Skip to main content
NCBI home page
Elsevier - PMC COVID-19 Collection logoLink to Elsevier - PMC COVID-19 Collection
. 2023 Feb 11;299:119649. doi: 10.1016/j.atmosenv.2023.119649

Inferred vehicular emissions at a near-road site: Impacts of COVID-19 restrictions, traffic patterns, and ambient air temperature

Dolly L Hall-Quinlan a,∗, Hao He a,b, Xinrong Ren a,c, Timothy P Canty a,d, Ross J Salawitch a,b,e, Phillip Stratton a, Russell R Dickerson a,b,e
PMCID: PMC9918323  PMID: 36816430

Abstract

Vehicles are a major source of anthropogenic emissions of carbon monoxide (CO), nitrogen oxides (NOx), and black carbon (BC). CO and NOx are known to be harmful to human health and contribute to ozone formation, while BC absorbs solar radiation that contributes to global warming and also has negative impacts on human health and visibility. Travel restrictions implemented during the COVID-19 pandemic provide researchers the opportunity to study the impact of large, on-road traffic reductions on local air quality. Traffic counts collected along Interstate-95, a major eight-lane highway in Maryland (US), reveal a 60% decrease in passenger car totals and an 8.6% (combination-unit) and 21% (single-unit) decrease in truck traffic counts in April 2020 relative to prior Aprils. The decrease in total on-road vehicles led to the near-elimination in stop-and-go traffic and a 14% increase in the mean vehicle speed during April 2020. Ambient near-road (NR) BC, CO, NOx, and carbon dioxide (CO2) measurements were used to determine vehicular emission ratios (ΔBC/ΔCO, ΔBC/ΔCO2, ΔNOx/ΔCO, ΔNOx/ΔCO2, and ΔCO/ΔCO2), with each ratio defined as the slope value of a linear regression performed on the concentrations of two pollutants within an hour. A decrease of up to a factor of two in ΔBC/ΔCO, ΔBC/ΔCO2, ΔNOx/ΔCO2, and in the fraction of on-road diesel vehicles from weekdays to weekends shows diesel vehicles to be the dominant source of BC and NOx emissions at this NR site. We estimate up to a 70% reduction in BC emissions in April 2020 compared to earlier years, and attribute much of this to lower diesel BC emissions resulting from improvements in traffic flow and fewer instances of acceleration and braking. Future efforts to reduce vehicular BC emissions should focus on improving traffic flow or turbocharger lag within diesel engines. Inferred BC emissions from the NR site also depend on ambient temperature, with an increase of 54% in ΔBC/ΔCO from −5 to 20 °C during the cold season, similar to previous studies that reported increasing BC emissions with rising temperature. The default setting of MOVES3, the current version of the mobile emission model used by the US EPA, does not adjust hot-running BC emissions for ambient temperature. Future work will focus on improving the accuracy of mobile emissions in air quality modeling by incorporating the effects of temperature and traffic flow in the system used to generate mobile emissions input for commonly used air quality models.

Keywords: On-road emissions, Air quality, COVID-19, Ambient temperature, Emission model, Black carbon, NOx, CO, CO2

1. Introduction

Vehicles emit substantial amounts of carbon monoxide (CO), nitrogen oxides (NO + NO2 = NOx), black carbon (BC), and carbon dioxide (CO2). Both CO and NO2 are criteria pollutants regulated by the United States Environmental Protection Agency (US EPA) due to their hazards to human health and role as precursors to the formation of ozone, O3 (Crutzen, 1973; Pusede and Cohen, 2012; Sather and Cavender, 2016). Fine particulate matter (PM2.5), defined as “particulate matter with a nominal mean aerodynamic diameter less than or equal to 2.5 μm” (US EPA, 2016), includes BC and all other particulate matter in this size range, and is regulated by the US EPA due to the adverse health and visibility impacts (US EPA, 2021b). Ozone, BC, and CO2 absorb radiation and can have adverse radiative forcing (US EPA, 2021a).

According to the 2017 National Emissions Inventory (NEI), the mobile sector is the dominant emitter of CO and NOx both nationally and within Maryland (US), the focus of this study. At the national scale, vehicles contribute to 44% of total CO (including wildfires) and 52% of total NOx emissions, and in Maryland vehicles generate 83% of total CO and 70% of total NOx emissions. The mobile sector is also the largest contributor to BC emissions, representing 42% of total emissions nationally and 58% in Maryland. Within the on-road mobile sector in Maryland, diesel engines are estimated to emit 75% of BC, 41% of NOx, but only 3.8% of CO, with similar percentages on the national scale (US EPA, 2022).

Lockdown mandates and other restrictions imposed in response to the COVID-19 public health emergency in early 2020 resulted in a sharp decline in the number of on-road vehicles, a decrease in traffic congestion, and lower vehicular emissions (e.g., Du et al., 2021; Harkins et al., 2021; Hudda et al., 2020; Laughner et al., 2021; Yang et al., 2021) and other sources (e.g., Ahn et al., 2022; Gaubert et al., 2021; Krüger et al., 2022; Lopez-Coto et al., 2022; Yadav et al., 2021; Zeng et al., 2022).

Diesel-fueled trucks and gasoline-fueled cars emit air pollutants in varying amounts (Grieshop et al., 2006; Hudda et al., 2020), primarily due to differences in engine technologies used by each type of vehicle. Diesel engines inject the fuel directly into the engine cylinders (known as direct injection) and tend to operate fuel-lean, while gasoline engines typically premix the air and fuel upstream of the engine cylinders and operate at lower air-to-fuel ratios and lower maximum temperature. The latter process results in higher emissions of CO and lower emissions of BC and NOx for gasoline compared to diesel-fueled vehicles (Ban-Weiss et al., 2008; Park et al., 2011).

To study emissions from gasoline and diesel vehicles in the near-road (NR) environment, here we use emission ratios, such as ΔNOx/ΔCO2, to indicate the amount of NOx emitted (ΔNOx, where Δ represents the change in concentration relative to the background) relative to the amount of CO2 emitted (ΔCO2). Emission ratios are useful to interpret the relative influence of gasoline and diesel vehicles in an air sample because each source has its own unique range of emission ratios. For example, exhaust emissions from diesel trucks contain approximately ten times higher ΔNOx/ΔCO2 emission ratios because diesel vehicles produce more NOx than gasoline vehicles and diesel engines typically run with greater fuel efficiency, resulting in lower CO2 than from a gasoline engine of comparable size (O'Driscoll et al., 2018). ΔNOx/ΔCO emission ratios from diesel-fueled compression-ignited vehicles (CIVs) are typically close to 1 and for gasoline-fueled spark-ignited vehicles (SIV) the ratios are closer to 0.1 (Bureau of Transportation Statistics, 2018). Fuel-specific BC emission factors (gBC/kgfuel) from CIVs are up to 50 times higher than those from light-duty SIVs (Ban-Weiss et al., 2008; Dallmann et al., 2013), while CO emission factors are roughly half of those from SIVs (Dallmann et al., 2013).

Vehicle speed and acceleration affects vehicular emissions of BC, CO, and NOx. Turbocharger lag, resulting from acceleration in diesel vehicles, produces a mixture of BC, metals, unburnt fuel and oil, and sulfates, a blend commonly referred to as soot (Rakopoulos and Giakoumis, 2009). Therefore, fewer instances of acceleration result in lower BC emissions (Giakoumis and Zachiotis, 2021). Prior work shows conflicting results of the effect of driving cycles on emissions. For example, PM emissions showed no clear trend with driving cycle in Dixit et al. (2017) and Durbin et al. (2008). Using dynamometer data, Dixit et al. (2017) showed that NOx emissions from heavy-duty (HD) diesel trucks were often more than twice as high for a stop-and-go driving cycle as for free-flow traffic. NOx emissions were also uniformly higher when exhaust aftertreatment systems (designed to remove NOx) were below the optimal operating temperature; see also (Jiang et al., 2022; McCaffery et al., 2021). These studies were focused on urban driving cycles and are not strictly comparable to traffic on an interstate highway where a substantial fraction of the vehicles move at speeds exceeding 27 m s−1 (for reference, 1 m s−1 equals 2.2 miles hr−1). Prior studies also indicate that some, but not all, HD vehicles show higher NOx emissions (per distance traveled) for the higher cruise speeds (>24.5 m s−1) than for 18 m s−1 cruise (Durbin et al., 2008). Batterman et al. (2010) found emissions of NOx and CO (per distance traveled) to vary by speed for the combined vehicle fleet, with emissions increasing with speeds above 18 m s−1.

Ambient temperature has been found to impact on-road vehicular emissions (Bishop et al., 2022; Grange et al., 2019; Hall et al., 2020; Li and Lü, 2021; Saha et al., 2018; Weber et al., 2019). Inferred mobile NOx emissions at a NR site in Maryland were lower at higher outdoor temperatures while CO and CO2 emissions remained relatively constant (Hall et al., 2020). A similar temperature sensitivity was reported at a NR site in North Carolina (US), with higher NOx emission factors in winter than in summer (Saha et al., 2018). Further studies examining the European vehicle fleet found declining mobile NOx emissions with rising ambient temperatures in light-duty diesel vehicles (Grange et al., 2019; Weber et al., 2019). Some publications suggest that mobile BC emissions peak at the highest ambient temperatures in diesel-fueled vehicles (Book et al., 2015; Chen et al., 2001; Kondo et al., 2006) and in “measureable emitters” (Wang et al., 2018). Other studies found little seasonal difference in the contribution of total PM2.5 mass concentration from diesel vehicles (Kim and Hopke, 2012); gasoline-powered vehicles may even emit higher BC emissions at colder temperatures, such as during cold-starts (Schauer et al., 2008).

Impacts of ambient temperature on vehicular emissions need to be properly represented in mobile emissions models and inventories. The NEI estimates mobile emissions using the MOtor Vehicle Emissions Simulator (MOVES) developed by the US EPA (US EPA, 2015b) that incorporates multiplicative factors to adjust mobile hot-running emissions of CO, NOx, and BC for ambient temperature. In both the previous version (MOVES2014b) and the latest version (MOVES3), the adjustment factors for CO and NOx are set to unity and running emissions from any gasoline- or diesel-powered vehicle are commonly not adjusted for temperature (US EPA, 2015a, 2020a). MOVES2014b includes an adjustment for hot-running BC emissions for gasoline vehicles of model year (MY) 2004 and earlier but does not modify running BC emissions from diesel-fueled vehicles (US EPA, 2015a). The temperature sensitivity in running BC emissions from MY 2004 and older gasoline vehicles was removed in MOVES3 (US EPA, 2020a); see also Hall et al. (2020). The absence of a temperature sensitivity in mobile BC emissions simulated by MOVES may also help to explain a previous study showing that the CMAQ model, which indirectly uses mobile emissions simulated by MOVES, underestimates ambient BC concentrations in the summer but not in the winter (Appel et al., 2008).

In this study we use ambient measurements of BC, CO, NOx, and CO2 collected at a fixed NR site located along Interstate-95 (I-95) in Maryland to infer mobile emissions using ratios of these measurements. Our goal is to quantify the impact of changes in on-road vehicle fleet composition and the near-elimination in traffic jams on highway emissions. We also present evidence for a temperature dependence in mobile BC emissions and compare the results to the temperature sensitivity of BC emissions from MOVES3.

2. Methodology

2.1. Interstate-95 near-road and traffic observations

Ambient pollutant concentrations and meteorological data were measured at a NR site located 12 m from the southbound leg of I-95, an eight-lane interstate highway in Howard County, Maryland (MD, Fig. S1; 39° 8′ 35.38″ N, 76° 50’ 45.53” W). Observations were collected between January 2017 and December 2018, then again from March 2020 through December 2020. The site is managed by the Maryland Department of the Environment (MDE) as part of the US EPA Air Quality System (AQS), and the monitoring site follows standard measurement collection and quality control protocol outlined in the Code of Federal Regulations (CFR) 40 part 58 (available at http://www.ecfr.gov/cgi-bin/ECFR?page=browse). Hourly-mean data were calculated from 10-s mean NOx and CO mixing ratios, 1-min mean BC concentrations ([BC]), and 2-s mean CO2 abundances. The measurement technique, analyzer used, and relative accuracy of the CO, NOx, CO2, BC, and meteorological measurements are provided in Table 1 and discussed in more detail in Section S1 of the Supplemental Information (SI). To isolate highway emissions, only hours when winds were blowing from the adjacent highway (25°–225° clockwise from North) were used in these analyses. Allowing wind from all directions had little impact on the results.

Table 1.

Instrument details for the CO, NOx, BC, CO2, and meteorological variables measured at the I-95 NR site used in this study. The accuracy is provided as a relative value with the appropriate reference given.

Pollutant Species Method of Detection/Analyzer Accuracy
CO Infrared Energy Absorption
Teledyne API 300U		 2.3% (See the SI)
NO/NO2/NOx Chemiluminescence
Teledyne API 200U		 4.7% (See the SI)
BC Filter-based, optical measurement technique
Magee Scientific TAPI M633 (AE33)
Aethalometer (880 nm)		 ∼5% (Grimes and Dickerson, 2021)
CO2 (2017 and 2018) Laser Absorption Spectroscopy
Los Gatos Research Fast Greenhouse Gas Analyzer		 0.089% (See the SI)
CO2 (2020) Laser Absorption Spectroscopy/Cavity Ring-down
Picarro Model G2301		 0.025% (See the SI)
T/P/RH/Winds Vaisala WXT 520 T: 0.3 °C
P: 0.5–1 hPa
RH: 3–5%
WS: 3–5%
WD: 3° (Vaisala, 2012)
Open in a new tab

Hourly traffic observations along I-95 were collected by an in-road traffic sensor, operated by the Maryland State Highway Administration. The sensor is located 8 km north of the ambient NR trailer and had a daily mean of 198,000 passing vehicles prior to March 2020. Vehicle information (count, type, and speed) was measured using a combination of two inductive loops and a class II piezo pressure sensor in each lane to count the number of and spacing between vehicle axles. This technique allows the site to count the hourly total number of vehicles, as well as vehicle type and speed from the number of and distance between axles (US Department of Transportation Federal Highway Administration, 2014). The sensors connect to an electronic equipment cabinet adjacent to the highway that contains two Diamond Traffic Products Phoenix II Classifiers with Integrated IRIS cellular modems to record the vehicle data.

2.2. Inferred emission ratios and speed and temperature sensitivity analysis

The proximity of the I-95 NR site to the highway provides an ideal setting to analyze mobile emissions. BC, CO, NOx, and CO2 share vehicles as a common source, resulting in similar patterns for ambient measurements of BC, CO, NOx, and CO2 in the NR environment (see, for example, Krecl et al., [2018]; Naser et al., [2009]; Krecl et al., [2018] and Naser et al., [2009]). Covariations in pollutants are useful to study properties in vehicular emissions through inferred emission ratios, which are often applied to trace specific sources in plume analysis. Here we calculate the hourly ΔBC/ΔCO, ΔBC/ΔCO2, ΔNOx/ΔCO, ΔNOx/ΔCO2, and ΔCO/ΔCO2 emission ratios from the slope parameter of linear geometric mean regression fits performed on 5-min mean mixing ratios of two pollutants sampled within that hour, similar to Anderson et al. (2014) and Hall et al. (2020). For example, the regression slope of 5-min mean [BC] and [CO] represents the hourly ΔBC/ΔCO. Only hours with more than 30 min of observations were used in the analysis. Table S1 in the SI shows good agreement between fuel-based BC and NOx emission factors calculated from the emission ratios in this study and emission factors reported in other references.

To investigate the relative importance of CIVs and SIVs to the traffic flow and temperature sensitivity in each time period, the analysis was performed for all days and hours, as well as by weekday and time-of-day (overnight, morning rush hours, and afternoon rush hours). We use Eastern Local Time (ET) for the analysis by hour of day, which follows Eastern Standard Time from early-November to mid-March and daylight savings for the remainder of the year.

2.3. MOVES3 temperature sensitivity simulations

The latest version of the US EPA MOVES model, MOVES3, was used to simulate mobile emissions for Howard County, MD. At the county-level scale, MOVES3 requires county-specific input data including vehicle speed, road type, and age distributions; vehicle miles traveled (VMT) by vehicle group, hour, day, and month; an average 24-h profile of temperature and relative humidity; vehicle population; inspection and maintenance; and fuel information (see Table S2). All input, except for fuel information, was supplied by the MDE. For input on fuel specifics, MOVES3 default data for Howard County, MD, was used as this information had not yet been finalized by MDE at the time of paper submission.

MOVES3 emissions were generated in the form of emission rates, representing the mass of pollutant per distance traveled. Emission rates generated by MOVES are a function of month, temperature, relative humidity, vehicle type, fuel, speed, road type, model year, and weekday/weekend. The choice of month impacts emission rates because the fuel information used by MOVES3 differs between winter and summer. A major difference between summer and winter months in the model and in the real-world is the volatility of the fuel, with winter fuel used from September through April and summer fuel used from May through August (US EPA, 2021c).

In this study, MOVES3 simulations were performed using 2017 input data, the latest year with available data, and for the month of December, when winter gasoline and diesel fuels are used. Repeating the analysis for summer yields little difference in emission ratios (Fig. S3). A detailed explanation of how weighted mean emission ratios were calculated using fleet composition data along I-95 is provided in Hall et al. (2020).

MOVES3 does not incorporate a temperature adjustment for hot-running BC emissions from either CIVs or SIVs. While neither MOVES3 nor the previous version, MOVES2014, adjusts hot-running BC emissions from CIVs, the default setting of MOVES2014 incorporates an adjustment for running BC emissions from pre-2004 model year SIVs for temperature, using a multiplicative adjustment factor (US EPA, 2015a, 2020a). To verify the documented absence of a temperature sensitivity in MOVES3 output, weighted mean emission ratios were calculated in 1 °C increments from −5 °C to 25 °C and the results presented in Section 3.2.

3. Results and discussion

3.1. Impact of COVID-19 traffic restrictions on vehicle fleet, traffic flow, and emissions

3.1.1. Vehicle fleet characteristics before and during the pandemic

Travel restrictions for the COVID-19 pandemic began in mid-March 2020 in the Baltimore-Washington Region (BWR) and led to a peak decrease in on-road vehicle counts in April 2020 (Fig. 1 and S4). The mean daily total number of vehicles passing the site was 192,000 pre-COVID (April 2017 and 2018) and decreased to 96,300 in April 2020. Light-duty passenger vehicle counts were 60% lower in April 2020 relative to pre-COVID counts, while the number of buses was down by 85%, single-unit trucks by 21%, and trailer (combination-unit) trucks by 8.6% (Fig. 1 ).

Fig. 1.

Fig. 1

Open in a new tab

Relative difference (%) in monthly 2020 traffic counts compared to 2017 and 2018 for the months of January through July at the I-95 NR site. Error bars represent the 95% confidence intervals (2σ) of the mean daily traffic count differences in each month.

Of the vehicles passing the traffic sensor, the typical pre-COVID fleet composition was 88% cars, motorcycles, and light-duty pick-up trucks; 1.3% buses; 3.0% single-unit trucks; and 7.7% tractor-trailer trucks. The highest fraction of CIVs (single-unit and tractor-trailer trucks) typically occurred overnight from 23:00 to 4:59 ET, with a mean value (±1σ) of 0.22 ± 0.072 (Table 2 and Fig. S5c). During the overnight period there were few passenger cars on the road. The fraction of CIVs dropped to 0.11 ± 0.027 during the morning rush hour (5:00 to 9:59 ET) and further declined to 0.073 ± 0.024 during the afternoon rush hour (14:00–20:59 ET). In terms of day-of-week, the fraction of CIVs decreased by a factor of 2 from 0.14 on weekdays to 0.074 on weekends (Table 2 and Fig. S5f).

Table 2.

Traffic composition of vehicles passing the I-95 NR traffic counter pre-COVID (April 2017 and 2018) and in April 2020 by weekday and local time. SIV represents spark-ignited vehicles and CIV indicates compression-ignited vehicles.

April 2017 & 2018 SIV (CIV) April 2020 SIV (CIV)
All Days and Hours 88% (12%) 79% (21%)
Weekdays Only (M-F) 86% (14%) 77% (23%)
Weekends Only (S–S) 93% (7%) 84% (16%)
Overnight Hours
(23:00–4:59 a.m. Local)		 78% (22%) 63% (37%)
Morning Rush Hours
(5:00–9:59 a.m. Local)		 89% (11%) 82% (18%)
Afternoon Rush Hours
(14:00–20:59 Local)		 93% (7%) 88% (12%)
Open in a new tab

The traffic composition passing the I-95 NR site in April 2020 exhibited higher fractions of CIVs for all time windows than in April of 2017 and 2018. In April 2020, the fleet consisted of 79% cars, motorcycles, and pick-up trucks; 0.010% buses; 4.9% single-unit trucks; and 16% tractor-trailer trucks (Table 2). The total number of on-road passenger cars declined much more than the number of trucks (Fig. S5), resulting in a doubling of the fraction of CIVs relative to SIVs from 12% pre-COVID to 21% in April 2020 (Table 2).

While CIVs represent the minority of vehicles passing the NR site, their contribution to highway emissions of BC and NOx can be substantial. As demonstrated in a tunnel study by Dallmann et al. (2013), medium- and heavy-duty trucks, composing as little as <1% of the fleet, contributed up to 45% of BC and 18% of NOx concentrations. With the fraction of CIVs along I-95 increasing from 12% in April 2017 and 2018 to 21% in April 2020 (Table 2), diesel trucks would have had a greater contribution to mobile emissions in April 2020 than in prior years.

3.1.2. Influence of pandemic-related travel restrictions on traffic flow and inferred vehicular emissions

Significant reductions in the number of on-road vehicles in April 2020 affected fleet composition, vehicle speed, and stop-and-go traffic. The weighted mean vehicle speed (μ) on I-95 was 14% higher in April 2020 (μ = 30.3 m s−1) compared to prior years (μ = 26.6 m s−1) and the fraction of vehicles driving slower than 22 m s−1 was nearly eliminated in April 2020 (Fig. 2 ). The threshold of 22 m s−1 was selected because the speed limit on this portion of the highway is 29 m s−1 and speeds slower than 22 m s−1 typically indicate traffic jams marked by more frequent braking and accelerating. Repeating this analysis by weekday and local time shows consistent results with higher speeds and virtually eradicated stop-and-go traffic in April 2020 relative to prior years (Table S3). The only time period with comparable mean vehicle speed and lack of traffic congestion between April 2020 and prior years was the overnight period (23:00–4:59 a.m. ET), marked by fewer on-road vehicles and little traffic congestion (Table S3).

Fig. 2.

Fig. 2

Open in a new tab

Mean hourly number of vehicles in each of 12 speed bins in April 2017 and 2018 (turquoise) and April 2020 (purple) at a traffic sensor located on I-95 in Howard County, MD. Provided in the top left-hand corner is the weighted mean speed and the fraction of vehicles driving slower than 22 m s−1 (frac<22mps) pre-COVID and in April 2020.

Fewer on-road vehicles in April 2020 resulted in reduced mobile emissions as well as lower ambient BC, CO, NOx, and CO2 mixing ratios compared to prior years at the I-95 NR site (Table S4). Rather than use absolute concentrations, affected by other processes including changing boundary layer height, here we use concentration ratios to infer emissions and to better understand the impact of fewer vehicle miles traveled and changing traffic patterns on the emission of pollutants.

A comparison of inferred emission ratios for April 2020 and pre-COVID by weekday and local time is presented in Table 3 . The difference in temperature between April 2020 and prior Aprils was small (2.2 °C) and its impact on our temperature analysis is discussed in Section 3.2.1.

Table 3.

Median inferred emission ratios for April 2017 & 2018 (“pre-COVID”) and April 2020 (”2020”) at the I-95 NR site. Also provided is the relative difference (%), with the statistical significance indicated by the p-value, between each pair of median values for each emission ratio and time period considered. Percentages of CIVs and SIVs provided are for typical pre-COVID traffic composition.

Temperature
(°C)		 ΔBC/ΔCO
(μg m−3) (μg m−3)−1 		ΔBC/ΔCO2
(μg m−3) (μg m−3)−1 		ΔNOx/ΔCO
ppbv
ppbv−1 		ΔCO/ΔCO2
ppbv ppmv−1 		ΔNOx/ΔCO2
ppbv
ppmv−1
All Days and Hours (Pre-COVID: 12% CIV, 88% SIV)
Pre−COVID 14.4 0.0104 5.11 × 10−5 0.247 8.02 2.05
2020 12.2 0.0076 3.75 × 10−5 0.276 8.40 2.37
Relative Change (%) −26.9% (p < 0.05) −26.6% (p < 0.05) 11.7% (p = 0.074) 4.74% (p = 0.41) 15.6% (p < 0.05)
Weekdays Only (Pre-COVID: 14% CIV, 86% SIV)
Pre−COVID 13.9 0.0125 5.54 × 10−5 0.282 7.69 2.21
2020 11.7 0.0096 4.23 × 10−5 0.313 7.19 2.49
Relative Change (%) −23.2% (p < 0.05) −23.6% (p < 0.05) 11.0% (p < 0.05) −6.50% (p = 0.26) 12.7% (p < 0.05)
Weekends Only (Pre-COVID: 7.4% CIV, 92.6% SIV)
Pre−COVID 15.3 0.0056 3.31 × 10−5 0.159 9.09 1.41
2020 12.2 0.0038 2.76 × 10−5 0.169 12.5 2.05
Relative Change (%) −32.1% (p < 0.05) −16.6% (p = 0.27) 6.29% (p = 0.68) 37.5% (p < 0.05) 45.4% (p < 0.05)
23:00–4:59 a.m. Local Time (Pre-COVID: 22% CIV, 78% SIV)
Pre−COVID 11.7 0.0151 6.39 × 10−5 0.439 6.62 3.03
2020 10.0 0.0128 4.52 × 10−5 0.549 5.26 2.94
Relative Change (%) −15.2% (p = 0.12) −29.3% (p < 0.05) 25.1% (p < 0.05) −20.5% (p = 0.21) −2.97% (p = 0.76)
5:00–9:59 a.m. Local Time (Pre-COVID: 11% CIV, 89% SIV)
Pre−COVID 11.7 0.0123 5.31 × 10−5 0.280 7.63 1.96
2020 10.0 0.0095 3.94 × 10−5 0.296 6.45 2.08
Relative Change (%) −22.8% (p < 0.05) −25.8% (p < 0.05) 5.71% (p = 0.17) −15.5% (p = 0.23) 6.12% (p = 0.76)
14:00–20:59 Local Time (Pre-COVID: 7.3% CIV, 92.7% SIV)
Pre−COVID 18.3 0.0076 4.25 × 10−5 0.189 8.93 1.63
2020 14.4 0.0038 3.04 × 10−5 0.169 12.8 2.48
Relative Change (%) −50.0 (p < 0.05) −28.5 (p < 0.05) −10.6 (p = 0.12) 43.3% (p < 0.05) 52.1% (p < 0.05)
Open in a new tab

Table 3 shows that ΔBC/ΔCO in April 2020 was 26.9% (p < 0.05) lower than in prior years, considering all days and hours. Likewise, ΔBC/ΔCO2 was 26.6% (p < 0.05) lower in April 2020. With SIVs (i.e., passenger cars) exhibiting a much larger decline in numbers than CIVs (i.e., trucks) in 2020, one initial hypothesis was that the reduction in mobile CO and CO2 emissions would be greater than that for BC and thus lead to an increase in ΔBC/ΔCO and ΔBC/ΔCO2 during April 2020. Instead, we observed consistently lower values of ΔBC/ΔCO in April 2020 compared to prior years. The ΔBC/ΔCO2 ratio exhibited a statistically significant decrease in all time periods except for weekends, which had a statistically insignificant decrease in the emission ratio. These findings indicate that the influence of vehicle fleet changes on ΔBC/ΔCO and ΔBC/ΔCO2 was overshadowed by a drastic change in another highway characteristic: stop-and-go traffic.

While the differences in mean traffic speeds were generally minor (<6 m s−1), the fraction of vehicles driving below 22 m s−1 (frac<22mps) dropped down from 23.6% in prior years to less than 2% in April 2020 (Fig. 2 and Table S3). The afternoon rush hour (14:00–20:59 ET), when the frac<22mps was highest of the day at 35.4% (Table S3), exhibited the largest decrease in ΔBC/ΔCO (−50.0%, p < 0.05) from April 2017 and 2018 to April 2020 (Table 3). The morning rush hour also saw a significant improvement in traffic flow, with the percentage of vehicles driving slower than 22 m s−1 decreasing from 24.1% pre-COVID to 1.23% in April 2020, and a concurrent decrease of 22.8% in ΔBC/ΔCO and 25.8% in ΔBC/ΔCO2 (Table 3). The drop in ΔBC/ΔCO during the overnight period (23:00–4:59 ET), when traffic flow did not differ much in April 2020 compared to previous years, was less remarkable and statistically insignificant (−15.2%, p = 0.12).

The difference in ΔBC/ΔCO and ΔBC/ΔCO2 between weekdays and weekends suggests that CIVs are the primary contributor to mobile BC emissions at the I-95 NR site. The median pre-COVID ΔBC/ΔCO ratio on weekdays was 0.0125 compared to the weekend ratio of 0.0056, a decrease of 55.2% (Table 3). Similarly, ΔBC/ΔCO2 decreased by 40.3% from 5.54 × 10−5 on weekdays to 3.31 × 10−5 on weekends. Note that for both ΔBC/ΔCO and ΔBC/ΔCO2, the units of ΔBC, ΔCO, and ΔCO2 (μg m−3) are the same and hence the mass ratios are expressed as unitless quantities. The decrease in both ratios from weekdays to weekends followed the change in the fraction of CIVs passing the NR site: the fraction of CIVs dropped from 14% on weekdays to 7.4% on weekends, resulting in a decrease of 47.1% (Table 2). These observations provide strong evidence that CIVs dominate the BC emissions at the I-95 NR site.

Fig. 2 shows that traffic congestion was nearly eradicated in April 2020. Two mechanisms may help to explain the decline in ΔBC/ΔCO and ΔBC/ΔCO2 during the time of little to no stop-and-go traffic: BC emissions from brake pad wear and acceleration in CIVs. Vehicular brake pads, composed of various metals and graphite, are known to emit BC and a variety of other species (Lyu and Olofsson, 2020). Significant alleviation of stop-and-go traffic in April 2020 thus led to lower BC emissions from brake pad wear. Black carbon is also released from the exhaust of CIVs during acceleration (Giakoumis and Zachiotis, 2021; Rakopoulos and Giakoumis, 2009), and emissions from tailpipes are expected to be higher than those from brake pad wear in the NR environment. With CIVs representing the dominant source of exhaust emissions of BC, the near-elimination of traffic jams in April 2020 produced a larger reduction in mobile BC emissions than the effect of fewer on-road SIVs on CO emissions. Given a decrease in the number of SIVs of 60% in April 2020 and assuming a comparable reduction in CO emissions, the decrease in ΔBC/ΔCO of 26.7% (considering all days and hours) indicates that BC emissions likely decreased by up to 70%, a number compatible with the decrease in the median BC concentration of nearly a factor of two (Table S4).

Table 3 presents the results for various other emission ratios; comparison of CO and NOx to CO2 provides better understanding of the effectiveness of exhaust-aftertreatment devices, specifically catalytic converters, installed in vehicles passing the NR site. Trucks generally emit less CO relative to CO2, resulting in lower ΔCO/ΔCO2 than cars (Fig. S6). With fewer on-road SIVs relative to CIVs in April 2020, one might expect lower ΔCO/ΔCO2 as well. The ΔCO/ΔCO2 analysis presented in Table 3 shows no significant effect considering all hours; however, we observe higher ratios on weekends and during the afternoon rush hour in April 2020 than in previous years. A plausible explanation for this observation is that vehicles emit more CO at high speeds (Hickman et al., 1999; OECD & ECMT, 2006; Wang et al., 2018; Fig. S7). With less traffic congestion on I-95 in April 2020, vehicles were driving 14% faster than usual (Fig. 2) and emitted more CO relative to CO2.

A statistically significant increase in ΔNOx/ΔCO2 was observed for weekdays, weekends, and during the afternoon rush hour. ΔNOx/ΔCO2 increased by 45.4% (p < 0.05) on weekends and 52.1% (p < 0.05) in the afternoon rush hours, compared to 12.7% on weekdays (Table 3). The increase in ΔNOx/ΔCO2 can be explained by elevated NOx emissions from SIVs at high speeds, as well as the higher fraction of CIVs in April 2020 than in April 2017 and 2018. Since the ΔNOx/ΔCO2 emission ratio from CIVs is approximately ten times higher than from SIVs (O'Driscoll et al., 2018), a greater fraction of on-road CIVs in April 2020 would have led to elevated ΔNOx/ΔCO2 (but lower NOx mixing ratios; Table S4) for all days and hours in April 2020 compared to April 2017 and 2018. The time periods that had the largest increase in ΔNOx/ΔCO2 (weekends and the afternoon rush hours) also had the highest fraction of SIVs compared to weekdays, overnight, and the morning rush hours, both in April 2020 and April of prior years. A plausible explanation for this finding is that gasoline vehicles emit more NOx at high speeds (Kean et al., 2003; OECD & ECMT, 2006), contributing to the higher ΔNOx/ΔCO2 observed at times when the fraction of SIV was at a maximum in April 2020. Fig. S8 shows a strong dependence of MOVES-estimated NOx emission factors (per distance traveled) on vehicle speeds; NOx emission factors from MOVES were 28% higher for SIVs traveling at speeds faster than 32.4 m s−1 than for those driving at 22.4 m s−1 (Fig. S8). In April of 2017 and 2018, 12% of the I-95 vehicle fleet drove faster than 32.4 m s−1, while in April of 2020 this fraction increased to 25% of the vehicle fleet. The shift toward faster speeds would have led to greater speed-related NOx emissions from SIVs in April 2020 than in April 2017 and 2018 and contributed to the observed increase in ΔNOx/ΔCO2. With CO emissions from SIVs also increasing with higher speeds (Fig. S7), a likely explanation for the increase in ΔCO/ΔCO2 from April 2017 and 2018 to April 2020 on weekends and during the afternoon rush hours is similar to that discussed for NOx emissions.

To evaluate the sensitivity of the results presented in Table 3, we repeated the analysis using a different method of estimating emission ratios, as well as for August 2020 when on-road traffic had increased. To test the impact of the method used to estimate emission ratios, we repeated the analysis using a background subtraction technique similar to that discussed in Brantley et al. (2014). The details of how background concentrations were calculated are provided in Section S2 of the SI. The results, shown in Table S7, are similar to those in Table 3 and do not change our conclusions. Differences in vehicle speeds, SIV/CIV fleet composition, and frequency of stop-and-go traffic between August 2020 and August 2017/2018 were smaller compared to differences between April 2020 and April 2017/2018 (Tables S8 and S9). More on-road vehicles in August 2020 than in April 2020 led to more frequent traffic congestion, resulting in less significant decreases in ΔBC/ΔCO and ΔBC/ΔCO2, as well as smaller increases in ΔNOx/ΔCO2 and ΔCO/ΔCO2 in August than in April (Table S10).

3.2. Temperature sensitivity of vehicular BC emissions

3.2.1. Near-road ambient observations

Previous studies reported a seasonal cycle in observed ΔBC/ΔCO at suburban and urban sites, with higher ΔBC/ΔCO emission ratios in the summer than in the winter (Chen et al., 2001; Kondo et al., 2006). Additionally, Wang et al. (2018) found higher BC fuel-based emission factors (mgBC kg−1 fuel) in the spring-summer than in the fall-winter using observations collected adjacent to a four-lane highway in Toronto, Canada. A similar seasonal pattern is evident at the I-95 NR site. Fig. 3 shows seasonal mean values of ΔBC/ΔCO ratios for 2017 and 2018 separately (colored by season) and 2017–2018 combined (filled black circles) as a function of ambient temperature. At the I-95 NR site, the highest ΔBC/ΔCO ratio occurred in summer and lowest ΔBC/ΔCO in winter. Seasonal mean ΔBC/ΔCO, calculated from the hourly emission ratios, were strongly correlated with ambient temperature (r2 = 0.96) and increased by 81% from winter (5 °C) to summer (25 °C). The seasonal changes in ΔBC/ΔCO and proximity of the NR site to a major highway suggests a temperature sensitivity in vehicular emissions, particularly those from diesel trucks.

Fig. 3.

Fig. 3

Open in a new tab

Seasonal variability in ΔBC/ΔCO inferred from observations collected at the I-95 NR site from January 2017 through December 2018. Each season contains two colored symbols for 2017 and 2018 separately, with the mean of the two shown by the filled black circles. Blue stars represent winter (December–February), green diamonds represent spring (March–May), red circles represent summer (June–August), and orange triangles represent fall (September–November). The mean emission ratio in each season is represented by the filled black circles, fitted using a linear geometric mean regression (solid black line) with regression statistics provided on the bottom of the plot.

To avoid potential impacts of seasonal fuel switching on emissions, we focus on the temperature dependence of ΔBC/ΔCO during the cold season. Fig. 4 shows that ΔBC/ΔCO and ΔBC/ΔCO2 exhibit a positive correlation with ambient temperature considering all hours and days of the week during the cold months, defined as November, December, January, and February, in 2017 and 2018. Hourly ΔBC/ΔCO and ΔBC/ΔCO2 were placed into ten equal-numbered bins, sorted by ascending temperature. Fig. 4 shows the median and 25th and 75th percentile values for each of these temperature bins, along with an ordinary linear least-squares regression fit to the median values. Considering all data, both emission ratios exhibited a statistically significant increase with temperature: 1.4 ± 0.40 × 10−4 °C−1 in ΔBC/ΔCO and 1.0 ± 0.27 × 10−6 °C−1 in ΔBC/ΔCO2, equivalent to an increase of 56% in ΔBC/ΔCO and 83% in ΔBC/ΔCO2 from −5 °C to 20 °C (Fig. 4). Using these regression parameters, the difference in temperature between April 2020 relative to the April of prior years (April 2020 was 2.2 °C cooler) can only account for a 3–4% decrease in the ΔBC/ΔCO and ΔBC/ΔCO2 quantified in Table 3. Similar temperature sensitivities were obtained when hourly emission ratios were calculated with an orthogonal linear regression instead of geometric mean regression. Repeating the analysis in Fig. 4 for the upwind region of the NR site (225–360°) results in statistically insignificant changes in both ΔBC/ΔCO and ΔBC/ΔCO2 with temperature (Fig. S10). The absence of a temperature effect in the upwind region supports the finding of Fig. 4 that direct vehicular emissions are sensitive to ambient temperature.

Fig. 4.

Fig. 4

Open in a new tab

Hourly ΔBC/ΔCO (a) and ΔBC/ΔCO2 (b) inferred from I-95 NR observations as a function of ambient temperature for the cold season in 2017 and 2018. Hourly ratios (light gray points) were estimated from the slope values of linear geometric mean regressions (±standard error) performed on 5-min mean data within each hour. Only hours with winds originating from the highway were included in the analysis. Hourly data were compiled into ten equal-sized bins, with the median value of each bin represented by the black diamonds, and the 25th and 75th percentile values by the dashed black line. An ordinary least-squares regression was fit to the binned median values (red line) and the details of the regression provided in the top left of each plot. The statistical significance of each regression was determined by the non-parametric Mann-Kendall test (Gilbert, 1987, Kendall, 1975, Mann, 1945).

As shown previously, vehicular running emissions of CO and CO2 do not appear sensitive to temperature during the cold season (Hall et al., 2020). Thus, the increase in ΔBC/ΔCO and ΔBC/ΔCO2 is likely due to an increase in BC emissions from mobile sources at higher temperatures. Some evidence suggests that gasoline vehicles utilizing direct injection and particulate filters appear to remove BC more efficiently at lower intake temperatures (Chan et al., 2013). Direct injection, the mechanism in diesel engines, may also be employed with gasoline engines and has been on the rise in recent years (US EPA, 2020b) to provide more power and increase fuel efficiency. Like in diesel engines, this mechanism of injecting fuel directly into the cylinder produces more BC emissions. However, diesel engines emit the vast majority of BC reaching the NR site and thus a more plausible explanation for the temperature sensitivity lies within diesel vehicles. Turbochargers on diesel engines compress intake air, leading to higher temperatures and lower densities. This can favor elevated BC production as the air-to-fuel ratio is reduced (Chen et al., 2001; Ghazikhani et al., 2013). Diesel trucks equipped with turbochargers typically also utilize charge-air coolers, or turbo coolers. In an air-to-air turbo cooler, the hot, compressed air is cooled via heat exchange with outside air (Joshi et al., 2009; Sher, 1998), resulting in greater air density and lower emissions of NOx and BC (Cipollone et al., 2017). Because compressed air is cooled by heat exchange with ambient air, and also because the coolers degrade over time (Joshi et al., 2009), higher ambient temperatures may lead to less cooling of the intake air and, in turn, higher BC emissions at warmer conditions. This process can lead to increasing BC emissions, and higher ΔBC/ΔCO and ΔBC/ΔCO2, with rising temperature, as observed in Fig. 4.

To further explore the contribution of the vehicle fleet composition on the temperature sensitivity, we repeated the analysis by day of week and at different times of the day. Both ΔBC/ΔCO and ΔBC/ΔCO2 exhibited a significant increase with rising temperature on weekdays (p < 0.05) absent on weekends (Table 4 ). Both the total truck traffic (1022 vehicles/hr on weekdays and 433 vehicles/hr on weekends) and the fraction of CIVs (14% on weekdays and 7.4% on weekends) dropped by a factor of two from weekdays to weekends. Traffic congestion marked by slow-moving vehicles (<22 m s−1) was greater on weekdays (24%) than on weekends (11%, Table S3), resulting in higher weekday BC emissions from diesel trucks due to the stop-and-go effect discussed in Section 3.1.2. With diesel trucks emitting the bulk of on-road BC emissions, the increase in ΔBC/ΔCO and ΔBC/ΔCO2 with rising temperature on weekdays suggests that the emission of BC from diesel vehicles is sensitive to ambient temperature.

Table 4.

Measured ΔBC/ΔCO and ΔBC/ΔCO2 as a function of ambient temperature during the months of November, December, January, and February in 2017 and 2018 at the I-95 NR site. Hourly emission ratios were placed into ten equal-sized bins and an ordinary linear regression fit to the median emission ratios of each bin, the same as in Fig. 4.

I−95 Fleet CIV/SIV (ΔBC/ΔCO)/T
(μg m−3) (μg m−3)−1 °C−1 		(ΔBC/ΔCO2)/T
(μg m−3) (μg m−3)−1 °C−1
All Days and Hours 12%/88% 1.42 ( ± 0.40) × 10−4
(p < 0.05) 		1.01 ( ± 0.27) × 10−6(p < 0.05)
Weekdays Only 14%/86% 1.64 ( ± 0.68) × 10−4(p < 0.05) 1.06 ( ± 0.19) × 10−6(p < 0.05)
Weekends Only 7.4%/92.6% −0.085 (±0.06) × 10−4
(p = 0.72)		 0.072 (±0.24) × 10−6
(p = 1)
Overnight
(23:00–4:59 a.m. ET)		 22%/78% 2.84 ( ± 0.84) × 10−4(p < 0.05) 0.96 ( ± 0.58) × 10−6(p < 0.05)
Morning Rush Hours
(5:00–9:59 a.m. ET)		 11%/89% 3.62 ( ± 0.83) × 10−4(p < 0.05) 1.48 ( ± 0.32) × 10−6(p < 0.05)
Afternoon Rush Hours
(14:00–20:59 ET)		 7.3%/92.7% 1.59 ( ± 0.43) × 10−4(p < 0.05) 1.00 (±1.48) × 10−6
(p = 0.074)
Open in a new tab

A positive correlation between ΔBC/ΔCO and temperature was evident in all three local time windows (Table 4). The slope with temperature (°C−1) was highest during the morning rush hours [3.62 (±0.83) × 10−4, p < 0.05], followed by the overnight period [2.84 (±0.84) × 10−4, p < 0.05], and was lowest in the afternoon rush hours [1.59 (±0.43) × 10−4, p < 0.05]. These results indicate that fleet make-up and driving conditions both contribute to BC emissions.

Patterns in the temperature dependence of ΔBC/ΔCO by local time follow the fraction of CIVs and vehicular traffic over the course of the day. The overnight period experienced the highest fraction of CIVs to total vehicles, even though the total number of trucks on the road was lower than during the rest of the day (Fig. S5). The temperature dependence in BC emissions observed in the overnight period was likely driven by the high fraction (22%) of diesel vehicles (Table 4).

The morning rush hour exhibited the largest slopes of ΔBC/ΔCO and ΔBC/ΔCO2 with temperature. While the fraction of CIVs to total vehicles in the morning (11%) was lower than overnight (22%), stop-and-go traffic was more common during the morning rush hour than during the overnight period (Table S3). A plausible explanation for the observed temperature sensitivity in the morning is that elevated BC emissions from CIVs due to frequent acceleration enhances the temperature effect from diesel engines.

The temperature sensitivity of ΔBC/ΔCO in the afternoon rush hours had the lowest slope considering all times of day [1.59 (±0.43) × 10−4 °C−1, p < 0.05] (Table 4). The slope of ΔBC/ΔCO2 was also at a minimum in the afternoon rush hours, but the regression was statistically significant only at the 90% confidence level (p = 0.074). Given that CIVs make up only 7.3% of the vehicles passing the I-95 NR site in the afternoon, the sensitivity in ΔBC/ΔCO likely originates from high BC emissions by CIVs due to a peak in stop-and-go traffic during the afternoon rush hour or a combination of CIV and SIV.

3.2.2. MOVES simulations

Here we compare the ΔBC/ΔCO and ΔBC/ΔCO2 temperature sensitivity in mobile emissions inferred at the I-95 NR site to MOVES3 output. MOVES3 was run using 2017 input data, the latest year with available data, supplied by the Maryland Department of the Environment for Howard County, MD, the county in which the I-95 NR site resides. Data used as input for the model include average speed distribution, meteorology, road type distribution, fuel information, and vehicle age, populations, and miles traveled by hour, weekday, and month.

Unlike prior versions, MOVES3 does not adjust hot-running exhaust of PM2.5 from any gasoline or diesel vehicle for ambient temperature. The temperature adjustment for gasoline vehicles of MY 2004 and earlier was removed in the development of MOVES3 (US EPA, 2020a; Table S6). Fig. 5 documents the lack of sensitivity; the temperature dependence simulated by MOVES in the I-95 weighted mean from −5 °C to 20 °C is negligible for both ΔBC/ΔCO (0.04%) and ΔBC/ΔCO2 (0.02%) and arises because of a small indirect adjustment to CO and CO2 emissions due to increased air conditioner use at higher temperatures that exists in both MOVES3 and MOVES2014 (Hall et al., 2020; US EPA, 2020a). The absence of a temperature effect is robust regardless of vehicle type and fuel use. Negating the temperature adjustment factors for hot-running emissions does not appear to represent conditions along I-95 well: the observed increase in emission ratios from −5 to 20 °C at the I-95 NR site, considering all weekdays and hours of the day, was 56% in ΔBC/ΔCO and 83% in ΔBC/ΔCO2 (Fig. 4). With the air conditioning effect at a minimum in the observed temperature range and inferred mobile emissions of CO and CO2 at the NR site exhibiting an insignificant temperature impact (Hall et al., 2020), the temperature effects seen in ΔBC/ΔCO and ΔBC/ΔCO2 are likely the result of an increase in BC emissions with rising temperatures that is not represented within MOVES as configured.

Fig. 5.

Fig. 5

Open in a new tab

MOVES3 simulated ΔBC/ΔCO and ΔBC/ΔCO2 as a function of temperature for the 13 different vehicle types modeled in MOVES. The dominant type of fuel (G for gasoline and D for diesel) is provided next to each vehicle type and is based on county-level data compiled by the Maryland Department of the Environment mobile emissions team. The weighted mean emission ratios were calculated using the fractions of various vehicle types derived from traffic counts collected at the traffic counter site along I-95. S–H indicates short-haul while L-H represents long-haul combination unit trucks.

4. Conclusions

Peak travel restrictions at the beginning of the COVID-19 pandemic in April 2020 enabled us to quantify the effects of traffic flow on BC emissions from diesel trucks. Lower ΔBC/ΔCO and ΔBC/ΔCO2 observed in April 2020 compared to prior years likely resulted from diesel trucks emitting far less BC in 2020 than under typical, congested traffic conditions with frequent acceleration. These results suggest that eliminating traffic jams could reduce highway BC emissions from diesel trucks by up to 70% (i.e., to 30% of normal BC emissions). Similar reductions in mobile BC emissions may be possible with diesel engine designs that reduce turbocharger lag.

Vehicular BC emissions are observed to exhibit a strong sensitivity to ambient temperature. Prior to COVID restrictions we observed a 56% increase in ΔBC/ΔCO as temperature rose from −5 to 20 °C. The overnight period, when the fraction of CIVs was at a maximum, had a strong increase in ΔBC/ΔCO and ΔBC/ΔCO2 with temperature, suggesting BC emissions from diesel trucks are sensitive to temperature. The observed sensitivity may result from emissions control devices used on diesel engines producing elevated BC emissions at either warmer ambient conditions and/or lower air density.

Finally, the default settings of MOVES3 do not adjust for the temperature sensitivity of BC emissions, for any type of vehicle. Future air quality simulations would benefit from incorporation of a realistic temperature dependence of BC emissions within the default settings of this model. The future improvement of air quality model representation of BC emissions as a function of traffic patterns and possibly turbocharger lag is needed to improve the accuracy of air quality simulations from on-road vehicles, and to properly guide emissions engineering and public health policies.

CRediT authorship contribution statement

Dolly L. Hall-Quinlan: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Data curation, Writing – original draft, Visualization. Hao He: Methodology, Conceptualization, Writing – review & editing, Supervision. Xinrong Ren: Methodology, Conceptualization, Investigation, Data curation, Resources, Writing – review & editing, Supervision, Funding acquisition. Timothy P. Canty: Conceptualization, Methodology, Resources, Writing – review & editing, Supervision. Ross J. Salawitch: Conceptualization, Methodology, Writing – review & editing, Supervision, Funding acquisition. Phillip Stratton: Investigation, Resources, Data curation. Russell R. Dickerson: Conceptualization, Methodology, Resources, Writing – review & editing, Supervision, Project administration, Funding acquisition.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

We want to thank the Maryland Department of the Environment for support of the Regional Atmospheric Measurement Modeling and Prediction Program (RAMMPP; Grant #U00P0600702), the National Institute of Standards and Technology (Grant #70NANB19H037) and the National Oceanic and Atmospheric Administration for support of greenhouse gas observations, and the EPA AQS program for the near-road measurements. We also want to thank Barry Balzanna, Abhay Nigam, and their team at the MD State Highway Administration for traffic observations; Joel Dreessen, Ryan Auvil, Aaron Ryan, and David Krask for their help with the NR observations; and Mohamed Khan and the MOVES team at MDE for providing input data and assistance with our MOVES3 simulations.

Footnotes

Appendix A

Supplementary data to this article can be found online at https://doi.org/10.1016/j.atmosenv.2023.119649.

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Multimedia component 1
mmc1.docx (2.5MB, docx)

Data availability

Data will be made available on request.

References

  1. Ahn D.Y., Salawitch R.J., Canty T.P., He H., Ren X.R., Goldberg D.L., Dickerson R.R. The U.S. power sector emissions of CO2 and NOx during 2020: separating the impact of the COVID-19 lockdowns from the weather and decreasing coal in fuel-mix profile. Atmos. Environ. X. 2022;14 doi: 10.1016/j.aeaoa.2022.100168. [DOI] [Google Scholar]
  2. Anderson D.C., Loughner C.P., Diskin G., Weinheimer A., Canty T.P., Salawitch R.J., Worden H.M., Fried A., Mikoviny T., Wisthaler A., Dickerson R.R. Measured and modeled CO and NOy in DISCOVER-AQ: an evaluation of emissions and chemistry over the eastern US. Atmos. Environ. 2014;96:78–87. doi: 10.1016/j.atmosenv.2014.07.004. [DOI] [Google Scholar]
  3. Appel W.K., Bhave P.V., Gilliland A.B., Sarwar G., Roselle S.J. Evaluation of the community multiscale air quality (CMAQ) model version 4.5: sensitivities impacting model performance; Part II-particulate matter. Atmos. Environ. 2008;42(24):6057–6066. doi: 10.1016/j.atmosenv.2008.03.036. [DOI] [Google Scholar]
  4. Ban-Weiss G.A., McLaughlin J.P., Harley R.A., Lunden M.M., Kirchstetter T.W., Kean A.J., Strawa A.W., Stevenson E.D., Kendall G.R. Long-term changes in emissions of nitrogen oxides and particulate matter from on-road gasoline and diesel vehicles. Atmos. Environ. 2008;42(2):220–232. doi: 10.1016/j.atmosenv.2007.09.049. [DOI] [Google Scholar]
  5. Batterman S.A., Zhang K., Kononowech R. Prediction and analysis of near-road concentrations using a reduced-form emission/dispersion model. Environ. Health. 2010;9(1) doi: 10.1186/1476-069X-9-29. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bishop G.A., Haugen M.J., McDonald B.C., Boies A.M. Utah wintertime measurements of heavy-duty vehicle nitrogen oxide emission factors. Environ. Sci. Technol. 2022;56(3):1885–1893. doi: 10.1021/acs.est.1c06428. [DOI] [PubMed] [Google Scholar]
  7. Book E.K., Snow R., Long T., Fang T., Baldauf R. Temperature effects on particulate emissions from DPF-equipped diesel trucks operating on conventional and biodiesel fuels. J. Air Waste Manag. Assoc. 2015;65(6):751–758. doi: 10.1080/10962247.2014.984817. [DOI] [PubMed] [Google Scholar]
  8. Brantley H.L., Hagler G.S.W., Kimbrough E.S., Williams R.W., Mukerjee S., Neas L.M. Mobile air monitoring data-processing strategies and effects on spatial air pollution trends. Atmos. Meas. Tech. 2014;7(7):2169–2183. doi: 10.5194/amt-7-2169-2014. [DOI] [Google Scholar]
  9. Bureau of Transportation Statistics . 2018. Estimated national average vehicle emissions rates per vehicle by vehicle type using gasoline and diesel.https://www.bts.gov/content/estimated-national-average-vehicle-emissions-rates-vehicle-vehicle-type-using-gasoline-and [Google Scholar]
  10. Chan T.W., Meloche E., Kubsh J., Brezny R., Rosenblatt D., Rideout G. Impact of ambient temperature on gaseous and particle emissions from a direct injection gasoline vehicle and its implications on particle filtration. SAE Technical Papers. 2013;2(2):350–371. doi: 10.4271/2013-01-0527. [DOI] [Google Scholar]
  11. Chen A.L.-W., Doddridge B., Dickerson R., Chow J., Mueller P., Quinn J., Butler W. Seasonal variations in elemental carbon aerosol, carbon monoxide and sulfur dioxide: implications for sources. Geophys. Res. Lett. 2001;28(9):1711–1714. [Google Scholar]
  12. Cipollone R., Di Battista D., Vittorini D. Experimental assessment of engine charge air cooling by a refrigeration unit. Energy Proc. 2017;126:1067–1074. doi: 10.1016/j.egypro.2017.08.226. [DOI] [Google Scholar]
  13. Crutzen P. A discussion of the chemistry of some minor constituents in the stratosphere and troposphere. Pure Appl. Geophys. 1973;106:1385–1399. doi: 10.1007/BF00881092. [DOI] [Google Scholar]
  14. Dallmann T.R., Kirchstetter T.W., DeMartini S.J., Harley R.A. Quantifying on-road emissions from gasoline-powered motor vehicles: accounting for the presence of medium- and heavy-duty diesel trucks. Environ. Sci. Technol. 2013;47(23):13873–13881. doi: 10.1021/es402875u. [DOI] [PubMed] [Google Scholar]
  15. Dixit P., Miller J.W., Cocker D.R., Oshinuga A., Jiang Y., Durbin T.D., Johnson K.C. Differences between emissions measured in urban driving and certification testing of heavy-duty diesel engines. Atmos. Environ. 2017;166:276–285. doi: 10.1016/j.atmosenv.2017.06.037. [DOI] [Google Scholar]
  16. Du J., Rakha H.A., Filali F., Eldardiry H. COVID-19 pandemic impacts on traffic system delay, fuel consumption and emissions. International Journal of Transportation Science and Technology. 2021;10(2):184–196. doi: 10.1016/j.ijtst.2020.11.003. [DOI] [Google Scholar]
  17. Durbin T.D., Johnson K., Miller J.W., Maldonado H., Chernich D. Emissions from heavy-duty vehicles under actual on-road driving conditions. Atmos. Environ. 2008;42(20):4812–4821. doi: 10.1016/j.atmosenv.2008.02.006. [DOI] [Google Scholar]
  18. Gaubert B., Bouarar I., Doumbia T., Liu Y., Stavrakou T., Deroubaix A., Darras S., Elguindi N., Granier C., Lacey F., Müller J.-F., Shi X., Tilmes S., Wang T., Brasseur G.P. Global changes in secondary atmospheric pollutants during the 2020 COVID-19 pandemic. J. Geophys. Res. Atmos. 2021;126(8) doi: 10.1029/2020JD034213. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Ghazikhani M., Ebrahim M., Mahian O., Sabazadeh A. Vol. 28. 2013. (Effects of Altitude on the Soot Emission and Fuel Consumption of a Light-Duty Diesel Engine). 2. [DOI] [Google Scholar]
  20. Giakoumis E.G., Zachiotis A.T. A comprehensive comparative investigation of a heavy-duty vehicle's performance, consumption and emissions during eight driving cycles. Int. J. Ambient Energy. 2021;42(1):29–45. doi: 10.1080/01430750.2018.1525578. [DOI] [Google Scholar]
  21. Gilbert R.O. Statistical Methods for Environmental Pollution Monitoring. Van Nostrand Reinhold; 1987. [Google Scholar]
  22. Grange S.K., Farren N.J., Vaughan A.R., Rose R.A., Carslaw D.C. Strong temperature dependence for light-duty diesel vehicle NOx emissions. Environ. Sci. Technol. 2019;53:6587–6596. doi: 10.1021/acs.est.9b01024. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Grieshop A., Lipsky E., Pekney N., Takahama S., Robinson A. Fine particle emission factors from vehicles in a highway tunnel: effects of fleet composition and season. Atmos. Environ. 2006;40:287–298. doi: 10.1016/j.atmosenv.2006.03.064. [DOI] [Google Scholar]
  24. Grimes C.D., Dickerson R.R. Evaluation of a filter-based black carbon (BC) instrument using a brown carbon (BrC) surrogate as well as pure and coated BC surrogates. Aerosol. Sci. Technol. 2021;55(5):501–511. doi: 10.1080/02786826.2021.1878096. [DOI] [Google Scholar]
  25. Hall D.L., Anderson D.C., Martin C.R., Ren X., Salawitch R.J., He H., Canty T.P., Hains J.C., Dickerson R.R. Using near-road observations of CO, NOy, and CO2 to investigate emissions from vehicles: evidence for an impact of ambient temperature and specific humidity. Atmos. Environ. 2020;232(x) doi: 10.1016/j.atmosenv.2020.117558. [DOI] [Google Scholar]
  26. Harkins C., McDonald B.C., Henze D.K., Wiedinmyer C. A fuel-based method for updating mobile source emissions during the COVID-19 pandemic. Environ. Res. Lett. 2021;16(6) doi: 10.1088/1748-9326/ac0660. [DOI] [Google Scholar]
  27. Hickman J., Hassel D., Joumard R., Samaras Z., Sorenson S. 1999. Methodology for calculating transport emissions and energy consumption.https://trimis.ec.europa.eu/sites/default/files/project/documents/meet.pdf [Google Scholar]
  28. Hudda N., Simon M.C., Patton A.P., Durant J.L. Vol. 742. Science of The Total Environment; 2020. (Reductions in Traffic-Related Black Carbon and Ultrafine Particle Number Concentrations in an Urban Neighborhood during the COVID-19 Pandemic). [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Jiang Y., Tan Y., Yang J., Karavalakis G., Johnson K.C., Yoon S., Herner J., Durbin T.D. Understanding elevated real-world NOx emissions: heavy-duty diesel engine certification testing versus in-use vehicle testing. Fuel. 2022;307 doi: 10.1016/j.fuel.2021.121771. [DOI] [Google Scholar]
  30. Joshi A., James S., Meckl P., King G., Laboratories R., Jennings K. Assessment of charge-air cooler health in diesel engines using nonlinear time series analysis of intake manifold temperature. J. Dyn. Syst. Meas. Control. 2009;131(4) doi: 10.1115/1.3023142. [DOI] [Google Scholar]
  31. Kean A.J., Harley R.A., Kendall G.R. Effects of vehicle speed and engine load on motor vehicle emissions. Environ. Sci. Technol. 2003;37(17):3739–3746. doi: 10.1021/es0263588. [DOI] [PubMed] [Google Scholar]
  32. Kendall M.G. Rank Correlation Methods. Charles Griffin; 1975. [Google Scholar]
  33. Kim E., Hopke P.K. Source apportionment of fine particles in Washington, DC, utilizing temperature-resolved carbon fractions. J. Air Waste Manag. Assoc. 2012;54(7):773–785. doi: 10.1080/10473289.2004.10470948. [DOI] [PubMed] [Google Scholar]
  34. Kondo Y., Komazaki Y., Miyazaki Y., Moteki N., Takegawa N., Kodama D., Deguchi S., Nogami M., Fukuda M., Miyakawa T., Morino Y., Koike M., Sakurai H., Ehara K. Temporal variations of elemental carbon in Tokyo. J. Geophys. Res. 2006;111(D12205) doi: 10.1029/2005JD006257. [DOI] [Google Scholar]
  35. Krecl P., Targino A.C., Landi T.P., Ketzel M. Determination of black carbon, PM2.5 particle number and NOx emission factors from roadside measurements and their implications for emission inventory development. Atmos. Environ. 2018;186:229–240. doi: 10.1016/j.atmosenv.2018.05.042. [DOI] [Google Scholar]
  36. Krüger O.O., Holanda B.A., Chowdhury S., Pozzer A., Walter D., Pöhlker C., Andrés Hernández M.D., Burrows J.P., Voigt C., Lelieveld J., Quaas J., Pöschl U., Pöhlker M.L. Black carbon aerosol reductions during COVID-19 confinement quantified by aircraft measurements over Europe. Atmos. Chem. Phys. 2022;22(13):8683–8699. doi: 10.5194/acp-22-8683-2022. [DOI] [Google Scholar]
  37. Laughner J.L., Neu J.L., Schimel D., Wennberg P.O., Barsanti K., Bowman K.W., Chatterjee A., Croes B.E., Fitzmaurice H.L., Henze D.K., Kim J., Kort E.A., Liu Z., Miyazaki K., Turner A.J., Anenberg S., Avise J., Cao H., Crisp D., et al. Societal shifts due to COVID-19 reveal large-scale complexities and feedbacks between atmospheric chemistry and climate change. Proc. Natl. Acad. Sci. USA. 2021;118(46) doi: 10.1073/pnas.2109481118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Li P., Lü L. Research on a China 6b heavy-duty diesel vehicle real-world engine out NOx emission deterioration and ambient correction using big data approach. Environ. Sci. Pollut. Res. Int. 2021 doi: 10.1007/s11356-021-15778-2. [DOI] [PubMed] [Google Scholar]
  39. Lopez-Coto I., Ren X., Karion A., McKain K., Sweeney C., Dickerson R.R., McDonald B.C., Ahn D.Y., Salawitch R.J., He H., Shepson P.B., Whetstone J.R. Carbon monoxide emissions from the Washington, DC, and Baltimore metropolitan area: recent trend and COVID-19 anomaly. Environ. Sci. Technol. 2022;56(4):2172–2180. doi: 10.1021/acs.est.1c06288. [DOI] [PubMed] [Google Scholar]
  40. Lyu Y., Olofsson U. On black carbon emission from automotive disc brakes. J. Aerosol Sci. 2020;148 doi: 10.1016/j.jaerosci.2020.105610. [DOI] [Google Scholar]
  41. Mann H.B. Nonparametric Tests Against Trend. Econometrica. 1945;13(3):245–259. [Google Scholar]
  42. McCaffery C., Zhu H., Tang T., Li C., Karavalakis G., Cao S., Oshinuga A., Burnette A., Johnson K.C., Durbin T.D. Vol. 784. Science of The Total Environment; 2021. (Real-world NOx Emissions from Heavy-Duty Diesel, Natural Gas, and Diesel Hybrid Electric Vehicles of Different Vocations on California Roadways). [DOI] [PubMed] [Google Scholar]
  43. Naser T.M., Kanda I., Ohara T., Sakamoto K., Kobayashi S., Nitta H., Nataami T. Analysis of traffic-related NOx and EC concentrations at various distances from major roads in Japan. Atmos. Environ. 2009;43(15):2379–2390. doi: 10.1016/j.atmosenv.2009.02.002. [DOI] [Google Scholar]
  44. O'Driscoll R., Stettler M., Molden N., Oxley T., Apsimon H. Real world CO2 and NOx emissions from 149 Euro 5 and 6 diesel, gasoline and hybrid passenger cars. Sci. Total Environ. 2018;621:282–290. doi: 10.1016/j.scitotenv.2017.11.271. [DOI] [PubMed] [Google Scholar]
  45. OECD. ECMT . 2006. Speed management.https://www.itf-oecd.org/sites/default/files/docs/06speed.pdf [Google Scholar]
  46. Park S.S., Kozawa K., Fruin S., Mara S., Hsu Y.-K., Jakober C., Winer A., Herner J. Emission factors for high-emitting vehicles based on on-road measurements of individual vehicle exhaust with a mobile measurement platform. J. Air Waste Manag. Assoc. 2011;61(10):1046–1056. doi: 10.1080/10473289.2011.595981. [DOI] [PubMed] [Google Scholar]
  47. Pusede S.E., Cohen R.C. On the observed response of ozone to NOx and VOC reactivity reductions in San Joaquin Valley California 1995-present. Atmos. Chem. Phys. 2012;12(18):8323–8339. doi: 10.5194/acp-12-8323-2012. [DOI] [Google Scholar]
  48. Rakopoulos C., Giakoumis E. 2009. Diesel engine transient operation. [DOI] [Google Scholar]
  49. Saha P.K., Khlystov A., Snyder M.G., Grieshop A.P. Characterization of air pollutant concentrations, fleet emission factors, and dispersion near a North Carolina interstate freeway across two seasons. Atmos. Environ. 2018;177(January):143–153. doi: 10.1016/j.atmosenv.2018.01.019. [DOI] [Google Scholar]
  50. Sather M.E., Cavender K. Trends analyses of 30 years of ambient 8 hour ozone and precursor monitoring data in the South Central U.S.: progress and challenges. Environ. Sci.: Process. Impacts. 2016;18(7):819–831. doi: 10.1039/C6EM00210B. [DOI] [PubMed] [Google Scholar]
  51. Schauer J.J., Christensen C.G., Kittelson D.B., Johnson J.P., Watts W.F. Impact of ambient temperatures and driving conditions on the chemical composition of particulate matter emissions from non-smoking gasoline-powered motor vehicles. Aerosol. Sci. Technol. 2008;42(3):210–223. doi: 10.1080/02786820801958742. [DOI] [Google Scholar]
  52. Sher E., editor. Handbook of Air Pollution from Internal Combustion Engines: Pollutant Formation and Control. first ed. Academic Press; 1998. Chapter 10: combustion-related emissions in CI engines. [Google Scholar]
  53. US Department of Transportation Federal Highway Administration . 2014. Verification, Refinement, and Applicability of Long-Term Pavement Performance Vehicle Classification Rules.https://www.fhwa.dot.gov/publications/research/infrastructure/pavements/ltpp/13091/002.cfm [Google Scholar]
  54. US EPA . 2015. Emission Adjustments for Temperature , Humidity , Air Conditioning , and Inspection and Maintenance for on - Road Vehicles in MOVES2014.https://www.epa.gov/moves/moves-technical-reports [Google Scholar]
  55. US EPA . 2015. MOVES2014a User Guide.https://www.epa.gov/moves/moves2014a-latest-version-motor-vehicle-emission-simulator-moves [Google Scholar]
  56. US EPA . 2016. Integrated Review Plan for the National Ambient Air Quality Standards for Particulate Matter.https://www3.epa.gov/ttn/naaqs/standards/pm/data/201612-final-integrated-review-plan.pdf [Google Scholar]
  57. US EPA . 2020. Emission Adjustments for Temperature, Humidity, Air Conditioning and Inspection and Maintenance for Onroad Vehicles in MOVES3.https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P1010M29.pdf [Google Scholar]
  58. US EPA . 2020. The 2019 EPA automotive trends report.https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P100YVFS.pdf [Google Scholar]
  59. US EPA . 2021. Basics of climate change.https://www.epa.gov/climatechange-science/basics-climate-change#aerosols [Google Scholar]
  60. US EPA . 2021. Particulate matter (PM) basics.https://www.epa.gov/pm-pollution/particulate-matter-pm-basics#effects [Google Scholar]
  61. US EPA . 2021. Regional Fuels and the Fuel Wizard in MOVES3; pp. 1–67.https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P10119R7.pdf [Google Scholar]
  62. US EPA . 2022. National Emissions Inventory (NEI) Data; p. 2017.https://www.epa.gov/air-emissions-inventories/2017-national-emissions-inventory-nei-data [Google Scholar]
  63. Vaisala . 2012. Vaisala Weather Transmitter WXT 520.https://www.vaisala.com/sites/default/files/documents/M210906EN-C.pdf [Google Scholar]
  64. Wang J.M., Jeong C.H., Hilker N., Shairsingh K.K., Healy R.M., Sofowote U., Debosz J., Su Y., McGaughey M., Doerksen G., Munoz T., White L., Herod D., Evans G.J. Near-road air pollutant measurements: accounting for inter-site variability using emission factors. Environ. Sci. Technol. 2018;52(16):9495–9504. doi: 10.1021/acs.est.8b01914. [DOI] [PubMed] [Google Scholar]
  65. Weber C., Sundvor I., Figenbaum E. Comparison of regulated emission factors of Euro 6 LDV in Nordic temperatures and cold start conditions : diesel- and gasoline direct-injection. Atmos. Environ. 2019;206:208–217. doi: 10.1016/j.atmosenv.2019.02.031. [DOI] [Google Scholar]
  66. Yadav V., Ghosh S., Mueller K., Karion A., Roest G., Gourdji S.M., Lopez-Coto I., Gurney K.R., Parazoo N., Verhulst K.R., Kim J., Prinzivalli S., Fain C., Nehrkorn T., Mountain M., Keeling R.F., Weiss R.F., Duren R., Miller C.E., Whetstone J. The impact of COVID-19 on CO2 emissions in the los angeles and Washington DC/Baltimore metropolitan areas. Geophys. Res. Lett. 2021;48(11) doi: 10.1029/2021GL092744. [DOI] [PMC free article] [PubMed] [Google Scholar]
  67. Yang J., Wen Y., Wang Y., Zhang S., Pinto J.P., Pennington E.A., Wang Z., Wu Y., Sander S.P., Jiang J.H., Hao J., Yung Y.L., Seinfeld J.H. From COVID-19 to future electrification: assessing traffic impacts on air quality by a machine-learning model. Proc. Natl. Acad. Sci. USA. 2021;118(26) doi: 10.1073/pnas.2102705118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  68. Zeng N., Han P., Liu Z., Liu D., Oda T., Martin C., Liu Z., Yao B., Sun W., Wang P., Cai Q., Dickerson R., Maksyutov S. Global to local impacts on atmospheric CO2 from the COVID-19 lockdown, biosphere and weather variabilities. Environ. Res. Lett. 2022;17(1) doi: 10.1088/1748-9326/ac3f62. [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Multimedia component 1
mmc1.docx (2.5MB, docx)

Data Availability Statement

Data will be made available on request.

Articles from Atmospheric Environment (Oxford, England : 1994) are provided here courtesy of Elsevier

RESOURCES