Evaluation of Roadside Air Quality Models

SATURN

Ranjeet S. Sokhi

Atmospheric Science Research Group (ASRG), Department of Environmental Sciences, University of Hertfordshire, College Lane,Hatfield Herts, AL10 9AB. UK
*CERC, Cambridge, UK
**School of Earth Sciences, University of Greenwich, UK
*** Transport Research Laboratory, Berkshire, UK


 

1. Summary

Numerous models are now available to predict the dispersion of traffic-related pollutants near roadsides. Within the UK models such as the USEPA approved CALINE4 and the ADMS-urban model are widely used to assess urban air quality. Each of these models have specific requirements with regard to input data which usually consists of air pollutant concentrations, meteorological parameters, traffic data and emission data. In order to identify areas where current state-of-the art models provide adequate predictions and where they require further development it is important to evaluate them with comprehensive and reliable datasets.
This study reports on an evaluation of three models GRAM, CALINE4 and ADMS-urban for CO, PM10 and NO2 prediction. The models have been compared with measured air quality data from roadside sites and where possible, traffic and meteorology data have also been collected. Comparisons of the predictions have been made in terms of statistical measures appropriate to UK national air quality standards.

2. Aim of the research

To estimate the status of current models for predicting roadside air quality and to highlight areas of improvement.

3. Activities during the year

Evaluation protocols have been identified.
Suitable datasets have been colated.
Intercomparison of three models (CALINE4, GRAM and ADMS) for CO, NO2 and PM10 have been conducted.
Future needs regarding datasets and models have been highlighted.
Results have been presented at the Harmonisation Conference, Rhodes, May 1998 (Sokhi et al., 1998).

4. Principle results

Models have been compared to CO, NO2 and PM10. Initial results are given in figures 1-4. Results of CO (8 hourly maximum running mean) show wide differences between the models. CALINE4 shows agreement with measured data for CO annual mean. The two advanced models CALINE4 and ADMS show closer agreement with PM10 data (99 percentile of 24 hr running mean). Surprisingly, CALINE4 deviates significantly from the PM10 annual mean. With regard to NO2 at the M25 (motorway) site CALINE4 and GRAM show reasonable agreement.
 

5. Main Conclusion

Significant disagreement has been observed between models.
Background choice is critical: ADMS and CALINE show that PM10 background is approximately the same as local contribution  (99%ile of 24 hr running means).
Temporal (and speed) dependence of emission factors needs to be considered.

6. Aim for the coming year

To use more comprehensive datasets.
Extend the range of models such other European models.
Review local scale atmospheric chemistry for modelling NO2.
Review modelling of vehicle turbulence.

7. Acknowledgements

This modelling work has relied on important contributions from Professor Bernard Fisher (University of Greenwich), Alaric Lester (CERC), Dr Ian McRae (Transport Research Laboratory, UK) and Surat Bualert (University of Hertfordshire).

8. References.

Sokhi , R.S., B.Fisher, A.Lester, I.McCrae, S.Bualert and N.Soontornstit. 1998.
Modelling of air qulity around roads. In the Proceedings of 5th International
Conference on Harmonisation within Atmospheric Dispersion Modelling for
Regulatory Purpose, 18-21 May 1998 . Rhodes, Greece. pp 492-497.
 


Concentration  (ppm)

                                  Site
 

Figure 1 Comparison of CO with measured data (maximum 8 hour running mean)
 
 

 Concentration  (ppm)

                                Site
 

Figure 2 Comparison of CO with measured data (annual mean)
 
 
 

Concentration  (ppm)

Figure 3 Comparison of PM10 at M25 site
 
 
 
 

Concentration  (ppb)

Figure 4 Comparison of NO2 with measured data at M25 site