Air quality assessments: garbage in, gospel out?

The UK’s new housing secretary, Steve Reed, has vowed to “build baby build” with a promise to build 1.5 million homes.

He even gave away free Maga-style baseball caps featuring the three-word motif.

This will be achieved, he says, “Through major planning reform and investment.” Adding: “We will break down the barriers to development.”

The major concern, therefore, is that one of these barriers will be environmental protection.

The impact of development

More housing creates more nutrient pollution in rivers and groundwater. But the focus of this column will be the impact of development (not just housing) on air quality.

Here, the real danger is that the planning system is already skewed in favour of development. Developers can easily generate models that indicate “no significant harm” to air quality.

The problem with air quality models is that they predict a post-development value for a pollutant, often with 1 or even 0.1 µg/m3 resolution. That's despite the fact that the inputs to the models are riddled with high levels of uncertainty.

Air Quality Assessments (AQAs) are necessary for proposals that represent a potentially significant threat to air quality.

They therefore form part of development proposals for large projects, for certain industrial processes and where there are sensitive receptors or an existing Air Quality Management Area.

Data for an AQA are drawn from multiple sources. But each of these comes with its own level of uncertainty.

When combined in a model that predicts air quality, these uncertainties are amplified.

However, the true levels of uncertainty in the predicted levels of say NO2 or PM2.5 are rarely, if ever, quoted.

Members of the public, councillors and even planners can be forgiven for believing that such predictions are ‘accurate’. So, what are the sources of error?

Sources of error

1. Some authorities benefit from (expensive) real-time reference monitoring stations such as the Automatic Urban and Rural Network (AURN). But these are sparsely located, fixed in position and are frequently not sited in the air quality locations of greatest concern. Reference monitors also have levels of uncertainty, which vary according to factors such as parameter, measurement method, sampling system, calibration etc. For example, there is a CEN standard for the chemiluminescent method (EN 14211:2005 Ambient air quality – Standard method for the measurement of the concentration of nitrogen dioxide and nitrogen monoxide) and its uncertainty is typically quoted as ± 15%
2. Frequently, monitoring data is drawn from passive diffusion tubes (PDTs), which DEFRA says: “May exhibit substantial under- or overestimation compared to the reference method.” As a consequence, accuracy is ± 25%.
   In an age when everyone has become accustomed to instant information, it is alarming that most local authorities still rely on a 1970s monitoring technology that can take many months to deliver data – largely because DEFRA has to apply a National Diffusion Tube Bias Adjustment (fudge) factor.
   PDTs only measure one parameter, usually NO2, and only provide a monthly average. So, they are useless for the determination of short-term exposure. Their main advantage is, guess what, they are cheap!
3. Background air quality data is drawn from DEFRA maps that have a 1km square resolution. Hardly ideal for modelling highly localised pollution hot-spots
4. Weather data is drawn from the nearest weather station, which can be many miles away, is therefore unlikely to be representative
5. Air quality consultants can choose Receptor locations at which to model air quality. But these may not be the locations with the most people, or where the air quality is likely to be the worst. The concern is therefore that consultants may choose Receptor locations that portray better air quality
6. Traffic models (with estimated vehicle emissions) have their own levels of uncertainty. These also feed into air quality models.

Summary

As a consequence of all of this accumulated uncertainty, developers’ AQAs should be treated with an appropriate level of cynicism.

In addition, it is important to keep in mind that air quality predictions are usually compared with National Air Quality Objectives (NAQOs). But even with these targets, 28-36,000 people are dying prematurely every year from air pollution.

It is well known that health effects occur below the NAQOs. So we cannot afford complacency.

Many will dismiss air quality concerns, claiming that the electrification of the fleet will resolve air pollution.

However, electrification will take many years. Also, electric vehicles are generally heavier, and have greater instantaneous torque. We are only just starting to understand the health and ecotoxicity effects of tyre and brake emissions, many of which are ultrafine and currently unregulated.

What’s the solution?

1. Local Authorities have a statutory duty to monitor, assess, and act to improve local air quality. But in many cases their power to do so is severely limited by a lack of funds. In the past, many relied on DEFRA’s air quality grant scheme, which was suspended last year and has still not been replaced. So, DEFRA needs to provide adequate funding for air quality monitoring
2. DEFRA needs to embrace so-called ‘low-cost sensors’. Like PDTs, these are also ‘indicative’ monitors. However, independent trials and certification schemes such as MCERTS have demonstrated the reliability and accuracy of the instruments from several manufacturers. These small devices are easily located in the pollution hot-spots. They are able to measure multiple parameters continuously and deliver data in real-time, so that maximum value can be extracted from the readings
3. AQAs should clearly define the accumulated levels of uncertainty in their models
4. Continuous monitoring should underpin AQAs to establish a baseline for post-development to evaluate predictions
5. The rules for the selection of AQA Receptors should be tighter and clearer
6. Planners should impose an air quality monitoring requirement on any approved development that presents a potential threat to air quality. The developer should bear the cost of doing so.

Conclusions

The trouble with modelling is that it can conceal a multitude of sins, which, in the case of AQAs, can be multiple sources of error.

The dilemma for members of the public, planning committee members and even planners, is that air quality is a technical matter. So it is vitally important that clear, accurate, reliable information is available.

One of the ways to achieve this is with continuous air quality monitoring with publicly available real-time data.

Twenty years ago, this would have been horrendously expensive. But with the development of low-cost sensors, there is an enormous opportunity for monitoring to become so much better.

London has proved this with the Breathe London network of hundreds of monitors. It is about time that every town and city had the funding to do the same.

DEFRA can obviously help. But there is an enormous opportunity for developers to fund air quality monitoring.

If they are saying they won’t harm air quality…let’s make them prove it!

NB. Apologies and thanks go to Dr Ashley Mills and Prof. Stephen Peckham for allowing me to plagiarise the title of their Paper: Garbage in, gospel out? – Air quality assessment in the UK planning system. Environmental Science and Policy, 101. pp. 211-220. ISSN 1462-9011