Nottingham Clean Air Zone Modelling and Target Determination Report ______

Report for City Council

ED 10107 | Issue Number 6 | Date 16/05/2018

Nottingham Clean Air Zone Modelling and Target Determination Report | i

Customer: Contact:

Nottingham City Council Guy Hitchcock

Ricardo Energy & Environment Confidentiality, copyright & reproduction: Gemini Building, Harwell, Didcot, OX11 0QR, United Kingdom This report is the Copyright of Nottingham City Council. It has been prepared by Ricardo Energy

& Environment, a trading name of Ricardo-AEA Ltd, under contract to Nottingham City Council t: +44 (0) 1235 75 3327 dated Click here to enter a date. The contents of e: [email protected] this report may not be reproduced in whole or in part, nor passed to any organisation or person without the specific prior written permission of Nottingham City Council. Ricardo Energy & Ricardo-AEA Ltd is certificated to ISO9001 and Environment accepts no liability whatsoever to ISO14001 any third party for any loss or damage arising from any interpretation or use of the information contained in this report, or reliance on any views Authors: expressed therein.

Michel Vedrenne, Anne Misra, Ancelin Coulon

Approved By:

Guy Hitchcock

Date:

16 May 2018

Ricardo Energy & Environment reference:

Ref: ED10107- Issue Number 6

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | ii

Table of contents

1 Introduction ...... 4 2 Modelling scope ...... 6

2.1 Outline scheme options ...... 6

2.2 Model domain ...... 6

2.3 Modelling years ...... 8

2.4 Background modelling ...... 9 3 Air Quality modelling assessment ...... 9

3.1 Domain ...... 9

3.2 Model selection...... 10

3.3 Air quality model receptor locations ...... 12

3.4 Base year modelling ...... 13

3.4.1 Base year and meteorological dataset ...... 13

3.4.2 Road traffic modelling ...... 13

3.4.2.1 Average daily vehicle flow and speeds ...... 13

3.4.2.2 Vehicle fleet composition ...... 13

3.4.2.3 Representation of road locations ...... 14

3.4.2.4 NOx/NO2 emissions assumptions...... 14

3.4.3 Non-road transport modelling and background concentrations ...... 14

3.4.4 Measurement data for model calibration ...... 16

3.5 Projected future year scenario modelling ...... 16

3.5.1 Road transport future year baseline ...... 16

3.5.2 Scheme option modelling projections ...... 17 4 Model results for 2016 base year and 2020 baseline ...... 18

4.1 Comparison with PCM ...... 18

4.2 Results for AQMAs and local exceedances ...... 32

4.3 Model uncertainty ...... 36

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | iii

Appendices

Appendix 1 Air Quality Modelling QA Table

Appendix 2 RapidAir street canyon equations

Appendix 3 Air quality model verification and adjustment

Appendix 4 Analytical assurance statement

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 4

1 Introduction

Nottingham, like many other urban areas, has elevated levels of Nitrogen Dioxide (NO2) due mainly to road transport emissions. Due to local air quality problems, Nottingham City Council (NCC) has designated 2 Air Quality Management Areas (AQMA) across the City where concentrations of NO2 breach Government, health-based air quality objectives and has undertaken reviews of current and predicted levels in the future, including assessments of measures to reduce pollution levels.

At the national level, the EU has commenced infraction proceedings against the UK Government and Devolved Administrations for their failure to meet the EU Limit Value for NO2. In 2015, the Supreme Court ordered the Government to consult on new air pollution plans that had to be submitted to the European Commission no later than 31 December 2015. As such DEFRA released plans1 to improve air quality, specifically tackling NO2, in December 2015. The Plans identify 5 cities outside London, including Nottingham, where the EU Limit Value for NO2 are not expected to be met by 2020. The Plans state that each of the cities identified could be legally required to introduce a formal charging-based Clean Air Zone (CAZ) for specified classes of vehicles and European Vehicle Emission Standards (Euro Standards) by 2020 or sooner.

Figure 1 Nottingham Air Quality Management Areas (AQMA)

1 https://www.gov.uk/government/publications/air-quality-in-the-uk-plan-to-reduce-nitrogen-dioxide-emissions

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 5

The assessment undertaken for the Nottingham Urban Agglomeration by DEFRA as part of the 2017 UK Air Quality Plans indicated that the annual limit value was exceeded in 2015 but is likely to be achieved in 2024 through the introduction of measures included in the baseline. When combined with the Class B CAZ envisaged by DEFRA, it is expected that this zone will achieve compliance by 2020. The key roads identified by the DEFRA plan that exceed in 2015 the annual limit value of NO2 are in AQMA No. 2, which is the large, horseshoe-shape area close to the city centre of Nottingham. Other roads in exceedance are the Western Boulevard, and the National Road at the West of the city.

Ongoing work by DEFRA to update its air quality plan is using more recent information on the expected real-world emission performance of vehicles. This latest analysis is suggesting that emission from vehicles will be higher than previously estimated and so breaches of the air quality limits are likely to persist for longer and over a wider area. It should also be noted that the source apportionment suggested that the exceeding roads in the Nottingham Urban Area are influenced notably by diesel cars, diesel LGVs and buses.

With the objective of designing and putting in place a Clean Air Zone for Nottingham, Nottingham City Council has commissioned a detailed feasibility study for the CAZ covering the transport and air quality impacts of the scheme, and developing a business case for implementation. This report sets out the modelling needs in relation to assessing the air quality impacts of the scheme.

Figure 2 Road transport NOx emissions source apportionment for the Nottingham Urban Area (2015). Source: DEFRA

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 6

2 Modelling scope 2.1 Outline scheme options

Nottingham has been requested by DEFRA to consider the implementation of a Clean Air Zone which is intended to restrict the use of the most pollutant vehicles. DEFRA have four classes of Clean Air Zone from “Class A” which is the least restrictive type covering buses and taxis only, through to “Class D” which covers all vehicle categories. Nottingham is most likely to deliver a “Class B” zone, which will restrict the use of buses, taxis and heavy goods vehicles (HGVs).

In addition to the CAZ, Nottingham’s primary approach to reducing NO2 concentrations from road traffic has been to implement schemes which encourage the use of alternative modes of transport to the private car. Examples of these are:

• the Local Taxi Strategy, which aims to get all taxis to meet a ULEV standard by 2025, with at least 40% of the fleet ULEV by 2020;

• the Go Ultra Low Programme which funds the installation of fast charging points and grants for electric vehicles;

• improving the use of sustainable forms of transport and cycling through the Nottingham City Cycle Ambition Programme;

• adopting improvements for bus fleets;

• the creation of a substantially electric, bus-based corridor that benefits from lengths of new bus lanes;

• and the implementation of a work place parking levy.

In going forward with developing a Clean Air Zone for Nottingham, it is therefore necessary to consider different options for a formal charging CAZ along with a package of support measures. In defining options for the charging CAZ, the approach proposed is to aim for a scheme that achieves compliance with the lowest level of cost both to the city council and the transport users in the city. On this basis, three initial boundary options are being considered along the following lines (Figure 3):

• A Clean Air Zone with a City-Centre boundary. • A Clean Air Zone with its outer boundary based on arterial routes to the outer ring road. • A Clean Air Zone with its outer boundary based on the City Area and extension into borough arterial road links.

Alongside the formal charging CAZ, a package of measures aimed at aiding compliance and wider reductions in emissions form part of the overall approach. A committed package of these measures will be assessed alongside the main charging CAZ feasibility study. The details of this likely package are currently being considered by the Council.

2.2 Model domain

In carrying out the modelling of the air quality impacts of the scheme a model domain is required that covers the scheme options, relevant AQMAs and potential diversion routes. Therefore, the proposed model domain shown in Figure 3 has been chosen to cover the following:

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 7

• All the AQMAs in Nottingham including the main roads of concern from the national modelling assessment;

• The wider transport network covering all the key diversion routes

Further details in relation to the model domain are provided in section 3.2 of the air quality modelling assessment.

Figure 2 Illustrative CAZ boundaries

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 8

Figure 3 Proposed model domain

2.3 Modelling years

There will be two key model years used in the modelling work as set out in Table 1 below. The base year is taken as 2016 as this covers the latest air quality and transport data. The target implementation year for CAZ is 2020. Intervening years can be assessed through simple interpolation to as necessary.

Table 1 Model years

Year Description

2016 Base year – using latest available data on air quality and transport

2020 Implementation year – year when CAZ scheme is due

The details of the forecast transport model years and the sensitivity assessment have been provided in the Transport Model Forecasting Methodology Report prepared by SYSTRA for Nottingham City Council.

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 9

2.4 Background modelling

The primary cause of the air pollution problems in Nottingham are related to traffic activity and the impact of the CAZ will be in relation to this traffic activity. As such the focus of the modelling is the transport emissions. However, there are several other background sources that are important in Nottingham and will need to be covered specifically in the modelling work and will cover industrial emissions related to the incinerator at the Queens Medical Centre, which is located close to an AQMA and the ring road, as well as the facilities of Wastenotts Reclamation Ltd. The details of how these sources will be treated and their relation to the wider background is described in section 3.4.3. 3 Air Quality modelling assessment 3.1 Domain

The core air quality model domain covers the area of Nottingham bounded by the M1 motorway and Chilwell to the west, Warren Hill and Redhill to the north, the A52 motorway and the Trent River to the east and south, as well as the Ruddington.

Displacement of traffic due to the implementation of CAZ measures is not expected to occur beyond the proposed model domain and the sub-regional traffic model proposed to support the study (discussed in section 4 and built and run by SYSTRA) has been chosen as it fully encompasses the affected areas.

A map showing the extent of the air quality receptor domain relative to the proposed CAZ zones and the associated traffic model network is presented in Figure 4. A map showing the model domain relative to roads included in the national Pollution Climate Mapping (PCM) model is presented in Figure 5. All road links in the PCM model pertinent to Nottingham are included in the model domain specification.

Nottingham City Council has declared 2 Air Quality Management Areas (AQMA’s) across the city to date, all of which are within the proposed model domain (Figure 1). Gedling Borough Council has declared one area encompassing part of the A60 where it meets the City’s administrative boundary, and Rushcliffe Borough Council has declared 2 contiguous areas, one including part of the A60 where it meets the City’s southern administrative boundary at Trent Bridge. Air Quality Management No. 2 (AQMA 2) was determined and formalised in 2010 after monitoring data and modelling identified an 3 exceedance of the NO2 annual mean objective of 40 μg/m . AQMA 2 covers an area encompassing the City Centre arterial routes roads and junction including the A60, A610, A6002, A612 and properties fronting onto these roads. Air Quality Management Area No. 1 was declared by the Rushcliffe Borough Council, being an area encompassing the Lady Bay Bridge/Radcliffe Road Junction, the Trent Bridge and the Wilford Lane/Loughborough Road/Melton Road Junction.

All of Nottingham City Council’s 2016 NO2 roadside measurements will be used in the air quality modelling assessment to verify the model outputs, assuming data capture and QA/QC are satisfactory for the 2016 baseline year. A map showing the sites at which NO2 concentrations were measured during 2016 is presented in Figure 7.

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 10

Figure 4: CAZ study domain and relationship SYSTRA’s sub-regional transport model links

3.2 Model selection

RapidAir will be used for the study- this is Ricardo Energy & Environment’s proprietary modelling system developed for urban air pollution assessment. The model is based on convolution of an emissions grid with dispersion kernels derived from the USEPA AERMOD2 model. The physical parameterisation (release height, initial plume depth and area source configuration) closely follows guidance provided by the USEPA in their statutory road transport dispersion modelling guidance3. AERMOD provides the algorithms which govern the dispersion of the emissions and is an accepted international model for road traffic studies (it is one of only two mandated models in the US and is widely used overseas for this application). The combination of an internationally recognised model code and careful parameterisation matching international best practice makes RapidAir demonstrably fit for purpose for this study.

The USEPA have very strict guidelines on use of dispersion models and in fact the use of AERMOD is written into federal law in ‘Appendix W’ of the Guideline on Air Quality Models4. The RapidAir model uses AERMOD at its core and is evidently therefore based on sound principles given the pedigree of the core model.

The model produces high resolution concentration fields at the city scale (1 to 3m scale) so is ideal for spatially detailed compliance modelling. A validation study has been conducted in London using the same datasets as the 2011 Defra inter-comparison study5. Using the LAEI 2008 data and the

2 https://www3.epa.gov/ttn/scram/dispersion_prefrec.htm#aermod 3 https://www.epa.gov/state-and-local-transportation/project-level-conformity-and-hot-spot-analyses 4 40 CFR Part 51 Revision to the Guideline on Air Quality Models: Adoption of a Preferred General Purpose (Flat and Complex Terrain) Dispersion Model and Other Revisions; Final Rule, Environmental Protection Agency, 2005 5 https://uk-air.defra.gov.uk/research/air-quality-modelling?view=intercomparison

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 11

measurements for the same time period the model performance is consistent (and across some metrics performs better) than other modelling solutions currently in use in the UK. A paper is currently being finalised for publication with our partners at Strathclyde University in a suitable journal (most likely Atmospheric Environment).

Figure 5: PCM model road links within the CAZ study domain 2015

The platform includes two very well-known street canyon algorithms with significant pedigree in the UK and overseas. The first replicates the functionality of the USEPA ‘STREET’ model. The code was developed by the Office of Mobile Source Air Pollution Control at the USEPA and published in a series of technical articles aimed at operational dispersion modellers in the regulatory community6,7. The STREET model has been used for many years and has been adopted in dispersion modelling software such as AirViro. The USEPA canyon model algorithms are essentially the same as those recommended by the European Environment Agency for modelling canyons in compliance assessment8.

The RapidAir model also includes the AEOLIUS model which was developed by the UK Met Office in the 1990s. The AEOLIUS model was originally developed as a nomogram procedure9. The scientific

6 Ingalls., M. M., 1981. Estimating mobile source pollutants in microscale exposure situations. US Environmental Protection Agency. EPA-460/3-81-021 7 USEPA Office of Air Quality Planning and Standards., 1978. Guidelines for air quality maintenance planning and analysis, Volume 9: Evaluating indirect sources. EPA-450/4-78-001 8 http://www.eea.europa.eu/publications/TEC11a/page014.html 9 Buckland AT and Middleton DR, 1999, Nomograms for calculating pollution within street canyons, Atmospheric Environment, 33, 1017-1036.

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 12

basis for the model is presented in a series of papers by the Met Office10,11,12,13,14. The model formulation shares a high level of commonality with the Operational Street Pollution Model1516 (OSPM) which in turn forms the basis of the basic street canyon model included in the ADMS-Roads software. Therefore, the AEOLIUS based canyon suite in RapidAir aligns well with industry standards for modelling dispersion of air pollutants in street canyons.

Gradient effects will be included for relevant road links during emissions calculations. LIDAR Composite Digital Terrain Model (DTM) datasets at 1m and 2m resolution are available over the proposed model domain17. Link gradients across the model domain can be calculated using GIS spatial analysis of LIDAR DTM datasets.

The method described in TG(16) provides a method of adjusting road link emission rates for gradients greater than 2.5%; it is applicable to broad vehicle categories for heavy vehicles only. As per the guidance and clarification provided by JAQU this adjustment has been applied to all pre-Euro VI HGVs and buses.

A limitation of this approach is that the method in TG(16) only adjusts emissions from older heavy vehicles; there are no gradient terms in the guidance that allow for correction of newer vehicles of any type. 3.3 Air quality model receptor locations

We would suggest that even a resolution of 10m is too coarse to reliably inform a complex compliance assessment of this nature. For the Nottingham domain, we will set RapidAir to model at a maximum of 2m resolution. The model can comfortably deal with about 500 million gridded locations which provides for over 20,000 cells in the x and y axes. We can therefore model 20km x 20km at 1m resolution, 40km x 40km at 2m resolution, 60km x 60km at 3m resolution and so on. The canyon model is set to the same resolution as the grid model so that they align perfectly spatially.

Nottingham has a wide network of monitoring locations comprising a mix of passive and active sampling. All available monitoring locations will be treated as receptors in the model as the 2016 NO2 annual mean measurements will be used for model verification and producing model performance statistics. We use the Openair software which is developed here at Ricardo to calculate model performance metrics- primarily we based our performance judgement on the coefficient of determination (R2), root mean square error (RMSE), mean bias (MB), index of agreement (IOA) and coefficient of efficiency (COE). We use model performance metrics to inform the iterative development of the model- that is to say we do not simply accept the first model run and make broad linear adjustments. Rather we use the metrics to investigate uncertainties as we iterate and improve the model over time.

As RapidAir produces concentration grids (in raster format) modelled NO2 concentrations can be extracted at receptor locations anywhere on the 2m resolution model output grid. For comparison with PCM model results, annual mean concentrations at a distance of 4m from the kerb can be extracted from the RapidAir data and presented as a separate model output file. This will also enable a selection

10 Middleton DR, 1998, Dispersion Modelling: A Guide for Local Authorities (Met Office Turbulence and Diffusion Note no 241: ISBN 0 86180 348 5), (The Meteorological Office, Bracknell, Berks). 11 Buckland AT, 1998, Validation of a street canyon model in two cities, Environmental Monitoring and Assessment, 52, 255-267. 12 Middleton DR, 1998, A new box model to forecast urban air quality, Environmental Monitoring and Assessment, 52, 315-335. 13 Manning AJ, Nicholson KJ, Middleton DR and Rafferty SC, 1999, Field study of wind and traffic to test a street canyon pollution model, Environmental Monitoring and Assessment, 60(2), 283-313. 14 Middleton DR, 1999, Development of AEOLIUS for street canyon screening, Clean Air, 29(6), 155-161, (Nat. Soc for Clean Air, Brighton, UK). 15 Hertel O and Berkowicz R, 1989, Modelling pollution from traffic in a street canyon: evaluation of data and model development (Report DMU LUFT A129), (National Environmental Research Institute, Roskilde, Denmark). 16 Berkowicz R, Hertel O, Larsen SE, Sørensen NN and Nielsen M, 1997, Modelling traffic pollution in streets, (Ministry of Environment and Energy, National Environmental Research Institute, Roskilde, Denmark). 17 http://environment.data.gov.uk/ds/survey/#/survey

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 13

of locations to be assessed according to the Air Quality Directive (AQD) requirements Annex III A, B, and C3.

Nottingham has two AQMAs all of which contain numerous residential receptors. RapidAir, by virtue of its very high-resolution outputs, will produce discrete estimates at every single residential property in Nottingham (every 2m ‘square’ in actual fact); any location where there is a risk of the objective being exceeded will therefore be included in the modelling and outlined during post processing.

To aid interpretation of the outcomes of the study when considering compliance with the air quality directive (AQD). Annual mean concentrations at the roadside exceedance locations identified in the PCM model can be extracted from the RapidAir dispersion model results and presented as a separate model output file. These receptor locations will be at a distance of 4m from the kerb and at 2m height.

Annex III of the AQD specifies that macroscale siting of sampling points should be representative of air quality for a street segment of no less than 100 m length at traffic-orientated sites. To provide results relevant to this requirement, for roadside locations where there is public access and the directive applies; road links with exceedances of the NO2 annual mean objective stretching over link lengths of 100m or greater can be presented as a separate GIS layer of model results.

Annex III of the AQD also specifies that microscale sampling should be at least 25 m from the edge of major junctions. When reporting model results relevant to compliance with the AQD, locations up to 25m from the edge of major junctions in the model domain will therefore be excluded. 3.4 Base year modelling

3.4.1 Base year and meteorological dataset

As described in section 2.2 we proposed to model a baseline year of 2016. We will use the 2016 annual surface meteorological dataset measured at East Midlands Airport processed in house using our own meteorological data gathering and processing system. We use open overseas met databases which hold the same observations as supplied by UK met data vendors. Our RapidAir model also takes account of upper air data which is used to determine the strength of turbulent mixing in the lower atmosphere; we will derive this from the closest radiosonde site and process with the surface data in the USEPA AERMET model. We will utilise data filling where necessary following USEPA guidance which sets out the preferred hierarchy of routines to account for gaps (persistence, interpolation, substitution). Our modelling will be supplied with full meteorological discussion and if required we can supply the computer code used to process the data and details of any data filling that was required.

3.4.2 Road traffic modelling

3.4.2.1 Average daily vehicle flow and speeds

Baseline and future year annual average daily traffic (AADT) link flows for each model link will be provided by SYSTRA using outputs from the Greater Nottingham Transport Model (GNTM) that covers the areas of Nottingham and its surroundings (Greater Nottingham).

Baseline daily average link speeds will be calculated using the DfT Traffic Master GPS measured datasets cross referenced with the Ordnance Survey ITN roads GIS dataset. This will provide observed average speed data over defined road links at a fairly well-resolved spatial resolution. It should also provide a reasonable representation of the change in emissions at locations where typical vehicle speeds are reduced e.g. approaching junctions. 3.4.2.2 Vehicle fleet composition

Vehicle emission rates for the vehicle categories Buses, taxis, coaches, rigid HGVs, articulated HGVs, LGVs, cars and motorcycles can be calculated using the latest COPERT v5 NOx emission functions. The traffic model will provide vehicle flows for four highway user classes which are: Car, HGV, LGV and

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 14

Buses. A further breakdown of the HGV into rigid and articulated categories and an estimate of the proportion of car traffic that are taxis can be conducted using local traffic count data and ANPR data. In this particular case, the ANPR data for HGVs was not differentiated between rigid and articulated categories so the split factor between the two was estimated from AADF values for major roads located within the boundaries of Nottingham, contained within the 2015 Streamlined-PCM emissions tool (v3.2.1)18.

Emission calculations for each vehicle category will be based on vehicle age split by Euro classification. Information on the baseline Euro standard mix (traffic composition & age) has been collected during ANPR surveys. The Euro 6 emission standard was disaggregated into the three Real Driving Emissions stages for light vehicles (passenger cars and LGVs) using the NAEI projections. An average distribution of Euro classifications calculated from the complete ANPR dataset can either be applied across the entire model domain; or road specific fleet age composition can be calculated. We are currently in the process of reviewing DfT/DVLA analysis of the ANPR surveys. 3.4.2.3 Representation of road locations

A realistic representation of road locations will be modelled by assigning emissions to the road links represented in the Ordnance Survey ITN Roads GIS dataset; it contains reasonably spatially accurate road centreline locations for various road categories e.g. Motorway, A road, B road, minor road, local street etc.

3.4.2.4 NOx/NO2 emissions assumptions

Link specific NOx emission factors will be calculated using the COPERT v5 emission functions for all vehicles up to and including Euro 6/VI. Emission rates will be calculated with our in-house emission calculation tool pyCOPERT as agreed by JAQU, which is fully consistent with COPERT v5 and links directly to our RapidAir dispersion modelling system.

JAQU recommend the use of data on primary NO2 emissions (fNO2) by vehicle type which is available via the NAEI website (based on 2014 NAEI) to provide a more detailed breakdown than the LAQM NOx to NO2 convertor. This suggests a link specific f-NO2 emissions estimate for use in the NO2 modelling. Based on this requirement, the pyCOPERT road emissions calculation tool now includes additional functionality to calculate fNO2 emission rates for each road link. Link specific fNO2 fractions can then be calculated for each link by dividing fNO2 by total road NOx emission rate.

Calculating link specific fNO2 emission rates also facilitates dispersion modelling of both road NOx and fNO2 across the entire model domain to produce separate concentration rasters, which can then be combined with background concentrations to calculate NO2 concentrations in each grid cell. The recently updated version (v5.3) of the LAQM NOx to NO2 conversion spreadsheet will be used to convert road NOx, fNO2 and background NOx into NO2 concentrations where results at discrete receptor locations are required. This currently includes all NO2 monitoring site locations and receptors placed at 4m from the PCM road links.

3.4.3 Non-road transport modelling and background concentrations

We proposed to model non-road transport sources of NOx emissions using the following types of emission (and background concentration) data.

1. The LAQM background maps (Updated background maps which have been adjusted to take into account new COPERT 5 emission factors will be provided by JAQU). The contribution from local road transport sources that are modelled explicitly will be subtracted from the background maps.

18 https://uk-air.defra.gov.uk/library/no2ten/2017-no2-projections-from-2015-data

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 15

2. Rail transport modelling. The contribution of rail transport to the air quality of Nottingham will be modelled by extracting the emissions reported by the NAEI for rail activities within the modelling domain. These emissions will be allocated to the main railway lines of the city and modelled as individual line sources with RapidAir. As an example, the NOx emissions reported by the NAEI for the city of Nottingham are presented in Figure 6.

Figure 6 NOx rail emissions reported by the NAEI in 2015. Source: DEFRA

© Crown copyright and database rights 2017 licenced under BEIS's Public Sector Mapping Agreement with Ordnance Survey (licence No. 100037028) and Defra's Public-Sector Mapping Agreement with Ordnance Survey (licence No. 100022861).

3. Nearby large point sources - We intend to model emissions from selected nearby industrial sources categorised as large point sources in the NAEI. This aims to provide a more resolved footprint of these sources contribution to background NOx/NO2 concentrations than available from the 1km LAQM background maps. We are interested in the following nearby large point sources currently listed on the NAEI. • Queens Medical Centre CHP Boiler. • Wastenotts Reclamation Ltd. • The Boots plc (two stacks). • FCC Environment Eastcroft Energy from Waste Facility (EfW). • Enviroenergy Nottingham District Heating and Power. • Nottingham University City Hospital (future replacement boiler plant – 5 stacks). Stack parameters (height and diameter) were provided for all the sources. Release conditions were provided for all the sources listed above except for the Queens Medical Centre CHP Boiler and the two stacks of The Boots plc. These two sources were modelled as elevated small area sources, which effectively implies that no buoyancy effects are being considered.

To avoid double counting of non-road transport sources, for any explicitly modelled sources, we will adjust the background maps by calculating the fraction of the total NAEI emissions being modelled explicitly, we can then subtract this fraction from the relevant source sector category in the NOx background maps.

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 16

3.4.4 Measurement data for model calibration

Nottingham City Council’s 2016 automatic and diffusion tube annual mean NO2 measurements from roadside sites will be used for model verification. Information on monitoring data QA/QC, diffusion tube bias adjustment factors etc. will be as presented in the Nottingham City Council 2017 LAQM Annual Progress Report.

Figure 7 Nottingham City Council NO2 monitoring sites 2016

3.5 Projected future year scenario modelling

3.5.1 Road transport future year baseline

Future year baseline scenarios will be modelled in the years 2018, and 2020 as described in section 2.2.

Each of the main modelling issues for the future year baseline scenarios are described in turn below:

• AADT flows for future baseline years will be provided from the SYSTRA sub-regional traffic model. Further information on how these traffic flows will be derived and how local growth in traffic will be

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 17

calculated as explained in the Transport Model Forecasting Methodology Report prepared by SYSTRA for Nottingham City Council.

• Projected fleet split (vehicle type): All future year scenarios will have the four core-vehicle category fleet splits provided from the traffic model in the same breakdown as provided for the 2016 base year. The further split of HGV’s into artic and rigid, and for taxis will use the same ratios as derived for the 2016 baseline.

• Projected fleet age composition (Euro class): It will be necessary to adjust the 2016 baseline Euro fleet composition to account for turnover in the local fleet in the future baseline years being modelled. The JAQU specification document states that they are working on providing a forecast of local fleet (with Euro class) composition tool for the EFT which will be available shortly. We are currently assuming that this tool will be available and will provide a standardised method of adjusting the local fleet age composition to future years. In the absence of this tool, we can derive our own method of adjusting the local ANPR derived fleet age to future baseline years; this will be based on the fleet turnover contained in the national fleet projections.

• Future year scenarios average vehicle speed data: Average link speeds for all future year scenarios will be calculated by adjusting the observed baseline speed data (Traffic Master) by the ratio of the CAZ scenario vs future baseline journey times calculated by the traffic model. For links in the traffic model where junction queuing is included, speeds will be adjusted by distributing extra time pro rata where time is spent on the links where queuing occurs.

• Projected vehicle NOx emission rates will be calculated using the latest COPERT v5 NOx emission functions applied to the projected average flows, fleet and vehicle age composition for each future baseline year being modelled.

3.5.2 Scheme option modelling projections

Table 2 JAQU assumptions on behavioural response to the CAZ

Proportions of non-compliant vehicle kilometres which react to the zone

Petrol Diesel Petrol Diesel RHGVs AHGVs Buses Coaches Cars Cars LGVs LGVs

Pay charge – Continue into 7.1% 7.1% 20.3% 20.3% 8.7% 8.7% 0.0% 15.6% zone

Avoid Zone – Vkms removed, modelled 21.4% 21.4% 10.0% 10.0% 0.0% 0.0% 0.0% 0.0% elsewhere

Cancel journey – vkms removed 7.1% 7.1% 6.0% 6.0% 8.7% 8.7% 6.4% 12.5% completely

Replace Vehicle – vkms replaced with 64.3% 64.3% 63.8% 63.8% 82.6% 82.6% 93.6% 71.9% compliant vkms

Source: JAQU, CAZ Technical working group minutes – 15/2/17

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 18

Following the traffic model run the compliant and non-compliant vehicles will be modelled as two separate fleets in the emission model with their own Euro standard distribution. The emissions from each of these fleets will then be added up for each link to link specific emission representing the mix of compliant and non-complaint vehicles on that link. All background concentration data will remain the same as in the baseline forecasts.

4 Model results for 2016 base year and 2020 baseline 4.1 Comparison with PCM

For comparison with national PCM model results, annual mean NO2 concentrations at the roadside locations assessed in the national compliance PCM model have been extracted from the RapidAir dispersion model results; the results have been presented in both tabular form and using graduated colours on a map of the study area.

Roadside receptor locations in the PCM model are at a distance of 4m from the kerb and at 2m height. To represent this in our city scale modelling, a subset of the OS Mastermap GIS dataset provided spatially accurate polygons representing the road carriageway; receptor locations were then placed at 50m intervals along relevant road links using a 4m buffer around the carriageway polygons.

Each PCM link has a unique Census ID number and a grid reference assigned which is typically the co- ordinates describing the location of the DfT traffic count points on each link; this location may not however be where the highest roadside concentrations are occurring along the entire link length when using a more detailed local scale modelling method with observed average vehicle speeds on shorter road sections. The PCM links within our model domain range in length from approximately 90m to 3.85km; we have therefore reported the highest of the modelled concentrations from the city scale model receptors, 4m from the carriageway.

A comparison of the local model maximum concentration modelled 4m from each PCM link and the corresponding PCM results (in the list provided by JAQU only) in 2016, 2018, and 2020 are presented in Table 3. Maps showing the predicted annual mean results in 2016, 2018 and 2020 are presented in Figure 8 to 13. These model results should be considered in context with the model uncertainty quantified during model verification (see Section 4.3)

The results show that:

• At locations where the PCM model indicates that NO2 annual mean concentrations will be less than the 40 µg.m-3 objective in 2020, the tabulated results of the local model are largely consistent with the PCM results with the exception of the following Census IDs:

o 28415 – A6008 King Street/Parliament Street section: where the local model indicates a significantly higher concentration in both 2016 and 2020. The differences are basically due to the significantly lower speeds that are considered in the local model (13 and 19 km/h against 36 km/h considered by the PCM). Due to these differences in speed, the average fleet-weighted emission factors for the two roads are 4.4 and 4.7 times higher than those originally considered in PCM for 2020. In addition to this, the local model considers significantly higher vehicle flows (2.5 and 3.7 times higher than PCM, respectively).

• At all locations where the PCM model indicates that NO2 annual mean concentrations will be in excess of the 40 µg.m-3 objective in 2020; the local model predicts compliance. These model results should also be considered in context with the model uncertainty quantified during model

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 19

verification. The relevant census ID points are where exceedance was predicted by the national model were:

o 7350 – A6514 Middleton Boulevard section: where the local model shows significantly lower concentrations than PCM in both 2016 and 2020 (the latter year in compliance). The difference in concentrations may be due to the lower speeds considered by the PCM compared to the local model (36 km/h and 55 km/h respectively). Two diffusion tubes are located along Census ID 7350: DT02 – Road and DT32 – Middleton Boulevard. In both cases, these diffusion tubes record NO2 concentrations that are significantly lower than those estimated by PCM (27.4 μg/m3 and 26.6 μg/m3 versus 49.6 μg/m3). Since the local model predicts values that are closer to those measured by the diffusion tubes, these are considered to be more representative.

o 17304 – A52 Clifton Boulevard section: where the local model predicted compliance in 2020 (31.8 μg/m3), as opposed to PCM (42.0 μg/m3). The difference in concentrations may be due to lower vehicle flows in the local model for this road (about half the value considered in PCM). One diffusion tube is located close to Census ID 17304: DT03 – Marlborough Street. The concentrations of NO2 measured by this diffusion tube was 25.1 μg/m3, while the outputs of the local model and PCM are 26 μg/m3 and 49.9 μg/m3 respectively. Since the local model predicted an NO2 concentration that is closer to that measured by the diffusion tube, it is considered to be more representative.

o 75216 – A6514 Middleton Boulevard section: where the local model predicted compliance in 2020 (27.9 μg/m3), as opposed to PCM (48.7 μg/m3). The difference in concentrations may be due to lower vehicle flows in the local model for this road (0.41 times the value considered in PCM). One diffusion tube is located close to Census ID 75216 and was not excluded from the statistical analysis: DT31 – Middleton Boulevard. 3 The concentrations of NO2 measured by this diffusion tube was 28.6 μg/m , while the outputs of the local model and PCM are 28.6 μg/m3 and 58.6 μg/m3 respectively. Since the local model predicted an NO2 concentration that is closer to that measured by the diffusion tube, it is considered to be more representative.

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 20

-3 Table 3: NO2 annual mean concentrations 2016 baseline year as well as 2018 and 2020 future baseline year - Comparison of PCM vs local model results (µg.m )

Census ID grid 2016 2018 2020 Road Census ID Length (m) Name X Y PCM Local PCM Local PCM Local

6541 A6011 2,138 460000 337460 33.6 27.8 31.4 25.9 28.7 24.1 7176 A6005 1,568 455800 339140 38.3 35.5 36.4 32.4 33.9 29.2 7342 A610 1,045 453000 343740 42.4 27.8 39.5 25.6 36.0 23.5 7349 A60 1,558 457960 345000 34.7 32.3 32.7 30.0 30.0 27.8 7350 A6514 1,527 454480 339300 49.6 31.4 45.7 29.3 41.3 27.3 7351 A60 272 457880 339000 38.0 38.4 36.0 35.7 33.4 32.9 7873 A52 2,411 452000 338100 34.3 31.4 32.2 29.9 29.5 28.4 7874 A611 3,007 455060 345500 27.1 30.2 25.3 27.8 23.1 25.4 7875 A6130 1,510 456000 341270 26.8 29.7 25.4 28.1 23.5 26.4 8155 A6520 505 458300 338150 34.3 30.3 32.3 27.9 29.6 25.5 8338 A6002 668 450000 338600 26.5 24.6 24.7 21.7 22.6 18.8 16521 A52 2,401 450000 337040 36.7 28.6 34.4 26.3 31.5 23.9 16572 A60 1,170 457822 344000 32.8 38.3 30.8 34.1 28.1 30.0 17297 A610 1,800 455000 341300 35.9 33.0 33.7 29.3 30.8 25.5 17303 A6514 2,131 456000 343400 35.4 35.0 33.4 32.6 30.8 30.3 17304 A52 1,688 455600 338000 49.9 38.9 46.2 35.4 42.0 31.8 17852 A60 346 457450 340470 38.4 36.7 36.2 33.9 33.3 31.1 17853 A60 148 458050 338300 40.0 35.5 37.6 32.1 34.4 28.7 17854 A609 950 455000 340100 31.8 29.2 29.8 26.9 27.3 24.6 17855 A6011 376 458000 338840 32.5 29.1 30.3 27.1 27.8 25.1

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 21

Census ID grid 2016 2018 2020 Road Census ID Length (m) Name X Y PCM Local PCM Local PCM Local

17856 A6130 745 455500 340000 32.4 31.3 31.0 29.1 29.1 26.9 17857 A6008 498 457600 339120 35.9 40.3 34.0 39.3 31.7 38.2 18325 A6008 276 457150 340200 27.8 37.3 26.0 34.3 24.0 31.2 18414 A6005 3,099 451630 335240 27.3 30.3 25.7 28.2 23.6 26.2 26550 A6200 1,043 455962 339964 36.6 45.5 34.1 41.7 31.2 37.8 26599 A60 4,498 457570 335000 28.9 24.5 27.2 22.2 24.9 19.8 27200 A6005 3,127 455000 338340 41.4 37.6 38.9 34.2 35.8 30.7 27360 A609 2,606 454000 340000 31.4 34.9 29.5 31.6 27.1 28.4 27363 A6002 1,837 454360 346100 33.2 25.5 31.0 23.6 28.2 21.6 27369 A6514 2,166 454450 341000 40.7 38.1 38.0 34.9 34.6 31.7 27898 A60 208 457880 339200 45.0 40.5 42.7 38.5 39.6 36.6 27899 A6464 1,216 452843 338029 28.2 28.6 26.4 26.9 24.1 25.1 28303 A60 203 457860 339400 44.4 34.7 42.3 32.4 39.3 30.0 28415 A6008 244 457200 340080 39.4 49.9 35.2 46.3 30.9 42.7 37407 A6008 352 457000 339800 44.3 47.7 41.1 42.6 37.6 37.5 37409 A610 1,607 454000 343260 41.7 37.3 38.6 33.1 35.1 28.9 37416 A52 660 454717 338620 46.4 37.1 43.0 33.7 39.2 30.3 37959 A6008 120 457747 340000 36.8 31.7 34.7 29.5 32.2 27.3 37960 A611 2,196 456890 342000 30.8 30.3 28.8 27.2 26.3 24.1 37961 A6005 459 455600 339000 36.6 32.7 34.6 30.5 32.1 28.3 37962 A6011 384 458620 338350 36.2 26.9 34.2 25.3 31.6 23.7

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 22

Census ID grid 2016 2018 2020 Road Census ID Length (m) Name X Y PCM Local PCM Local PCM Local

37963 A6130 828 455430 340500 28.4 27.3 27.0 25.3 25.1 23.4 38358 A6008 223 457450 340100 33.0 44.3 29.9 40.6 26.6 36.9 38423 A6002 579 451013 340600 33.2 24.1 31.2 21.9 28.6 19.7 46610 A6008 154 457850 340000 37.4 34.2 35.3 31.9 32.7 29.6 47381 A52 879 457200 335000 41.0 28.6 38.4 26.3 35.1 24.0 47439 A60 1,714 458260 338000 32.3 35.4 30.2 32.5 27.6 29.5 47798 A6007 1,232 450000 338340 26.8 22.5 25.1 20.6 23.0 18.6 47925 A453 2,448 456900 338911 38.4 36.3 36.1 34.3 33.2 32.2 47927 A6130 611 455600 339200 34.2 33.8 32.5 31.6 30.3 29.4 48221 A60 825 457196 340631 36.3 41.4 33.3 37.7 30.0 34.0 48407 A6002 1,414 452790 345100 32.7 19.7 30.2 18.2 27.2 16.8 48497 A60 369 457320 340160 29.6 39.6 27.1 35.3 24.2 31.1 56182 A60 636 456930 341700 37.8 42.5 35.2 37.4 31.9 32.2 56188 A52 1,394 456190 336700 42.6 25.5 39.8 23.8 36.4 22.1 56557 A52 1,619 454000 338860 44.3 39.1 41.2 34.2 37.4 29.4 56838 A609 850 456000 340180 31.7 30.2 29.6 27.5 27.0 24.8 56857 A610 1,033 456000 340520 34.2 34.9 31.9 32.3 29.1 29.7 56903 A453 1,551 455000 335035 37.6 36.2 34.1 33.0 30.7 29.8 57380 A6514 1,537 457400 344000 33.7 28.1 31.7 25.7 29.1 23.3 57482 A6008 393 457060 339140 35.4 38.5 33.7 37.3 31.4 36.1 57731 A6008 208 457959 339800 39.4 39.6 37.3 36.6 34.7 33.5

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 23

Census ID grid 2016 2018 2020 Road Census ID Length (m) Name X Y PCM Local PCM Local PCM Local

57733 A6008 381 457299 338970 31.1 36.0 29.2 35.3 26.9 34.5 57890 A6008 377 456900 340000 32.2 31.9 30.1 30.2 27.7 28.4 57924 A6200 1,215 455000 339220 38.9 44.0 36.3 39.9 33.3 35.9 57958 A60 712 457040 341100 40.8 46.3 37.9 40.4 34.3 34.5 58064 A610 739 456600 340110 35.5 29.9 32.9 27.3 29.7 24.7 70291 A60 415 457920 338550 34.4 34.8 32.9 31.6 30.6 28.4 70292 A6011 336 458100 338590 35.5 29.6 33.2 27.6 30.4 25.6 70293 A6011 267 458300 338810 32.4 26.2 30.3 25.1 27.8 23.9 70294 A6008 321 457400 339390 44.9 44.4 41.6 38.8 37.9 33.3 70296 A6008 188 457840 339720 35.1 37.8 33.4 35.4 31.1 33.0 73856 A52 1,029 456459 335966 39.9 32.8 37.2 30.2 33.9 27.6 73858 A60 429 458200 338210 40.3 38.5 37.8 33.5 34.6 28.5 73859 A6011 325 458410 338820 36.2 25.7 34.2 24.4 31.6 23.1 73860 A60 1,192 457111 342229 32.3 32.2 30.1 28.4 27.3 24.6 73863 A6002 727 452600 344500 33.0 20.7 30.4 19.3 27.4 18.0 73864 A610 380 453480 343580 42.7 22.6 39.7 20.8 36.2 19.0 73865 A6002 1,189 451700 343490 35.5 24.9 33.1 22.8 30.2 20.7 73870 A6002 1,446 450795 340170 32.6 23.7 30.5 21.2 27.9 18.7 73871 A6002 241 451150 341015 30.4 22.8 28.4 21.0 26.0 19.1 73872 A6002 405 451220 341185 33.3 22.9 31.2 20.9 28.5 18.9 73873 A6002 508 451420 341500 30.4 23.5 28.5 21.7 26.0 19.9

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 24

Census ID grid 2016 2018 2020 Road Census ID Length (m) Name X Y PCM Local PCM Local PCM Local

73874 A6002 1,225 451550 342400 31.7 28.2 29.7 26.1 27.1 24.0 74204 A6008 541 457210 339420 42.0 47.4 39.2 42.1 35.9 36.9 75213 A52 703 456360 336330 33.5 32.4 31.3 29.7 28.7 27.1 75215 A453 1,871 456320 336280 25.5 31.6 23.9 29.4 21.9 27.2 75216 A6514 839 454350 340050 58.6 31.6 54.1 29.7 48.7 27.9 75217 A610 349 456650 340140 33.0 28.4 30.7 26.2 27.9 24.1 75218 A610 352 456800 340070 32.9 29.4 30.5 26.8 27.7 24.2 75219 A60 190 457550 340300 36.5 32.7 34.4 30.7 31.6 28.8 75220 A60 256 457600 340170 27.9 32.2 26.0 29.9 24.0 27.7 75221 A6008 218 457600 340110 43.1 31.1 40.1 28.6 36.6 26.1 75222 A6008 245 457700 340100 31.2 27.9 29.2 25.9 26.8 23.9 75223 A6008 243 457850 339940 34.5 35.7 32.7 33.1 30.6 30.5 75225 A6008 516 457860 339560 45.0 39.7 42.8 36.6 39.8 33.6 77427 A606 2,389 459000 336280 22.4 30.8 20.9 28.1 19.1 25.5 81226 A6211 1,186 462443 341580 29.0 22.0 27.4 20.8 25.3 19.6 99028 A611 854 454600 347000 36.9 32.7 34.5 29.8 31.5 26.9 99032 A612 2,366 461500 340280 35.2 27.7 32.7 25.9 29.8 24.1 99033 A612 1,843 459995 339622 41.0 26.1 37.7 24.5 34.1 23.0 99034 A612 1,159 458250 339560 34.5 30.2 32.3 28.3 29.6 26.4 99035 A612 362 458000 339900 32.6 40.6 30.3 37.9 27.6 35.2 99037 A60 211 457470 340250 30.1 33.1 28.4 30.4 26.3 27.7

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 25

Census ID grid 2016 2018 2020 Road Census ID Length (m) Name X Y PCM Local PCM Local PCM Local

99038 A60 225 457700 340150 34.5 33.0 32.3 30.9 29.6 28.8 99039 A6011 303 458440 339000 39.6 28.5 37.1 27.6 34.1 26.8 99115 A60 511 457850 344300 36.4 31.5 34.1 29.1 31.2 26.7 99116 A6211 801 458000 344310 30.7 28.2 29.0 26.0 26.6 23.7 99117 A6211 911 459300 344500 21.8 21.8 20.4 20.2 18.8 18.7 99118 A6211 1,501 459750 344380 22.4 28.7 21.2 26.6 19.5 24.5 99119 A6211 1,563 461500 342890 28.3 25.9 26.5 24.1 24.3 22.3 99120 A6211 888 462230 342300 25.0 28.7 23.4 26.5 21.5 24.4 99702 A6008 219 457300 339470 41.0 44.3 38.3 35.7 35.2 27.2

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 26

Figure 8: Local modelled NO2 annual mean concentrations 2016 base year – PCM links

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 27

Figure 9: Local modelled NO2 annual mean concentrations 2016 base year – PCM links (Inner domain)

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 28

Figure 10: Local modelled NO2 annual mean concentrations 2018 base year – PCM links

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 29

Figure 11: Local modelled NO2 annual mean concentrations 2018 base year – PCM links (Inner domain)

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 30

Figure 12: Local modelled NO2 annual mean concentrations 2020 base year – PCM links

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 31

Figure 13: Local modelled NO2 annual mean concentrations 2020 base year – PCM links (Inner domain)

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 32

4.2 Results for AQMAs and local exceedances

Annual mean NO2 concentration measured in 2016 and predicted annual mean NO2 concentrations at each monitoring site in 2020 are presented in Table 4. We have presented two sets of results in 2020; one using the global/domain wide road NOx adjustment factor, and the other using a site-specific road NOx adjustment factor. The site-specific adjustment factor results aim to provide an indication of when compliance may be achieved at each monitoring site without any of the bias introduced by using an average road NOx adjustment factor across the entire domain. Further information on model verification and uncertainty is presented in Appendix 3 and Section 4.3 respectively.

The results indicate that in 2020, using the model wide adjustment factor, compliance with the 40 µg/m3 NO2 annual mean objective will be achieved at most of current monitoring locations, except for diffusion tube 23 (DT) which is located in Mansfield Road. However, when using the site-specific adjustment factor a number of locations look likely to exceed in 2020.

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 33

Table 4: Predicted NO2 annual mean concentrations at monitoring site locations in 2020

-3 NO2 annual mean (µg.m )

Monitoring site name Site ID Site type Modelled 2020 Modelled 2020 (site Measured (using global NO specific NO adjust 2016 x x adjust factor) factor) Woodside Road DT01 Roadside 34.6 25.1 29.8 Derby Road DT02 Roadside 27.4 26.7 23.6 Marlborough Street DT03 Roadside 25.1 22.1 21.5 Beeston Rd DT04 Roadside 31.7 25.7 25.7 Abbey Street DT05 Roadside 33.4 28.2 27.9 Western Blvd DT06 Roadside 29.9 24.5 25.8 Wilkinson St DT07 Roadside 31.0 15.8 26.6 Nottingham Rd DT08 Roadside 38.0 26.5 32.9 Nottingham Road DT09 Roadside 41.7 23.8 36.0 Valley Road DT10 Roadside 38.0 27.8 32.6 Danethorpe Vale DT11 Roadside 28.5 21.0 24.3 Bentinck Rd - Sch DT12 Roadside 37.6 29.1 30.5 Ilkeston Road DT13 Roadside 34.8 23.0 27.2 Ilkeston Road DT14 Roadside 46.9 23.4 37.3 Gregory Blvd DT15 Roadside 31.0 23.2 26.0 Castle Gardens DT16 Roadside 36.7 28.8 30.3 Castle Boulevard DT17 Roadside 39.0 28.6 32.0 LP 3P17 Park Road DT18 Roadside 28.9 25.6 25.0 Castle Blvd DT19 Roadside 32.7 23.8 24.7 Canning Terrace DT20 Roadside 49.3 25.1 38.9 Maid Marian Way DT21 Roadside 43.4 32.4 31.8

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 34

-3 NO2 annual mean (µg.m )

Monitoring site name Site ID Site type Modelled 2020 Modelled 2020 (site Measured (using global NO specific NO adjust 2016 x x adjust factor) factor) LP 80A01 Alfreton Rd DT22 Roadside 46.5 26.2 39.7 Mansfield Rd DT23 Roadside 45.7 40.1 35.9 St Andrew's Rd B DT24 Roadside 35.6 34.3 28.8 St Andrew's Rd M DT25 Roadside 36.1 34.3 29.2 St Andrew's Rd T DT26 Roadside 33.5 34.3 27.2 Cartergate DT27 Roadside 39.9 30.8 33.1 Trent Bridge DT28 Roadside 38.9 28.8 27.7 Meadow Lane DT29 Roadside 45.0 24.9 42.9 LP 53890 Manvers Street DT30 Roadside 45.1 26.2 39.4 Middleton Boulevard DT31 Roadside 28.6 24.5 25.0 Middleton Boulevard DT32 Roadside 26.6 19.5 23.3

Lace Street NOx RTA DT33 Roadside 34.1 26.7 27.6 24 Bradbourne Avenue DT34 Roadside 27.2 24.4 23.2 Lampost 70 Wollaton Road DT35 Roadside 42.4 22.0 34.1 Western Boulevard Pedestrian Crossing DT36 Roadside 48.7 28.0 42.1 Wollaton Road Subway DT37 Roadside 40.0 30.2 34.8 Lampost 180 Wollaton Road DT38 Roadside 34.1 20.4 28.7 Wollaton Rd/Crown Island Crossing DT39 Roadside 43.2 22.1 36.4 Lampost Middleton Boulevard/Crown Island DT40 Roadside 38.3 22.0 33.3 Crossing Middleton Boulevard/Crown Island DT41 Roadside 55.0 27.7 48.0 Inside 328 Derby Road DT42 - 32.7 30.1 27.6 Outside 328 Derby Road DT43 Roadside 29.9 30.1 25.6 Parking Post Cliff Road DT44 Roadside 39.4 27.7 30.6

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 35

-3 NO2 annual mean (µg.m )

Monitoring site name Site ID Site type Modelled 2020 Modelled 2020 (site Measured (using global NO specific NO adjust 2016 x x adjust factor) factor) 66 London Road DT45 Roadside 52.5 32.8 41.4 Hicking Building Queens Road DT46 Roadside 45.5 31.0 44.4 Sheriffs Way/ Railway Station DT47 Roadside 45.3 33.1 39.8

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report | 36

4.3 Model uncertainty

Some clear outliers were apparent during the model verification process, whereby we were unable to refine the model inputs sufficiently to achieve good model performance at these locations. There are a number of reasons why this could be the case e.g.:

• A site located next to a large car park, bus stop, boiler flue, or taxi rank that has not been explicitly modelled due to unknown activity data. • Sites located underneath trees or vegetation i.e. unsuitable locations for diffusion tubes to measure NO2 concentrations effectively • Sites located in the middle of traffic islands or other locations excessively exposed to queue road traffic that has not been explicitly modelled • Uncertainties in the traffic model outputs (please refer to the traffic model validation report for further information on this) • Potential inaccuracies in diffusion location co-ordinates

To evaluate model performance and uncertainty, the Root Mean Square Error (RMSE) for the observed vs predicted NO2 annual mean concentrations was calculated, as detailed in LAQM.TG(16). In this case the RMSE was calculated at 5.22 µg.m-3.

More information on model performance and uncertainty is presented in Appendix 3.

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report

Appendices

Appendix 1: Air Quality modelling QA table

Appendix 2: RapidAir street canyon equations

Appendix 3: Air quality model verification and adjustment

Appendix 4: Analytical assurance statement

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report

Appendix 1 – Air quality modelling QA table

Ref Requirement Proposal Air Quality model specification Model selection 1.1.1 Details of air quality dispersion model to be RapidAir will be used for the study- this is Ricardo’s proprietary modelling system developed for urban used air pollution assessment. The model is based on convolution of an emissions grid with dispersion kernels derived from the USEPA AERMOD19 model. The physical parameterisation (release height, initial plume depth and area source configuration) closely follows guidance provided by the USEPA in their statutory road transport dispersion modelling guidance20. AERMOD provides the algorithms which govern the dispersion of the emissions and is an accepted international model for road traffic studies (it is one of only two mandated models in the US and is widely used overseas for this application). The combination of an internationally recognised model code and careful parameterisation matching international best practice makes RapidAir fit for purpose for this study. The model produces high resolution concentration fields at the city scale (1 to 3m scale) so is ideal for spatially detailed compliance modelling. Further details given in section 3.2 of the main report 1.1.2 Canyon effects included? Yes. The model includes a canyon treatment based on the USEPA ‘Stanford’ model21. The canyon model algorithms are essentially the same as those recommended by the European Environment Agency for modelling canyons in compliance assessment22. Our model has terms to deal with canyon height, width, vehicle length, receptor height, emission strength, wind speed and direction (taken from the same met record as the main RapidAir model). Further details given in section 3.2 and appendix 3 of the main report. 1.1.3 Gradient effects included? Further to the update/clarification of the gradient method in TG16 we confirm that we will apply the gradient impact to all pre-Euro VI HGVs in the emissions processing step. In order to do this, we will carry out a GIS gradient analysis of our modelling domain to identify any road links with gradients

19 https://www3.epa.gov/ttn/scram/dispersion_prefrec.htm#aermod 20 https://www.epa.gov/state-and-local-transportation/project-level-conformity-and-hot-spot-analyses 21 USEPA., Estimating Mobile Source Pollutants in Microscale Exposure Situations, EPA-460/3-81-021 22 http://www.eea.europa.eu/publications/TEC11a/page014.html

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report

greater 2.5%. The gradient adjustment will then be applied to the proportion pre Euro VI HGV movements on identified links.

Air Quality model domain 1.2.1 Please provide a map showing model domain See Figure 5 in main report

in relation to exceedance locations identified in PCM model 1.2.2 Locally identified exceedance locations Yes, the high-resolution nature of RapidAir and its inclusion of street canyons will make the model included? outputs naturally align with hotspots/exceedance locations. See Figure 1 in main report for model domain and location of AQMAs 1.2.3 Domain includes displacement routes? Yes. See description of model on main report and relationship between proposed traffic model and modelling domain in Figure 4

Air Quality model receptor locations 1.3.1 Details of receptor grid size For the Nottingham domain (which is reasonably small) we can set RapidAir to model down to 1 m. The model can comfortably deal with about 500 million locations which provides for over 20,000 cells in the x and y axes. So we can model 20km x 20km at 1m resolution, 40km x 40km at 2m resolution, 60km x 60km at 3m resolution and so on. The canyon model is set to the same resolution as the grid model so that they align perfectly spatially. See section 3.3 of main report for further details. 1.3.2 Details of receptors at monitoring site Nottingham has a wide network of monitoring locations comprising a mix of passive and active locations sampling. RapidAir run time is not sensitive to the number of receptors so all available monitoring locations will be included.

1.3.3 Details of receptors at exceedance locations For comparison with PCM model results, annual mean concentrations at the roadside exceedance identified in PCM model (include distance locations identified in the PCM model can be extracted from the RapidAir dispersion model results from kerb and height above ground level) and presented as a separate model output file. These receptor locations will be at a distance of 4m from the kerb and 2m height.

1.3.4 Details of receptors at locally identified Nottingham has four AQMAs all of which contain numerous residential receptors. RapidAir, by virtue exceedance locations, if any of its very high-resolution outputs, will produce estimates for every single residential property in Nottingham, so any receptors at exceedance locations will naturally be included. Also see Figure 1 for model domain and AQMAs

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report

1.3.5 Methods to be used to assign subset of Annex III of the AQD specifies that macroscale siting of sampling points should be representative of receptors for AQD assessment requirements air quality for a street segment of no less than 100 m length at traffic-orientated sites. To provide results relevant to this requirement, for roadside locations where there is public access and the directive applies; road links with exceedances of the NO2 annual mean objective stretching over link lengths of 100m or greater can be presented as a separate GIS layer of model results.

Annex III of the AQD also specifies that microscale sampling should be at least 25 m from the edge of major junctions. When reporting model results relevant to compliance with the AQD, locations up to 25m from the edge of major junctions in the model domain will therefore be excluded. Base Year modelling General 2.1.1 Base year to be used The modelling base year will be 2016 in line with the latest traffic and air quality data and the base year of the proposed transport model. 2.1.2 Details of Meteorological data to be used We will use surface meteorological data from East Midlands Airport processed in house using our own meteorological data management system. Our RapidAir model also takes account of upper air data which is used to determine the strength of turbulent mixing in the lower atmosphere- we will derive this from the closest radiosonde site and process in the USEPA AERMET model. We will utilise data filling where necessary following USEPA guidance which sets out the preferred hierarchy of routines to account for gaps (persistence, interpolation, substitution). Our modelling will be supplied with full meteorological discussion and if required we can supply the computer code used to process the data and details of any data filling that was required. Traffic input data 2.2.1 Source of traffic activity data The key source of traffic data will be the Greater Nottingham Transport Model (GNTM) for Nottingham. Details of this are provided in the Transport Model Forecasting Methodology Report prepared by SYSTRA for Nottingham City Council. The transport model data will be complemented by local traffic counts, ANPR data and traffic master data in the base year. The split between articulated and rigid HGVs was taken from SL-PCM 2015 (v3.2.1). This is described in detail in section 3.4.2 of the main report 2.2.2 Vehicle types explicitly included in air quality The core vehicle categories will be cars, taxis, LGVs, rigid HGVs. Artic HGVs and buses. The emissions and concentrations modelling standard Euro and technology categories will be used in line with COPERT 5. Details in section 3.4.2 of the main report.

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report

2.2.3 Details of representation of road locations See Figure 4 in main report for map of transport model road network. All modelling links will be (achieved through use of a georeferenced snapped to the OS ITN road network for the best spatial representation. transport model or another approach?) 2.2.4 Source of vehicle fleet composition Detailed fleet composition data will be derived from an ANPR survey. This will be complemented by information (local/EFT) local cunt data and NAEI fleet data as necessary.

2.2.5 Source of vehicle speed information Traffic speeds will be taken from the traffic master data set for the base year and will be adjusted for future years in relation to changes in link travel times from the transport model. This is described in section 3.5.1 NOx/NO2 emissions assumptions 2.3.1 Source of emission factors for NOx COPERT 5 data either in the form of an update EFT or with JAQU’s agreement our in-house emission calculation tool pyCOPERT which is fully compatible with COPERT 5.

2.3.2 Source of primary NO2 emission fractions (f- Defra f-NO2 fractions which we understand will be released in time to support this work, See also NO2) section 3.4.2.4 in the main report. 2.3.3 Details of method used to calculate See section 3.4.2.4 projections for f-NO2 2.3.4 Details of methods to be used to calculate The Defra NOx:NO2 model will be used. See section 3.4.2.4 for details. NO2 concentrations from NOx concentrations Non-road transport modelling 2.4.1 Details of modelling for non-road transport Three key local background sources will be modelled explicitly: sources • Queens Medical Centre CHP Boiler. • Wastenotts Reclamation Ltd. • The Boots plc (two stacks). • FCC Environment Eastcroft Energy from Waste Facility (EfW). • Enviroenergy Nottingham District Heating and Power. • Nottingham University City Hospital (future replacement boiler plant – 5 stacks).

In addition to this, the contribution of rail will be modelled separately.

Details of these are provided in section 3.4.3

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report

Measurement data for model calibration 2.5.1 Details of the date, locations and type of Air quality monitoring data collected by Nottingham City Council for 2016. Details of this are still monitoring data (automatic and/or diffusion being collated. See Figure 7 for location and type of monitoring points. tubes) used for the model calibration Projections modelling

Baseline projections modelling 3.1.1 Years to be modelled (to include 2020; Modelling years are: please include explanation for any additional • 2020 – CAZ implementation year years) See section 2.3 for full details. 3.1.2 Details of method for projected vehicle fleet See section 3.4.2 for base year fleet data composition See section 3.5 for forecast fleet data

3.1.3 Details of method for projected vehicle Future vehicle traffic will be derived from the transport model described in the Transport Model activity Forecasting Methodology Report prepared by SYSTRA for Nottingham City Council. 3.1.4 Impact of RDE included? This is included only in relation to the COPERT emissions data. But could consider use of remote sensing data sets as an additional piece of work. With measures projections modelling 3.2.1 Years to be modelled 2020 as described in section 2.3 in main report

3.2.2 Details of method for projected vehicle fleet The fleet composition will be assessed separately for complaint and non-compliant vehicles. See composition section 3.5.3 in main report.

3.2.3 Details of method for projected vehicle Project vehicle traffic will be done by the traffic model. Within the traffic model the vehicle matrices activity will be split between complaint and non-complaint vehicles so that the behaviours of these groups will be modelled separately. The details of this is provide in section 4.3.2 of the main report.

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report

Appendix 2 – RapidAir street canyon equations

The formulations for both models are described below.

USEPA STREET model The STREET model assumes that the concentration of pollutants within a street canyon location consist of the urban background concentrations and a concentration from vehicle emissions within the street being modelled. The recommendation by the USEPA is to use the concentration from the model at 3m height as background concentrations at the actual receptor height being modelled. Since the canyons are expected to be well mixed over longer averaging periods it is sensible that we use the RapidAir kernel model to provide boundary conditions to the STREET model. Concentrations on the leeward (CL) and windward (CW) side of the canyon are calculated in this method, using the equations below:

퐾 ∗ 푄 퐶퐿 = 2 2 1⁄ (푈 + 0.5) ∗ [(푥 + 푧 ) 2 + 퐿0] 퐾 ∗ 푄 ∗ (퐻 − 푧) 퐶푊 = 푊 ∗ (푈 + 0.5) ∗ 퐻

Where K is an empirical constant (usually set between 10 and 14); Q is the emission rate (g/m/s); U is the wind speed (m/s); L0 is the length of individual vehicles (set to 3 m in this case); W is the width of the canyon (m); H is the average building height of the canyon (m); x is the distance from emission source to receptor (m); and z is the receptor height.

AEOLIUS/OSPM

There are three principal contributions in the AEOLIUS model, a direct contribution from the source to the receptor, a recirculating component within a vertex caused by winds flowing across the top of the canyon, and the urban background. The RapidAir model only take the recirculating component from the canyon and sums this with the kernel derived concentrations.

The RapidAir implementation of AEOLIUS is written in python 2.7 and uses the same equations described in the referenced Met Office papers.

During the coding of the canyon model we tested the outputs of our code with calibration data provided with the FORTRAN version of AEOLIUS. Our implementation agrees almost (R2 = 0.97) perfectly with the version supplied by the Met Office (which is in any case now out of circulation).

The AEOLIUS model is more complex than the STREET model. Concentrations are calculated for the windward and leeward sides of the road using the equations detailed below (based on equations from the Met Office). The leeward and windward concentrations described below are only calculated for streets that were perpendicular to the direction of the wind. Concentrations calculated in ppb, and for 3 NOx/NO2 models are converted to µg/m by multiplication by 1.91. The system of equations in RapidAir’s implementation of the AEOLIUS model are shown below.

Inputs:

Emission rates (Q, µg/m/s); traffic speeds (vt, mph), traffic density (f, vehicles per hour), % of cars and heavy good vehicles (fc and fh respectively), wind speed at roof level (ur, m/s), street canyon width (w, m), street canyon height (h, m), and angle of street (θ).

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report

Leeward concentrations:

The leeward concentrations = sum(Cdlee + Crec) where Cdlee is the direct contribution from vehicles and Crec is the pollution associated with recirculation.

Direct contribution (Cdlee):

푅푒푐𝑖푟푐푢푙푎푡𝑖표푛 푧표푛푒 (푙푟) = min (푤, 푙푣 ∗ sin(휃)) (meters) Where:

푣표푟푡푒푥 푙푒푛𝑔푡ℎ (푙푣) = 2 ∗ 푟 ∗ ℎ (meters)

And r = wind speed dependence factor = 1 if ur > 2 m/s and = ur/2 otherwise.

If the recirculation zone is greater than the width of the canyon:

2 푄 휎푤 ∗ 푤 퐶푑푙푒푒 = √ ∗ ∗ ln [( ) + 1] 휋 (푤 ∗ 휎푤) ℎ표 ∗ 푢푠

Where:

2 2 σw = mechanical turbulence from wind and traffic (m/s) = √(휆 ∗ 푢푠) + 휎푤표 λ = constant for removal at the top of the canyon = 0.1

푣푡∗푓푐∗푠푐+푣푡∗푓ℎ∗푠ℎ σwo = traffic-created turbulence (m/s) = 푏 ∗ √ 푤

2 where sc = mean surface area of cars (4 m ), sh = mean surface area of heavy vehicles (16 m2) and b = aerodynamic constant (0.18)

ℎ ln( 표) 푧표 us = wind speed at street level (m/s) = 푢푟 ( ℎ ) (1 − 푑 ∗ sin(휃)) ln( ) 푧표

ho = effective height of emissions (2 m)

zo = effective roughness length (0.5 m)

d = model dependence (0.45)

If the recirculation zone is less than the width of the canyon:

푑6 ℎ + 휎 ∗ −휔 푑 2 푄 휎 ∗ 푑 표 푤 휎 ( 푡 7) 푤 1 푢푠 푤 푢 ℎ 퐶푑푙푒푒 = √ 푙푛 [( ) + 1] + 푅 ∗ ln ( ) + [1 − 푒 푠 ] 휋 (푤 ∗ 휎푤) ℎ표 ∗ 푢푠 휎푤 ∗ 푙푟 휔푡 + ℎ표 [ 푢푠 ] Where:

d1 (m) = min(w, lr)

R = max(0, Cang)

Cang = cos(2*r* θ)

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report

d6 (m)= min(max(lmax, lr), x1)

lmax = w/sin(θ)

푢푠(ℎ− ℎ표) x1 = vertical distance (m) at which pollutants can escape canyon = σw

2 2 ωt = removal at top of the canyon (m/s) = √(휆 ∗ 푢푟) + 0.4(휎푤표)

d7 (m) = max(lmax, x1)-x1

Recirculation contribution (Crec): 푄 [( ) 푑 ] 푤 1 퐶푙푒푒 = 휔푡 ∗ 푑2 + 휔푠 ∗ 푑3 Where

d2 (m) = min(w, 0.5*lr)

2푤 d3 (m) = 푙푠 (max (0, − 1) 푙푟

2 2 ls (m) = √(0.5 ∗ 푙푟) + ℎ

2 2 ωs = removal speed at the side of the canyon (m/s) = √푢푠 + 휎푤표

Windward concentrations (Cdwind):

Final windward concentrations = Cdwind + Crec. Cdwind = 0 if lr ≥ w, else:

−휔 푑 2 푄 휎 + 푑 휎 ( 푡 5) 푤 4 푤 푢 ℎ 퐶푑푤𝑖푛푑 = √ [푙푛 ( + 1) + [1 − 푒 푠 ]] 휋 푤 ∗ 휎푤 푢푠 + ℎ표 휔푡

d4 (m) = min[(w – lr), x1]

d5 (m) = [max[(w – lr),x1]]-x1

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report

Appendix 3 – Air quality model verification and adjustment

Verification of the model involves comparison of the modelled results with any local monitoring data at relevant locations; this helps to identify how the model is performing and if any adjustments should be applied. The verification process involves checking and refining the model input data to try and reduce uncertainties and produce model outputs that are in better agreement with the monitoring results. This can be followed by adjustment of the modelled results if required. The LAQM.TG(16) guidance recommends making the adjustment to the road contribution of the pollutant only and not the background concentration these are combined with.

The approach outlined in LAQM.TG(16) section 7.508 – 7.534 (also in Box 7.14 and 7.15) has been used in this case.

All roadside automatic and diffusion tube NO2 measurement sites in Nottingham have been used for model verification.

It is appropriate to verify the performance of the Rapid Air model in terms of primary pollutant emissions of nitrogen oxides (NOx = NO + NO2). To verify the model, the predicted annual mean Road NOx concentrations were compared with concentrations measured at the various monitoring sites during 2016.

The model output of Road NOx (the total NOx originating from road traffic) was compared with measured Road NOx, where the measured Road NOx contribution is calculated as the difference between the total NOx and the background NOx value. Total measured NOx for each diffusion tube was calculated from the measured NO2 concentration using the latest version of the Defra NOx/NO2 calculator issued for use in the CAZ cities (v6.0).

The initial comparison of the modelled vs measured Road NOx identified that the model was under- predicting the Road NOx contribution at most locations. Refinements were subsequently made to the model inputs to improve model performance where possible.

The gradient of the best fit line for the modelled Road NOx contribution vs. measured Road NOx contribution was then determined using linear regression and used as a global/domain wide Road NOx adjustment factor. This factor was then applied to the modelled Road NOx concentration at each discretely modelled receptor point to provide adjusted modelled Road NOx concentrations. A linear regression plot comparing modelled and monitored Road NOx concentrations before and after adjustment is presented in Figure A3.1.

The total annual mean NO2 concentrations were then determined using the NOx/NO2 calculator to combine background and adjusted road contribution concentrations.

Some clear outliers were apparent during the model verification process, whereby we unable to refine the model inputs sufficiently to achieve acceptable model performance at these locations. There are a number of reasons why this could be the case e.g.

• A site located next to a large car park, bus stop, boiler flue, or taxi rank that has not been explicitly modelled due to unknown activity data. • Sites located underneath trees or vegetation i.e. unsuitable locations for diffusion tubes to measure NO2 concentrations effectively

• No traffic model road link included where the NO2 sampler is located, or not all road links included e.g. at a junction

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report

• Uncertainties in the traffic model outputs (please refer to the traffic model validation report for further information on this) • Sites located in the middle of traffic islands or other locations excessively exposed to road traffic • Uncertainties in the traffic model outputs (please refer to the traffic model validation report for

further information on this)

13 out of 49 diffusion tube sites were considered as outliers and were therefore excluded from the verification process. A primary NOx adjustment factor (PAdj) of 1.0974 based on model verification using the remaining 2016 NO2 measurements was derived and applied to all modelled Road NOx data prior to calculating an NO2 annual mean.

A plot comparing modelled and monitored NO2 concentrations before and after adjustment during 2016 is presented in Figure A3.2.

Figure A3.1 Comparison of modelled Road NOx Vs Measured Road NOx before and after adjustment 2016

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report

Figure A3.2: Modelled vs. measured NO2 annual mean 2016

Model performance

To evaluate the model performance and uncertainty, the Root Mean Square Error (RMSE) for the observed vs predicted NO2 annual mean concentrations was calculated, as detailed in Technical Guidance LAQM.TG(16). The calculated RMSE is presented in Table A3.1.

In this case the RMSE when outliers were excluded was calculated at 5.22 µg.m-3.

Table A3.1: Root mean square error

NO2 monitoring site Measured NO2 annual Modelled NO2 annual Difference mean concentration mean concentration measured – 2016 (µg.m-3) 2016 (µg.m-3) modelled (µg.m-3) DT01 34.6 29.1 5.5 DT02 27.4 31.0 -3.6 DT03 25.0 26.0 -0.9 DT04 31.6 31.7 0.0 DT05 33.3 33.8 -0.4 DT06 29.9 28.5 1.5 DT08 38.0 30.7 7.3 DT09 41.7 27.7 14.0

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report

NO2 monitoring site Measured NO2 annual Modelled NO2 annual Difference mean concentration mean concentration measured – 2016 (µg.m-3) 2016 (µg.m-3) modelled (µg.m-3) DT10 37.9 32.4 5.6 DT11 28.5 24.6 3.9 DT12 37.5 35.8 1.8 DT13 34.8 28.7 6.2 DT15 31.0 27.6 3.5 DT16 36.6 34.5 2.1 DT17 39.0 34.3 4.8 DT18 28.8 29.6 -0.8 DT19 32.6 30.4 2.2 DT21 43.3 44.8 -1.4 DT23 45.6 51.4 -5.7 DT24 35.6 42.5 -6.8 DT25 36.1 42.5 -6.4 DT26 33.5 42.5 -8.9 DT27 39.8 36.8 3.1 DT28 38.8 40.6 -1.8 DT31 28.6 28.1 0.6 DT32 26.6 22.4 4.2 DT33 34.1 33.0 1.1 DT34 27.2 28.6 -1.4 DT37 39.9 34.7 5.3 DT42 32.6 36.0 -3.4 DT43 29.8 36.0 -6.2 DT44 39.3 34.3 5.0 DT45 52.5 40.2 12.3 DT47 45.3 37.6 7.8 DT48 27.8 22.9 5.0 DT49 50.5 44.2 6.3 RMSE (excluding clear outliers) 5.22

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report

Appendix 4 – Analytical assurance statement

Analytical Assurance Statement for transport and air quality modelling.

1. Limitations of the Analysis • Has the Analysis been constrained by time or cost, meaning further proportionate analysis has not been undertaken? • Could the further analysis that could be done lead to different conclusions? • Does the analysis rely on appropriate sources of evidence? • How reliable are the underpinning assumptions?

2. Risk of Error / Robustness of the Analysis • Has there been sufficient time and space for proportionate levels of quality assurance to be undertaken? • Have sufficient checks been made on the analysis to ensure absence of errors in calculations? • Have sufficiently skilled staff been responsible for producing the analysis?

3. Uncertainty • What is the level of residual uncertainty (the level of uncertainty remaining at the end of the analysis)?

4. Use of analysis

• Does the evidence provided support the business case? • Is there evidence the agreed target will be achieved?

1. Limitations of the Analysis

• Has the Analysis been constrained by time or cost, meaning further proportionate analysis has not been undertaken?

The analysis has been constrained by time and cost. However, this has not constrained proportionate analysis from being undertaken. In fact, the modelling of air pollutants has been carried out in a much greater level of detail and complexity than is typical of studies of comparable size. This particularly relates to the background sources which have been considered in detail.

• Could the further analysis that could be done lead to different conclusions?

No. Any further analysis would not be expected to lead to different conclusions, but may provide a greater understanding and/or greater refinement of the results presented.

• Does the analysis rely on appropriate sources of evidence?

The best available evidence has been used as specified by JAQU. Both the traffic model and emissions and air quality model meet the quality criteria provided by JAQU. Local fleet

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report

composition data was derived from an analysis of ANPR data from sites across the city. This data indicated that the Euro distribution was consistent across the modelling domain.

• How reliable are the underpinning assumptions?

For the purposes of this CAZ study SYSTRA have prepared a transport model review note for the Nottingham Area Transport Model entitled ‘Stage 2a Transport Model Review Document v1.5’. This note has been assessed by JAQU/DfT and NATM has been approved as being ‘Fit for Purpose’ to assess the highway impacts of the Nottingham CAZ and other Air quality proposals.

In terms of the detailed fleet composition data this was derived from local ANPR data, with forward projection done on the basis of a methodology agreed with JAQU using national trends. In addition, JAQU provided data on the proportion of vehicles that will upgrade to meet the standard. There is some uncertainly around whether this generic figure will match the response to local schemes. However, with not local information on behavioural response these provide the best assumptions available.

The emissions and air quality modelling follow current best practice. The emissions data is the latest COPERT V data which represents latest information on the real-world emission performance of vehicles. The average speed approach adopted by the COPERT emissions data set is a simplification but is the most practical approach over a modelling domain of this size.

2. Risk of Error / Robustness of the Analysis

• Has there been sufficient time and space for proportionate levels of quality assurance to be undertaken?

Yes. All modelling work is reviewed internally before issuing the results. This internal review includes sense checking, review of the assumptions, and checking of both the modelling inputs and outputs.

• Have sufficient checks been made on the analysis to ensure absence of errors in calculations?

The checks undertaken have been proportionate to the scale and complexity of the modelling analysis.

• Have sufficiently skilled staff been responsible for producing the analysis?

The air quality modelling team at Ricardo have significant experience of developing, assessing and recommending measures to reduce emissions and improve air quality at the city scale, including extensive expertise in air pollution modelling from the development of inventories and baselines to modelling the future impacts of abatement scenarios. The team is led by a Project Director who holds over 20 years of experience of working on transport and emissions reduction projects. His key areas of expertise include vehicle emissions modelling, low emission vehicle technologies, sustainable transport measures and local air quality management and policy and he has worked on a number of LES, LEZ and CAZ projects in the UK including in , Derby, Nottingham, Oxford, London, and South Oxfordshire. The modelling work is led by an experienced atmospheric scientist with a strong focus on modelling transport and industrial emissions and characterising their effects on ambient air quality who is an advanced user of ADMS, ADMS-Roads, ADMS-Urban, AERMOD, CALPUFF, Envi-Met CFD, ArcGIS, QGIS and other air dispersion modelling tools as well as meteorological modelling software such as WRF, and has also developed Ricardo’s in-house dispersion modelling suite

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6 Nottingham Clean Air Zone Modelling and Target Determination Report

(RapidAir). The modelling lead is supported by a team of experienced consultants specialising in air quality impact assessment and atmospheric dispersion modelling.

3. Uncertainty

• What is the level of residual uncertainty (the level of uncertainty remaining at the end of the analysis)?

To evaluate model performance and uncertainty of the air quality model, the Root Mean Square Error (RMSE) for the observed vs predicted NO2 annual mean concentrations was calculated, as detailed in Technical Guidance LAQM.TG(16). In this case the RMSE was calculated at 5.7 µg.m-3 Appendix 3.

4. Use of analysis

• Does the evidence provided support the business case?

To be completed once CAZ options have been modelled.

• Is there evidence the agreed target will be achieved?

To be completed once CAZ options have been modelled.

Ricardo in Confidence Ref: Ricardo/ED10107/Issue Number 6

The Gemini Building Fermi Avenue Harwell Didcot Oxfordshire OX11 0QR United Kingdom t: +44 (0)1235 753000 e: [email protected] ee.ricardo.com