a collection of peer-reviewed articles was published [
11,
13,
38], confirming the necessity of evaluating emissions from on-road sources and how they contribute to adverse health effects resulting from acute and chronic exposure.
Numerous articles include the results from the two most recently published MNRISKS studies based on emissions for 2008 [
11] and 2011 [
13]. Using these studies, Pratt et al. [
11] showed that on-road mobile sources contributed to the highest cancer risks within the 2008 MNRISKS study, with a cancer risk of 3.70E-04, which is equal to an increase in cancer incidence of “370 in 1 million people”. Additionally, exposure to MSATs contributed to the highest noncancer risk.
Pratt et al. [
13] compared different source groups and concluded that nontraditional sources, such as on-road mobile sources, were important sources of human health risk. Furthermore, on-road mobile source contribution to cancer and noncancer endpoints was higher than point sources. The analysis of Pratt et al. [
13] showed that formaldehyde, acrolein, and diesel particulate matter (DPM) were the inhalation cancer risk drivers, and these air toxics are found in motor vehicle exhaust. The maximum DPM inhalation cancer risk modeled using MNRISKS-AERMOD was equal to an increase in cancer incidence of “900 in 1 million people”.
One area for improving the MNRISKS study is the better characterization of emissions and health impacts from mobile sources by introducing spatiotemporal considerations, i.e., changes in space over time. An existing limitation in current air toxics risk analyses is the practice of only employing annual averages to classify motor vehicle emissions in space but with time averaged out. Annual averages miss important pathways for mobile toxic emissions. By adding hourly spatiotemporal capabilities to existing mobile source simulations, a better ability to predict specific area and time health impacts is achievable. Lastly, it is observed that all reviewed literature employs deterministic risk models. These deterministic models include average exposure assumptions, such as an adult body weight value of 70 kg, meaning the results do not capture the range of variability in cancer or noncancer health risk.
The literature reviewed in this section is summarized in
Table 1.
2. Materials and Methods
This work presents a novel and verified methodology to quantify the myriad health risks associated with exposure to on-road mobile emissions. Methodology verification is based on validated air dispersion results and observed trends being as expected. The methodology is summarized as follows:
Estimate emissions from on-road vehicles using current U.S. regulatory models such as MOVES or similar vehicle emissions simulators.
Conduct air dispersion modeling using current U.S. regulatory air dispersion models, such as AERMOD or similar, to predict Mobile Source Air Toxics (MSATs) air concentrations and deposition to media, including soil and water.
Determine critical air concentrations for each MSAT of interest.
Identify through a multi-pathway fate-and-transport analysis the concentration of the MSATs in various exposure media, to evaluate indirect exposure pathways (e.g., deposition from air to soil, water, sediment, plants, and animal tissue) at sensitive locations of potential high impact.
At each sensitive location, determine the total dose to humans from all pathways, i.e., direct inhalation and scenario-relevant indirect pathways.
Conduct cumulative human health risk assessment to determine the total cancer and noncancer health effects for all-pathways based on the air toxics doses at each sensitive location (including acute and chronic exposure and direct and indirect exposure pathways).
Figure 1 shows the flowchart of the methodology.
Three MSATs, namely benzene, formaldehyde, and benzo(a)pyrene, were selected for this study. Though this a very small fraction of the MSATs, these three chemicals were selected due to their potency and the abundance of cancer and noncancer assessments in the literature.
Under the guidelines for carcinogen risk assessment, benzene is classified as a known human carcinogen [
39]. Moreover, the international agency for research on cancer (IARC) has determined that exposure to benzene causes acute myeloid leukemia [
40,
41].
Formaldehyde is classified as a human carcinogen as per the findings published by IARC [
42], and more recently, the United States Department of Health and Human Services (HHS) has categorized formaldehyde as a known human carcinogen [
43]. IARC and EPA both classify benzo(a)pyrene (BaP) as carcinogenic to humans [
44,
45].
Due to the proximity of human activity near busy and congested roads, humans are exposed to high ambient concentrations of MSATs such as benzene and other potent Polycyclic Aromatic Hydrocarbons (PAHs) [
46] for extended periods.
Furthermore, traffic emissions are particularly rich in ultrafine and nanoparticles that can easily enter the bloodstream and affect multiple organ systems. They can even directly penetrate the brain via the olfactory bulb [
47].
2.1. On-Road Vehicle Emissions Inventory
In this work, the vehicle emissions inventory is prepared using the EPA’s state-of-the-science Motor Vehicle Emission Simulator called MOVES [
48]. The version used in this work is MOVES2014b, embedded in the TRAQS system [
49].
MOVES2014b estimates the emission factors and emission inventories for the following pollutants: particulate matter (PM10 and PM2.5), nitrous oxides (NOx), sulfur dioxide (SO2), carbon monoxide (CO), greenhouse gases (GHGs), metals, and MSATs, which are the focus of this work. MOVES2014b is capable of modeling continuous releases of MSATs with over 50 different exhaust and evaporative species. Moreover, the simulator accounts for various fuel types (e.g., diesel, gasoline, E85) used by various mobile sources, such as cars, buses, motorcycles, and haul trucks. Furthermore, the simulator includes different emission rates for each combination of sources, age groups, and operating modes, and it accurately reflects the various vehicle operating processes, such as running exhaust, crankcase running exhaust, cold start, or extended idle, and provides estimates of bulk emissions or emission rates. It is important to note that the availability and completeness of an on-road emissions inventory are crucial to assess the deleterious effects of MSATs.
Vehicle emissions for this work were generated at the road-segment level. MOVES2014b requires data on road segments as well as vehicle classes, including motorcycles, light-duty vehicles, buses, single-unit trucks, and combination trucks. Additionally, the simulator requires information such as fuel specifications and usage, vehicle age distribution, traffic volume, and meteorology.
2.2. Roadway Geometry Characterization
One major impediment in quantifying health impacts from on-road emissions is roadway geometry characterization. MOVES2014b is geospatially unaware and does not require geographic coordinates such as the latitude and longitude to define the road segments in the real world. However, coordinates are required to perform the air dispersion analysis of the MSATs away from the roads towards human receptors. Air dispersion models like AERMOD require precise coordinates to characterize terrain, receptors, and roads, which are considered the emission source rather than the individual vehicles.
The Transportation Air Quality System (TRAQS) [
49] is used in this methodology to address roadway geometry characterization. TRAQS addresses the roadway geometry characterization limitation by enabling the user to define the roadway segments within a geographic information system (GIS), automatically saving the precise coordinates in preparation for air dispersion modeling.
Figure 2 shows the TRAQS graphical user interface (GUI) and the road segments analyzed in this work.
2.3. Air Dispersion and Deposition Modeling
The air dispersion and deposition modeling subsection is composed of two parts. The first part describes the air dispersion and deposition modeling of MSATs. In the second part, a description of the meteorological files and how they were prepared is presented.
2.3.1. Air Dispersion and Deposition Modeling
Air dispersion modeling is defined as the mathematical description of pollutant transport in the atmosphere. Air dispersion modeling is performed to simulate the transport, diffusion, and deposition of MSATs in the ambient air once emitted by the various emissions processes, such as running exhaust, extended idle exhaust, crankcase running exhaust, or fuel vapor venting.
Air dispersion models are used to predict the downwind concentrations of an emission source, provided the model is fed with all necessary data, such as source parameters like coordinates; physical characteristics like stack height, exit temperature, exit velocity, and diameter; and project-specific data like meteorological conditions, geophysical data layers (e.g., terrain, land use), and receptor information (including sensitive population and ambient air monitor locations).
Air dispersion modeling is needed to:
Predict air concentrations and deposition on vegetation, land, and water.
Determine on-road mobile source emission contributions to ambient air quality.
Better understand how MSATs amplify the deleterious effects of air pollution.
Aid in ambient air monitoring stations selection, such as identifying target road segments for near-road monitoring.
In this paper, MSATs dispersion and deposition are simulated by an extensively validated Gaussian EPA model, AERMOD (American Meteorological Society (AMS)/United States Environmental Protection Agency (EPA) Regulatory Model). AERMOD is the EPA’s preferred dispersion model for regulatory applications [
50]. In this work, a commercialized version of AERMOD called AERMOD View [
51] is used to calculate the hourly and annual air concentrations and deposition fluxes resulting from on-road emissions. AERMOD View utilizes EPA’s AERMOD model version 19191.
Our proposed methodology is flexible in incorporating more sophisticated air dispersion models, such as non-steady-state puff models, which simulate three-dimensional complex wind fields and derive meteorology from a combination of upper-air launches and point-based surface observations to estimate ambient air concentrations and deposition.
2.3.2. Meteorological Files
The Weather Research and Forecasting (WRF) model [
52] was used to generate the AERMOD meteorological files. WRF is a prognostic meteorology model developed in a collaborative partnership between the U.S. National Center for Atmospheric Research (NCAR), the National Centers for Environmental Prediction (NCEP), the Naval Research Laboratory, and others. The WRF model is a limited-area, non-hydrostatic, terrain-following sigma-coordinate model designed to simulate or predict mesoscale and regional-scale atmospheric circulation.
The air dispersion meteorological files were generated by processing the AERMET-Ready data files output by the Mesoscale Model Interface Program [
53] through the most recent version of the U.S. EPA’s AERMET meteorological pre-processor executable (Version 19191). This includes the use of the MMIF-generated AERSURFACE output file for Stage 3 surface characteristics.
The two generated meteorological data files are: surface data and profile data. The surface data file contains the hourly boundary layer parameter estimates, and the profile data file contains multiple-level observations of wind speed, wind direction, temperature, and standard deviation of the fluctuating wind components. A wind rose for the study area is shown later in the paper in the Case Study subsection.
2.4. Human Health Risk Assessment
Human health risk assessment is the scientific process that estimates the probability that adverse health effects may occur in humans resulting from exposure to environmental stressors, now or in the future. The human health risk assessment process includes the completion of four steps: (1) hazard or stressor identification, (2) exposure assessment, (3) dose-response assessment, and (4) risk characterization. Risk is an estimate of the “chance” that the exposure will cause or contribute to harmful human health effects. Risk characterization is the process whereby human health risk estimates are used to establish the presence or absence of risk and quantification of expected outcomes. In addition, risk characterization strives to identify the key factors contributing to risk in such a way that regulators, risk analysts, or air quality engineers are armed with actionable information to guide decisions regarding what actions, if any, should be taken to mitigate risk. Risk characterization includes calculating the incremental lifetime cancer risks and noncancer hazards by combining the toxicity benchmarks and exposure quantities [
54].
Cancer risk is defined as the probability of contracting cancer based on a unique set of exposure and toxicity assumptions throughout a lifetime, which is assumed to be 70 years for this research. For example, a risk of 5.0E-05 is interpreted to mean that an individual has up to a five in 100,000 increased chance of developing cancer during their lifetime from the evaluated exposure. Future research will employ a probabilistic approach for the lifetime variable (e.g., lifetime probability distribution). Noncancer hazard is an estimate of the likelihood that a human will experience noncancer health effects (e.g., neurological, cardiovascular, reproductive, respiratory, etc.) as a result of exposure to MSATs through the scenario-specific exposure pathways and routes. A hazard is calculated as a ratio of the receptor’s potential exposure relative to a standard exposure level such as a reference dose or reference concentration.
There are two primary hazard classifications based on the length of the exposure duration: acute for short term, and chronic for long term.
The multi-pathway and cumulative human health risk assessment in this paper was performed following the U.S. EPA 2005 Final Human Health Risk Assessment Protocol (HHRAP) [
26]. The HHRAP provides guidance for completing a human health risk assessment study from modeled air concentrations and deposition fluxes. HHRAP was developed for evaluating human health risks from hazardous waste combustion facilities such as incineration plants or energy generation plants. This research (utilizing source characterization for mobile sources) extended (added utilizing source characterization for mobile sources; adapted to mobile sources) HHRAP in evaluating health risks associated with exposure to MSATs. The chronic and acute dose-response values were obtained from the following sources: (1) U.S. EPA’s Integrated Risk Information System [
55], (2) California’s EPA (CalEPA) chemical database [
56], (3) CDC’s Agency for Toxic Substances and Disease Registry Toxic Substances Portal [
57]. For MSATs with multiple dose-response values from the above sources, the more conservative value was selected.
To quantify the potential health risks associated with exposure to MSATs, target levels are commonly established by the regulatory agency to gauge the magnitude of risk, which in turn influences decisions regarding the management of risk. The HHRAP does not define target risk levels, as this responsibility is left to local and state regulatory agencies, who are better equipped to consider other factors such as background concentrations. However, the U.S. EPA, through its Region 6 Office [
58], has established target risk levels within the Agency, which range as follows: (1) for carcinogenic risks, a low and likely to be a negligible risk is described as an incremental lifetime cancer risk of 1.0E-06 to less than 1.0E-05; (2) a potentially elevated risk is an incremental lifetime cancer risk greater than 1.0E-04; (3) for noncancer hazards, a low and likely to be a negligible risk is defined as being a hazard quotient ranging between 0 and 0.25. In this work, the following target levels are used to interpret the magnitude of cancer risks and noncancer hazards: a cancer risk not to exceed a target risk level of 1.0E-05, and a hazard quotient not to exceed a target risk level of 0.25.
The target risk levels are not a discrete indicator of observed harmful effects. If the modeled risk values fall within the accepted target risk levels, it might be concluded, without further analysis, that a proposed project does not present an unacceptable risk. A risk calculation that exceeds the target values triggers further consideration of the underlying scientific basis and uncertainties associated with the risk calculation.
The general equations used to calculate the cancer risk and noncancer risk in this work are:
Equation (2) is used to calculate the noncancer risk from oral exposure, whereas Equation (3) evaluates the noncancer risk from inhalation exposure. Equation (4) is used to calculate the acute (short-term) inhalation hazard quotient.
2.4.1. Cumulative Risk
Humans may be exposed to multiple carcinogenic air toxics via a single exposure pathway such as inhalation. The total cancer risk for a single pathway is calculated from the sum of cancer risks for each Mobile Source Air Toxic (MSAT) (e.g., benzene, formaldehyde, and benzo(a)pyrene). In addition, exposure may come via multiple pathways (e.g., inhalation, produce, animal tissues, water); therefore, the cumulative cancer risk is calculated from the sum of total cancer risks for each exposure pathway.
2.4.2. Risk Modeling
Due to the complex nature of MSATs fate and transport in different media such as air, water, soil, and atmospheric deposition, more than 78 fate, transport, and exposure equations are used to compute media concentrations, potential cancer risk, and noncancer health effects. Many of these equations are non-linear, such as the calculation of the cumulative soil concentration, calculation of MSATs loss due to runoff, or the calculation of aboveground produce concentration due to direct deposition of particle-phase MSATs such as benzo(a)pyrene or diesel particulate matter.
A hybrid modeling approach was used to calculate the cancer and noncancer risks and cumulative risk. The hybrid model consists of a computer program known as IRAP-h View [
59] and a Python program written by the authors. IRAP-h View follows the methodologies published in the Human Health Risk Assessment Protocol (HHRAP). IRAP-h View, also known as the risk engine or risk modeling platform in this work, was used to calculate the cancer and noncancer risks encapsulating all the guidance and procedures contained in HHRAP. The cumulative risk was calculated using the Python code.
In addition to modeled ground level air concentrations, our proposed methodology is flexible in utilizing in situ measured concentrations from near-road monitoring and air monitoring networks to estimate risk resulting from exposure to MSATs through the inhalation exposure pathway.
This paper fills gaps in the mobile emissions health impact literature, as described below:
Mobile Emissions Inventory Data Quality—This work generates the on-road vehicle emissions inventory using a state-of-the-science simulator instead of using outdated emission factors. Additionally, the methodology presented in the paper is flexible in incorporating spatiotemporal considerations to improve the characterization of emissions and health impacts.
Spatial Accuracy—The Vehicle Emission Simulator (MOVES) lacks spatial information. To fill this gap, the present work geospatially links the emissions results to air dispersion and deposition models, air toxics fate and transport analysis, and a risk modeling platform to estimate the impacts of MSATs on humans.
Allocation of Source Impacts—Previous work accounted for a lumped air concentration and could not apportion sources to specific human impacts. This work conducts a multi-pathway fate-and-transport analysis of MSATs, source by source. This allows allocating the contribution of each source’s deleterious effects from on-road mobile sources, revealing the full extent of exposure to MSATs.
4. Discussion
In this section, we discuss the results of our comprehensive multi-pathway human health risk assessment and examine the validation of our methodology and model performance. We then discuss the strengths and limitations of the methodology and the likelihood of the scenario being realistic. Finally, we conclude by examining the variability and uncertainty associated with our analysis.
4.1. Comprehensive Multi-Pathway Human Health Risk Assessment
The cumulative human health impact results based on the evaluation of modeled ambient air concentrations and deposition fluxes at the three sensitive receptor locations indicate cancer risks in excess of the target risk level 1.0E-05 (Normalized value = 1.00). A summary of the comprehensive multi-pathway human health risk assessment is presented in
Table 15 below.
It is important to note that there are uncertainties associated with computing and interpreting the cumulative ELCRs and noncancer risks. The cumulative risk results represent a high-level overview of the potential of cancer risk (gross cancer levels) and noncancer hazards from cumulative exposure to MSATs. These uncertainties in the cumulative risk computations are attributed to the relationship between exposure, critical effect systems, and the origin or production of a malignant or benign tumor, as each MSAT targets different organs within the human body. For example, the critical effect of benzene exposure is observed in the immune system [
73], whereas formaldehyde targets the gastrointestinal tract [
74], and benzo(a)pyrene targets the following systems: developmental, immune, and reproductive [
45].
4.2. Validation of Methodology
The near-road air monitoring site used to validate the air dispersion model is the Saint Paul—Ramsey Health Center. The Saint Paul—Ramsey Health Center monitoring site is located at 44°57′2.50″ N, 93° 5′54.60″ W and was chosen to be the project center. The station base elevation is 251 m above sea level, and the monitors are positioned on the north side of the building, approximately 60 m south of the Interstate-94 corridor and interchange with Interstate-35E. The station measures the following pollutants: PM
10, PM
2.5, metals, CO, carbonyls, O
3, SO
2, NO
x, asbestos, and volatile organic compounds (VOCs) [
75].
This monitoring site was chosen to validate the air dispersion model because of the
Proximity to the modeled high-traffic roadway segments;
Absence of other sources within the project area. This increases the confidence of attributing the concentrations to the high-traffic roadway segments;
Availability of model to monitor ratio data for MSATs.
Figure 13a shows the location of the Saint Paul—Ramsey Health Center monitoring site. The monitoring equipment is displayed in
Figure 13b.
The MSATs included in the validation study are benzene and formaldehyde. After completing the air dispersion modeling, the modeled results were compared with the monitored concentrations at the near-road monitoring site. Comparing modeled and monitored concentrations is critical in evaluating how the air dispersion model performs. Using data from air monitoring stations to validate the air dispersion model performance is known as the validation of the model or model calibration. Calibrating the model is an iterative process used to improve the model performance and lower the uncertainty associated with the data.
Table 16 summarizes the modeled on-road air concentrations compared to annual mean monitored concentrations at the Saint Paul—Ramsey Health Center monitoring site.
The ratios presented in
Table 16 conform with the often-quoted factor of two accuracy, a ratio recognized in the air dispersion modeling field. The validation of the air dispersion model demonstrated reasonable agreement between the modeled concentrations and the available measurements at the near-road air monitoring site, even though results varied among benzene and formaldehyde, and the modeled concentrations were higher than the measurements. The MPCA monitored concentrations and source contribution percentages were obtained from MPCAs’ MNRISKS Model-Monitor Tool [
76].
4.3. Air Dispersion Model Performance
In the Validation of Methodology subsection, it was determined that the air dispersion model’s performance is deemed acceptable because the annual modeled concentrations of benzene and formaldehyde are within a factor of two [
77] of the corresponding observed concentrations. Additionally, two statistical indicators were calculated to assess the model’s performance: fractional bias (FB) and normalized mean square error (NMSE) [
78]. FB and NMSE were calculated using the equations shown below, and the values of these statistical indicators are summarized in
Table 17.
where
C0 represents the observed concentration, and
Cp represents the modeled concentration.
An ideal air dispersion model would have both fractional bias and normalized mean square error equal to zero [
78]. The fractional bias was selected as a performance measure because it is symmetrical and bounded, as referenced in [
79]. A negative value of fractional bias indicates model overprediction.
4.4. Strengths and Limitations of the Methodology
In this subsection, we will discuss the strengths and limitations of the research methodology used in this study. Key strengths of the presented methodology are as follows:
Integration of the results from three different types of models (i.e., on-road vehicle emissions inventory, air dispersion models, and risk estimate models), enabling the estimation of cumulative human health impacts from multiple exposure routes including direct inhalation and indirect ingestion of MSATs. Heath impact assessments include chronic cancer (i.e., risk) and noncancer health effects (i.e., hazard) as well as shortly term acute exposure to MSATs.
Incorporation of spatiotemporal considerations to improve the characterization of emissions and resulting health impacts to represent real-world exposures more realistically.
It is worth noting that the methodology being presented has certain limitations, which are summarized below:
It uses average values for key exposure parameters, such as body weight and inhalation rate, which may not accurately reflect the variability in the population. To address this limitation, follow-up research is being conducted to develop a new stochastic risk characterization model (discussed in the Conclusions section), enabling the assessment of risks across a range of population groups (e.g., body weight by age groups), including those most susceptible to the effects of MSATs exposure. This will enable more targeted resource allocation and abatement measures for improved public health outcomes.
The current exposure assessment does not consider exposure to non-mobile-source air toxics, such as indoor emissions from sources like wood-burning stoves, or air toxics formed by secondary reactions, such as the secondary formation of formaldehyde and acetaldehyde. While this type of exposure assessment is not currently planned, it would be valuable for future work.
4.5. Determining the Existence of Exposure Scenarios in the Study Area
Human health impacts from MSATs exposure are calculated and analyzed for multiple developed scenarios in this paper. The analyzed scenarios include Farmer Adult, Farmer Child, Urban Resident Adult, and Urban Resident Child. In this subsection, the likelihood of a developed scenario being realistic is discussed.
As revealed in the Risk Results, the farmer scenario drives the total individual cancer risk in the case of benzo(a)pyrene and the cumulative cancer risk at the Saint Paul—Ramsey Health Center and the Anderson Office Building sensitive receptors. However, both sensitive receptors (e.g., Saint Paul—Ramsey Health Center and Anderson Office Building) are in highly developed commercially zoned urban areas, meaning the likelihood of the farmer exposure scenario driving the risk at these locations is very low.
In addition, the exposure pathways for a farmer scenario are not likely to occur because the following exposure pathways are improbable in a highly developed urban area:
Incidental ingestion (e.g., from hand-to-mouth contact) of soil.
Ingestion of homegrown produce such as lettuce, potatoes, onion, or tomatoes.
Ingestion of homegrown beef, chicken, eggs, or home-reared pork.
It is important in the human health risk assessment process to take a common-sense approach in analyzing the developed scenarios within a project domain and describing the probability of that scenario occurring or not, because if the exposure scenario does not occur in the study area, the relevant exposure pathways that represent the exposure scenario do not exist, and therefore no risk occurs. In this paper, the farmer scenario was included to demonstrate the different scenarios that can be applied and analyzed within the proposed methodology. Furthermore, the farmer scenario can be easily applied in other exposure settings, such as evaluating human health risks in a project area situated in agricultural land with major heavy diesel commercial motor vehicle operations, such as the Interstate 5 and California 99 intersection in the San Joaquin Valley, the United States’ most productive farming area.
4.6. Variability and Uncertainty Discussion
Several exposure parameters used in the risk and hazard characterization were represented by average values corresponding to the 50th percentile of the exposure factor distribution, such as body weight and metabolic rates. These values misrepresent the natural variability in a population and, therefore, will introduce uncertainty in the risk characterization and influence the results. Three dominant exposure parameters, along with their associated uncertainties, are described in this subsection. These parameters are body weight, inhalation rate, and consumption rate of animal tissue. These parameters are used in quantitatively estimating cancer risk and noncancer hazards.
4.6.1. Body Weight
The body weight (BW) parameter is used to calculate the MSATs intake via the ingestion exposure pathway [
26], and the ratio of the consumption rate to the body weight is termed the dose.
The U.S. EPA’s recommended default BW values in deterministic human health risk assessments are 70 and 15 kg for adults and children, respectively [
26], and these values were used in the case study presented in this paper. The values for this parameter are project-specific and might be higher or lower depending on the project location and the assessed population. Therefore, the default BW values might not accurately reflect the population’s actual body weights due to the intra-population variability. However, the degree of variation of the BW is not expected to impact the risk results significantly because, in the case of adults, the default value is within 25% of the true value for most adults [
80].
4.6.2. Inhalation Rate
The inhalation rate parameter is used to estimate the average daily MSATs intake via the inhalation exposure pathway. The U.S. EPA’s recommended default values are 0.83 m
3/h for adults and 0.3 m
3/h for children [
26]. The uncertainties associated with this parameter include varying inhalation rates amongst the population due to many factors, such as age, activity level, genes, and lifestyle. For example, a farmer or fisher exposure scenario may have an increased respiratory rate due to the strenuous activities involved in their jobs, such as heavy carrying and lifting. Moreover, farmers may have prolonged exposures to MSATs due to the proximity of their homes to their workplaces (e.g., farms).
4.6.3. Consumption Rate of Animal Tissue
The consumption rate of animal tissue is a critical parameter in estimating the daily intake of MSATs from the ingestion of animal tissue. This parameter represents the ingestion of pork, beef, milk, poultry, or eggs. The unit for this parameter is expressed in kg/kg-day on a fresh weight (FW) basis. The recommended default consumption rate values [
26] used in this study are summarized in
Table 18.
The uncertainties associated with this parameter include employing default average values that are not representative of project-specific conditions. Moreover, these default values do not consider the relatively sizeable inter-person variability in dietary intake [
81]. The variability associated with this parameter may underestimate or overestimate the studied population’s consumption rate of animal tissue and thus impact the exposure calculations.
As described in this subsection, significant sources of uncertainty in human health risk assessment include using default average values for dominant exposure parameters to represent the population without considering the natural variability in a studied population. As a result, default average values provide inadequate information about variability surrounding the risk from exposure.
Variability can be addressed using a stochastic model to improve the analysis by reducing the variability of data uncertainty. Moreover, reducing and characterizing the variability in a population can assist in focusing the analysis on segments of the population that may be at higher risk from environmental exposure. Future research is needed to address the variability issues by developing a new stochastic analysis-based risk characterization to describe better the risk posed to human health and the surrounding ecosystem.
5. Conclusions
This paper presents a novel and validated methodology to quantify the numerous potential health risks associated with exposure to Mobile Source Air Toxics emitted from on-road sources for the first time.
This methodology overcomes many existing gaps and limitations in the literature and current practices, such as roadway geometry characterization between different models, employing average emissions to calculate human health impacts, and assessing human health impacts from inhalation routes of exposure only. Not evaluating non-inhalation exposures does not reveal the full extent of health impacts from exposure to MSATs, especially with air toxics like benzo(a)pyrene, which builds up in the tissues of organisms.
This methodology integrates several academic disciplines, including the estimation of mobile source emissions of thousands of air toxics, the atmospheric dispersion and deposition of these toxics, the fate and transport of toxics through various media, the assessment of the dose on humans, and the computation of the cumulative cancer risks and noncancer hazards.
The methodology supports the identification of human health risk impacts from exposure to mobile source air toxics by using validated models to determine if the predicated direct (e.g., inhalation) and indirect (e.g., ingestion) risks in existing or proposed transportation projects exceed target risk levels. Moreover, the result of this work provides regulators and stakeholders with the information to evaluate the magnitude of potential human health risks.
Three MSATs were evaluated in this paper, and their modeled air concentrations were validated against monitored data. The validation of the air concentrations demonstrated reasonable agreement between the modeled concentrations and the available measurements at the Saint Paul—Ramsey Health Center monitoring site.
Risk results using the hybrid risk modeling approach show that exposure to benzo(a)pyrene has the potential to cause adverse human health effects based on the evaluation of the Farmer Adult scenario at the Saint Paul—Ramsey Health Center and Anderson Office Building sensitive receptor locations. As shown in
Table 12, for the farmer scenario, the calculated total benzo(a)pyrene cancer risk is equal to 4.95E-05 at the Saint Paul—Ramsey Health Center and 2.02E-05 at the Anderson Office Building. While the Farmer Adult scenario was not applicable in this study area, other transportation corridors with traffic flows of the same magnitude exist where the Farmer Adult scenario would apply (e.g., Interstate 5 and California 99 intersection in the San Joaquin Valley). This conclusion is based on the fact that the estimated cancer risks using modeled air toxics concentrations at the referenced sensitive receptors are above the target risk levels published in
Table 5. Synthesizing the results from the air dispersion and deposition model and the hybrid risk model leads to inferences that:
On-road mobile sources have the potential pose serious health concerns due to the proximity of these emission sources to human receptors;
Human receptors are exposed to MSATs via direct pathways such as inhalation and indirect pathways, where the MSATs pass through another environmental medium, such as the ingestion of contaminated food. Risks from indirect pathways further contribute to risk already attributed to the inhalation exposure pathway. Risks from non-inhalation pathways are likely higher than inhalation risks.
Evaluating human health risk using default average estimates for important dominant exposure variables such as body weight or inhalation rates does not provide sufficient information about the variability surrounding the human health risk from exposure. Additional research is recommended to address the variability in the population by developing a new stochastic analysis-based risk characterization to better describe the risk posed to human health and the surrounding ecosystem. Moreover, it is recommended to implement the first deterministic ecological health assessment methodology for Mobile Source Air Toxics to understand the magnitude of damage done to the ecosystem as a result of air toxics exposure. Lastly, this research lends itself to incorporating risk driver analyses to aid in detecting and uncovering exposure scenarios with adverse impacts that are not well-known or well-studied, such as exploring the risks from endocrine disruptors or the consequences of DNA methylation.