• About this site|
• Site index|
• Contact us

Good Practice Guide on Assessing Discharges to Air from Land Transport

Foreword by the Ministry

Acknowledgements

Key Data Sources & Models

1 Introduction

2 Legislative and Policy Context

3 Assessing Discharges to Air from Transport

4 The Assessment Process

5 Air Quality Criteria

6 Tier 1 Screening Assessment Process

7 Tier 2 Conservative Assessment Process

8 Tier 3 Detailed Assessment Process

9 Assessment of Community Impacts and Comparison of Project Options

10 Mitigation Options

11 Monitoring Project Effects

References

Appendix 1: Calculations and Data

Appendix 2: Traffic and Transport Data

Appendix 3: Vehicle Emissions and Effects

Appendix 1: Calculations and Data

A1.1 Method for estimating concentrations

Atmospheric dispersion modelling has previously been used to determine dispersion curves for New Zealand conditions (Scoggins et al, 2004; Bluett and Fisher, 2005). As part of this project, the curves have been normalised for emissions factors and number of vehicles to provide an estimate of concentration per g/km of emissions as a function of distance from the roadside. To determine concentration, the road emissions are estimated based on the number of vehicles (for the relevant averaging period) and the fleet-weighted emissions factor. There are a number of assumptions and simplifications that need to be made, all described here.

None of these calculations account for existing or background concentrations. These must be added if a cumulative assessment is required. The method only applies the effects due to the transport corridor under consideration. This is relatively straightforward for CO, but can become more complex for NO₂ and PM₁₀ because of atmospheric chemistry processes which depend on background concentrations that may not be known.

The dispersion curves for all other gaseous pollutants (here mainly NO₂ and PM₁₀, but also any others that may be needed) are equivalent to the CO dispersion curves. This is a simplified approach, because the dispersion curves for each pollutant are not identical, due not only to chemical processes but also to subtle differences in dispersion and deposition. However, it is argued that the differences are not great, and that the CO dispersion is a reasonable representation for the other pollutants over the distances and times scales of interest. This is also the methodology followed by Austroads in Australia (Austroads, 2004) and DEFRA in the UK (National Roads Authority and DEFRA, 1992/ 2003).

All measurements in this document are taken from the edge of the road (some methods take the measurements from the road centre line).

A1.2 Dispersion curve basics

The basic dispersion curve is derived from the modelling and experimental measurement of CO. The one-hour dispersion curve for CO was developed based on measured roadside concentrations at an Auckland monitoring site. To determine how the CO concentration varied with distance from the carriageway, Caline4 was used (eg, Scoggins et a., 2004). NZTER emissions factors were used to estimate emissions. Using a best-fit algorithm, the estimated variation of maximum predicted one-hour concentration with distance from the carriageway can be modelled as:

Maximum 1-hour CO concentration (µg/m³ per veh) = 4 exp (-0.3 d^0.5)

This equation is based on an assumed NZTER fleet-weighted CO emission factor of 12.302 g/km. Therefore, an alternative normalised equation is:

Maximum 1-hour CO concentration (µg/m³ per g/vkt) = 0.325 exp (-0.3 d^0.5)

To evaluate the conservatism of this dispersion curve, a similar normalised curve has been developed using meteorological parameters recommended for screening assessments by Austroads (2004). For screening assessments, the Austroads review recommends the use of Caline4, assuming a one m/s wind speed, stable atmospheric conditions, sigma-theta of 10 and worst-case wind angle. The Austroads review compared maximum concentrations predicted using these assumptions with maximum measured concentrations for a freeway and arterial road. The analysis showed that this approach is adequately conservative for screening assessments. The most significant sources of uncertainty were background concentration and the atmospheric chemistry rates, where these are relevant (eg, in applying the relationship to NO₂, see below).

Figure A1-1 shows a comparison of the normalised dispersion curve based on the parameters recommended by Austroads and the methods assumed here. The Austroads curve predicts significantly lower concentrations (approximately half) than the Scoggins et al curve within 50 m of the roadside. This suggests that the Scoggins et al curve may be most appropriate in situations where worst-case wind speeds of less than one m/s are likely. The emphasis of the method developed here is on worst-case circumstances, and these can occur just about anywhere in New Zealand - even in very exposed places, albeit for shorter times. If this assessment is critical, then specific modelling (say, using Caline4 and on-site meteorological data) should be used.

The curve chosen, being the more conservative and applicable to very light wind speeds, is the upper one (Scoggins et al, 2004). This tends to be conservative relative to the Austroads recommendation within 100 m or so of the road, but allows concentrations to fall off more quickly at distances further than 100 m.

A1.3 Averaging periods

The eight-hour and 24-hour averages are based on a linear extrapolation of the one-hour dispersion curve. This is a conservative approach, because it assumes that worst-case meteorology applies for the entire eight- or 24-hour averaging period.

A1.4 Generalised dispersion curves

The sections above describe how the dispersion relationship is derived for CO, for one-hour, eight-hour and 24-hour periods. This is now assumed to extend to NO₂ and PM₁₀. A detailed validation of this assumption is beyond the scope of this document, and requires a fairly intensive field programme of measurements that cover the light wind speed conditions of interest. This was partially undertaken as part of the derivations (Bluett and Fisher, 2005) but needs to be further investigated. However, as part of a checking process, the dispersion rates used in one of the more commonly applied models were examined. Using the latest (2006) version of Caline4, through its GUI processor CalRoads View, a series of model runs were made using the New Zealand fleet-weighted average emission rates and a meteorological field with light wind speeds (1 m/s) in Auckland.

Figure A1-2 shows the results, all normalised for a concentration of one at the edge of the carriageway. This figure shows the Caline4 dispersion curves for CO, NO₂, and PM₁₀. Also shown is the generalised dispersion curve based on CO described earlier.

The data show that the choice of the generalised curve is a reasonable representation of what Caline4 uses (although some caution needs to be applied when modelling NO₂ and PM₁₀). The atmospheric transformation of NO₂ is accounted for in Caline4 by making assumptions about ozone. Secondary production of PM₁₀ is not, however this is not significant on the short-time scales involved here.

Therefore, to the first order and as a very good approximation, the calculation for any gaseous pollutant is made using the following relationship:

Maximum concentration over period T (µg/m³) = 0.325 exp (-0.3 d^0.5) x (N/T) x EF

where:

d = distance from roadside in metres

N = number of vehicles in the period T

T = time period assessed (one to 24 hours)

EF = emission factor in g/km

Note that this relationship is reliable within 250 m of the road, but needs to be used with caution for PM₁₀ and NO₂ at greater distances. In practice, this is not a serious restriction as compliance would usually be required within 250 m, and cases where non-compliance at 250 m or further would be acceptable would be uncommon.

A1.5 CO eight-hour

Thus the eight-hour CO concentration is calculated based on the one-hour dispersion curve using average one-hour traffic for the eight-hour period.

Maximum 8-hour CO concentration (µg/m³) = 0.325 exp (-0.3 d^0.5) x (N/8) x EF

where:

d = distance from roadside in metres

N = number of vehicles in eight hours (worst eight hours)

EF = emission factor in g/km.

A1.6 PM10 24-hour

In an identical fashion, the 24-hour PM₁₀ concentration is calculated based on the one-hour CO curve.

Maximum 24-hour PM₁₀ concentration (µg/m³) = 0.325 exp (-0.3 d^0.5) x (N/24) x EF

where:

d = distance from roadside in metres

N = number of vehicles in 24 hours (high traffic day)

EF = PM₁₀ emission factor in g/km.

A1.7 PM10 annual

Annual average PM₁₀ concentrations at the roadside (20 m from the edge of the road) are derived from the annual average CO curve using a scaling argument based on Auckland conditions (Scoggins et al, 2004). This is a difficult factor to generalise because it is strongly dependent on climate factors (eg, the frequency of calm conditions) at the particular site of interest. The method described here is a first-order approach, used in the absence of any more detailed data, and it must be recognised that it is based on Auckland conditions and only applicable at an indicator location 20 m from the roadside edge.

Roadside annual average PM₁₀ (µg/m³) = 0.007 x N x EF

where:

N = number of vehicles in one hour

EF = PM₁₀ emission factor in g/km.

A1.8 NO2 one-hour

In a similar fashion, the one-hour NO₂ concentration is calculated based on the one-hour CO curve. However, since most emissions factors are given for NOx, it is necessary to convert the concentration to NO₂.

Maximum one-hour NOx concentration (µg/m³) = 0.325 exp (-0.3 d^0.5) x N x EF

where:

d = distance from roadside in metres

N = number of vehicles in one hour (peak)

EF = NOx emission factor in g/km.

Maximum one-hour NO₂ concentration (µg/m³) = NOx x C

where: C = conversion factor from NO_x to NO₂, here assumed to be 0.2
(See detailed discussion later in this Appendix).

This estimation is similar to that used by Austroads, and is discussed in detail below. There are more complex and detailed models that can be used to improve accuracy, based on roadside measurements made in Auckland (eg, Scoggins et al, 2004, Bluett and Fisher, 2005). However, these are not applied here, with the emphasis instead being on establishing a conservative case envelope, which is covered by the 20 percent conversion factor used. This is maximised at 20−50 m from the roadside, and is lower at closer or further distances.

A1.9 NO2 24-hour

In an identical fashion again, the 24-hour NO₂ concentration is calculated based on the one-hour NO₂ curve.

24-hour NO₂ concentration (µg/m³) = 0.325 exp (-0.3 d^0.5) x (N/24) x EF x C

where:

d = distance from roadside in metres

N = number of vehicles in 24 hours (high traffic day)

EF = NO_x emission factor in g/km

C = NO_x to NO₂ conversion factor, 0.2.

A1.10 NOx to NO2 conversion

Emissions of oxides of nitrogen are generally quantified in terms of NOx, which is mainly composed of NO and NO₂. For most combustion sources, the emission is mainly NO, which oxidises to NO₂ in the atmosphere. The contaminant of interest, from the point of view of health effects, standards and guideline compliance, and degradation of visibility, is NO₂.

Given NOx emissions, models can fairly easily simulate the dispersion of NOx as if it were an inert gas. But the determination of the fraction of NO₂ requires a model that simulates chemical transformations, or some empirically determined formula for the NO₂:NOx ratio. Even when a sophisticated model is used to simulate the oxidation of NO to NO₂, knowledge of the oxidants taking part in such reactions (eg, ozone, volatile hydrocarbon products, and oxidants such as OH) may still not be well quantified. Thus the NO₂ concentrations depend not only on the NO_x emissions and the distance from the source, but also the absolute quantity of these emissions and the amount of oxidant in the air, which is not usually known.

In the United Kingdom, the problem of modelling NO₂ has been addressed by air quality expert groups on behalf of the Department for Environment Food and Rural Affairs (National Roads Authority and DEFRA, 1992/2003). Using air quality data, they developed relationships between the annual mean NO₂ and NOx.

When deriving a relationship specific to New Zealand conditions, there are several considerations.

1. The temporal and spatial variation in the NO₂−NOx relationship needs to be assessed. NO_x may be emitted as NO, but is transformed to NO₂ further downstream, affected by both the distance the emissions travel and the time they have to react with ambient air.
2. The maximum possible NO₂:NOx ratio is one, but the oxidants may be depleted before all NO is transformed to NO₂. In such situations, more NOx may be emitted as NO, but little more NO₂ will result.
3. The NO₂-NOx relationship must be applicable to the whole region rather than to individual emission sources, because NOx from all sources becomes mixed in the atmosphere.

A simple universal ratio of NO₂ to NOx was proposed 15 years ago (Janssen, 1988) to change linearly with distance from the source:

[NO₂] / [NOx] = 0.2 x

where x is the distance from the source in km, and the concentrations are in units of either parts per billion by volume (ppbv) or µg/m3 as NO₂. However, this does not account for the NO₂ fraction near the source (x = 0), nor the limiting value at larger distances (ie, the ratio should always be ≤ 1).

Considering air quality observations in Auckland at Khyber Pass, the [NO₂]/[NOx] ratio varies from close to 1.0 to quite low values (down to 0.1 during winter), with day-to-day variations of the order of 0.1. At Musick Point, 10 km away, there is a range from around 1.0 to 0.2, with a median value of 0.86. Detail of these ratios from data from the Penrose site is shown in Figures A1-3a and b.

A further analysis conducted on annual ratios from a number of stations throughout Auckland shows a similar pattern (see Figure A1-4).

It can be seen from these figures that the NO_x:NO₂ ratio is strongly dependent on the concentration, with more complete conversion (ratio close to 1.0) occurring at lower concentrations. At the higher end (the concentrations of most interest for standards and health effects) the ratio tends to be lower, levelling out at around 0.2 for the longer-term, 24-hour and annual values.

These values are for the "urban plume" as a whole, but are still applicable to emissions from transport as these are part of the urban air.

In order to make use of such information in assessing transport emissions, however, a distance-based approach must be developed. From the observations, a formula can be derived empirically, as:

[NO₂] / [NOx] = 0.1 + 0.2 (x / (1 + x/5))

The ratio starts at 0.1 when x = 0, increases at a rate of 0.2 per km (as given by Janssen, 1988), but then levels off. The factor of 5 in the formula leads to a ratio of 0.20 at x = 0.5 km, 0.60 at x = 5 km, and 1.0 in the far field at x > 40 km. This relationship holds well for longer-range transport, but is not strictly applicable on the roadside scales of a few tens of metres. As shown in Figure A1-4, the conversion rate can vary from 0.1 to close to 1.0 depending on the concentration.

The Austroads (2004) review recommended a maximum conversion of NO to NO₂ of 15 percent at 30 m from the curb (ie, a ratio of 0.15).

The data in Figures A1-3 and A1-4 can be used to derive an appropriate rate for use in Auckland (and by implication other New Zealand areas affected by traffic emissions). At low concentrations, although the conversion ratio is high, the NO_x value is low and the resulting NO₂ concentration is low. At high concentrations, the conversion rate is low but the NO_x value is high, resulting in higher NO₂ concentrations. The point at which the concentration tends to achieve its conservative maximum is when the NO_x is around 1,000 µg/m3 or greater, which gives rise to a maximum one-hour NO₂ concentration of 200 µg/m3 - the one-hour standard. This is illustrated in Figure A1-5.

This maximum conversion occurs when the conversion rate is 0.2 (or 20 percent). This pattern also occurs in data from Christchurch, and in the 24-hour data (not shown here). Thus the appropriate NO to NO₂ conversion rate to use within the 20-50 m zone away from a roadside is 20 percent, or a ratio of 0.2.

A similar best-fit relationship between measured NO₂ and NO_x was developed by Scoggins et al using annual average NO₂ and NOx data from a number of Auckland monitoring sites for the period 1995 to 2002. This showed that for high values of NOx, the amount of conversion to NO₂ is small. This is due to the other components (such as ozone) being used up. For lower amounts of NOx, the conversion to NO₂ can be almost complete, since enough ozone is present to complete all the reactions.

As for previous discussions and recommended methodologies, the figures given here are conservative default values that should be revised in cases where more accurate, site-specific data are available.

A1.11 Thresholds for tier 1 assessment

The tier 1 assessment thresholds are intended to identify proposals that should be assessed using either the tier 2 or tier 3 assessment procedures, and (more importantly) to identify projects that do not require assessment. These thresholds are intentionally conservative because they need to capture relatively small projects that may have significant effects because of their location.

Assessments to date have generally focused on arterial roads and state highways. The Transit definition of an urban arterial road is "arterial and collector roads within urban areas carrying traffic volumes of greater than 7,000 vehicles/day". This was taken as a starting point for developing an assessment threshold.

The tier 2 assessment method was used to predict maximum contaminant concentrations for a range of scenarios. The assumptions and results for a road with 3,000 vehicles under congested conditions, and a road with 7,000 vehicles under free-flow conditions, are summarised in Table A1-1.

Table A1-1: Caline4 model results showing effects due to two traffic flow scenarios at various distances from the road

Table A1-1 shows the basis for selecting the 5 m / 50 m criterion in the tier 1 process. It shows model calculations for the most conservative concentrations of various pollutants at various distances, based on the two scenarios (the detailed calculation examples are shown below). The grey shaded figures are below the significance criteria as discussed in the tier 2 methodology.

The assumption is there is a distance beyond which the effects might be considered insignificant, relative to the standard. These values are shown highlighted, with breakpoints at 50 m for pollutants assessed at 24 hours (PM₁₀), and 5 m for pollutants assessed at one hour (NO₂). (Note: an additional criterion could have been selected for eight-hour CO [at 10 m], but this appears to fit within the other two and so would add little value. That is, if either of the other criteria are met, then the eight-hour CO will also be met).

The selection of these breakpoint criteria is subjective, but basically is very tight for PM₁₀, at 1 µg/m3, being two percent of the standard, and slightly more relaxed for CO and NO₂, at one mg/m3 and 20 µg/m3 respectively, being 10 percent of the standard for each. The criterion for PM₁₀ is tighter because this is a pollutant of concern throughout the country, has greater health effects, and has more strict enforcement provisions under the Standards.

A1.12 Benzene emission factors

The impact of fuel specification on benzene emissions has been assessed for the Review of Fuel Quality Requirements for Australian Transport (Environment Australia, 2000b). According to this review, petrol engine exhaust is responsible for the bulk of benzene emissions, with relatively small contributions from petrol evaporative losses and diesel engine exhaust. Incomplete combustion of fuel benzene and other aromatics results in exhaust emissions of benzene. The introduction of catalyst technology has led to large reductions in benzene emissions.

Relationships to estimate benzene emissions provided by the Environment Australia review are as follows.

Petrol catalyst equipped vehicles:

Percentage benzene of exhaust VOC = 1.077 + 0.7732 x % benzene (vol) + 0.0987 x {% aromatics (vol) − % benzene (vol)}

Petrol non-catalyst vehicles:

% benzene of exhaust VOC = 0.8551 x % benzene (vol) + 0.12198 x % aromatics (vol) − 1.1626

Evaporative:

% benzene (vapour, New Formulation) = % Benzene (liquid, New Formulation) x 0.9/2.6

Benzene emission factors have been estimated based on these relationships, and New Zealand fuel specifications are shown in Table A1-2.

Table A1-2: The fraction of total aromatics and benzene in New Zealand petrol fuels, including estimates of the percentage of volatiles (VOCs) emitted through exhaust and evaporation

For the sake of simplicity, it is recommended that the highest benzene percentage for any one year be adopted.

A1.13 Comparison of New Zealand and Australian emission factors

Table A1-3 shows ratios of emissions estimates based on different emissions factors compared with factors used in Australia (Victoria EPA). Values less than one imply Ministry of Transport emissions estimates are higher than the EPA's, indicating that the New Zealand Ministry of Transport estimate is more conservative, in part reflecting the tighter standards applying in Australia (Auckland Regional Council, 2005a). Most are close to one, showing reasonable consistency between Australia and New Zealand, with the notable exception of SO₂ since New Zealand has adopted lower sulphur levels, and of VOC, since New Zealand also uses less aromatic petrol.

Table A1-3: Comparison of the emissions factors used in Australia with those used in New Zealand (by the Ministry of Transport)

EPA = Environment Protection Authority, Victoria; MoT = Ministry of Transport, New Zealand Transport Emissions rate database.