are predictive models, but there is some doubt about their reliability. With each of these three impacts, much then depends on the ability of the transport model to predict the necessary causal factors.
The most aggregate sketch planning and strategic models will typically not provide link-based data. The most that they can do is to provide estimates by zone of total traffic (in veh-km) and average speed. They are not therefore able to distinguish the effects of acceleration and deceleration, cold starts, link type or built form from the first table, except in a very approximate way. They will thus provide approximate estimates of impacts, and these will be particularly approximate in situations where there are marked differences in speed across the network. Equally, they are unable to estimate effects for individual links and types of built form, and hence say little about the distribution of impacts. However, such models usually provide estimates for more than one time period, and thus enable the effects of traffic throughout the day to be estimated. They also typically provide information on changes in vehicle movements for a wider range of modes, including bus and rail, although the relationships between impacts and causal factors are typically less well developed for these other modes.
Conventional network models perform rather better than this, in that they do estimate speed by link, and thus provide more accurate estimates of impacts. They also identify effects by link, and can therefore indicate the distribution of impacts. Few, however, estimate acceleration and deceleration directly, and identification of cold start traffic, while possible, may be complex. They often also focus solely on peak period conditions. Their main weakness is that they provide so much data that it becomes difficult to assimilate. Microsimulation models are able to estimate levels of acceleration and deceleration directly, but are too detailed for appraisal of city-wide strategies.
Typically, national governments have set the limit to 55 dB(A) for noise perceived outdoor during daytime. For night time the corresponding limit is 45 dB(A). An indoor limit might also be defined: 30 dB(A) with doors and windows shut. Since the noise level is fluctuating, it is usually transformed to find the equivalent continuous 24 hours sound pressure, dB(A) Leq. Weighting the Leq to account for the fact that noise is more annoying during the night than during the day produces various measures of the weighted equivalent sound pressure level. One of these is EFN, which weights noise between 00.00 and 06.00 ten times (10 dB) more than during the day. The EU working group on indicators recommend two indicators, the Leu and the Leu(n). The first divides the day into three periods, of which the 4-hour evening period is given a penalty of 5 dB relative to the daytime, and the 8-hour night period is given a penalty of 10 dB. The Leu(n) consists only of the (unweighted) night period.
Noise from traffic is a complicated matter in many respects. The perceived level of noise depends partly on the volume and composition of traffic and on how far away from the source of noise the individual is located. Topography also matters. Further, a change in the actual emission of noise may be perceived differently depending on the original noise level.
Obviously, some sort of unit cost related to traffic volume is a very simplistic approach. A scheme for noise cost calculation should preferably include the aspect of existing noise level, how many persons are affected and how far away from the source they are located. Further, one would need some knowledge on how a change in traffic volume would affect these variables. Finally, a valuation in money terms is needed.
The principles of willingness to pay and willingness to accept is relevant here – see Section 9.7 about compensating and equivalent variation.
A pragmatic approach is to use a unit cost per vehicle kilometre which is different for the different vehicle types and which also distinguish between urban and rural areas.
When there are many sources of noise, the resulting noise level is by no means the sum of them. Instead, the difference between the noise levels of the two sources determines the total impact. Only if this difference is small does the second source contribute appreciably to the total noise level.
Similarly, the effect on the noise level of removing one of many equally strong noises is small. This would make the marginal cost of noise per vehicle kilometre small in highly congested or heavy traffic areas. On the other hand, there will usually be more people exposed in such areas. These two counteracting influences on the noise impact and cost may make the assumption of a constant unit cost per vehicle kilometre acceptable as a very crude approximation, except for sparsely populated (rural) areas, where the unit cost is much closer to zero.
12.3.2 Air pollution
Emissions from traffic have impacts on the local and regional environment and on global warming. The global impacts are predominantly related to emissions of CO2 and will be treated by a separate indicator described in Chapter 16.
Regarding the local and regional emissions, we want to establish indicators of the cost of air pollution for the pollutants CO, NOx, VOC, and particulates (PM). Even if the smallest particulates (less than 2.5 micrometres) are the most harmful, it will probably be easier to establish emission rates and valuation for particulates less than 10 micrometres. If needed, we should also be able to derive an indicator of SO2 costs from fuel consumption and the sulphur content of fuels. The indicators will be denominated in monetary units, and so they may be presented singly or aggregated.
In principle, we need to consider emission, the dispersion of the harmful substances through the air, their chemical reactions and interaction with pollutants from other sources, the resulting air quality at different places in the city, and the number of people (buildings, crops) exposed to these air conditions. We then need to assess the damage inflicted on the recipients in monetary terms. Since there is no way we can do this in the strategic analyses we aim for here, we will have to rely on knowledge produced elsewhere and on reduced forms of such air pollution modelling. In particular, we want to be able to compute emissions as accurately as possible, while relying on other sources for the average cost per emitted kilogram of the pollutants in the particular conditions prevailing in urban areas like ours.
Air pollution costs are a typical case of the need to be able to supplement analyses at the strategic level with more detailed analyses from time to time, to ensure that the simplified relationships of the strategic models are broadly in line with the results from detailed air pollution modelling in each particular city.
Since in many cases, the urban area we study will also be composed of less densely inhabited areas, the sources that will be most useful to us will include costs per emitted kilogram in both urban and rural conditions. We might then subdivide the urban area in urban, rural and intermediate areas, each with their own cost of emission.
To the extent that the emission indicators admit of spatial disaggregation, our cost of air pollution indicators will do the same, and this may form the basis for presentation of results in the form of maps.
The general form of the indicators will be
(12.1) APC =C⋅EF⋅A
where APC is the cost of air pollution indicator, C is the cost per emitted kilogram of substance (incorporating in a very simplified way the dispersion of the pollutants in the air, the dose and the numbers of exposed receptors, the damage done to them and the cost per damage), EF is the emission factor in kilograms per vehicle kilometre and A is the activity in vehicle kilometres.
It must be pointed out that there will be considerable uncertainty surrounding these indicators, stemming from the emission factors, the problems of integrating emission models and transport models, the transferability of the unit costs, and the uncertainties inherent in the underlying dispersion modelling, the dose-response functions and the costs of damage estimates.
With respect to the emission factor EF, there are basically two options open to us. The first is to assume constant average emission rates per vehicle kilometre for the different types of vehicle and fuel. The rates will nevertheless change due to technological development and its rate of penetration in the vehicle fleet, which are factors that belong to the scenario assumptions. They might also be differentiated across classes of road (urban, rural, highway). This refinement will require some extra programming for most models.
More and more it is recognised that the effects of congestion on air pollution and global warming merit at least as much attention as the traditional problems of delays and time losses. As congestion increases, emissions to air can increase very rapidly.
To be useful for the evaluation of strategies, our indicators of energy use and air pollution should therefore be sensitive to speed. The other option, then, is to make emission rates a function of average vehicle speed.
The MEET project (Hickman et al 1999) provides the appropriate functional relation- ships. However, since the functions are estimated from full real-world driving cycles, the application of this methodology must be based on average conditions in fairly large areas or zones. The MEET methodology cannot be applied at the link level. By and large, there is no easy relationship between transport models and emission models, which have developed separately. Applying the methodology to the output from transport models will therefore be somewhat experimental.
Consequently, we divide the urban areas into areas of suitable size. Presumably, homogeneous driving conditions in an area will produce better results. Some consideration should also be given to the need to define the areas such that the volume of walking and cycling can be had, and such that the conditions with respect to interaction between slow and motorised modes are similar throughout the area. From the transport model (or from empirical evidence) we compute vehicle kilometres and average speed for the different types of vehicle in each area in each strategy. This might require some programming. Finally, each area is characterised by its population density as being urban, rural or intermediate. This is done to make the right choice of unit cost per emitted pollutant.
Applying the MEET methodology, there is also scope for further refinement. For instance, changes in the number of cars can be used to compute changes in evaporation (only gasoline, not diesel), and changes in the number of trips can be used to assess emission from cold starts.
Improvements in fuel efficiency and cleansing technology and changes in the composition of the vehicle fleet will obviously matter for the emission factors. This must be built into the description of the scenarios. The ensuing shifts in the emission factors are uncertain, but some clues are contained in Hickman et al (1999) and other sources. Unless it can be derived from the network model, assumption will have to be made about the shares of different public transport modes in the total public transport supply.
Next, we turn to the unit costs. Table 12.3 shows some possible sources for the unit costs and their values. The sources are Eyre et al (1997), EUNET (as reported in Grant-Muller et al (2001)), ECMT (1998), Eriksen et al (1999) and SIKA (2000).
Prices are 1995-1999 prices and it is assumed that 1 euro = 0.625 pounds = 8.25 NOK
= 8.80 SEK. The value of life is assumed to be 2 million pounds in the Eyre et al study.
Table 12.3. Costs in Euro per emitted kilo of pollutants from transport
SO2 NOx VOC PM10
Urban Rural Urban Rural Urban Rural Urban Small town Rural
Eyre 52 7 13 9 3 3 92 14
EUNET 1.7 4.5 185*
ECMT 8 4 8 4 0
Eriksen 9 2 8 4 8 4 206 25 0
SIKA 27 2.3 10 6.8 8.5 3.4 864 216 0
* PM2.5
The "rural" category of Eyre et al is probably somewhere in-between the "small town"
and "rural" categories of Eriksen et al, which makes the PM values comparable. An explanation of the high "urban" PM value of Eriksen et al might perhaps be the considerable problems in Norway with high levels of PM, not so much because of exhaust emissions as because of studded tyres. The NOx and VOC values of Eriksen et
al are taken from ECMT. The urban values shown for SIKA apply to Stockholm inner city, which might explain the very high PM10 value. The SIKA small town value of PM10 applies to Stockholm’s surroundings, but is very similar to the corresponding value for a small to medium sized town.
Some other valuable sources of unit costs of air pollution will be Nellthorp et al (1998) and Watkins et al (2001).
12.3.3 Non-transport sources of emission and backward linkages
Modelling total emissions in the city is beyond the scope of the rather coarse strategic planning that we consider in this guidebook. Considering total emissions could involve environmental input-output modelling. In particular, that would be useful if we intend to apply a dispersion model to draw detailed conclusions about the local air quality or noise in each part of the city, or to study environmental equity impacts.
The strategies we are testing and assessing involve only land use and transport policies. Unless such policies are very different from each other with respect to the amounts of money that consumers must use for transport and housing, they will not influence general consumption of other goods and services very much. The level of general consumption is rather given by the scenario assumptions. So except for differences due to energy use and emissions from housing and transport, the indicators from environmental input-output analysis will mainly be indicators of the sustainability of scenarios, not strategies. Furthermore, as long as the stock of houses or the forms of energy used in homes do not change much, emissions from housing will also be fairly constant, even if households relocate within the given stock of houses.
Consequently, as a first approximation, we propose to take non-transport emission as exogenously given. The question then arises if we should somehow take account of backward linkages (the emissions from the production of fuel and vehicles) in our transport and land use planning, or if they can be ignored altogether.
With respect to greenhouse gas emissions and energy use, we want in principle to include life-cycle emissions, since the point of emission does not matter. But except for the greenhouse gases, any indicators of emissions from backward linkages must be kept separate from the direct emissions. The reason is that we do not know where these emissions occur. Probably they are irrelevant for dispersion modelling and for the assessment of the costs of local and regional air pollution in the particular city we are studying.
By consulting sources such as Hickman (1999) or Eyre et al (1997), we can fairly easily find the emissions due to the production of the fuels. Estimates of life-cycle costs of the production of cars and houses can also be found in the literature.
However, considering the other challenges of establishing environmental indicators, except for CO2, we do not make it a priority to keep track of these backward linkages at present.
If dispersion models are used, there will be a need to include emissions from the local production sector in them. Thus we cannot totally abstract from the production sector and the question of what industries are located where.