For lack of data on actual pollution concentration in areas near roads. I need to do an estimate of the air quality in these areas. All I have is traffic data. Can anyone lead me to a study, perhaps using linear regression to estimate air pollutant concentration (e.g. PM10, NO2, CO, etc.) with traffic data as predictor, preferably the average annual daily traffic (AADT)?
Yes, you can always do an estimate of ambient pollution levels. Lack of data just leads to poor accuracy of your estimate. Modelling ambient pollution requires three models: emission, transmission and immission.
Traffic emissions are driven by number of vehicles and specific emission per vehicle. The specific emission is governed by vehicle type and driving situation, e.g. speed, gradient, traffic disturbances (free flowing vs. stop+go or urban traffic).
Transmission describes the dilution of pollutants between emittor and receptor. For traffic pollution, transmission is governed by wind speed, direction and atmospheric turbulence. Buildings close to the road may play an important role. For some pollutants, a chemical reaction may have to be considered.
Immission is the sum of the transsmitted pollution and the background pollution level. The background pollution includes emission from sources that are not modelled explicitly. For traffic pollutants, this may include other traffic emissions as well as industrial or household sources. In an ideal model you include all sources or you have data from local measurements. And the immission part includes the statistical evaluation of the pollution level - annual mean, 95-percentile of hourly mean values etc.
For emission and transmission models there is some data/software available in the public domain. But it really depends on the dispersion situation and pollutant if the data suits your question. For wind data and background pollution, you'll need local information that might be quite difficult to find.
I fully agree with Ingo Riess ... It is always possible to make an estimate...
However, emission data-base needs to be taken carefully when it is derived for a road (or city) and then applied to other roads (or cities). For example, in some countries (for example middle east and low-income countries) the car conditions and service are not well-controlled. Therefore, emissions do not match what is already derived for EU and US conditions, where most likely car service and conditions are well-controlled based on well-defined protocols...
In order to tackle this lack of information about emissions, you can build up information learning about car emissions. For instance, applying long-term monitoring and intensive measurement campaign can provide you with air quality data. Combine this air quality data and weather conditions and assume that the emissions trend is not changing significantly during a 24 months, then you can run probabilistic modeling to predict air quality based on weather predictions.
We have tested that in several cities. The air quality trends can in cities is very well-connected to the traffic activities. Therefore, it is easy to model based on traffic pattern.
See for example:
[1] Alghamdi MA, Al-Hunaiti A, Arar S, Khoder M, Abdelmaksoud AS, Al-Jeelani H, Lihavainen H, Hyvärinen A, Shabbaj II, Almehmadi FM, Zaidan MA, Hussein T, Dada L. A predictive model for steady state ozone concentration at an urban-coastal site. International Journal of Environmental Research and Public Health 2019, 16, 256.
[2] Mølgaard B, Birmili W, Clifford S, Massling A, Eleftheriadis K, Norman M, Vratolis S, Wehner B, Corander J, Hämeri K, Hussein T. Evaluation of a statistical forecast model for size-fractionated urban particle number concentrations using data from five European cities. Journal of Aerosol Science 2013, 66: 96–110.
[3] Mølgaard B, Hussein T, Corander J, Hämeri K. Forecasting Size-Fractionated Particle Number Concentrations in the Urban Atmosphere. Atmospheric Environment 2012, 46: 155–163.
[4] Clifford S, Low Choy S, Hussein T, Mengersen K, Morawska L. Using the generalised additive model to model the particle number count of ultrafine particles. Atmospheric Environment 2011, 45: 5934–5945.
[5] Hussein T, Karppinen A, Kukkonen J, Härkonen J, Aalto PP, Hämeri K, Kerminen V-M, Kulmala M. Meteorological dependence of size fractionated number concentrations of urban aerosol particles. Atmospheric Environment 2006, 40: 1427–1440.@
Agree with Ingo Riess. The trick here is that you will need a well documented fleet emission factor for your traffic data. Beside that you will need a validated dispersion model to provide you the rate of dispersion.Our group is currently doing long term near roadway monitoring. In 2019, we found that both emission and dispersion are a function of vehicle flow rate and driving condition. Please check our recent publication for more information.
This doens't fully address your question (ambien air quality), but here's a relevant study that uses traffic GPS and camera data, as well as emission factors and other data sources, to estimate tailpipe emissions in a whole city ( Article Spatial-temporal inference of urban traffic emissions based ...
). Relating emissions to pollutant concentration is a different story, as traffic alone doesn't fully account for air quality in a complex urban environment. Perhaps mobile air quality sensing could offer better observations of the spatial-temporal variations of air quality with traffic being the primary drive (Preprint Deep-MAPS: Machine Learning based Mobile Air Pollution Sensing
).- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India