1. Guidance on the use of models for the European Air Quality Directive Activity of WG1 FAIRMODE Forum for air quality modelling in Europe Bruce Denby Bruce Denby 1, Emilia Georgieva 6, Steinar Larssen 1, Cristina Guerreiro 1, Liu Li 1, John Douros 2, Nicolas Moussiopoulos 2, Lia Fragkou 2, Michael Gauss 3, Helge Olesen 4, Ana Isabel Miranda 5, Panagiota Dilara 6, Philippe Thunis 6, Sari Lappi 7, Laurence Rouil 8, Anke Lükewille 9, Xavier Querol 10, Fernando Martin 11, Martijn Schaap 12, Dick van den Hout 12, Andrej Kobe 13, Camillo Silibello 14, Keith Vincent 15, John Stedman 15, María Gonçalves 16, Guido Pirovano 17, Luisa Volta 18, Addo van Pul 19, Alberto González Ortiz 20, Peter Roberts 21, Dietmar Oettl 22

2. Presentation General guidance document NO 2 guidance document Work plan, needs and discussion

3. Terms of reference of FAIRMODE To provide a permanent European forum for AQ modellers and model users To produce guidance on the use of air quality models for the purposes of implementation of the AQ Directive and in preparation for its revision To study and set-up a system (protocols and tools) for quality assurance and continuous improvements of AQ models To make recommendations and promote further research in the field of AQ modelling

4. General guidance document Aimed at modellers and authorities, providing interpretation and guidance in the application of models for the AQD Current version (v6.1) available on FAIRMODE home page Input from 2 nd FAIRMODE plenary November 2009 Some new examples in, some old examples out Comments from the implementation group ‘Does not include direct guidance on modelling itself’ (Dick van den Hout)

5. CONTENT: General guidance (pp. 99) 1. Introduction 2. Summary of the 2008 AQ Directive 3. Interpretation of the AQ Directive in regard to modelling 4. Reporting and public information when using models 5. Model quality assurance and evaluation 6. Applications of models for assessment 7. Application of models for air quality planning 8. Special topics Annexes with examples

6. EXAMPLES: General modelling guidance Spatial representativeness and modelling (p. 17) For industrial sites concentrations should be representative of a 250 x 250 m area For traffic emissions the assessment should be representative for a 100 m street segment and monitoring should be carried out less than 10 m from the kerbside Urban background concentrations should be representative of the exposure of the general urban population (‘several square kilometres’)

7. EXAMPLES: General modelling guidance Consequences of spatial representativeness when modelling traffic Models used for assessing near road concentrations are Gaussian based models (street canyon or open road) Positioning of receptors has impact on the modelled concentrations Model receptors should be positioned at kerbside (AQD ‘valid everywhere’) and within the breathing zone (1.5 – 4m) Model receptors on both sides of a road every 100 m is sufficient for roads longer than 100 m.

8. EXAMPLES: General modelling guidance Interpretation of the Directive quality objective (p. 20) Quality objectives for modelling provided in Annex I Most modellers present results in terms of some interpretation of these objectives No standard interpretation exists Relative Percentile Error Relative Directive Error Uses observed concentrations at the percentile Uses concentrations closest to the limit value

9. EXAMPLES: General modelling guidance Interpretation of the Directive quality objective (p. 20)

10. EXAMPLES: General modelling guidance Interpretation of the Directive quality objective RDE ”reasonable” for percentiles or when O LV RPE ”reasonable” for annual means when O > LV but unnecessarily stringent when O < LV Recommendations: Review these criteria for the following Directive

11. FAIRMODE general guidance work plan General guidance document will be updated regularly, and available from FAIRMODE website. Some comments received in 2010 still to be addressed (not possible to address all!) Current version seen as complete but improvable Will be published as an ETC/ACC technical paper after a final update this year.

12. CONTENT: NO 2 modelling guidance 1. Introduction 2. Dispersion modelling 3. Chemistry modelling 4. Emission data and inventories 5. Meteorological data 6. Quality control 7. Modelling requirements for notification of postponement

13. CONTENT: NO 2 modelling guidance Modelling chapters include Description Application Examples Recommendations Focus of the modelling is on street and urban applications Focus of emissions is on traffic

14. STATUS: NO 2 modelling guidance First draft version available (v2.2) on FAIRMODE homepage Aimed at authorities, providing background information, links and recommendations on modelling methods and applications for NO 2 Presentations at the NO 2 ‘postponement’ workshop in Brussels and Harmo13 conference Focus on the actual modelling rather than the general application of models for the AQD

15. Dispersion: Examples Gaussian models Open roads (Caline 4, CAR-FMI, OML-Highway, …) Street canyons (OSPM, IMMIScpb, SEP_SCAM) Urban (URBIS, IMMIS luft, ADMS-urban, …) Lagrangian particle models Street level (GRAL) Urban (AUSTRAL, SPRAY, FLEXPART) Eulerian models Obstacle resolving (Fluidyn, VADIS, MIMO, …) Urban (CAMx, TAPM, CMAQ, WRF-CHEM, CHIMERE, MATCH, MEMO, …)

16. Dispersion: Recommendations Transport and dispersion Gaussian modelsLagrangian particle models CFD or LES modelsUrban scale Eulerian models Open roads Computationally more expensive Computationally expensive, no obstacles Inappropriate resolution Street canyon Computationally moer expensive Computationally expensive Inappropriate resolution Urban scale Requires homogenous meteorology Computationally expensive Not computationally feasible Chemistry Non-reactive with deposition and decay Possible but not operational Full chemistry possible State of the art Conditionally applicable Not fit for purpose ’Fitness for purpose’ of dispersion models for Directive related applications

17. Chemistry: NO 2 dependence The total NO x emission The primary NO 2 emission The VOC emission The existing chemical balance in the atmosphere The available ozone (and other oxidants) The distance from the source (time) The degree of turbulent mixing

18. Chemistry: local scale modelling How do local scale models represent the chemistry? Majority use empirical functions relating NO 2 to NO x (dependent on year, city, site) Some (~4) use photostationary approximation (only valid far from sources) Some (~3) use parameterised ’distance from source’ dependent solutions (more realistic) Some ( ~ 2) use parameterised ’limitted mixing’ dependent solutions (reflects the turbulent mixing)

19. Chemistry: urban scale modelling How do urban scale models represent the chemistry? Empirical functions relating NO 2 to NO x Some (~3) use photostationary approximation (only really valid when hydrocarbons are not involved) Some (~2) use ’reduced’ photochemical schemes (e.g. Generic Reaction Scheme) A number use ’complete’ photochemical schemes (based on regional scale CTMs)

20. Chemistry: empirical functions Bächlin W., R. Bösinger, 2008: Untersuchungen zu Stickstoffdioxid-Konzentrationen, Los 1 Überprüfung der Rombergformel. Ingenieurbüro Lohmeyer GmbH & Co. KG, Karlsruhe. Projekt 60976-04-01, Stand: Dezember 2008. Gutachten im Auftrag von: Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein--Westfalen, Recklinghausen.

21. Chemistry: European NO 2 v’s NO X A=30 B=42 C=0.26 991 traffic station data 1539 background station data

22. Chemistry: European NO 2 v’s NO X 93% of stations within +/- 30%

23. Model performance for NO 2 Generally the results can be quite reasonable on the local and urban scale. Published validations tend to be biased towards ‘publishable’ results Some models will perform ‘better’ for NO 2 than for NO x due to: decreased sensitivity of NO 2 with increasing NO X the use of measured ozone as a boundary condition the use of empirical NO X conversions Results are always dependent on input data: emissions, meteorology and background.

24. Model results: OSPM (Copenhagen) Berkowicz, R., Hertel, O., Larsen, S. E., Sørensen, N. N. and Nielsen, M. (1997): Modelling traffic pollution in streets. National Environmental Research Institute, Roskilde, Denmark. ISBN: 87-7772-307-4. NO x NO 2

25. Ingrid Sundvor, Leiv Håvard Slørdal and Scott Randall (2009) Dispersion and Exposure Calculations of PM10, NO2 and Benzene in Oslo and Trondheim for 2007. NILU: OR 9/2009 Model results: HIWAY-2 (Oslo two sites)

26. V. Diegmann, S. Wurzler (2009) QUALITY CONTROL IN DISPERSION MODELING: VALIDATION OF A SCREENING MODEL FOR PM10 AND NO2. Proceedings of the 12’th Harmonisation conference, Croatia Model results: IMMIS luft (N. Rhine Westphalia) Annual mean NO 2

27. Sokhi, R.S. et al. An integrated multi-model approach for air quality assessment: Development and evaluation of the OSCAR Air Quality Assessment System, Environmental Modelling & Software, v.23 n.3, p.268-281, March, 2008 Model results: CAR-FMI (Helsinki and London) NO x NO 2 HelsinkiLondon

28. Figure 1: NO 2 model validation at background sites for 2008 Figure 2: NO 2 model validation at roadside sites for 2008. Model results: PCM - Grice et al., 2009 (UK)

29. Chemistry: recommendations Chemistry Empirical schemes Photostationary and ozone limitting schemes Distance from source and mixing schemes Reduced photochemical schemes Full photochemical schemes Assessment Street level Given sufficient observations Overestimates NO 2 Difficult to apply at this scale Urban scale Given sufficient observations Suitable for winter or low hydrocarbons Regional scale Missing significant chemistry Planning Street level Not sensitive to changes in ozone or NO 2 emissions Sensitivity to ozone and NO 2 emissions represented Difficult to apply at this scale Urban scale Not sensitive to changes in ozone or NO 2 emissions Suitable with low light or hydrocarbons Regional scale Missing significant chemistry State of the art Conditionally applicable Not fit for purpose ’Fitness for purpose’ assessment for NO 2 chemistry

30. The NO 2 modelling guidance is under development, 80% complete. NO 2 guidance will also be used in a trial web based guidance scheme to be tested at the end of 2010. Need community involvement for contributions, review and discussion Example tables Modelling experience Development of recommendations FAIRMODE NO 2 guidance work plan

31. ChapterContent (sub-chapters)Review 1. Introduction 2. Dispersion modellingRIVM 3. Chemistry modellingRIVM 4. Emission data and inventories Borges and Lumbreras (all) Christian Nagl UBV (all) Christian Nagl (UBV) 5. Meteorological data 6. Quality control 7. Modelling requirements for notification of post.

32. http://fairmode.ew.eea.europa.eu/ For information and contributions contact Bruce Denby bde@nilu.no