We can’t simply predict air pollution at street level based on the volume of pollution emitted. In a city, this depends on the buildings, traffic, wind, turbulence, street configuration and people’s movement through urban space.
This is why street-level air pollution can change quickly over short distances and time-scales: two close streets can experience different levels of pollution from the same nearby traffic route; a junction can behave differently than a straight road; and a momentary shift in wind direction can change where pollution accumulates, disperses or recirculates.
This difficult-to-predict element was one of the drivers behind DAPPLE’s early work on urban dispersion, which explored the interaction of air, traffic, people and pollutants in an urban context.
Pollutants Do Not Disperse Uniformly Through the Street Network
At city scale, air pollution is relatively even. It is higher close to busy roads, lower in open spaces, and modified to some extent by general regional weather conditions. Street scale is a different matter.
Buildings affect air movement. Roads affect the movement of emissions. A junction involves moving traffic turning at road junctions, stationary traffic, pedestrian movement and complex turbulence. In other words, in the city street a local situation may channel, dilute, hold, mix, displace or move pollutants in a range of specific conditions.
Street canyons are a simple example: air trapped between rows of buildings does not behave like the air over open land. Wind can blow over rooftops while lower down, the air is slower and more turbulent. Pollution can therefore linger at road surface level inside the canyon, at least for some time.
So a question that is central to prediction is not just how much pollution is present, but what happens to pollutants once released?
Buildings Modify the Flow Field
Buildings affect the wind flow. They do not just block it. In a street, air has to move around, accelerate and separate, form wakes and recirculation zones, and in an urban environment all these elements occur simultaneously.
The influence of a single street on a city block is modified by its own roof height, gaps between buildings, streets running in at different angles, courtyards and buildings behind them, junctions, trees, pedestrian furniture and other local features.
This is why it is crucial that urban dispersion studies examine the flow field first. Without this, we do not fully understand where and how pollutants move.
DAPPLE therefore focused on street and neighbourhood scale flows, using a combination of field measurements, wind tunnel and computer modelling experiments to study the flow of air and the dispersion of pollutants in it.
Traffic Is a Moving, Rather Than a Static Source of Pollution
Traffic is not static; it varies, accelerates, idles, brakes and turns. These emissions are affected by speed, congestion, vehicle type and drivers’ behavior, the latter being potentially significant as they may influence turbulence at the surface.
At a junction, the complexity is increased as vehicles may wait at traffic lights, accelerate away from a stop or turn into a side road. Emissions are therefore not constant in time or in space.
As a consequence, traffic emissions are not uniform. They vary at the micro scale, and a general traffic count may not adequately describe peaks. A prediction model may be accurate enough for general pollution but may not describe exposure that occurs during congestion, or that depends on traffic lights or variable traffic flow.
This problem was addressed in DAPPLE-EPSRC with research on vehicle movement and detailed local emission inventories. In this manner, traffic emissions were incorporated into the model as a dynamic variable rather than static background input.
The Same Street Changes Under Different Weather Conditions
The same street in the city changes its behavior under different weather conditions. For example, a wind along a street can transport pollutants along the street, while a wind perpendicular to it will lead to recirculation within the street canyon.
Furthermore, the transport of pollutants along a road will occur at low speeds of the air, while pollutants may increase at corners, junctions and near buildings, as dispersion is favored at higher speeds.
Weather parameters, such as temperature, air stability, atmospheric turbulence and boundary layer conditions, also contribute to the dispersion of pollutants. Weather conditions that occur above the roof can be very different from those that occur below. A measurement of conditions at a roof or tower, while providing useful information, does not represent local conditions at the street level.
In the later DAPPLE-HO project, this issue was addressed by looking at short-range dispersion, urban canopy flow and local meteorological conditions to find conditions that may lead to very high exposure.
Exposure Depends on Where People Actually Are
While air pollution modeling focuses on concentration fields, the interest of public policy is in exposure, and a concentration map provides a first step. However, exposure is a complex process since people do not stay in one location for long time periods.
People travel in the street, through side streets, through junctions, wait at a bus stop or cross a road to work in an office. All these people will be exposed to different conditions in time and space.
This is why street-level exposure is difficult: one has to consider not just pollutant transport but also the interaction with human behavior, including an understanding of what places people visit, how long they stay there, and the routes they use.
This problem was addressed within DAPPLE by including exposure measurements made by project participants, as well as fixed instruments and traffic data. This was important since fixed instrumentation alone cannot capture what happens in the dynamic city.
Short-Term Peaks Are Hard to Capture
Policy makers rely on averages; hourly, daily and annual average data are common. These are useful averages but can obscure short-term peaks.
Peaks may also be important at street level in the urban environment. A momentary change in signal timing, a brief period of congestion, a vehicle queue, a temporary change in wind direction or a release may all cause exposure conditions which are not represented in an average over time.
This was one of the key areas of investigation explored in the original DAPPLE project: causes of peak exposure and the actions available to mitigate peak exposure.
The issue of peaks is important not only because of technical difficulties, but because the way in which data are measured and used can change significantly if short-term peaks are not taken into account. This affects monitoring interpretation, model evaluation, and how urban design or traffic decisions are made.
For policy purposes, this raises the question of whether to use broad averages alone or whether to be aware that there may be other exposure conditions of more local significance which may be most important for people in the street.
Field Measurements Are Necessary but Limited
Field measurements provide the most direct evidence from the real urban environment. Field measurements can capture elements which are not possible to recreate under laboratory conditions: actual traffic conditions, real weather, realistic building shapes, pedestrian movement and real local background pollution levels.
However, field measurements cannot recreate ideal conditions. Wind direction, wind speed, traffic and background concentrations all vary at street level. Repeat measurements can never be made in exactly the same circumstances.
In terms of model development this leads to uncertainty. If a model fails to reproduce a field data set, this may be due to a modelling error, an uncertainty in the emission factors, changes in the local meteorology or a local effect not captured in the input data.
Therefore, in DAPPLE field campaigns, field measurements were considered alongside other forms of evidence such as wind tunnel and numerical modelling.
Wind Tunnel Studies Add Control
In the case of wind tunnel modelling, the flow and dispersion over an urban area can be controlled, as a physical model of the study region is placed under repeatable wind conditions. This provides the advantage of multiple repeat experiments from which repeat data may be obtained.
Repeat experiments enable a clearer picture to be drawn as to the dispersion processes involved and to determine which features are important. This is of particular value in the study of short-range dispersion, where it may be expected that concentration values can differ significantly between repeat releases.
However, wind tunnel models are limited in that not all real-world elements can be recreated. Pedestrian movement, thermal effects, changes in background concentration and full-scale urban features may be absent or simplified.
This emphasises the need for evidence from field measurements to be considered together with evidence from wind tunnel modelling. This is an integral element in DAPPLE and is also one of the areas of focus in the broader physical modelling work developed through PHYSMOD 2013.
Models Must Deal With Uncertainty
The need for modelling comes from the necessity to make predictions and assessments. Not all measurements can be made at all times and in all places. Models are required to test different scenarios, predict concentrations, assess impacts, and inform decisions about policy, planning or emergency situations.
However, urban dispersion models function in an uncertain environment. Inherent in them is input data on emissions, buildings, meteorology, traffic and boundary conditions. They must also represent real physical processes using various degrees of simplification.
Some models are quite detailed, such as those applying Computational Fluid Dynamics or CFD, while others are simpler, using more empirical formulations. Each kind of modelling tool has its strengths and limitations.
The DAPPLE technical documentation referred to an entire suite of models, including both RANS CFD and LES modelling, canopy modelling, traffic modelling and simple empirical methods.
The key question is not which model is the most accurate in some general sense, but which is appropriate to the particular decision or issue being addressed. A model to be used in a broad-scale screening exercise does not need the same level of detail as a model which is to be used for assessing short-term exposure in the immediate vicinity of a road junction.
A model to be used for emergency response may need speed and robustness as much as detailed accuracy.
Implications for Urban Air Pollution Policy
Predicting urban air pollution at the street scale is difficult because cities are very complex environments. The physical nature of a city in relation to air pollution involves buildings of diverse size and shapes, traffic flows which can generate both primary pollutants and modify flow conditions, and people who are moving within the same space.
Cities are also physically complex in terms of air flow because of local meteorological conditions. Urban emissions enter an air flow environment that changes rapidly both in time and space, through wind and building flows, but also through local changes in temperature and stability. Urban air pollution is experienced by people who also live and move around a city in dynamic ways.
There are some implications for policy which flow from the complexity of urban air pollution prediction:
- Urban air pollution monitoring is not always straightforward. The measurement of air pollution at a single fixed site in a city may yield valuable data about pollution levels, but because of the dynamic nature of urban pollution, it may not represent a particular street in that city or the way in which individuals in that street are actually exposed.
- Traffic and urban planning decisions should consider local dispersion. Policy decisions can take into account the specific ways in which pollutants are emitted, their interactions with air flow, and how they are subsequently experienced as personal exposure.
- Model use should be supported by model validation. Models need to be validated by comparison with field and laboratory evidence. If those models are to be used for policy, emergency planning or design purposes, then it is essential that they are well evaluated before being used to inform decisions.
- Short-term and local urban exposures should not be neglected. In many cities around the world, people are not only subject to city-scale pollution levels but are also exposed at specific locations to pollution generated in the immediate locality.
Understanding these short-term, local exposures is as important as the city-wide concentrations because people are subject to a variety of local pollution environments as they move through cities.
A Continuing Research Challenge
The problem of predicting air pollution at the street scale is not one which can be solved by improved measurement or more sophisticated models alone, but arises from the complexity of the urban environment itself.
DAPPLE sought to tackle the issues of urban air pollution at this scale through direct and complementary studies: field campaigns involving air pollution and wind measurements, field studies involving tracer measurements of dispersion in urban street canyons, measurements of exposure using mobile monitoring techniques, and the use of a wide variety of computational models of urban air pollution dispersion.
The reason that the DAPPLE research programme was able to make advances in street-level air pollution prediction is because it considered this problem at several different scales simultaneously: at the level of the urban district and at the level of the street.
The challenge for urban air pollution prediction is not going away. It is a reflection of the challenge that will be faced in predicting the dispersion of pollutants in any complex environment. However, advances in measurement of air pollution levels, improved modelling techniques, and better availability of information on buildings and traffic flow will all contribute to further improvements in urban air pollution models.
The central issue, however, is not only about the tools to be used but the questions which need to be answered in order to understand how emissions of air pollution interact with urban air flows and urban forms, and how people may be exposed to those emissions.
Urban air pollution research needs to be a technical research programme, but it also needs to have an impact upon the policy agenda. Understanding and predicting urban air pollution at the street scale is not simply about having better maps. It is about having the ability to make better informed and safer decisions in places where people actually live, work and move around.
