Air quality represents a big challenge for our society. It is well known that ambient air pollution has an impact on human health and ecosystems. Air quality mapping is a public information vector and enables action levers to be identified to reduce pollutant emissions. The reference stations operated by the Approved Air Quality Monitoring Associations (AASQA) allow for reliable and standardised observations of pollutant concentrations, but their deployment over a vast territory such as France remains subject to constraints of cost and maintenance which limit the coverage of this network. Where there are no measurements, models are generally used to calculate pollutant concentrations at a regional and local level. They take pollutant emissions (traffic, industry, agriculture, residential...) and meteorological conditions into account in order to estimate in a more or less complex way the dispersion and chemical transformation of the pollutants. Nevertheless, these models present uncertainties because of the quality of the input data, their spatial resolution or imperfect representation of the processes; they should therefore be corrected using observations as much as possible.
Low-cost sensors for air quality measurement
In the last 5 years, new miniaturised and low-cost measurement instruments have made their appearance. These sensors are being used more and more frequently today. They can be deployed in great numbers to obtain an observation of the spatio-temporal variability of pollutants with a high degree of resolution. They can be installed on street furniture to provide fixed measurements, or on service vehicles, ambulances, driving school cars or bicycles to measure the concentrations of pollutants while in motion. This then opens new horizons to improve air quality mapping at an urban level. However, using the sensors raises a great number of challenges. Because of their miniaturisation and their simplified metrology, these sensors are associated with greater measurement uncertainty, sometimes higher by one order of magnitude than those taken in reference stations. To this is added the effect of mobility whose impact on the measurements is yet to be studied in depth. Furthermore, the quantity of data produced means that new data processing methods are needed, reliant on big data.
Considering the challenges linked to the emergence of these new technologies, the National Institute for Industrial Environment and Risks and the National Metrology and Testing Laboratory (LNE, Laboratoire national de métrologie et d'essais) have committed to developing an "AIR Quality Sensor" voluntary certification process to validate the level of metrological performance of the sensors in accordance with regulatory criteria.
The SESAM tool
The merging of low-cost sensor data and simulations from dispersion models is a unique opportunity to improve the mapping of air quality at an urban level. For the first time, the SESAM tool makes it possible to merge the observations from fixed and mobile sensors with the estimates of an urban model, whilst taking the intrinsic uncertainty of the sensors into account in the mapping. A SESAM tool application was developed in Nantes by merging data from sensors installed by AtmoTrack with output from the ADMS-Urban dispersion model, supplied by Air Pays de la Loire. As illustrated on the diagram below, based on a first estimate across the area (here an annual average), the application makes it possible to take into account data reported at a given moment by sensors on the ground. This application was the subject of a scientific publication in the Environment International journal. A user guide for the tool has also just been published by the Central Laboratory for Monitoring Air Quality (LCSQA). The development efforts continue to sustain innovation in this direction with a thesis underway at the National Institute for Industrial Environment and Risks, co-directed with the Mines ParisTech Geostatic Centre.
The SESAM tool is in application in Nantes, using sensors placed on service vehicles. These vehicles moved about and measured at roughly 10-second intervals the concentration of fine particles in ambient air between 8 and 9 a.m. of the morning of 29 November 2018. The concentrations thus measured and the calculations of the dispersion model (here the 2016 annual average presented in the background) are merged to obtain a more precise mapping of PM10 concentrations in the Nantes conurbation.