Modeling Spatial Variations of Black Carbon Particles in ... · quality studies have characterized open highway conditions, i.e., large roads without any major structures inthe close

Published: November 15, 2011

r 2011 American Chemical Society 312 dx.doi.org/10.1021/es201938v | Environ. Sci. Technol. 2012, 46, 312–319

ARTICLE

pubs.acs.org/est

Modeling Spatial Variations of Black Carbon Particles in an UrbanHighway-Building EnvironmentZheming Tong,† Yan Jason Wang,† Molini Patel,‡ Patrick Kinney,‡ Steven Chrillrud,§ and K. Max Zhang*,†

†Sibley School of Mechanical and Aerospace Engineering, Cornell University, Ithaca, New York 14853, United States‡Department of Environmental Health Sciences, Mailman School of Public Health, Columbia University, 60 Haven Avenue, New York,New York 10032, United States§Lamont-Doherty Earth Observatory, Columbia University, Palisades, New York 10964, United States

bS Supporting Information

1. INTRODUCTION

In recent years, epidemiological and toxicological studiesworldwide have suggested an association between human ex-posure to traffic-related air pollutants and a range of adverserespiratory and cardiovascular health effects.1 Many near-road airquality studies have characterized open highway conditions, i.e.,large roads without any major structures in the close vicinity.2�4

Nevertheless, the presence of buildings near highways, referredto here as the highway-building environment, is common inurban areas. In a highway-building environment, roadway con-figuration and building geometry both affect the flow, whichdiffers from a street canyon environment, where buildingsdominate the overall flow.

Thus, in order to accurately assess human exposure to traffic-related air pollutants in the highway-building environments, weneed to address the following two questions: 1) How dobuildings affect the transport and transformation of traffic-relatedair pollutants near roadways? And 2)What are the horizontal andvertical profiles of traffic-related air pollutants near buildingsurfaces? This paper represents the first effort to answer thosetwo questions. We applied and improved the Comprehensive

Turbulent Aerosol Dynamics and Gas Chemistry (CTAG)model to simulate the spatial variation of near-road black carbon(BC) concentration in a highway-building environment nearhighway I-87 in the South Bronx, NY, and compared themodeling results with the measurements conducted by Pateland co-workers in 2004.5

Among the five boroughs of NYC, the Bronx has rankedhighest in both asthma hospitalizations and deaths in recentyears. Between 1990 and 2000, the asthma rates decreased only3% in the Bronx, as compared to 39% and 35% reduction inBrooklyn and Manhattan during the same period according toNew York City Department of Health and Mental Hygiene.6

Although the origin of asthma is multifactorial, recent studies linkasthma to exposure to diesel particulate matter from heavy-dutydiesel traffic in the area.7�9 It is estimated that 66% of thepopulation in the Bronx is living within 150 m of major roads,

Received: June 7, 2011Accepted: November 15, 2011Revised: November 7, 2011

ABSTRACT: Highway-building environments are prevalent in metropolitan areas.This paper presents our findings in investigating pollutant transport in a highway-building environment by combing field measurement and numerical simulations. Weemploy and improve the Comprehensive Turbulent Aerosol Dynamics and GasChemistry (CTAG) model to simulate the spatial variations of black carbon (BC)concentrations near highway I-87 and an urban school in the South Bronx, New York.The results of CTAG simulations are evaluated against and agree adequately with themeasurements of wind speed, wind directions, and BC concentrations. Our analysissuggests that the BC concentration at the measurement point of the urban schoolcould decrease by 43�54% if roadside buildings were absent. Furthermore, wecharacterize two generalized conditions in a highway-building environment, i.e.,highway-building canyon and highway viaduct-building. The former refers to thecanyon between solid highway embankment and roadside buildings, where thespatial profiles of BC depend on the equivalent canyon aspect ratio and flow recirculation. The latter refers to the area between ahighway viaduct (i.e., elevated highway with open space underneath) and roadside buildings, where strong flow recirculation isabsent and the spatial profiles of BC are determined by the relative heights of the highway and buildings. The two configurations mayoccur at different locations or in the same location with different wind directions when highway geometry is complex. Our studydemonstrates the importance of incorporating highway-building interaction into the assessment of human exposure to near-road airpollution. It also calls for active roles of building and highway designs in mitigating near-road exposure of urban population.

313 dx.doi.org/10.1021/es201938v |Environ. Sci. Technol. 2012, 46, 312–319

Environmental Science & Technology ARTICLE

including interstate, state, and county highways, access ramps,and arterials, and 91% within 300 m.10 BC particles are not only atracer of diesel traffic but are also associated with cardiopulmo-nary health effects.11�14 As BC particles cause a direct positiveradiative acting as a short-lived climate forcer, control measuresof particulatematter (PM) emissions that lead to reduction of theBC component will likely have a positive impact on humanhealth, resulting in a cobenefit of PM reductions.

This paper is organized as follows. We start with a descriptionof our numerical modeling approach, followed by a brief intro-duction of the field measurement. Then, we compare thepredicted and measured wind speed, wind direction, and BCconcentrations at the sampling site. Next, we elaborate on howhighway configurations and buildings affect the spatial variationof BC concentrations based on our numerical simulations.Finally, we discuss the implications of our study in terms ofhuman exposure assessment and highway design.

2. BC PROPERTIES, FIELD MEASUREMENTS, ANDSIMULATED CASES

In this study, black carbon is defined as the primary constitu-ent of PM2.5 that is responsible for light absorption of particles inthe atmosphere.15 We model BC particles as inert species, whichexperiences no chemical transformation within the time scale ofconcern (seconds to minutes). Because most BC occurs assubmicrometer particles, they are expected to adopt the flowvelocity very quickly (i.e., with small Stokes number), it isreasonable to assume that they will disperse like gaseousspecies.13,14 Due to their relatively low concentrations, we furtherassume that BC particles do not affect the turbulent flow in theatmosphere.16

A detailed description of the field measurements was providedby Patel and co-workers.5 A brief summary is presented here. Thefield measurements were conducted in an urban school, referred

as U2, in the South Bronx, from February to March, 2004. U2 islocated approximately 30 m east of I-87, a highway with annualaverage daily traffic of 85,000 vehicles. BC concentrations werecollected using a dual-beam Aethalometer (Model AE-21, MageeScientific, Berkeley, CA) operated at 4 LPM and using a sizeselective cyclone for PM2.5 (KTL cyclone, BGI, Waltham, MA).The unit was placed in an empty classroom with the samplinginlet located 0.9 m outside from the school wall, and the samplinginlet was protected from rain with a stainless steel rainhat. Thusthe BC concentrations weremeasured at a fixed height and a fixeddistance from I-87. A weather station (Vantage Pro Model6150C, Davis Instrument Corp., Hayward, CA) was installedon the school rooftop to monitor and record temperature,relative humidity, barometric pressure, wind speed and direction,and precipitation. At U2, traffic data were collected using a videocamera adjacent to highway. Vehicles counts were obtainedmanually by watching daytime videos. Two categories of vehiclesare counted: Category I, mostly gasoline powered vehiclesincluding passenger cars, vans, sport utility vehicles, pick-uptrucks, and small double-axle trucks; and Category II, primarilydiesel powered vehicles including large trucks with more thantwo axles, and buses.

As we are primarily interested in the effects of highwaypollution on the nearby environment, we focus only on condi-tions with wind blowing from I-87 to the U2 site, i.e., with winddirection varying from south southwest (SSW or 202.5�) tonorth northwest (NNW or 337.5�). For the I-87 segment next toU2, some portion is elevated with solid embankment, while theother portion is elevated with open space (Figure 1). Thiscomplex geometry, combined with different wind directions,leads to distinctly different spatial distributions of BC in thishighway-building environment (discussed in Section 4.2). As themeasured BC concentrations are only available as hourly aver-aged data, we selected 34 cases with stable wind and trafficconditions, modeled using steady-state simulations. These 34

Figure 1. Sketch of the modeling domain. Encircled are close views of the transition portions of the highway configuration. There are two decliningramps, which are solid embankments, at both ends of I-87 in the modeling domain. The middle portion of I-87 is also elevated but with open spaceunderneath, referred to as a highway viaduct.



cases represent different hours from 6 a.m. to 4 p.m. betweenMarch 12th and 22nd, 2004, among which 14 cases were duringmorning rush hours. Details of wind directions/speed, trafficvolumes, and BC emission rates are listed in Table S1 in theSupporting Information.

3. MODELING METHOD

The CTAGmodel simulates the transport and transformationof exhaust particles from points of emissions to ambientbackground.2�4 Figure S1 in the Supporting Information illus-trates the structure and components of the CTAG model. Thetransport portion of the CTAG model is called CFD-VIT-RIT,which has been applied to study the effects of highway config-uration on near-road air quality under the open highway condi-tions. VIT stands for vehicle induced turbulence, and RIT standsfor road induced turbulence.2 In this study, we expand thecapability of CFD-VIT-RIT to simulate near-road air qualityin highway-building environments. ANSYS FLUENT 12.1 isemployed as a turbulence solver for the CTAG model.17

We compared several turbulence models including standardk-epsilon, RNG k-epsilon, and realizable k-epsilon models, whichshows that three models yield similar results (Figure S2). Weselected the steady standard k-epsilon turbulence model as it hasbeen demonstrated more computationally stable and less inten-sive for isothermal flow.18 Several user-defined functions (UDF)are created to simulate the highway-buildings environment.3.1 Modeling Domain. The modeling domain for our study

was chosen as an approximate 800 m� 800 m� 150 m block inthe South Bronx (Figure 1). It consists of highway I-87, a surfacestreet, U2, several low-rise buildings, and a high-rise buildingabout 45 m tall. More details regarding the modeling domain canbe found in the Supporting Information (Section S2).3.2. Boundary Conditions.The ambient wind and turbulence

profiles representing the urban lower atmospheric boundary arecreated using surface meteorological measurement data recordedat La Guardia Airport (6.4 km from the U2 site) and upper airdata provided by the National Oceanic and Atmospheric Admin-istration’s Radiosonde Database. Detailed discussions on theboundary conditions can be found in the Supporting Information(Section S3).3.3. BC Emission Rate and Emission Zone Modeling. The

BC emission rates are estimated by the following procedures.First, the PM2.5 emission factors reflecting the vehicle conditionsin New York are obtained from the New York State Departmentof Transportation (NYSDOT). These emission factors weregenerated using the MOBILE6.2 and are required for microscaleair quality analyses in NYSDOT projects.19 Second, PM2.5

emission factor for each vehicle type is multiplied by its corre-sponding traffic volumes. The two vehicle types, Categories I andII (defined in Section 2), are further classified by vehicle type

distribution from NYSDOT.19 Next, the relative fractions of BCin PM2.5 emissions from mobile sources were obtained fromSPECIATE (Version 4.0), USEPA’s repository of particu-late matter (PM) speciation profiles of air pollution sources.SPECIATE reports elemental carbon (EC)measurements ratherthan BC. EC is typically defined as refractory carbon by thermal/optical method, while BC is defined as light-absorbing carbon byoptical methods.20 Typically, BC optical and EC thermo-opticalmethods are highly correlated.21,22 A recent study by Yan et al.showed a strong correlation between the two methods in NewYork City.23 As there is no simple conversion factor between BCand EC, many studies assumed that BC and EC are equivalent,the same assumption adopted in this study.21,24,25 The averageBC/PM2.5 fraction is 20% for Category I, and 65% for CategoryII, which are consistent with the findings from a number ofexperimental studies.20,26,27

A uniformly mixed traffic emission zone was created bymerging the turbulence zone on the surface of I-87. The heightof the emission zone was estimated by the ratio of recorded trafficvolume of Category I to Category II.3 VIT is generated bymodeling real-shaped vehicles in a traffic stream based on thetraffic volumes and the ratio of Category I to Category II. Toestimate RIT, appropriate surface roughness lengths are selectedfor the highway surface and estimated temperature differencesbetween highway surface and air is applied based on the weatherdata in New York.2 Values of turbulent kinetic energy are thencalculated due to velocity gradients.3.4. Background BC Concentrations. The background BC

concentrations for the modeling period are taken as the values ofelemental carbon (EC) concentrations at the ambient air mon-itoring station operated by New York State Department ofEnvironmental Conservation (NYSDEC), located at a New YorkCity Intermediate School (IS 52), 681 Kelly Street off ProspectAvenue in the South Bronx.28 With our assumption that EC andBC are equivalent, the daytime average BC concentration isestimated to be 0.64 μg m�3 for the corresponding period of thesimulation.

4. RESULTS

4.1. Model Evaluation. Our model is evaluated againstmeasured wind speeds, wind directions, and BC concentrationsunder three types of highway-building environments (i.e., high-way-building canyon, highway viaduct-building, and transitionalcondition, which will be elaborated in Section 4.2), respectively.The performance metrics include Mean Normalized Error(MNE), Mean Normalized Bias (MNB), Mean Fraction Error(MFE), andMean Fraction Bias (MFB). UnlikeMNB andMNE,MFE and MFB do not assume that the field measurement isabsolutely truth as the error and bias are normalized by the

Table 1. Summary of Performance Metrics for Model Evaluation

highway viaduct-buildinga highway-building canyonb transitional conditionc

statistical methods wind speed wind direction BC wind speed wind direction BC wind speed wind direction BC

Mean Normalized Error (MNE) 9.99% 4.12% 21.24% 9.79% 4.30% 16.83% 7.42% 4.90% 16.44%

Mean Normalized Bias (MNB) 8.19% 1.54% �16.44% 3.09% 3.53% �8.20% �4.72% 1.24% �12.25%

Mean Fraction Error (MFE) 7.62% 2.65% 15.87% 6.80% 2.77% 11.77% 5.40% 3.17% 12.08%

Mean Fraction (MFB) 6.47% 0.90% �14.08% 2.70% 2.24% �6.46% �3.66% 0.67% �9.38%a 15 cases. b 13 cases. c 6 cases out of total 34 cases.



average of the model and observation. It is considered as a moreappropriate way to judge the performance of themodel.29 Table 1summarizes the performance metrics for predictions of windspeeds, wind directions, and BC concentrations. To the best ofknowledge, our study is the first one that has introduced MFEand MFB into local-scale air quality modeling.Wind Speed and Directions. Figure 2a and b compares the

simulated and measured wind speed and directions, which showsa good agreement. Listed in Table 1, theMFEs andMFBs for thewind speed prediction are 5.40�7.62% and �3.66�6.47%,respectively. For direction prediction, MFEs and MFBs are2.65�3.17% and 0.67�2.24%, respectively. This suggests thatthe CTAG model is capable of capturing the flow fields in thehighway-building environment. It should be noted that theresolution of the measured wind direction is 22.5�, which resultsin multiple predicted wind directions corresponding to the samemeasured values (Figure 2b). The discrepancies between thesimulated results and measurements are due to factors such asidealized building and roadway geometries, estimated ambientwind profiles of the lower atmospheric boundary layer, and theturbulence model. The capability of capturing the flow fields iscritical to our study. Modeled as an inert species, BC’s transportis solely governed by turbulence mixing and meteorology. Inother words, the absolute values of BC concentrations vary withBC emission factors, but the shapes of horizontal and vertical BCprofiles are determined by the flow fields, which are the emphasisof this paper.BC Concentrations. Figure 2c compares the simulated and

measured BC concentrations. The MFEs and MFBs for theCTAG model are �6.46 to 11.77% and �6.46 to �14.08%,respectively. The relative narrow ranges of the two metricssuggest the model performances for the three types of conditionsare similar. Boylan and Russell proposed the criteria for accept-able performance for regional-scale air quality modeling as thatboth the MFE is less than or equal to +75% and the MFB is lessthan or equal to (60% for major components of PM2.5.

29 Inaddition, the MNEs (from 16.44 to 21.24%) and MNBs (from�8.20 to �16.44%) in BC concentrations are comparable tothose reported by Stein and co-workers in modeling local-scalebenzene concentrations (MNEs from 46.37 to 79.87%; MNBsfrom 9.18 to 34.07%).30 It should be noted that both BC andbenzene were modeled as inert species. Therefore, the simulatedBC concentrations are in an adequate agreement with themeasurements. The further analysis of our modeling resultssuggests that BC concentrations are systematically underpre-dicted for a majority (11 out of 14) of the morning rush hourscases shown as solid circles in Figure 2c. A likely contributingfactor to the difference between the real-world and modeled datais rooted from our adoption of NYC-regionally average vehicletype distribution in our emission estimates. The traffic videosindicate a large fraction of heavy-duty diesel vehicles in theCategory II vehicles during the morning rush hours compared tothe assumed average vehicle type distribution. However, thequality of the traffic videos due to the poor viewing angle makes itvery difficult to further classify the vehicle size classes. Inaddition, morning rush hours may lead to transient drivingconditions, which can result in higher emissions than cruisingconditions. But their effects are expected to be minor because theperiods we selected hadmostly cruise conditions. As illustrated inthe Supporting Information (Section S9), increasing BC emis-sion rates during the morning rush hours significantly improvesthe modeling performance.

In summary, the CTAG model is adequate to resolve theflow fields and BC concentrations in the highway-building

Figure 2. a) Simulated vsmeasuredwind speed, b) simulated vsmeasuredwind direction, and c) simulated vs measured BC concentration.



environment under steady-state conditions in spite of the uncer-tainties in BC emissions. To generalize the results, the BCconcentrations in the following sections are normalized by theconcentrations on the edge the highway, which to a large extenteliminates the effects of uncertainties in emission estimates.4.2. Characterizations of the Highway-Building Environ-

ment. The evaluated CTAG model provides a valuable numericaltool for characterizing the highway-buildings environment, whichwecategorize into twomain conditions: 1) highway-building canyon and2) highway viaduct-building. Highway-building canyon describes thecanyon between solid, elevated highway embankment and roadsidebuildings, where prevailing airflow is forced to flow over the highway(Figure 3a). Highway viaduct-building, on the other hand, refers tothe space between an elevated highway supported by columns (i.e., ahighway viaduct) and roadside buildings, where wind can streamfrom the space underneath the highway (Figure 3b). Both condi-tions occur at the U2 site due to the evolving highway geometry ofI-87 combined with different wind directions.Street Canyon. Highway-building canyon is analogous to the

familiar street canyon terminology. There are two aspect ratiosassociated with a street canyon, H/W and L/H, where H is theaverage height of building, W is the canyon width, and L is lengthof the building. Based on H/W and L/H, the characteristics ofthe flow field and pollutant transport over street canyon can beclassified into three categories: isolated roughness flow, wakeinterference flow, and skimming flow.31 Typically, for canyonswhere H/W < 0.3, only the wakes are disturbed (isolatedroughness flow). For narrower canyons where H/W≈ 0.5, thereis a strong interaction between the lee-vortex and windward

vortex. Thus, the downward flow of the leeward eddy isreinforced by the windward vortex. In the case of H/W ≈ 1,where isolated roughness flow transitions to skimming flow,most of the flow does not enter the canyon, and it is similar tosimple driven cavity flow where a single vortex is developed.32 Interms of pollutant dispersion, field studies have shown thepollutant concentration is directly linked to the flow field inthe canyon, which is described in detail in the SupportingInformation (Section S7).Highway-Building Canyon. The highway-building canyon con-

figuration modeled resembles a step-up condition of the streetcanyon in which the height of urban school U2 is greater than theelevation of the highway embankment.33 Since H and L are notequal on the upwind and downwind sides, we estimate H/W ≈0.65 by using the average height of highway and U2, and L/H > 6based on the geometry of U2 alone. Thus, the highway-buildingcanyon can be classified somewhere between wake interferenceflow and skimming flow.31 Figure 3a illustrates the vortexcirculation due to flow separation inside the highway-buildingcanyon between the highway and the urban school (U2) at anambient wind direction of 225� (Figure 3b will be described inthe Highway Viaduct-Building discussion). The flow field iscomparable to previous studies on skimming flow over a streetcanyon.34 Figure 4a depicts the vertical profiles of BC concen-tration inside the highway-building canyon between the highwayand the front gate of U2. The height is normalized by theelevation of the highway, and the BC concentration is normalizedby the concentration at the edge of the highway. On the leeward

Figure 3. Flow field between highway and urban school (U2) varies with wind direction.Highway-Building Canyon conditions (a) occurmostly betweensolid highway embankment and U2, and Highway Viaduct-Building conditions (b) occur mostly between viaduct (middle section of highway shown inFigure 1 with open space underneath) and U2.



wall (2 m from the highway), the peak concentration occurs atthe level slightly greater than the emission zone. This smallrise can be explained by the buoyancy of exhaust produced bythe greater temperature on the road surface and the barrier at theedges of highway. This vertical profile is consistent with thefinding in Wang and Zhang.2 The vertical BC concentrationgradient at the core of vortex circulation (∼10m from the edge ofthe highway) is nearly zero below the highway elevation,indicating a well-mixed region. At the windward wall (the frontgate of U2), the maximum concentration occurs at the groundlevel. The vertical profile of BC concentration can be approxi-mated as a simple exponential function.Highway Viaduct-Building. Illustrated in Figure 3b, the flow

field for the highway viaduct-building condition is very differentfrom the highway-building canyon condition. There is no primarycirculation vortex as shown in Figure 3a. The bulk of the flowfrom the bottom of highway is deflected upward once it hits the

windward wall of U2, and a small windward eddy is formed(Figure 3b). Figure 4b depicts the normalized vertical BCconcentration as a function of distance from the highway. Thedeflected airflow results in the maximum BC concentrationslightly above the top of U2 (“School Front Gate” in Figure 4b).However, the vertical profile of the BCconcentration is comparableto highway-building canyon near the edge of highway.Transitional Condition. At ambient wind direction around

270�, the highway-building environment is characterized by atransitional condition between the highway-building canyon andhighway viaduct-building conditions (Figure 1). Illustrated as“Baseline” in Figure 5, the vertical profile at the front gate of U2shows that maximum BC concentration occurs around the sameheight of the highway (∼8 m), between the ground level, wherethe maximum concentration occurs in the highway-buildingcanyon condition, and the height of U2, where the maximumconcentration occurs in the highway viaduct-building condition.

Figure 4. Simulated vertical BC concentration as a function of distance away from the edge of the highway for a) highway-building canyon and b)highway viaduct-building conditions. Only highway emission source is considered. BC concentration is normalized by the concentration at the edge ofhighway, and vertical height is normalized by the elevation of the highway modeled. Wind direction is 225� for a) and 292.5� for b).

Figure 5. Simulated normalized vertical BC profiles at the front gate of the urban school (U2) under the transitional condition with a) increasing aspectratio H/W (increasing the height of U2 from a baseline of 15 m to 25 m, 35 m, and 45 m) and b) decreasing aspect ratio H/W (translate U2 furtherhorizontally from a baseline of 30 m away from highway to 40 m, 50 m, and 60 m). All simulations are under the ambient wind direction of 270�. BCconcentration is normalized by the concentration at the edge of the highway, and the vertical height is normalized by the elevation of the highwaymodeled.



For this transitional condition, the aspect ratio of a highway-building canyon significantly affects pollutant dispersion. Wevaried the aspect ratios (H/W) to further explore its impact. Thefirst set of simulations vary the height of U2 from 15 m to 25 m,35 m, and 45 m, corresponding to aspect ratios of 0.58, 0.75, and0.92 respectively, which redefines the flow regime to skimmingflow. Figure 5a shows the vertical profiles of BC concentration atthe windward wall (i.e., the front gate of U2). The variation of theflow fields shown in velocity vectors is available in the SupportingInformation (Figure S4). As illustrated, BC concentration at thewindward wall decreases as height increases, and maximumconcentration of all the profiles occurs at the ground level dueto circulation vortex. Similar to a step-up canyon, this occursbecause the strength of eddies in the canyon grows as windwardheight increases. Growing eddies increases the dilution andventilation rate, thus reducing the pollutant concentration.33 Itshould be noted that this trend is different from that observed inthe case of even street canyons (buildings with similar heights onboth sides of the streets), where large aspect ratios (H/W)typically hinder the entrainment of ambient air, resulting inelevated pollutant concentrations.35,36

The second set of simulations increase the distance betweenthe urban school and highway to 40m, 50m, and 60m, effectivelyreducingH/W from 0.42 to 0.31, 0.25, and 0.21 correspondingly.The flow field evolves toward isolated roughness flow, while theinteraction of the leeward and windward vortex weakens. AtH/W = 0.21, the flow field resides in the regime where there arecorotative vortexes, a leeward eddy near highway embankmentdue to cavity of low pressure, and a smaller windward eddy infront of U2. The maximum normalized BC concentration at theschool front gate declines from 0.24 to 0.10 as H/W decreases to0.21. Further decreasing H/Wwould result in independent wakeflow, which substantially reduces BC concentration. The varia-tion of the flow fields shown in vectors is available in theSupporting Information (Figure S5).Open Highway. In order to estimate how different the BC

concentrations are with buildings relative to a situation withoutairflow obstructions, we conducted simulations of the 34 casesdiscussed in Section 2 by removing all the roadside structures inour modeling domain while keeping everything else intact,referred to as open highway environments. As shown in Figure 4,the vertical BC profiles in the open highway environments aredistinctly different than those in the highway-building environ-ments (for both highway-building canyon and highway viaduct-building). It needs to be emphasized that all concentrationspresented in Figure 4a and b are normalized by the sameconcentration (i.e., at the edge of the highway). For the 34 cases,the reduction in BC concentrations at the measurement pointresulting from building removal varies from 43 to 54%, indicatinga significant effect of roadside buildings on near-road air quality.

5. IMPLICATION

The CTAG simulations combined with field measurementsindicate that flow fields and pollutant transport can vary drama-tically in highway-building environments. The CTAG modeldemonstrates adequate agreement with measurements at U2,and it is able to predict the flow field and spatial gradient of traffic-related air pollutants in complex highway-building environmentsunder steady-state conditions. Under unsteady-state conditionssuch as those with intermittent wind speed, direction, or rapidchanging traffics, steady-state profiles may not be achieved. The

capability of the CTAG model in capturing the unsteady-stateconditions has not been tested yet due to the limitation in thefield data sets.

We show that the spatial variation of roadside pollutantconcentration in the highway-building environment is domi-nated by the configuration of the highway, buildings, the relativedistances between them, and the prevailing wind conditions. Weevaluated two main configurations in a highway-buildings en-vironment: highway-building canyon and highway viaduct-building,which lead to distinctly different spatial variations of air pollu-tants, vertically and horizontally. The two configurations mayoccur at different locations or even in the same location withdifferent wind directions when highway geometry is complex.Furthermore, our simulation results suggest that BC concentra-tion reaches its background level at height about 5 times theheight of the highway regardless of types of highway-buildingconditions.

Our study implies that human exposure to traffic-related airpollutants in a highway-buildings environment, prevalent inurban areas, is governed by both highway and building config-urations. Disregarding roadside buildings or highway configura-tions in a highway-building environment to assess near-roadexposure may lead to significant discrepancies. The findings fromthis study also demonstrate the need for a close collaborationbetween urban and transportation planners to tackle the near-road air pollution problems in highway-building environments.This kind of collaboration will help toward creating a healthyenvironment for an urban population.

’ASSOCIATED CONTENT

bS Supporting Information. Figures S1�S6, Tables S1�S3,and text. This material is available free of charge via the Internet athttp://pubs.acs.org.

’AUTHOR INFORMATION

Corresponding Author*E-mail: [email protected].

’ACKNOWLEDGMENT

Cornell researchers would like to acknowledge New YorkState Energy Research and Development Authority (NYSERDA)for funding support andwould also like to thank Pruek Pongprueksa,formerly at Cornell University, for his early contribution to thisstudy, Oliver Rattigan, Shida Tang, and Brian Frank at New YorkState Department of Environmental Conservation for providingus the ambient elemental carbon data. The Columbia researchersthank support from the National Institute of EnvironmentalHealth Science (ES11379, ES015905, and ES09089).

’REFERENCES

(1) Traffic-Related Air Pollution: A Critical Review of the Literature onEmisions, Exposure, and Health Effects, 2010.

(2) Wang, Y. J.; Zhang, K. M. Modeling Near-Road Air QualityUsing a Computational Fluid Dynamics Model. Environ. Sci. Technol.2009, 43 (20), 7778.

(3) Wang, Y. J.; DenBleyker, A.; McDonald-Buller, E.; Allen, D.;Zhang, K. M. Modeling the chemical evolution of nitrogen oxides nearroadways. Atmos. Environ. 2011, 45 (1), 43.

(4) Wang, Y. J.; Zhang, K. M. Simulating transport and trans-formation of multi-pollutants in complex environments: Part I:



Development of the Comprehensive Turbulent Aerosol Dynamics andGas Chemistry (CTAG) Model, under review, 2012.(5) Patel, M.M.; Chillrud, S. N.; Correa, J. C.; Feinberg, M.; Hazi, Y.;

Deepti, K. C.; Prakash, S.; Ross, J. M.; Levy, D.; Kinney, P. L. Spatial andtemporal variations in traffic-related particulate matter at New York Cityhigh schools. Atmos. Environ. 2009, 43 (32), 4975.(6) Garg, R.; Karpati, A.; Leighton, J.; Perrin, M.; Shah, M. Asthma

Facts, Second Edition; 2003.(7) Spira-Cohen, A.; Chen, L. C.; Kendall, M.; Lall, R.; Thurston,

G. D. Personal Exposures to Traffic-Related Air Pollution and AcuteRespiratory Health among Bronx Schoolchildren with Asthma. Environ.Health Perspect. 2011, 119, 4.(8) Patel, M.M.; Chillrud, S. N.; Correa, J. C.; Hazi, Y.; Feinberg, M.;

Kc, D.; Prakash, S.; Ross, J. M.; Levy, D.; Kinney, P. L. Traffic-RelatedParticulate Matter and Acute Respiratory Symptoms among New YorkCity Area Adolescents. Environ. Health Perspect. 2010, 118, 9.(9) Spira-Cohen, A.; Chen, L. C.; Kendall, M.; Sheesley, R.; Thurston,

G. D. Personal exposures to traffic-related particle pollution among childrenwith asthma in the SouthBronx,NY. J. Exposure Sci. Environ. Epidemiol.2010,20 (5), 446.(10) NYSERDA,Assessment of Carbonaceous PM2.5 for New York and

the Region; Final Report 08-01; 2008; Vol. 2.(11) Jansen, K. L.; Larson, T. V.; Koenig, J. Q.; Mar, T. F.; Fields, C.;

Stewart, J.; Lippmann, M. Associations between Health Effects andParticulate Matter and Black Carbon in Subjects with RespiratoryDisease. Environ. Health Perspect. 2005, 113 (12), 1741.(12) Mordukhovich, I.; Wilker, E.; Suh, H.; Wright, R.; Sparrow, D.;

Vokonas, P. S.; Schwartz, J. Black Carbon Exposure, Oxidative StressGenes, and Blood Pressure in a Repeated-Measures Study. Environ.Health Perspect. 2009, 117, 11.(13) Wang, X.; Westerdahl, D.; Chen, L. C.; Wu, Y.; Hao, J.; Pan, X.;

Guo, X.; Zhang, K. M. Evaluating the air quality impacts of the 2008Beijing Olympic Games: On-road emission factors and black carbonprofiles. Atmos. Environ. 2009, 43 (30), 4535.(14) Westerdahl, D.; Wang, X.; Pan, X.; Zhang, K. M. Characteriza-

tion of on-road vehicle emission factors and microenvironmental airquality in Beijing, China. Atmos. Environ. 2009, 43 (3), 697.(15) Horvath, H. Atmospheric light absorption�A review. Atmos.

Environ., Part A 1993, 27 (3), 293.(16) Gerashchenko, S.; Good, G.; Warhaft, Z. Entrainment and

mixing of water droplets across a shearless turbulent interface with andwithout gravitational effects. J. Fluid Mech. 2011, 668, 293.(17) ANSYS FLUENT 12.1 User’s Guide; ANSYS, 2010.(18) Finlayson, E. U.; Gadgil, A. J.; Thatcher, T. L.; Sextro, R. G.

Pollutant dispersion in a large indoor space. Part 2: Computational fluiddynamics predictions and comparison with a scale model experiment forisothermal flow. Indoor Air 2004, 14 (4), 272.(19) PM Emission Factor Table C for all the counties in NYSDOT

Region 10 and 11 and Rockland and Westchester counties in NYSDOTRegion 8; New York State Department of Transportation, 2008.(20) Schauer, J. J. Evaluation of elemental carbon as a marker for diesel

particulate matter. J. Exposure Anal. Environ. Epidemiol. 2003, 13 (6), 443.(21) Chow, J. C.; Watson, J. G.; Doraiswamy, P.; Chen, L.-W. A.;

Sodeman, D. A.; Lowenthal, D. H.; Park, K.; Arnott, W. P.; Motallebi, N.Aerosol light absorption, black carbon, and elemental carbon at theFresno Supersite, California. Atmos. Res. 2009, 93 (4), 874.(22) Liousse, C.; Cachier, H.; Jennings, S. G. Optical and thermal

measurements of black carbon aerosol content in different environ-ments: Variation of the specific attenuation cross-section, sigma([sigma]). Atmos. Environ., Part A 1993, 27 (8), 1203.(23) Yan, B.; Kennedy, D.; Miller, R. L.; Cowin, J. P.; Jung, K.-h.;

Perzanowski, M.; Balletta, M.; Perera, F. P.; Kinney, P. L.; Chillrud, S. N.,Validating a nondestructive optical method for apportioning coloredparticulate matter into black carbon and additional components. Atmos.Environ. In Press, Corrected Proof.(24) Watson, J. G.; Chow, J. C.; Chen, L. -W. A. Summary of organic

and elemental carbon/black carbon analysis methods and intercompar-isons. Aerosol Air Qual. Res. 2005, 5 (1), 65.

(25) Jeong, C.-H.; Hopke, P. K.; Kim, E.; Lee, D.-W. The compar-ison between thermal-optical transmittance elemental carbon andAethalometer black carbon measured at multiple monitoring sites.Atmos. Environ. 2004, 38 (31), 5193.

(26) Lena, T. S.; Ochieng, V.; Carter, M.; Holguin-Veral, J.; Kinney,P. L. Elemental carbon and PM2.5 levels in an urban community heavilyimpacted by truck traffic. Environ. Health Perspect. 2002, 10009.

(27) Kim Oanh, N. T.; Thiansathit, W.; Bond, T. C.; Subramanian,R.;Winijkul, E.; Paw-armart, I. Compositional characterization of PM2.5emitted from in-use diesel vehicles. Atmos. Environ. 2010, 44 (1), 15.

(28) Rattigan, O. V.; Dirk Felton, H.; Bae, M.-S.; Schwab, J. J.;Demerjian, K. L.Multi-year hourly PM2.5 carbonmeasurements inNewYork: Diurnal, day of week and seasonal patterns. Atmos. Environ. 2010,44 (16), 2043.

(29) Boylan, J. W.; Russell, A. G. PM and light extinction modelperformance metrics, goals, and criteria for three-dimensional air qualitymodels. Atmos. Environ. 2006, 40 (26), 4946.

(30) Stein, A. F.; Isakov, V.; Godowitch, J.; Draxler, R. R. A hybridmodeling approach to resolve pollutant concentrations in an urban area.Atmos. Environ. 2007, 41 (40), 9410.

(31) Oke, T. R. Street design and urban canopy layer climate. EnergyBuildings 1988, 11 (1�3), 103.

(32) Vardoulakisa, S.; Fisherb, B. E. A.; Pericleous, K.; Gonzalez-Flescac, N. Modelling air quality in street canyons: a review. Atmos.Environ. 2003, 37 (2), 155.

(33) Hoydysh, W. G.; Dabberdt, W. F. Kinematics and dispersioncharacteristics of flows in asymmetric street canyons. Atmos. Environ.(1967-1989) 1988, 22 (12), 2677.

(34) Xie, X.; Chun-Ho, L.; Dennis, Y.; Leung, C. Impact of BuildingFacades and Ground Heating onWind Flow and Pollutant Transport inStreet Canyons. Atmos. Environ. 2007, 41 (39), 9030.

(35) Chan, T. L.; Dong, G.; Leung, C. W.; Cheung, C. S.; Hung,W. T. Validation of a two-dimensional pollutant dispersion model in anisolated street canyon. Atmos. Environ. 2002, 36 (5), 861.

(36) Boarnet, M. G.; Houston, D.; Edwards, R.; Princevac, M.;Ferguson, G.; Pan, H.; Bartolome, C. Fine particulate concentrationson sidewalks in five Southern California cities. Atmos. Environ. 2011,45 (24), 4025.

Documents

Modeling Spatial Variations of Black Carbon Particles in ... · quality studies have characterized open highway conditions, i.e., large roads without any major structures inthe close