Urban air pollution policy commonly utilises averages. Daily means, annual objectives, and longer-term monitoring data are all important indicators of general air quality and can be used to compare trends, evaluate standards and measure broad policy changes.
Averages are, however, only part of the picture.
Street-scale conditions, and thus exposure, can change within quite short time periods. Short-duration congestion, changes in wind direction, traffic queues, an incident at a junction or a short duration of poor ventilation within a street canyon may each produce significant short-duration increases in concentrations.
While such short-term changes may only be important for a few minutes, they may still be of concern to people in the immediate vicinity.
Peak exposure therefore should not be dismissed as of peripheral concern; rather it is a component of the evidence base required to understand the exposure to pollution experienced within the urban environment.
Averages Can Mask Important Episodes
Averages are useful because they reduce a data set to a single measure. They can allow comparisons between locations to be made, changes over time to be identified and compliance with standards to be checked. But they can also mask short-term variation.
A street may have a mean concentration within the target but may still have brief spikes of higher pollution that occur at different times during the day. The hourly value from a monitoring site may meet an objective yet, during a few minutes of peak congestion or poor dispersion, nearby individuals are exposed to higher concentrations.
A pollution model may replicate a pattern but fail to reproduce the short-duration peaks that are of relevance to street-level concentrations.
This is particularly important at the street scale, especially in dense urban areas, where buildings and traffic interact with local meteorology to influence the movement of pollutants through the street network. Pollutants do not move evenly across the street. Concentrations can be locally high, disperse and recirculate within or move from the local street network, in an unpredictable way, depending on the wind and other local flow conditions.
The short duration of peaks of pollution experienced at the street level therefore means that long-term monitoring of concentration levels alone cannot tell us the exposure of individuals to peaks of pollution. We need more information on how and why short-duration peaks may occur.
The Street Scale Matters
At city level, policy is often focused on understanding whether levels of pollutants are increasing or decreasing. At street scale, the questions are different. When are people exposed to air pollution, where do people experience higher concentrations and under what conditions?
Waiting to catch a bus at a bus stop can result in a short-term episode that might be different to that experienced when walking down a side street. A cyclist that passes through an intersection may experience a peak that would not be captured from an average value in that area.
A building entrance, a pedestrian crossing or a traffic light could all be locations of interest because individuals may remain in the street environment long enough for the short duration to be relevant.
That is precisely why we must build a complete picture of urban pollution from multiple lines of evidence. Only by combining results from field campaigns, monitoring networks, tracer studies, models, and controlled experiments will we be able to piece together a clear and comprehensive view of what drives peak exposure.
Peak exposure occurs specifically at the intersection of these different sources. Knowing the average exposure level, or even the average pollution intensity across an entire study site, is not sufficient. We need to know why peak events have occurred, and what we might expect regarding future peaks.
Potential Triggers for Peak Exposure
A number of different factors may interact to drive peaks. The obvious driver is traffic. Vehicles may be idling, accelerating, braking, or queuing at junctions. Emissions are not uniform in time or space. A short burst of congestion can cause peaks in exposure close to where people are walking or cycling.
Another driver is meteorology. Even a modest change in wind direction may mean that the pollution plume flows right along a street, is diverted into an adjacent side street, or stays trapped within the street canyon. Low wind speeds can reduce dilution of pollutants in a local area, or certain wind directions can create localised recirculation close to buildings in narrow streets and roads.
Urban geometry is also key. The street canyon, intersections, corners of buildings, and gaps between buildings can all direct air flows that determine where a pollution plume will travel, potentially causing a peak in exposure not necessarily where the source is at its closest, but where the air flow directs the plume.
This is part of the wider reason why street-level air pollution is difficult to predict: the local built environment can change both the direction and intensity of exposure.
We must also consider how long the emission continues for. In the context of an emergency response, an airborne release of short duration might lead to very short-duration conditions of exposure. For regular urban air quality, short-duration releases are not as severe as those of an emergency scenario, but they are nonetheless an important aspect of local exposure levels.
Short-Term Exposure: Monitoring Challenges
We need to understand short-term peaks to assess how well air quality is meeting legal standards, and whether the environment is providing a healthy place for people to walk, cycle, run or drive.
The standard monitoring site measures the concentration of pollutants in the air at the fixed location of the monitor. This will give a reliable indication of local levels but not necessarily peak levels across the localised area of an intersection or adjacent streets.
A network of closely spaced monitors in a small area will provide a better view of how concentrations change across space and time, but it does not provide a full answer either.
This means that if we were to record high pollutant concentrations on a fixed monitor, we would not necessarily be able to tell whether that high concentration reflects a specific pathway of the pollutant plume, the passing of a source, or the local effects of short-term recirculation in an area. The concentration of pollutants at another point, in the immediate vicinity, may well be lower.
Well-designed sensor networks can help to better understand and assess the concentration gradients across space and time. They provide additional information to tell us how pollutants behave and interact across an area.
A sensor network can tell us if that peak is occurring repeatedly at that spot, is related to a specific direction of wind, or is just one event in a general pattern of high pollution across the surrounding area. There is, therefore, the possibility that we may need a larger network of sensors to better assess local air quality levels.
But it is important to remember that adding sensors is not the complete answer. The location of the sensors in question, the calibration and response of those sensors, and how we are going to use their data all come into play.
To determine the effectiveness of a specific sensor network, we need to ask what its specific objective is, such as to detect the localised peaks mentioned above, to evaluate the pathways of air pollution around a building, to evaluate whether model predictions of local air quality are accurate, or even if it is needed to support emergency response to pollution incidents.
Why Models Fall Short at Peak Events
Predicting the broader contours of a pollution plume is relatively straightforward for a model. Predicting a short-lived peak is not. In part, this is because peak exposure depends on detail, which may be hard to model accurately.
Details like precise source timing, local traffic, local turbulence, street morphology and rapidly varying meteorological conditions all matter for predicting peak exposure.
A model might predict average concentration well while missing a peak near a junction. It might predict plume direction but not plume deviation into a side street. It might miss variability, averaging it out to the point that short-duration exposure disappears from the results.
This is why model validation becomes a policy issue. If a model is being used to inform decisions around exposure assessment, planning or emergency response, it should be validated against evidence that reflects the time and space scales relevant to the question.
Model validation should ask if a model can predict peaks as well as average concentrations. It should also ask if the model is being used for a type of decision it should be used for.
Field Evidence and Controlled Experiments
Because of the short duration of events and changing conditions, it is harder to study peak exposure. We need field measurements to capture peak exposure in the real urban environment, but it is hard to replicate the same field conditions.
Field measurement might show that peak exposure was driven by traffic, or wind direction, or source position, or ambient concentration, or building effects. Without additional evidence, it is hard to know.
That is where controlled wind tunnel experiments help. We can test the effects of source position, wind direction and urban form in a controlled environment. The experiments can tell us if peaks are likely to appear, and why.
Wind tunnel experiments are not meant to replace field evidence but to help us identify physical mechanisms that might produce peak exposure.
Digital models also have a role to play. A combination of physical and digital modelling can help us evaluate uncertainty, test scenarios and explore conditions that we cannot easily measure all at once.
If we are to study and predict peak exposure, evidence generated from combining field observations, wind tunnel experiments and modelling is likely to be more convincing than any of those things in isolation.
Peak Exposure and Public Space
Peak exposure matters in particular places where people might be spending time or might pause.
A short pollution event might be unimportant for the road where a pedestrian walks by but important for the bus stop where someone waits three minutes. A junction may matter because pedestrians pause for longer to cross the road. A building entrance may matter either because people linger there or because of the building’s effects on the wind flow around the entrance.
It is hard to say a priori which places are most relevant, but transport interchanges, school entrances, narrow pedestrian areas and active travel routes may all be of concern.
Much of the urban policy is directed to emissions, and this is as it should be. Reducing emissions is only one aspect of the problem; the design of streets, the control of traffic and where people walk, wait and cycle will also affect exposure.
It does not mean that all short-term peaks should necessarily be mitigated; only that local design and management decisions should be based on evidence about where such peaks are likely to occur and who will be affected.
Implications for Urban Policy
There are several implications for urban air quality policy arising from peak exposure.
- Monitoring strategies should not rely solely on averaging. Averaging is still important but should be supplemented with evidence that identifies short-term and spatially localised events.
- Traffic management should consider time and location. Short-term exposure at a pedestrian area may be influenced by congestion, queuing and acceleration in the vicinity regardless of whether broader emissions trends are improving.
- Urban design should consider ventilation and local dispersion. Form, street width, junction layout and enclosure of spaces all play a part in determining whether pollutants dilute or accumulate.
- Models underpinning policy decisions should be validated at the scale of the decision. A model used to represent street-level exposure will require different evidence than a model used for screening a whole city.
- Communication should be transparent about uncertainty. Peak exposure will always be uncertain because it varies in time. Those using policy information should have a clear idea of what is known, what is still uncertain and the potential effect that uncertainty might have.
From Short Events to Better Decisions
Short-term events matter because people live with air pollution in real time, real space and real environments. They do not have direct access to an annual average exposure. What they are exposed to in practice is the air at a crossing, in a street, near a building, near traffic or along a corridor across a city.
This does not render the value of long-term averages null and void. They are still important for policy. They should not be the only form of evidence, however, that helps us to better understand exposure to urban pollution.
Peak exposure can add to the available body of evidence. It highlights the role of timing, pathways and local conditions in determining exposure. It also encourages greater integration between monitoring, modelling, field campaigns and controlled experiments.
Urban policy is better informed when it has regard for the average but also the short-term event. The aim is not to make every exposure event known and understood in detail. It is to build enough knowledge to support improved policy on where and when people are exposed in urban streets and junctions, buildings and open spaces.
This is why peak exposure matters: the city is not experienced as an average. It is experienced, second by second, where people live, work, wait and travel.
