Technical Guidance for the Assessment of Local Air Quality Under the UK National Air Quality

Executive Summary

Technical guidance for the assessment of local air quality under the UK National Air Quality

Strategy (NAQS) has identified that previous studies on road-traffic-related air pollution

often may not have considered the effects of junctions adequately (LAQM-TG3, 2003). The

influence of road traffic junctions in urban areas is particularly significant, as these areas

often have enhanced rates of emission due to traffic congestion and rates of atmospheric

dispersion are often reduced, due to the effects of the surrounding buildings on wind flows

(Vardoulakis et al., 2003). The combination of these two effects often results in areas being

identified as pollution hotspots through monitoring studies.

This project compares the predictions of the Gaussian Plume and Street Canyon models,

routinely used by Local Authorities for modelling air quality, with the predictions of a

Computational Fluid Dynamics model capable of including the complex topographies that

occur at urban intersections. Five case study sites were identified in order to determine the

influence of different modelling techniques on the prediction of air concentrations for Local

Air Quality Management. These study sites were proposed by each of the local authorities

involved in this project and were areas of known pollution hotspots, recognised as being

difficult to model using conventional tools.

An initial report (Hill et al., 2005) provided a discussion of the modelling methodologies,

detailed the case study areas, provided the model input meteorological and emissions data

and presented the urban air pollution monitoring data that the models were to be compared

with. This final report presents the predictions of the atmospheric dispersion models and

compares these with data collected using automatic monitoring equipment at each of the

selected junctions. Initial sections of this report present the methodologies used to model the

chemical conversion of NOx to NO2 in the atmosphere and the statistical methods used to

provide a quantitative assessment of the performance and uncertainties in the modelling.

The simulations conducted in this report have identified that urban buildings affect the

predictions of local NOx concentrations significantly. Where streets are flanked by tall

buildings these effects have long been recognised and Street Canyon models are typically

used for modelling such situations. The modelling assessments for Coventry and Leicester,

areas that are similar to the geometries for which Street Canyon models were developed,

showed that these models provide the most suitable tools for predicting air concentrations.

Interestingly, realistic predictions were also obtained using the AEOLIUS model at the

Sheffield and Leeds case study sites, areas that have very different topographies to those

typical of a street canyon. Further studies should focus on the applicability of street canyon

models and investigate methods to improve their configuration for realistic street geometries.

Model predictions from the CFD code for the Birmingham, Leeds and Sheffield sites also

showed that particular air concentration hotspots occurred in the wakes of large buildings and

at building faces along the roadside. The model runtimes encountered in these simulations

and the occasional significant overprediction of concentrations, would suggest that the

operational use of CFD at these resolutions for NAQS modelling may be prohibitively timeconsuming

and may not necessarily reduce the uncertainties in modelling assessments.

However, it may be appropriate to consider the types of feature shown by the CFD model to

result in local pollution hotspots when developing monitoring strategies in such areas.

030426/02 i Second Issue

WSC 12 May 2005

Table of Contents

Executive Summary i

1. Introduction 1

2. Methods to model the conversion of NOx to NO2 2

3. Methods for the evaluation of atmospheric dispersion models 3

4. Comparison of NOx and NO2 simulations at case study locations 4

4.1 Bristol Street (Birmingham) 4

4.1.1 Comparison of NOx time-series arising from local roads 4

4.1.2 Comparison of horizontal resolutions 5

4.1.3 Comparison of concentration maps 5

4.1.4 Statistical comparison of modelled and measured NOx and NO2 data 6

4.2 Ironmonger Row (Coventry) 6

4.2.1 Comparison of NOx time-series arising from local roads 6

4.2.2 Comparison of concentration maps 7

4.2.3 Statistical comparison of modelled and measured NOx and NO2 data 7

4.3 Haslewood Close (Leeds) 8

4.3.1 Comparison of NOx time-series arising from local roads 8

4.3.2 Comparison of horizontal resolutions 8

4.3.3 Comparison of concentration maps 9

4.3.4 Statistical comparison of modelled and measured NOx and NO2 data 9

4.4 Oxford Street (Leicester) 9

4.4.1 Comparison of NOx time-series arising from local roads 10

4.4.2 Comparison of concentration maps 11

4.4.3 Statistical comparison of modelled and measured NOx and NO2 data 11

4.5 Lowfields (Sheffield) 12

4.5.1 Comparison of NOx time-series arising from local roads 12

4.5.2 Comparison of concentration maps 12

4.5.3 Statistical comparison of modelled and measured NOx and NO2 data 13

5. Summary 13

5.1 Birmingham 13

5.2 Coventry 14

5.3 Leeds 14

5.4 Leicester 14

5.5 Sheffield 15

6. Recommendations 15

6.1 The influence of building effects 15

6.2 The influence of Gaussian Plume Model resolution 16

6.3 Other uncertainties highlighted in this study 16

7. References 17

8. Tables 18

9. Figures 25

030426/02 ii Second Issue

WSC 12 May 2005

List of Tables

Table 1: Statistical comparison of model results at Birmingham. 18

Table 2: Statistical comparison of model results at Coventry for original emission data. 19

Table 3: Statistical comparison of model results at Coventry for revised emission data. 20

Table 4: Statistical comparison of model results at Leeds. 21

Table 5: Statistical comparison of model results at Leicester for NOx. 22

Table 6: Statistical comparison of model results at Leicester for NOx for all available

data. 22

Table 7: Statistical comparison of model results at Leicester for NO2. 23

Table 8: Statistical comparison of model results at Sheffield. 24

List of Figures

Figure 1: Comparison of measured NOx concentrations arising from local roads at

Bristol Street (Birmingham) with the simulation of the Airviro model at 25 m

and 5 m horizontal resolution (AV25 and AV5) and the CFD model. 25

Figure 2: Transect across the model domain (shown by the red line) for two different

resolutions (coarse: c; fine: f) using the Airviro Gaussian Plume (AV) and

CFD models for Birmingham for 1900 hrs. 26

Figure 3: CFD model simulations for Bristol Street (Birmingham) 27

Figure 4: Model results from Airviro for Birmingham 28

Figure 5: CFD model flowfield for Bristol Street Birmingham for 1900 hrs. 29

Figure 6: Comparison of measured NOx concentrations arising from local roads at

Ironmonger Row (Coventry) with the simulation of the Airviro model (AV),

AEOLIUS model (AE) and CFD model (CFD) for original (-O) and revised

(-R) emissions datasets. 29

Figure 7: CFD model simulations for Ironmonger Row (Coventry) 30

Figure 8: Model predictions using the Airviro Gaussian Plume model for Coventry 31

Figure 9: CFD model predictions for the Burges Street canyon 0700. 32

Figure 10: Comparison of measured NOx concentrations arising from local roads at

Haslewood Close (Leeds), with the simulation of the Airviro model for two

horizontal resolutions, 25m and 5m (AV 25 and AV 5), AEOLIUS model

(AE) and CFD model (CFD). 33

Figure 11: Transect across the model domain (shown by the red line) for two different

resolutions (coarse: c; fine: f) using the Airviro Gaussian Plume (AV) and

CFD models for Leeds for 0600 hrs. 34

Figure 12: Air concentrations modelled using CFD for Leeds. 35

Figure 13: Air concentrations modelled using Airviro at a 25 m resolution for Leeds 36

Figure 14: Air concentrations modelled using Airviro at a 5 m resolution for Leeds 37

Figure 15: Comparison of measured NOx concentrations arising from local roads at

Oxford Street (Leicester) with the simulation of the Airviro Gaussian model

(AV), AEOLIUS model (AE) and CFD model (CFD) 38

Figure 16: CFD model simulations for Oxford Street (Leicester). 39

Figure 17: Model results from Airviro for Leicester without street canyon modelling. 40

Figure 18: Model results from ADMS URBAN for Leicester without the OSPM street

canyon model. 41

Figure 19: Model results from ADMS URBAN including the OSPM Street Canyon model 42

030426/02 iii Second Issue

WSC 12 May 2005

Figure 20: Comparison of measured NOx concentrations arising from local roads at

Lowfields (Sheffield) with the simulation of the Airviro Gaussian model

(AV), AEOLIUS model (AE) and CFD model (CFD). 43

Figure 21: CFD model simulations for Sheffield 44

Figure 22: Air concentrations modelled using Airviro at Sheffield 45

List of Appendices

Appendix A: The case study areas A1

Appendix B: Urban air pollution monitoring data B1

Appendix C: Model input meteorological data C1

Appendix D: Model input emissions data D1

Appendix E: CFD model setups E1

030426/02 iv Second Issue

WSC 12 May 2005

1. Introduction