Transportation Research Part D 46 (2016) 157–165
Contents lists available at ScienceDirect
Transportation Research Part D journal homepage: www.elsevier.com/locate/trd
Urban roadside monitoring and prediction of CO, NO2 and SO2 dispersion from on-road vehicles in megacity Delhi Rajeev Kumar Mishra a,⇑, Ankita Shukla b, Manoranjan Parida c, Govind Pandey d,1 a
Department of Environmental Engineering, Delhi Technological University, Delhi 110 042, India Department of Environmental Science, B.B. Ambedkar University, Lucknow 226025, India c Department of Civil Engineering, Indian Institute of Technology, Roorkee 247 667, India d Civil Engineering Department, Madan Mohan Malaviya University of Technology, Gorakhpur 273 010, India b
a r t i c l e
i n f o
Article history:
Keywords: Dispersion model Gaseous emission MS sheet
a b s t r a c t The study inspects the traffic-induced gaseous emission dispersion characteristics from the urban roadside sites in Delhi, India. The concentration of pollutants viz. CO, NO2 and SO2 along with traffic and ambient atmospheric conditions at five selected local urban road sites were simultaneously measured. A developed General Finite Line Source Model (GFLSM) was used to predict the local roadside CO, NO2 and SO2 concentrations. A comparison of the observed and predicted values emission parameters using GFLS model has shown that the predicted values for SO2, CO and NO2 at all the selected local urban roadside locations are found to lie within the error bands of 5%, 6%, and 7% respectively. A high level of agreement was found between the monitored and estimated CO, NO2 and SO2 concentration data. From the study, it has also been established that the developed model exhibits the capability of reasonably predicting the characteristics of gaseous pollutants dispersion from on-road vehicles for the urban city air quality. Ó 2016 Elsevier Ltd. All rights reserved.
Introduction The environmental impacts are particularly severe in urban environment due to high population, traffic levels, intense motor vehicle use, driving patterns, vehicle characteristics and complex urban geometry (Allen et al., 2009; Beelen et al., 2009; Davies et al., 2009; Gurjar et al., 2008; Weber et al., 2008). In Delhi, the total number of vehicles has increased from 3.62 million in 2001-02 to 8.83 million in 2014-15 (Table 1). The rapidly increasing number of motor vehicles has resulted in alarming levels of traffic congestion, air pollution, noise, and traffic danger. Transportation sector is accountable for about 50% of the emissions of nitrogen oxide and 90% of the carbon monoxide (Nagurney, 2000). As a result, urban environmental quality has been deteriorated in most of the cities of developing countries. Different models have been developed and used to investigate the source of emission from the vehicles in urban cities. Dispersion modelling is a common technique for inferring a quantitative deterministic correlation amid pollutant release and ambient concentrations. Various review studies on vehicular emission models have been carried out by different researchers across the country (Gokhale and Khare, 2004; Nagendra and Khare, 2002; Sharma and Khare, 2001). ⇑ Corresponding author. Tel.: +91 011 27871045. E-mail addresses:
[email protected] (R.K. Mishra),
[email protected] (A. Shukla),
[email protected] (M. Parida), pandey_govind@ rediffmail.com (G. Pandey). 1 Tel.: +91 511 2272272. http://dx.doi.org/10.1016/j.trd.2016.03.019 1361-9209/Ó 2016 Elsevier Ltd. All rights reserved.
158
R.K. Mishra et al. / Transportation Research Part D 46 (2016) 157–165
Table 1 Growth scenario of motor vehicles in Delhi (million). Source: Economic Survey of Delhi (2014–15). Year
Scooters and motor cycles
Cars and Jeeps
Ambulance
Auto rickshaws
Taxies
Buses
Other passenger vehicles
Tractors
Goods vehicles
Others
All motor vehicles
2001–02 2005–06 2009–10 2014–15
2.34 3.08 4.07 5.68
1.05 1.47 2.02 2.79
0.002 0.002 0.002 0.002
0.07 0.07 0.09 0.08
0.01 0.02 0.05 0.08
0.02 0.03 0.03 0.02
0.01 0.02 0.02 0.01
0.005 0.005 0.005 0.002
0.11 0.13 0.19 0.16
0.006 0.006 0.005 0.00003
3.62 4.83 6.47 8.83
These studies have signified that deterministic and semi-numerical based models showed better results when applied to estimate air quality near roadways (Gokhale et al., 2003; Gokhale and Raokhande, 2008; Gokhale and Khare, 2004; Gokhale and Khare, 2005; Khare and Sharma, 1999). Some of the most widely used models are Gaussian based viz. CALINE-3 model (Benson, 1979), EPA’s HIWAY-2 model (Peterson, 1980) and GM model (Chock, 1978). General Finite Line Source Model, which was conceptualised by Luhar and Patil (1989), is also widely used. Shukla et al. (2010) employed the GFLS model to study the pollutant concentration at five locations in Lucknow city of India and observed its suitability in those particular locations. Wang et al. (2006) also applied GFLSM model to predict the concentration of traffic induced gaseous and particle emission from roadside sites in Hon Kong and found the better capability of this model to foretell the characteristics of the CO and PM2.5 pollutant. Ganguly et al. (2009) conducted a comparative study between GFLSM and CALINE4 models in the city of Dublin. On the basis of field study observations, it was found that the GFLSM, an analytical model, is quite impressive in comparison to CALINE4. The user friendly functionality of the GFLSM suggests that it can be readily incorporated in integrated transportation–environmental models. In the present study, the general finite line source model has been developed for simulating the CO, NO2 and SO2 dispersion from the selected urban road sites in Delhi. General Finite Line Source Model General Finite Line Source Model consists of model evaluation for gaseous pollutant dispersion and model formulation in Excel sheet. Model equation for gaseous pollutant dispersion Due to user friendly property of this worksheet system, it is very easy to model the concentration of gaseous pollutants like CO, SO2 and NOx. To calculate the concentration of pollutants the following equation has been incorporated in the model (Luhar and Patil, 1989):
2 2 Q 1 zh0 0 þ exp 12 zþh C ¼ 2pffiffiffiffi exp rz rz 2 2prz ue i h sin hðL=2þyÞþx cos h cos h pffiffi pffiffi erf sin hðL=2yÞx þ erf 2r 2r y
ð1Þ
y
here C = concentration of the pollutant (lg/m3) Q = emission rate per unit length (g/m-sec) ue = effective wind speed (m/s) h0 = plume centre height distance x from the road (m) z = height of the receptor above the ground (m) ry = standard deviation of the pollutant in horizontal cross wind (m) rz = standard deviation of the pollutant in vertical direction (m) L = length of source (m) h = angle between the ambient wind and the road erf = error function x = distance of receptor from the line source (m) y = receptor distance from the roadway centre line along the line source (m) The emission rate per unit length can be evaluated as
Q ¼ Ef V h where Ef = emission factor (g/km/vehicle) Vh = vehicle density (number of vehicles/h)
ð2Þ
159
R.K. Mishra et al. / Transportation Research Part D 46 (2016) 157–165
The effective wind speed can be calculated by;
U e ¼ U sin h þ U 0
ð3Þ
where U = mean ambient wind speed at source height (m/s) U0 = wind speed correction due to traffic wake (m/s) h = wind road angle The plume centre height can be calculated by;
h0 ¼ h þ hp
ð4Þ
where h = line source height (m) hp = plume height (m) The lateral and vertical dispersion coefficients
ry and rz can be calculated by;
ry ¼ axb
ð5Þ
rz ¼ cxd þ f
ð6Þ
where x = downwind distance (km) and the value of a, b, c, d and f for different categories are the functions of downwind distance x as given in Table 2. The values of ry and rz depend upon atmospheric stability. The typical atmospheric stability classifications are given in Table 3. Model formulation in MS excel sheet General Finite Line Source model has been used for this study. It is very difficult and time consuming process to predict the concentration of different gaseous pollutants in each given hour at different locations manually. To resolve such kind of problem, the General Finite Line Source Model has been developed in MS EXCEL worksheet system (Fig. 1). The reason behind the selection of such a programme is its flexibility, compatibility and user friendly properties. This model does not require lengthy mathematical calculations and it can be used at any level for the prediction of concentration of gaseous pollutants. Only by using specified inputs like hourly traffic volume, emission factor, temperature, wind direction etc., one can get the predicted concentration of pollutants within no time. Thereafter, by changing emission factor, the single model can be used to predict the concentration of various pollutants like CO, NO2 and SO2.
Table 2 Atmospheric stability constants as a function of downwind distance. Source: Martin (1976). Stability
A B C D E F
x < 1 km
x < 1 km
a
c
d
f
c
d
f
213 156 104 68 50.5 34
440.8 106.6 61.0 33.2 22.8 14.35
1.941 1.149 0.911 0.725 0.678 0.740
9.27 3.3 0 1.7 1.3 0.35
459.7 108.2 61.0 44.5 55.4 62.6
2.094 1.098 0.911 0.516 0.305 0.180
9.6 2.0 0 13.0 34.0 48.6
Table 3 Atmospheric stability classification. Source: Turner (1970). Surface
Day Incoming solar radiation
Night Cloudiness
Wind speed (m/s)
Strong
Moderate
Slight
Cloudy
Clear
<2 2–3 3–5 5–6 >6
A A–B B C C
A–B B B–C C–D D
B C C D D
E E D D D
F F E D D
160
R.K. Mishra et al. / Transportation Research Part D 46 (2016) 157–165
Fig. 1. Print screen view of developed local GFLSM in MS Excel sheet.
Traffic-induced air pollutants concentration from vehicles using developed GFLSM During urban roadside measurements, traffic-induced CO, NO2 and SO2 concentration data were monitored continuously at five selected sampling sites for a certain sampling period. The local meteorological parameters like ambient temperature, wind speed and wind direction at the selected sites, were collected continuously as the input parameters for the developed local general finite line source model. Along with this, the traffic information such as average vehicle speed and the average traffic flow rate during the monitoring time interval was also collected. Sampling setup at urban roadside locations The roadside measurements from the traffic induced emissions were carried out at five selected local urban road sites, namely Rithala, Pitampura, Kashimiri Gate, Jhilmil and Panchsheel Enclave. The land use patterns of the selected sites are mostly commercial-residential except Kashmiri Gate which completely has commercial land use pattern. The traffic flow rate is found relatively high and steady at the selected sites. All the sampling sites approximate the breathing level of human body i.e. 1.5 m above the ground level. The concentrations of NO2 and SO2 were measured using high volume sampler (Model DPM 460 DXNL, Envirotech Instruments). Corresponding concentrations of CO were measured using sampling bags at all the sites and analysed using Non-Dispersive Infra Red (NDIR) spectroscopy method.
161
R.K. Mishra et al. / Transportation Research Part D 46 (2016) 157–165
To study the traffic-induced emissions dispersion characteristics for different local urban road traffic conditions, 12 h (8:00 to 20:00 h) sampling period was designated, which include the morning and evening peak hours. At each selected urban road site, the concentrations of CO, NO2 and SO2 were measured for a period of 12 h. Table 4 illustrates the average local traffic flow and weather conditions at each selected local urban roadside location. The average traffic flow density at Jhilmil site was the highest and at Rithala was the lowest. The traffic volumes of the road were quite high for these selected local urban locations. Vehicle emission rate and emission factor Emission rates depend on the volume of traffic, its composition and the operating modes of the vehicles. The emission rate per unit length (g/m/sec) can be calculated by
Q ¼ Ef V h
ð7Þ
where Ef is emission factor in g/km/vehicle and Vh is vehicle density i.e. number of vehicles/h. The vehicle emission factor of CO, NO2 and SO2 were based on those developed by ARAI (2007). Table 5 shows the emission factors for CO, NO2 and SO2 for each category of vehicle. These emission factors were used to calculate the aggregate emission rate from each category of vehicle at each urban roadside location. Analysis of data, results and discussion Comparison of the monitored and predicted emission concentrations The average concentrations of CO, NO2 and SO2 for each sampling location during monitoring period were compared with the predicted concentrations using the developed local GFLSM as shown in Figs. 2–4. The temporal variation depicted in these figures shows higher concentrations of CO, NO2 and SO2 in morning and evening peak traffic flow hours (8:00– 11:00 h and 17:00–20:00 h) whereas, comparatively low concentrations were observed during off peak hours (11:00– 17:00 h). Similar peak hour and off peak hour trend of the predicted CO, NO2 and SO2 concentration data was also calculated from selected five local urban roadside locations based on the developed local GFLSM. The results demonstrate that the combination of vehicle fleets, average traffic flow speed, geometrical and meteorological conditions play a significant role on the dispersion characteristics of the traffic induced gaseous emission at the selected roadside locations. The comparison of the measured and predicted gaseous emissions of CO, NO2 and SO2 shows that the developed local general finite line source model has a reasonable prediction performance for gaseous emissions with an average prediction error within 3% for CO, NO2 and SO2 at selected five local urban roadside locations. The developed model overpredicts the concentration of CO within 2% at Rithala, 3% each at Jhilmil and Panchsheel Enclave and 4% and 6% at Pitampura and Kashmiri Gate respectively. Similarly, the model overpredicts the concentration of NO2 within 1%, 2%, 3% and 7% at Pitampura, Jhilmil, Panchsheel Enclave and Rithala correspondingly whereas the concentration of SO2 within 1% at Jhilmil, 2% each at Rithala and Pitampura and 5% each at Kashmiri Gate and Panchsheel Enclave respectively. The developed GFLS model neither
Table 4 Traffic and weather conditions at the selected local urban roadside locations.
a
Roadside site
Average traffic flow (vehicles/h)
Average traffic speed (km/h)
Source length (m)
Wind speeda (m/s)
Ambient tempa. (K)
Rithala Pitampura Kashmiri Gate Jhilmil Panchsheel Enclave
2758 4158 4635 6667 5328
40 35 34 30 31
462.86 456.56 542.67 460.28 454.83
0.9 1.1 1.1 1.1 1.2
292 293 294 296 292
Source: India Meteorological Department.
Table 5 Vehicle emission factor. Source: ARAI (2007). Vehicles
Emission factor (gm/km-h) CO
NO2
SO2
Car Two wheelers Mini bus & mini truck Auto-rickshaw Bus Truck
0.891 1.036 3.66 0.845 3.72 6
0.304 0.31 2.12 0.345 6.21 9.3
0.053 0.013 0.029 0.013 0.15 0.15
162
R.K. Mishra et al. / Transportation Research Part D 46 (2016) 157–165
Fig. 2. Comparison of predicted and monitored CO concentrations at the selected local urban roadside locations.
Fig. 3. Comparison of predicted and monitored NO2 concentrations at the selected local urban roadside locations.
Fig. 4. Comparison of predicted and monitored SO2 concentrations at the selected local urban roadside locations.
Table 6 Emission and average statistical parameters for model evaluation at the selected local urban roadside locations. Rithala CO
Pitampura SO2
NO2
CO
Kashmiri gate SO2
NO2
SO2
Jhilmil NO2
Panchsheel enclave SO2
NO2
SO2
NO2
1058 1028 129 101
33 31 3 2
95 92 8 8
0.954 1.061 7.391
0.795 1.137 111.2
0.871 1.305 7.73
0.869 0.86 15.44
2.67 26.32 12 11 0.35 0.05 2.179 0.9839
29.9 114.34 12 11 0.91 0.05 2.179 0.9104
1.74 3.03 12 11 1.99 0.05 2.179 0.8527
2.62 8.08 12 11 1.12 0.05 2.179 0.9361
(lg/m3)
Summary measures Observed mean Predicted mean Observed deviation Predicted deviation
1396 1365 246 214
41 40 7 6
83 77 14 14
1243 1196 151 157
36 36 5 5
110 108 16 13
1231 1162 195 179
36 34 4 5
98 98 5 9
2039 1981 305 301
49 49 6 7
166 164 28 26
Linear regression Corerelation coefficient Slope Intercept
0.946 1.118 130.3
0.918 1.065 1.702
0.917 0.991 6.449
0.902 0.91 154.6
0.96 1.051 1.112
0.821 1.091 8.49
0.8 0.978 93.55
0.731 0.783 9.321
0.739 0.46 53.06
0.981 1.004 49.47
0.869 0.831 8.87
30.84 225.88 12 11 0.47 0.05 2.179 0.9764
0.92 6.32 12 11 0.5 0.05 2.179 0.9703
5.75 14 12 11 1.42 0.05 2.179 0.9342
47.31 152.63 12 11 1.07 0.05 2.179 0.9490
0.71 5.1948 12 11 0.48 0.05 2.179 0.9836
1.39 14.32 12 11 0.34 0.05 2.179 0.9399
68.66 186.31 12 11 1.28 0.05 2.179 0.907
1.88 4.61 12 11 1.41 0.05 2.179 0.8774
0.42 7.15 12 11 0.206 0.05 2.179 0.8366
58.54 297.89 12 11 0.68 0.05 2.179 0.9853
0.68 6.25 12 11 0.37 0.05 2.179 0.9589
Sd N D.O.F tcalc
a
ttabulated Index of agreement
(lg/m3)
CO
Emission (Unit)
Summary of T-test d
(lg/m3)
CO
(lg/m3)
CO (lg/m3)
R.K. Mishra et al. / Transportation Research Part D 46 (2016) 157–165
Roadside sites
163
164
R.K. Mishra et al. / Transportation Research Part D 46 (2016) 157–165
overpredicts nor underpredicts the concentration of NO2 at Kashmiri Gate site. Various atmospheric factors like ambient temperature, wind speed, wind directions together with estimation of traffic induced emission are responsible for the prediction performance of the developed model. In the present study, the traffic and weather conditions at selected local urban roadside sites are presented in Table 4, which shows the highest traffic density at Jhilmil location with average speed of 30 km/h. The wind speed at this location was 1.1 m/s with ambient temperature of 292 K whereas the minimum traffic flow was found at Rithala site with traffic flow of 2758 vehicles/h. Analysis of the GFLS model evaluation The emission and average statistical analysis parameters for all the five selected roadside urban locations are presented in Table 6. The values of index of agreement range from 0.91 to 0.98 for CO concentration, from 0.84 to 0.98 for NO2 concentration and from 0.85 to 0.98 for SO2 concentration. All of these values correspond to a fairly good agreement between the predicted and measured data at the selected urban roadside locations. The linear regression analysis for the predicted and monitored concentrations shows high coefficient values ranging from 0.8 to 0.98 for CO concentration, 0.73 to 0.95 for NO2 concentration and from 0.73 to 0.96 for SO2 concentration at all selected urban roadside sites. This indicates that the developed model predicts the shape of CO, NO2 and SO2 concentration profile within a fair degree of accuracy. A significance test is also applied to check the consistency of the observed data with predicted data. The table shows that the range of calculated value of t is found to be from 0.47 to 1.28 for CO, 0.21 to 1.42 for NO2 and from 0.37 to 1.9 for SO2 concentration at all the selected local urban roadside locations. All these values were found less than the tabulated values t i.e. 2.179, for degree of freedom 11 and level of significance 0.05, which supports the suitability of this model for selected roadside local urban locations. The developed model provided a reliable and convenient tool to evaluate the urban air quality at roadside locations in Delhi. Thus the present study sets up a research methodology in estimating the vehicular traffic induced emissions characteristics for local urban air quality at roadside for carrying out similar studies at other different roadside local urban locations. Conclusions The urban roadside monitoring data of traffic induced CO, NO2 and SO2 at five sampling sites namely Rithala, Pitampura, Kashmiri Gate, Jhilmil and Panchsheel Enclave are subjected to the analysis and prediction by GFLS model. The monitoring has revealed that the concentrations of CO, NO2 and SO2 in morning and evening hours during peak traffic flow condition are found to be higher than those during off peak traffic flow hours. A comparison of the observed and predicted values of SO2, CO and NO2 emissions using GFLS model has shown that the predicted values for SO2, CO and NO2 at all the selected local urban roadside locations are found to lie within the error bands of 5%, 6% and 7% respectively. Even though, certain degree of over-prediction and under-prediction is reflected by GFLS model at the selected roadside monitoring stations. Yet, the average production error is found within 10%. Assessment of the performance of an air quality model normally focuses on evaluating the accuracy of the model prediction relative to observed concentrations (Mittal et al., 2003). Developed GFLS model during this study, is found to exhibit good prediction performance and may prove to be a reliable and convenient tool for the evaluation of urban air quality at roadside locations in Delhi. It is also demonstrated that the combination of vehicle fleet, average speed and geometrical and meteorological conditions play significant role in dispersion characteristics of traffic induced gaseous emissions at roadside locations. The predictive approach developed in the study may be generalised for the dispersion of air pollutants from on-road vehicles in other cities too within a fair degree of accuracy. References Allen, W.R., Davies, H., Cohen, A.M., Mallach, G., Kaufman, D.J., Adar, D.S., 2009. The spatial relationship between traffic-generated air pollution and noise in 2 US cities. Environ. Res. 109 (3), 334–342. ARAI, Project Rep No.: AFL/2006-07/IOCL/Emission Factor Project/Final Rep dt. August 17, 2007. Beelen, R., Hoek, G., Houthuijs, D., van den Brandt, P.A., Goldbohm, R.A., Fischer, P., Schouten, L.J., Armstrong, B., Brunekreef, B., 2009. The joint association of air pollution and noise from road traffic with cardiovascular mortality in a cohort study. Occup. Environ. Med. 66 (4), 243–250. Benson, P.E., 1979. CALINE-3, A Versatile Dispersion Model for Predicting air Pollutant Levels near Highway and Arterial Roads. Final Report, FHWA/CA/TL79/23. California Department of Transportation, Sacramento, C A. Chock, D.P., 1978. A Simple line source model for dispersion near roadways. Atmos. Environ. 12 (4), 823–829. Davies, H.W., Vlaanderen, J.J., Henderson, S.B., Brauer, M., 2009. Correlation between exposures to noise and air pollution from traffic sources. Occup. Environ. Med. 66 (5), 347–350. Economic Survey of Delhi, 2014–15. Available at
. Ganguly, R., Broderick, B.M., O’Donoghue, R., 2009. Assessment of a general finite line source model and CALINE4 for vehicular pollution prediction in Ireland. Environ. Model. Assess. 14 (1), 113–125. Gokhale, S., Khare, M., Pavageau, M., 2003. Modelling distributions of air pollutant concentrations from vehicular exhausts in urban environment: a hybrid approach, PHYSMOD2003. In: International Workshop on Physical Modeling of Flow and Dispersion Phenomena, Prato, Italy 3–5, 2003. September. Gokhale, S., Raokhande, N., 2008. Performance evaluation of air quality models for predicting PM10 and PM2.5 concentrations at urban traffic intersection during winter period. Sci. Total Environ. 394 (1), 9–24. Gokhale, S., Khare, M., 2004. A review of deterministic, stochastic and hybrid vehicular exhaust emission models. Int. J. Transport Manage. 2 (2), 59–74.
R.K. Mishra et al. / Transportation Research Part D 46 (2016) 157–165
165
Gokhale, S., Khare, M., 2005. A hybrid model for the prediction of carbon monoxide from vehicular exhausts in urban environments. Atmos. Environ. 39 (22), 4025–4040. Gurjar, B.R., Butler, T.M., Lawrence, M.G., Lalieveld, J., 2008. Evaluation of emissions and air quality in megacities. Atmos. Environ. 42 (7), 1593–1606. India Meteorological Department. Available at . Khare, M., Sharma, P., 1999. Performance evaluation of general finite line source model for Delhi traffic conditions. Transp. Res. Part D 4 (1), 65–70. Luhar, A.K., Patil, R.S., 1989. A general finite line source model for vehicular pollution prediction. Atmos. Environ. 23 (3), 555–562. Martin, D.O., 1976. The change of concentration standard deviations with distance. J. Air Pollut. Control Assoc. 26 (2), 145–147. Mittal, N., Parida, M., Jain, S.S., 2003. ‘‘A Modelling Framework for Transport Related Air Pollution Prediction for Urban Areas”, Urban Transport Journal (Institute of Urban Transport, India), vol. 4(1), pp. 227–238. Nagendra, S.M.S., Khare, M., 2002a. Line source emission modeling – review. Atmos. Environ. 36 (13), 2083–2098. Nagurney, A., 2000. Congested urban transportation networks and emission paradoxes. Transp. Res. Part D 5 (2), 145–151. Peterson, W.B., 1980. User’s Guide for HIWAY-2. Highway Air Pollution Model, pp. 3–17, EPA-60018-80-018. Sharma, P., Khare, M., 2001. Modeling of vehicular exhausts – a review. Transp. Res. Part D 6 (3), 179–198. Shukla, A., Parida, M., Mishra, R.K., 2010. Analysis and PREDICTION OF VEHICULAR POLLUTION: A CASE STUDY. Int. J. Environ. Pollut. Control Manage. 2 (1), 47–60. Turner, D.B., 1970. Workbook of Atmospheric Dispersion Estimates. Washington, DC, USEPA Report, 26. Wang, J.S., Chan, T.L., Ning, Z., Leung, C.W., Cheung, C.S., Hung, W.T., 2006. Roadside measurement and prediction of CO and PM2.5 dispersion from on-road vehicles in Hong Kong. Transp. Res. Part D 11 (4), 242–249. Weber, S., Litschke, T., 2008. Variation of particle concentrations and environmental noise on the urban neighbourhood scale. Atmos. Environ. 42 (30), 7179–7183.