View all publications

8 Tier 3 Full Assessment

This section of the guide provides an outline of the issues to be considered, and recommends good practice protocols for undertaking a Tier 3 assessment. An overview of the Tier 3 assessment process is provided in Figure 8.1.

The aim of a Tier 3 assessment is to provide reasonably accurate estimates and a detailed assessment of the likely air quality impacts associated with a project. This is usually done through the use of detailed traffic information, emission factors, ambient air quality data and dispersion modelling. For any aspect of the assessment where detailed information is not available, or is not necessary, it is appropriate to adopt the conservative assumptions discussed for Tier 2 assessments.

Figure 8.1: Tier 3 assessment process

A Tier 3 assessment should be carried out by experienced practitioners. This may necessitate input from a multidisciplinary team, including experts in traffic, vehicle emissions and air quality. A Tier 3 assessment should be undertaken for the current year and for future years, both with and without the proposal so that the effect of the proposal can be assessed. The future assessment years may depend on the availability of traffic demand models but should include a long-term forecast (at least 10 to 20 years).

8.1 Characterising the discharges to air

Characterising the discharges to air from a transport project typically includes:

  • a detailed description of the proposal

  • estimating the effect of the proposal on traffic

  • estimating fleet composition.

The following sections provide guidance on estimating and reporting traffic effects and emission factors for the purposes of an air quality assessment.

8.1.1 Description of the proposal

The description should be sufficient to enable a full understanding of the proposal from an air discharge viewpoint, and should also provide sufficient information to ascertain whether any consents are required. The depth of information required will vary depending on the type of activity, but the following should generally be provided:

  • a detailed site plan, including maps with all relevant features and the location of sensitive receptors

  • a written summary of the project proposal focusing on aspects that give rise to emissions to air

  • details of the alternatives considered and the reasons for their rejection

  • details of any relevant historical information, including past changes to the activity

  • details of any proposed changes to an existing activity

  • any relevant timeframes or constraints for undertaking the activity

  • any mitigation and/or preventive measures undertaken to manage discharges to air

  • details of traffic data and emission factors, as discussed in the following sections.

8.1.2 Traffic data requirements

A key component of transport project assessment is the calculation of changes to traffic as a result of the project. Traffic and transport models provide traffic flows and speeds, which are used to estimate shifts and changes in pollutant loadings. Traffic and transport models are described in Appendix 2.

Traffic measurements or models used to determine traffic volumes and composition for air quality impact assessments should generally comply with the requirements of the Transfund Project Evaluation Manual (Transfund, 2004, available at manuals.html).

Traffic modelling results should be reported concisely in an air quality assessment report. The results need to be communicated in a way that can be understood by people who may not be experienced in interpreting traffic modelling, and the results need to be relevant to the air quality impacts being assessed.

Documentation for the air quality assessment should include:

  • a summary of the basis of the model, including its validation, performance and forecasting

  • a summary of how the area of influence was determined, to demonstrate that all roads affected by the proposal are being assessed

  • sensitivity analysis for traffic speeds and traffic volumes, particularly model results, growth rates, and the assessment of diverted and induced traffic.

Traffic analysts need to be particularly aware of the need for conservatism in air quality assessments. In particular, traffic projections developed in accordance with the Project Evaluation Manual generally assume that projects do not induce new trips − which may not always be the case. This is a common criticism of traffic projections produced for environmental assessments, and should be addressed in a Tier 3 assessment. If appropriate, an allowance for induced traffic should be included in projections.

The assessment should be undertaken for the opening year and for future years, both with and without the proposal. The future assessment years may depend on the availability of traffic demand models but should include a long-term forecast (for at least 10 to 20 years). Assumptions made in deriving projections should be clearly stated, including the basis for the assumed traffic growth rate and any induced traffic.

The error associated with current and predicted traffic volume and composition estimates should also be estimated and stated in the report.

The traffic data requirements will depend on the type of emission factor model being used. (The types of emission factor models are discussed further below.) However, most assessments will be based on NZTER. The following traffic information is generally required for a Tier 3 assessment using NZTER:

  • one-hour, eight-hour and 24-hour traffic flows on the links being evaluated (including confidence limits, where possible)

  • LOS / average speed on each link

  • vehicle class composition (if it is likely to be different to the default).

If detailed traffic data are not readily available, a Tier 3 screening assessment could be based on adjusted annual average daily traffic (AADT) and worst-case congestion levels, as discussed for the Tier 2 assessment.

It is critical that the assessor be absolutely clear about the basis and meaning of any traffic data within the confines of the assessment. Many traffic studies provide estimates of daily and peak-hour traffic for an average day, whereas air quality studies must estimate worst-case air quality. The main aim of an air quality assessment is to determine the effects of a roading proposal on air quality. The national ambient air quality standards only allow between one and 24 hours of exceedance per year, depending on the pollutant. To assess whether these criteria are likely to be exceeded, traffic data for a 'high traffic emissions day' are needed. High traffic emissions will occur on days with high traffic flow and associated low speed / LOS. Where there are adequate data, confidence limits associated with the traffic estimates should be reported.

Analysts must use their judgement to define a high traffic emissions day. For instance, in some cases the most appropriate 'day' will be an average one; in others it might be a particular weekday, and in yet others it may be the day of a special event (say peak traffic near a local stadium when a major event is occurring).

Average speed / level of service

Most emission factors (including NZTER) provide estimated average emissions in grams per kilometre (g/km) for a typical drive cycle. This means the emission factor represents average emissions from an average vehicle over a typical length of road. For example, at low speeds (or congested LOS), the drive cycle includes some allowance for stopping and starting, queuing at intersections, etc. Vehicle emissions are highly speed dependent, and most emission factor models require an estimate of average speed on each link.

NZTER emission factors are provided for three different LOS. LOS is a representation of traffic congestion, and includes:

  • free - little or no vehicle impedance, with warm running (LOS A or B)

  • interrupted - moderate vehicle interaction, with warm running (LOS C or D)

  • congested - severe vehicle interactions such as traffic jams, with warm running (LOS E or F).

These LOS are described in terms of road type and average speed in section 8.1.3. NZTER emission factors provide estimated average emissions for a typical New Zealand drive cycle, including intersection delays. This means average speeds for each link should be determined from the total link travel time and intersection delay at the downstream end of the link. So if the traffic model explicitly represents intersection delay at the end of the link rather than as part of the link travel time, the downstream intersection delay needs to be added to the link travel time before calculating the level of service on the link.

For assessing air quality at a specific location (eg, a receptor close to an intersection), the emission factor should be selected based on the expected speed or LOS at that location, as opposed to the average for the link.

For some projects, much more detailed traffic information is available. Micro-simulation traffic models represent the travel behaviour of individual vehicles, and with an appropriate emissions model can provide much more accurate estimates of emissions from a transport corridor. This is discussed further in section 8.1.3 (micro-simulation of emissions).

Vehicle fleet composition

Vehicle fleet composition is an important parameter in air quality assessments because of the high degree of variation in emissions from different sizes and types of vehicles, especially for particulates. The health effects of air pollution are dominated by particulates, and particulate emissions from transport are dominated by heavy commercial vehicles (HCVs).

Traffic composition can be estimated from traffic surveys, traffic models, or from vehicle size classification information associated with automated traffic counts. Wherever possible, site-specific data, or information from nearby locations, should be used to estimate the proportion of HCVs. Vehicle size classification information may be available from territorial local authorities (for local roads), Transit NZ (for state highways), or the Ministry of Transport (for national/ regional estimates).

There are a number of vehicle classification systems in use in New Zealand, and unfortunately there is limited consistency across these. The vehicle classification systems commonly used in New Zealand are summarised in Appendix 2.

Site-specific data should be estimated for any projects where the proportion of HCVs is expected to be significantly higher than the default values provided in Table 8.1 (eg, in busways, port access roads and quarry access roads).

Table 8.1: Percentage of VKT, by vehicle class5 (VFM) for New Zealand


LCVs HCVs Buses

3.5−7.5 t

7.5−15 t

15−30 t

> 30 t

3.5−12 t

> 12 t

Fleet year













































Where relevant size classification or other composition data are available, these should be used to adjust the national average default values provided in Table 8.1. The most conservative approach assumes the highest proportion of HCVs based on either local information or the default data. Where the estimated proportion of HCVs (or buses) is higher than the default, the most conservative way to adjust the data would be to simply reduce the proportion of light-duty vehicles by the corresponding amount.

For urban areas, the regional council will generally have information on the amount of bus travel associated with public transport. However, these figures generally will not include private coaches, so some care is needed in estimating the proportion of buses from these figures.

The Ministry of Transport Vehicle Fleet Emissions Model (VFEM) is a computer model that estimates vehicle kilometres travelled, by vehicle class. Table 8.1 shows the estimated proportion of VKT by vehicle class from the VFEM for 2004, 2011 and 2021. Where site-specific or local data are not available, VFEM data should be used to provide default fleet composition data. The Transfund Project Evaluation Manual data may then be used to adjust the VFM data for different road types, and time periods, if appropriate. The Project Evaluation Manual provides traffic composition data for different road types but does not include projections and does not differentiate between petrol and diesel vehicles.

The Research Unit of NZ Transport Agency (formerly Land Transport Safety Authority) plans to report national estimates of vehicle kilometres travelled by vehicle type on a regular basis. Any updated national estimates should be used to provide default data as they become available.

The Motor Vehicle Register is another source of data that can provide regional vehicle classification information. However, care must be taken when using vehicle classification data to ensure the proportions are adjusted to reflect utilisation. Commercial vehicles tend to travel greater distances than cars, so it is generally not valid to assume that the proportion of vehicle kilometres travelled by a vehicle class is the same as the proportion of vehicles in each class.

8.1.3 Emission factors

Emission factors provide an estimate of emissions from individual vehicles. When combined with traffic information, they are used to estimate discharges of contaminants from roads, as follows:

Emission rate = ∑v ∑r ∑d ∑p (VKTv,r,d,p x emission factorv,r,d,p)

where: v = vehicle type, including fuel type

r = road type

d = driving conditions

p = emission process (exhaust, evaporation or tyre wear).

Emission rates should be calculated using emission factors determined for the New Zealand fleet. The Vehicle Fleet Model and associated NZTER are currently the most readily available emissions information sources for the New Zealand fleet.

Using NZTER exhaust emission factors

The NZTER database includes emission factors dependent on road type, LOS and vehicle class. The emission factors provide average emissions in g/km for a typical drive cycle for the type of road and LOS selected. Descriptions of the road types are as follows.

  • Central urban: routes with a combination of one-block links, essentially all with signalised intersections, an approximate minimum of six per kilometre; maximum speed limit 50 km/h.

  • Motorway: a multi-lane divided freeway without intersections but with the frequency of on/off ramps reflecting routing through an urban zone; maximum speed limit 100 km/h.

  • Suburban: a mixed route representing daily commuter / local trip traffic (outside of the CBD) with a high frequency of intersections, mainly uncontrolled and relying on queuing and gap acceptance to change or join traffic flows; maximum speed limit 50 km/h.

  • Rural highway: routes in rural areas6 with uncontrolled intersections, with overtaking dependent on oncoming traffic; maximum speed limit 100 km/h.

LOS are described as follows.

  • Free flow: little or no vehicle impedance with warm running; LOS level A/B, defined as having speed ≥ 80 km/h on motorways and ≥ 45 km/h on central urban and suburban roads.

  • Interrupted: moderate vehicle interaction with warm running; LOS level C/D, defined as 67 km/h ≤ speed < 80 km/h on motorways and 25 km/h ≤ speed < 45 km/h on central urban and suburban roads.

  • Congested: severe vehicle interactions such as traffic jams, with warm running; LOS level E/F, defined as speed < 67 km/h on motorways and < 25 km/h on central urban and suburban roads.

  • Cold start: this is not a congestion level. Cold start describes the emissions from vehicles that have not warmed up. This can be assumed to be the first three minutes of any trip. In urban areas it is conservative to assume that 20% of vehicles are operating under cold-start conditions on average. This proportion could be significantly higher or lower at any specific location; for example, on motorways there are very few vehicles operating under cold start.

Fleet-weighted NZTER emission factors for the default vehicle class composition defined above are provided for the Tier 2 assessment procedure. These emission factors are appropriate for a Tier 3 assessment unless better information is available, or the vehicle class composition is expected to be different to the default.

Issues relating to the use of NZTER in effects assessment

Users have raised a number of concerns regarding the use of NZTER for assessing air quality impacts. These are documented in section 3.2.1 of the Good Practice Guide for Atmospheric Dispersion Modelling (Ministry for the Environment, 2004). Updated emission factors are being developed, but until these become available NZTER is the recommended tool. Alternative emission factors are discussed below.

Emissions calculated using NZTER have been compared to emissions calculated using Australian emission factors (Auckland Regional Council, 2005a). This analysis shows that Australian emission factors result in significantly lower emissions estimates, particularly for future years. However, this analysis does give some confidence that NZTER emission factors are conservative.

Non-exhaust emission factors

The NZTER database does not include non-exhaust emission factors. However, tyre wear, brake wear and evaporative VOC emission factors have been developed for the Auckland Regional Council emissions inventory (Auckland Regional Council, 2005b). A summary of these emissions factors follows.

VOC evaporative emission factors

Volatile organic compound (VOC) emission factors may be required for assessments of airshed air quality impacts (see section 8.4.3), or for an assessment of local impacts that includes consideration of one or more VOCs (eg, benzene).

Evaporative emissions are highly variable. The emission factors provided take into account some of the main variables involved, and provide an indication of relative emissions from petrol vehicles. Diesel vehicles are often not considered because their evaporative emissions are small compared with petrol vehicles (in the order of 0.2 g/km). The emission factors are summarised in Table 8.2.

The evaporative VOC emission factors for petrol vehicles are expected to decrease over time due to reduced petrol vapour pressure and improved vehicle technology. Key assumptions are:

  • the evaporative emissions from vehicles with no control technology are in the range of 3 to 6 g/km

  • the evaporative emissions from vehicles with control technology are less than 0.08 g/km

  • the maximum vapour pressure for summer grade petrol was 85 kPa, reducing to 65 kPa in 2006

  • the proportion of vehicles in the fleet with evaporative emissions control is predicted to be:

  • none in 1993

  • 5% by 1998

  • 55% by 2011

  • 100% by 2021.

Table 8.2: Evaporative VOC emissions

Vehicle type LOS / congestion level Evaporative VOC (g/km)





Petrol car
















Cold start




Petrol LCV*
















Cold start





* HCVs and buses were assumed to have the same evaporative VOC emissions as LCVs. HCV = heavy commercial vehicle; LCV = light commercial vehicle.

Source: Auckland Regional Council, 2005a

Tyre-wear and brake-wear emission factors

Tyre-wear and brake-wear emission factors have been derived from Ministry of Transport studies on potential water contamination from road transport. These emission factors for PM10 assume that 15% of tyre-wear particulate emissions are PM10 and 40% of brake-wear particulate emissions are PM10. It can be assumed that 80% of brake and tyre wear PM10 is PM2.5. These emission factors are similar to those used internationally. However, there is a great deal of uncertainty about tyre and brake wear emissions, and the factors are based on very few measurements. Tables 8.3 and 8.4 show the emission factors.

Table 8.3: Tyre wear PM10 emission factors

Vehicle type Tyre wear PM 10 (g/km)












HCV (small)




HCV (medium)




HCV (large)




Bus (medium)




Bus (large)








Notes (Tables 8.3 and 8.4): LCV = light commercial vehicle; HCV = heavy commercial vehicle.

Source: Auckland Regional Council, 2005a

Table 8.4: Brake wear PM10 emission factors

Vehicle type Brake wear PM 10 (g/km)












HCV (small)




HCV (medium)




HCV (large)




Bus (medium)




Bus (large)








Source: Auckland Regional Council, 2005a

Previously, brake- and tyre-wear emissions have not generally been included in assessing transport effects, mainly due to the lack of data on these factors. For future Tier 3 assessments it is recommended they be considered, because for busy roads these can be a significant source of PM10. However, due to the high level of uncertainty associated with these emission factors it is recommended that sensitivity analysis be undertaken.

A recent review by the UK Department for Environment, Food and Rural Affairs (Air Quality Expert Group, 2004) suggested that actual tyre-wear emission factors could be anywhere between 10% and 1,000% of published results. In New Zealand, analysis has been undertaken to attempt to quantify non-tailpipe PM10 (Kuschel and Bluett, 2002). This analysis found that dispersion modelling results, based on NZTER exhaust PM10 emission factors, correlated reasonably well with measured PM10. Although further work is required, this analysis suggests that non-exhaust PM10 emission factors are not significant compared to exhaust emissions. For the purposes of sensitivity analysis it should be assumed that non-tailpipe emission factors are between 10% and 200% of the values quoted above.

Benzene and 1,3-butadiene emission factors

Benzene emission factors for petrol and diesel vehicles have been estimated based on fuel properties (see Appendix 1). It is assumed evaporative emissions from diesel vehicles are negligible. Recommended benzene emission factors are provided in Table 8.5. 1,3-butadiene emission factors recommended by the Australian National Pollutant Inventory (Environment Australia, 2000a) are also provided in Table 8.5.

Table 8.5: Benzene and 1,3-butadiene emission factors

  Petrol vehicles Diesel vehicles

% of exhaust VOC

% of evaporative VOC

% of exhaust VOC



















Rail emission factors

The Ministry of Transport has developed a spreadsheet for calculating emission rates from the New Zealand rail fleet based on the locomotives used and the tracks that travel throughout the country. The publication, Impacts of Rail Transport on Local Air Quality (Ministry of Transport, 1999) and accompanying database provide data and information on the calculations and emission factors.

The emission factors are based on US Environmental Protection Agency (US EPA) data for rail-sector emissions inventories. The Ministry of Transport spreadsheet should be used for determining emissions from rail, but the user needs to be aware that any new or upgraded fleet may differ from that defined in the spreadsheet.

Bus emission factors

The Auckland Regional Council has developed a Bus Emissions Prediction Model (BEPM) (Auckland Regional Council, 2005b). This model is based on overseas emissions databases, and allows the user to define:

  • average speed

  • fleet profile (percentage of each vehicle type by country of origin, technology and year of manufacture)

  • fuel (sulphur content, biodiesel, water-blend fuel)

  • percentage retrofit of diesel-oxidising catalysts or particle traps.

The BEPM provides fleet average-emission factors for the defined fleet and average speed, and allows the user to compare scenarios.

Alternatives to NZTER emission factors

Auckland Regional Council is developing a Vehicle Emissions Prediction Model (VEPM), which is similar to the BEPM but will be applied to other vehicle classes. This will enable updated emission factors to be developed for the New Zealand fleet. In the meantime it is possible to estimate emission factors based on reported overseas databases and New Zealand fleet information. This will mean the assumptions and errors associated with the emission factors are well understood.

However, it is important that fleet composite emission factors from overseas are not used. The New Zealand fleet has much higher fleet average emissions than most comparable countries (like Australia) because of the delayed introduction of vehicle emissions standards in New Zealand. This means any emission factors must be based on detailed New Zealand (or local) fleet information, including the proportion of vehicles by class, and emissions build standard (or year of manufacture and country of origin). This information is available from the Ministry of Transport Vehicle Fleet Emissions Model.

Estimation of emission factors should be undertaken by professionals with specific vehicle emissions expertise. The BEPM is an example of how overseas emissions databases can be used to predict emissions from the New Zealand (in this case, bus) fleet.

Intersections and idle emissions

As discussed above, most emission factors (including NZTER) provide an estimate of emissions for a typical drive cycle. For example, for the congested level of service, the drive cycle includes some allowance for stopping and starting, idling at intersections, etc. This effect is also implicitly considered in most speed-related emission factors (like BEPM), since the variability of speed during a trip is closely related to the average speed. Slow-speed journeys in towns involve frequent speed changes in response to the traffic conditions, while higher-speed trips are normally driven more smoothly (National Roads Authority and DEFRA, 1992/2003). Nevertheless, for two trips at the same average speed (or for the same LOS/road combination in NZTER), emissions can vary substantially depending on speed variability.

This effect is not usually significant over the averaging times being considered. For an air quality assessment, average emissions over an hour or longer are needed, so an emission factor for an average drive cycle is generally appropriate.

However, emissions at locations such as intersections may give cause for concern because of the high proportion of vehicles idling and accelerating. To address this, the congested/central urban emission factors from NZTER should be used. This will provide a reasonably good representation of emissions around intersections, because it includes a high proportion of idling and acceleration/deceleration. Similarly, a speed-dependent emission factor for low average speed (say 10 km/h) should be reasonably representative of average emissions around intersections.

Micro-simulation of emissions

In circumstances where more detail about the variability of emission factors is required, micro-simulation of emissions can be undertaken. The use of micro-simulation traffic models, which represent the travel behaviour of individual vehicles, is becoming more common in New Zealand. Many of these models provide for the micro-simulation of emissions, but at the time of writing these have not been calibrated for the New Zealand fleet.

The accuracy of any assessment based on micro-simulation of emissions is likely to be limited by the availability and accuracy of local fleet information, and background air quality and meteorological data. Micro-simulation of emissions should therefore only be used in special circumstances, and then only with detailed justification. For an air quality assessment, average emissions over an hour or longer are needed, so an emission factor for an average drive cycle is generally appropriate.

As for any development of emission factors, this should only be undertaken by professionals with specific vehicle emissions expertise.

8.2 Characterising the receiving environment

8.2.1 Existing air quality

To assess the cumulative effect of any proposal on air quality, the predicted traffic-derived concentration of air pollutants must be added to existing air quality, including the concentrations discharged by other sources.

Existing air quality data are of critical importance in any assessment. Ambient air quality monitoring to establish existing air quality is strongly advised for any major projects where there is no nearby or relevant monitoring data. This should be considered at the planning stages of any major project because of the time involved in obtaining data (a full year of data is recommended).

When existing air quality data are required

Existing air quality should be considered in all assessments of discharges to air. The level of detail and accuracy required are influenced by the:

  • anticipated air quality – areas that are anticipated to have poor air quality due to a combination of existing emission sources and/or adverse terrain or meteorology would be expected to require a more robust definition of the existing air quality

  • sensitivity of the receiving environment – where discharges have the potential to affect highly sensitive receiving environments (see Table 6.2), existing air quality would be expected to be well defined.

It is the combination of these considerations that determines the extent to which existing air quality should be addressed. A small increase in emissions within a commercial/light industrial area, for example, might only require a qualitative statement on existing air quality, which identifies the reasons existing air quality is anticipated to be good. Conversely, a large-scale transport development with the potential to affect residential suburbs might be expected to provide good-quality, representative and quantitative air quality data.

Identifying existing data

Pre-existing air quality data can be obtained from a range of sources. Usually the regional council will have the best knowledge of the full range of data available within its region. The council will also be able to provide an opinion on whether the pre-existing data are sufficient for the assessment proposed.

The range of air quality data sources is identified in section 6.4.1 of the Ministry for the Environment’s Good Practice Guide for Atmospheric Dispersion Modelling (Ministry for the Environment, 2004).

Reviewing existing data

Guidance on assessing the quality of currently available ambient air quality data as well as quality assurance and control procedures for the collection of new air quality data is provided in the Ministry for the Environment’s (2000) Guide to Air Quality Monitoring and Data Management. The assessment criteria often include requirements for the air quality monitoring technique to be used. For example, the Standards contain specific requirements for the monitoring of pollutants within gazetted airsheds. The Ministry for the Environment’s Ambient Air Quality Guidelines also contain recommended monitoring methods.

The use of such methods reduces uncertainty and minimises inaccuracy. Before using existing air quality data in an assessment, it is important that the monitoring technique and protocols be audited against the requirements of the Standards and the Guide to Air Quality Monitoring and Data Management to demonstrate that the existing air quality data are of appropriate quality.

The location of a monitoring site and the time of monitoring also affect how representative existing air quality data might be. The site should be representative in terms of location (ideally, within the affected airshed), but also representative in terms of land use and physical setting. The specific location of the monitoring site (eg, its proximity to major sources such as roads) will also be important.

The time of the monitoring is also relevant, in that data collected at the site in previous years may not be representative if the character of the area has changed markedly since monitoring was last undertaken. For example, historical data from a roadside monitoring site in an area that has experienced significant traffic growth would no longer be representative of current levels.

Trends in air quality should be considered, and it is preferable for several years of data to be analysed so that any improvement or deterioration of the air quality of an area can be ascertained. As a minimum, one year of data could be used if there are other longer-term monitoring sites in similar locations which can be used to provide an indication of long-term trends. Ideally, 10 years of data should be used to determine trends.

As noted above, monitoring data should be reviewed with reference to the monitoring method identified in Schedule 2 of the Standards and other Ministry for the Environment guidance (Ministry for the Environment, 2000a, 2002 and 2004).

When local existing air quality data are not available

Section 6.4 of the Good Practice Guide for Atmospheric Dispersion Modelling (Ministry for the Environment, 2004) includes options for estimating existing air quality (without the project) when local data are not available. These options include:

  • comparing the location with somewhere similar

  • making a worst-case assumption

  • modelling other sources in the area

  • starting a new monitoring programme.

The most straightforward options are to compare the location with somewhere similar or make a worst-case assumption. If the area does not have significant large sources and does not have any complex geographical or meteorological features, then it can be assumed that the air quality will be similar to another area of similar population density, emission sources and meteorology. This method requires that such an area can be identified and that monitoring data are available.

Making worst-case assumptions

In the absence of air quality data from local or similar areas, it might be necessary to simply estimate the existing air quality. The safest approach is to assume a concentration at the upper end of what might be feasible.

The most conservative (and straightforward) approach to estimating existing air quality without the project is to make worst-case assumptions, as outlined in Figure 8.2 below. This approach is very conservative, and a more accurate assessment of existing air quality, or a health risk assessment, may be required if this results in a prediction of unacceptable air quality effects.

Figure 8.2: Determining existing air quality with conservative assumptions

Some examples of worst-case assumptions are outlined in Table 8.6. Once again, it needs to be stressed that this approach is very conservative, and in areas where air quality is likely to be poor, ambient air quality monitoring may be required. It is important to ensure the 'existing air quality without project' concentration selected is appropriate and relevant to the location of the proposed project. For example, to avoid double counting it is important to select existing air quality data from an area that is not influenced by traffic, if the assessment includes dispersion modelling to predict the impact of any existing traffic as well as traffic associated with the proposal.

Table 8.6: Examples of existing NO2, PM10 and CO concentrations 'without project'

Area where estimate of background air quality is required


Value to assume

Justification for worst-case assumption, based on review of data to 2004 (extracted from various council monitoring reports and website data in mid-2005)

An urban area with a significant wood- or coal- burning problem

1 hr

150 (µg/m3)

10-year average of maxima, Packe St Christchurch = 124.

3-year average of maxima, Coles Pl, Christchurch = 110.

1-year maximum, Fire Station, Nelson = 148. Christchurch and Nelson represent the worst case for areas with significant domestic heating pollution.

24 hr

100 (µg/m3)

Christchurch, Nelson, Timaru, Masterton, Mosgiel, Arrowtown, Richmond and Kaiapoi have all recorded peaks of over 100 (the highest is 252 in Christchurch in 2002).

8 hr

8 (mg/m3)

The highest values recorded in Christchurch have been slightly above 8.
Area with poor dispersion (eg, urban canyon) within 5 m of a busy intersection or congested area (with over 10,000 vehicles per day and/or wood or coal burning).

1 hr

340 (µg/m3)

4-year average of maxima Khyber Pass = 343. Khyber Pass is a peak traffic monitoring site for NO2(traffic approx. 30,000 vehicles/day, air quality monitoring < 5 m from roadside).

24 hr

80 (µg/m3)

Even smaller centres that have poor dispersion can record high values (Reefton 55, Nelson 165, Wainuiomata 57, Upper Hutt 60).

8 hr

10 (mg/m3)

The highest values recorded in Auckland have been slightly above 10.

Area within 20 m of vehicle routes of over 10,000 per day, or within 100 m of a motorway.

1 hr

140 (µg/m3)

10-year average of maxima, Auckland Penrose = 139.

2-year average of maxima, Peachgrove Rd, Hamilton = 133. Penrose and Peachgrove Road have the highest maximum NO2levels of all data reviewed, except for Khyber Pass.

24 hr

70 (µg/m3)

There are not many sites in this category with monitoring results, but Auckland’s Khyber Pass has recorded 81, almost certainly largely due to traffic.

8 hr

5 (mg/m3)

4-year average of maxima, Peachgrove Rd, Hamilton = 4.75.

Maxima at peak traffic sites in Rotorua and Tauranga are also less than 5.

Urban area that doesn’t have significant wood-burning problem and no vehicle routes of over 10,000 vehicles per day within 20 m, or motorways within 100 m.

1 hr

50 (µg/m3)

Hastings, less than 1 year of data, maximum = 36.

Napier, less than 1 year of data, maximum = 66.

Wellington, all sites, all years, maximum = 53. These sites have some traffic influence, so represent a worst-case assumption for urban areas without significant traffic.

24 hr

40 (µg/m3)

Residential neighbourhood monitoring sites in Hawke’s Bay and Bay of Plenty have recorded occasional exceedances of the PM10 standard, although averages of maxima taken over several years tend to be lower than 40.

8 hr

2 (mg/m3)

Maximum concentrations measured at residential neighbourhood sites in Upper Hutt, Lower Hutt and Masterton are typically 2 or less.
Rural area, or urban area that is very open with low population density.

1 hr

15 (µg/m3)

Masterton 2-year average of maxima = 13.5. There are no results available from rural monitoring sites. Masterton is the lowest result for a 'residential neighbourhood’ site, so this is a worst-case assumption for a rural area.

24 hr

15 (µg/m3)

This is a typical maximum concentration when no obvious sources occur upwind.

8 hr

0 (mg/m3)

With no local sources, CO concentrations are generally very low, and can be taken as effectively zero.

Model other sources in the area

Atmospheric dispersion modelling may be the preferred approach for estimating existing air quality where:

  • there are a small number of existing emission sources in the area for which reliable emission data are available

  • any contribution to ambient levels from other hard-to-characterise sources (such as vehicle emissions, domestic fires or dust from wind erosion) is negligible.

This situation would be unlikely to occur in urban areas of New Zealand. Again, the Good Practice Guide for Atmospheric Dispersion Modelling provides advice on the appropriate application of dispersion models.

Commission a monitoring programme

For significant projects, where the issue is likely to be of importance and when local data are not available, it is recommended that an air quality monitoring programme be undertaken. Monitoring should determine concentrations of the critical indicator contaminants (generally CO, NOx and PM10, and possibly benzene) as well as meteorology, and traffic counts.

Comprehensive guidance on setting up ambient air quality monitoring stations is provided in the Guide to Air Quality Monitoring and Data Management (Ministry for the Environment, 2000). It is recommended that the regional council be consulted before undertaking any new ambient air quality monitoring.

There are two general categories of monitoring site that can be used for background monitoring of ambient air quality concentrations for transport developments.

  • Receptor monitoring sites: a monitoring station can be established in the grounds of a receptor located within the potential zone of influence of the proposed project. The aim of this monitoring is to measure the existing or baseline concentrations of selected contaminants so that the impact of the project on that specific receptor can be determined by ongoing monitoring at the same site after the project has been established. It may be suitable to select a particularly sensitive receptor for this type of site, such as a school, residential dwelling or area where concerns have been raised during a submission process.

  • Representative monitoring sites: these sites are usually located in the area a proposed development will go through, where the aim is to obtain a representative measure of the existing ambient air concentrations over most of the area of concern.

Site-specific monitoring should be undertaken over a sufficient period to obtain representative existing air quality data. The assessor should justify any monitoring period that is less than 12 months.

8.2.2 The built environment

Sensitivity to air quality impacts will vary with land-use type. For example, residential land use (including schools) will typically have greater sensitivity than an industrial setting.

The land use surrounding a proposed development should be reviewed and described in any assessment of air quality impacts. Any facilities where there are likely to be sensitive groups of people (young, old, unwell), such as hospitals and schools, should be identified. The review will provide both an indication to any developer of the likely acceptability of, or objections to, a proposal and a guide to the depth of consultation required. The sensitivity will also be one factor that influences the level of assessment of environmental effects required for a proposed development. Table 6.2 (Tier 1) provides a general classification of the sensitivity of various land uses to discharges of contaminants into air.

Conversely, when considering changes to land use in areas surrounding transport corridors, 'reverse sensitivity' should be considered. The separation distance between transport corridors and sensitive receiving environments should be maintained for the duration of the corridor. Maintaining the existing separation distance should be a consideration in any notice of requirement for new corridors.

8.3 Exposure estimates

8.3.1 Dispersion modelling

Atmospheric dispersion models are used to estimate contaminant concentrations downwind of a discharge source. The information generated by dispersion models can be used in a number of ways, including to:

  • assess the potential adverse effects of proposed activities or changes to existing activities - dispersion modelling is usually the only way to assess the potential effects of an activity that has not yet been constructed

  • predict the effects of changes in emission rates or parameters over time (eg, a change in traffic flows)

  • estimate the influence of factors such as terrain, buildings and meteorology on an activity and their effect on the dispersion of the contaminants discharged by the activity

  • estimate the effect of any mitigation options.

Local and community effects will need to be assessed near the roads by atmospheric dispersion modelling (using a 'near-road' model). In New Zealand, the most common dispersion model used for this type of assessment is CALINE4.

The Good Practice Guide for Atmospheric Dispersion Modelling (Ministry for the Environment, 2004) gives extensive guidance on dispersion models, including those used for transport assessments. They include Gaussian line source models such as CALINE4, and those that can model line sources or account for regional airshed dispersion and secondary reactions, such as CALPUFF/CALGRID and TAPM. All dispersion modelling should be carried out in accordance with the recommendations of the Good Practice Guide for Atmospheric Dispersion Modelling.

8.4 Assessing the effects

8.4.1 Incorporating background data

Background air quality data and predicted pollutant concentrations must be considered together against the selected assessment criteria. Adding the background data and predicted results to provide an estimate of the cumulative impact for comparison with the selected assessment criteria is reasonable for annual average concentrations. For short-term concentrations this simplistic approach is appropriate where the criteria are not exceeded, although it is a very conservative approach and a more accurate assessment may be necessary where compliance is an issue.

The above approach can lead to an overly conservative assessment due to issues relating to the spatial and temporal co-incidence of background and predicted concentrations, as follows.

  • Spatial co-incidence problems - it is often difficult to know whether the background data are representative of the point at which the modelled peak occurs. In general, they will not be located in the same place, so adding the two will overestimate actual future concentrations.

  • Time co-incidence problems - both modelled and background concentrations vary with the time of day due to factors such as meteorological patterns and changes in background emission sources (eg, domestic heating or industrial peaks). In some cases, the peak caused by the transport emissions in question may not occur at the same time as the background peak, so adding the two together may again overestimate the future concentration.

For the highest percentiles (ie, concentration values close to the peak short-term concentration of a year’s worth of such concentration predictions), simple addition can overestimate the source contribution, and in general the overestimate is more severe for the higher percentiles. The best predictive assessment technique is to use hourly, sequential ambient air quality monitoring data that are recorded in the airshed of interest, and then add the hour-by-hour predicted concentrations. These predicted concentrations should be made using meteorological data recorded at the same time as the recorded air quality data. Where data are available, such an approach is recommended.

Careful examination of the locations and conditions under which measured and modelled peaks occur may help determine whether the peaks are likely to occur at the same time and place. Any alternative to adding hour-by-hour data, or adding peak background and peak predicted concentrations, should be fully explained and justified.

8.4.2 Comparing model results with air quality criteria

It is generally sufficient to assess health effects by comparing modelled predictions (including background air quality) with appropriate health assessment criteria.

As noted in section 5, it is important to ascertain both the short-term and the long-term impacts of discharges to air. For most assessments this will necessitate the use of both the national ambient air quality standards (one-hour and 24-hour standards) and the national ambient air quality guidelines (annual guidelines). In all cases, as noted in section 5, the selection of air quality criteria should be justified. In doing so the purpose of the standard or guideline should be clearly stated. None of the criteria provided in section 5 are levels that may be 'polluted up to'. It is also important to recognise the limitations of dispersion modelling when predicting impacts.

For relevant averaging times, the model results for maximum, 99.9th percentile and 99.5th percentile concentrations should be given. As a rule of thumb, modelling using less than a 24-hour average (eg, one hour) should present maximum and percentile concentrations, whereas averaging times of 24 hours or more should only show the maximum concentration levels.

The recommended procedure in the Good Practice Guide for Atmospheric Dispersion Modelling requires consideration of meteorology when assessing worst-case impacts. (This is reproduced in full below.)

Recommendation 53

For the purpose of comparing modelling results to an evaluation criterion:

  1. run the model for the minimum period of one full year of meteorological data where possible (ie, 8,760 hours)
  2. identify the receptor(s) that are most highly impacted and those that are most sensitive
  3. for the receptor(s), report the 99.9 percentile value of the predicted ground-level concentration as the maximum ground-level concentration likely to occur.

Provide an indication of the representativeness of the 99.9 percentile value ground-level concentration by also presenting a number of other percentile values (eg, maximum, 99.5th and 99th percentile values).

Use the frequency of exceedances to indicate the frequency of “pollution events” that exceed the evaluation criterion being used.

Source: Ministry for the Environment, 2004.

The above recommendation applies to one-hour time averages only. Following are a few examples.

  • For sulphur dioxide, nitrogen dioxide and ozone, compare the highest 99.9 percentile predicted concentrations for all receptors with the one-hour national ambient air quality standards. A number of other percentile values (eg, 99.5th and 99th percentiles) should also be reviewed. The maximum predicted concentrations for all receptors should then be compared with the relevant eight-hour, 24-hour or annual average guidelines.

  • For carbon monoxide, compare the maximum predicted concentrations for all receptors with the national ambient air quality standard of 10 mg/m3 as an eight-hour average.

  • For PM10, compare the maximum predicted concentrations for all receptors with the national ambient air quality standard of 50 µg/m3 as a 24-hour average. Similarly, compare the maximum predicted annual concentration for all receptors (if more than one year of meteorological data used) with the national ambient air quality guideline of 20 µg/m3 as an annual average.

The review of other percentile values is very important because it furthers understanding of how meteorology (and other factors) affect the maximum downwind concentrations.

As discussed in section 3.1, transport activities have the potential to discharge a large number of different contaminants. However, unless there are special circumstances (such as elevated background concentrations of another compound), most assessments will provide adequate coverage of all potential effects if based on CO, PM10 and NO2. As we have seen, these three do not explicitly cover a range of other toxics, in particular carcinogenic compounds such as benzene and 1,3 butadiene. The concentrations and effects of these other compounds are still being researched, but results from the New Zealand research under the Health and Air Pollution in New Zealand programme (further details can be found at suggest the overall health effects are substantially less than those due to CO, PM10 and NO2. In other words, unless there are very specific local circumstances, transport proposals that result in acceptable effects for CO, PM10 and NO2 will also most likely have acceptable effects for other toxics.

It is therefore recommended that an assessment of the effects of benzene and 1,3 butadiene only be considered for major projects or in locations where elevated levels are known or considered likely. Consultation with the regional council should clarify the level of assessment required.

8.4.3 Assessment of overall network effects

The Tier 3 assessment procedure has so far focused on the assessment of local effects. Most transport projects are designed to improve safety and traffic flows, and will therefore have an overall air quality benefit for the community. For significant projects that increase emissions in some locations and decrease emissions in others, it is recommended the assessment include an estimate of the likely air quality costs and benefits. Methods for quantifying community health impacts are recommended in Appendix 4. The method selected will depend on the level of detail required.

This guide recommends methods for assessing community impacts, but it is not possible to provide guidance on the relative importance of community versus local impacts. For example, construction of a new link may ease congestion and reduce overall vehicle kilometres over a wide area, but air quality immediately adjacent to the new link may be degraded. Although the net effect of the project would be to maintain or improve overall air quality, an assessment of local impacts is still required to determine whether any localised negative impacts are within acceptable limits, and whether mitigation can or should be applied. In cases where air quality is significantly degraded across the community, the project might require mitigation (depending on the state of air quality in the airshed).

It is recommended that the regional council be consulted to determine assessment requirements for any proposal that will result in a significant net increase in emissions.

8.4.4 Assessment of photochemical smog or other regional-scale impacts

In some circumstances, urban airshed modelling may be necessary to assess photochemical smog or other regional-scale impacts, such as the formation of secondary particulates. Figure 8.3 illustrates the process for modelling regional-scale air quality impacts.

Regional-scale assessments will be more commonly used for policy development. However, a very large scheme or transport strategy may require an assessment of regional effects if there is likely to be a significant increase in emissions. This could involve working with regional council air quality staff to revise the regional emissions inventory.

In cases where a significant net increase in emissions is expected, early consultation with the regional council is recommendedto determine whether an assessment of regional-scale impacts is required.

Some general guidance on assessing the regional effects of transport projects is provided in the Guidelines for Conducting Air Quality Studies (Austroads, 2000). Further guidance on emission inventories is provided in the Good Practice Guide for Preparing Emissions Inventories (Ministry for the Environment, 2001), and the Good Practice Guide for Atmospheric Dispersion Modelling (Ministry for the Environment, 2004) provides some guidance on urban airshed modelling.

Figure 8.3: Modelling of regional-scale air quality impacts7

8.4.5 Non-human health considerations

Air pollution effects on ecosystems

Section 4 of the Ambient Air Quality Guidelines (Ministry for the Environment, 2002) provides critical levels for protecting ecosystems from sulphur dioxide, sulphate particulate, nitrogen dioxide, ammonia, ozone and fluoride. The Effects of Air Contaminants on Ecosystems and Recommended Critical Levels and Critical Loads (Ministry for the Environment, 2000b), provides some guidance on methods for calculating pollutant deposition rates from predicted or ambient monitoring results, and guidance for assessing whether a discharge is likely to cause adverse effects on ecosystems.

In Europe and North America the effects on sensitive ecosystems of acid deposition and elevated pollutant concentrations from industrial and other anthropogenic sources have been subject to legislative controls for some decades. Similar effects in New Zealand are not so evident, although they have been reviewed in a number of technical reports (eg, Ministry for the Environment, 2000b).

Much of the work has drawn on knowledge of the effects on non-New Zealand (North American and northern European) plant species. The recommended critical levels and loads and the provisional guidance on assessing deposition and its effects given in The Effects of Air Contaminants on Ecosystems is based on this knowledge. There is scant information on either the effects of air pollutants on native New Zealand species or the current level of pollutant deposition or concentration in New Zealand’s natural environments. The robustness of any assessment of air pollution effects on ecosystems in New Zealand is therefore very vulnerable to these knowledge gaps. Despite these limitations, it is good practice to assess potential effects on ecosystems for any significant source that may have an impact on sensitive ecosystems.

8.5 When a health risk assessment is required

In some situations it may be necessary to undertake a more comprehensive air pollution health risk assessment as part of a detailed study. For example, this may be required when the predicted effects exceed ambient air quality criteria, and decision-makers require more specific information about the likely health effects of this.

An air pollution health risk assessment would include determining exposure and dose via a number of different pathways (inhalation, dermal, ingestion, etc), assessment of dose-response data, and characterisation of health risks from the exposure/dose assessments.

The national ambient air quality standards are health-based standards, which are intended to provide a guaranteed level of protection for the health of all New Zealanders. In most circumstances it is appropriate to assess the potential health effects of discharges to air from transport by comparing model predictions with the national ambient air quality standards, any regional council assessment criteria, and (where appropriate) the national ambient air quality guidelines.

Typically it is only when national ambient air quality standards or guidelines are exceeded that an air pollution health risk assessment is required.

Health risk assessments are very specialised tasks, and are often designed to account for very specific local concerns. It is recommended that expert assistance be sought for any full health risk assessment.

8.6 Accuracy

The Good Practice Guide for Atmospheric Dispersion Modelling (Ministry for the Environment, 2004) includes guidance on accounting for, and reporting of, model error and uncertainty. This includes consideration of source characteristics, meteorological data, terrain, model performance and the misapplication of models.

Careful consideration of accuracy is especially important in transport assessments, which usually rely on models to estimate traffic and vehicle emission factors as well as dispersion. With three or more levels of modelling there is plenty of room for error and - perhaps more importantly - mistrust of the results. It is important errors and uncertainty be considered throughout the assessment, and that the report clearly demonstrates this.

Standard quality control techniques, such as the use of calculation checking and aspects as basic as checking that model data are correctly input, are extremely important and should be used. It is also recommended that within an assessment report each aspect of an assessment be auditable and repeatable so that the approach, assumptions and calculations can be independently reviewed.

Ideally, the uncertainty associated with each aspect of the assessment should be estimated and stated. Where this is not possible, sensitivity analysis should be carried out. The results of the sensitivity analysis could be used to present a 'worst-case' estimate and a 'most likely' estimate of the effects. For example, a worst-case estimate might be based on unusually high traffic levels, high emission factors, a high proportion of heavy commercial vehicles, high background concentrations and poor meteorological conditions. In reality, it is unlikely that all of these variables will be 'worst-case' at the same time, so a more realistic “'most likely' scenario is also recommended.

8.7 Reporting a Tier 3 assessment

The results of a Tier 3 assessment should be documented for inclusion into any assessment of environmental effects. The report should summarise the findings of the assessment, including the basis of the traffic information, air quality information, any assumptions, and their justification. Section 4.5 describes the recommended content of an assessment report. The size and nature of the report will depend on the project, but it is anticipated that for any Tier 3 assessment each of the sections described will be included.

Tier 3 detailed assessment − recommendations

A full Tier 3 assessment would generally only be required for large or significant projects, or very sensitive receiving environments. However, some aspects of the Tier 3 assessment may be required for other projects. For example, a project that exceeds the Tier 2 significance criteria might only require consideration of existing air quality to determine whether the effect on air quality is actually significant in the location under consideration.

For any aspect of the assessment where detailed information is not available, or is not necessary, it is appropriate to adopt the conservative assumptions discussed for Tier 2 assessments.

5 Percentages are based on vehicle kilometres travelled by vehicle class in the Ministry of Transport Vehicle Fleet Emissions Model (VFEM). Model outputs were provided by Stuart Badger in August 2007. VKT for all petrol HCVs and buses have been added to petrol LCVs (because these classes are not included in the NZTER).

6 The distinction between urban and rural roads is that made in the annual Roading Statistics, published by Transit NZ, which distinguishes state highways from roads controlled by territorial local authorities.

7 Based on Guidelines for Conducting Air Quality Studies, Austroads, 2000, Chart 4.