PREV'AIR pollution

Introduction >> Modèle

Air quality modeling

The three-day forecasts and air quality maps published on a daily basis on the PREV’AIR server are the result of numerical simulations carried out with the help of so-called 3D eulerian deterministic models ("chemistry-transport" models). For periods of time ranging from several days to several months, these tools allow to calculate changes in photochemical and specific pollution in the lower layer of the atmosphere, on different spatial scales.

The models used in the PREV’AIR system have been developed by the project’s partners:

Three-dimensional Deterministic Modelling

A three-dimensional area includes the low troposphere above the region being studied. The changes in pollutant concentrations in this three-dimensional area over the chosen period - influenced by weather conditions and emissions of pollutants into the atmosphere - are calculated in a deterministic manner, i.e., by linking temporal variations in such concentrations to:

Numerical Resolution

A system of partial differential equations (PDE) translates - or "models" - this complex of phenomena into mathematical terms.

The PDE system that describes the transport and the physics-chemistry of atmospheric pollution is then numerically solved on calculating machines, with the help of a numerical scheme adapted to the type of PDE to be processed.

In the eulerian three-dimensionnal approach, EDPs are projected in each spatial direction..

The numerical resolution of the EDPs entails spatial discretisation of the considered three-dimensional area: this is described by a vertical and horizontal grid with a spatial resolution that depends on the following factors:

  • Size of the area,
  • Properties - particularly the lifespan - of the pollutants,
  • Computer resources (performance of the calculating machines, as regards operating speed but also as regards storage of results),
  • input data (particularly emission registers).

The concentrations calculated by the models in each grid cell are the average concentrations that would be observed if there were a perfect mix in each sector, which is rarely the case.

Likewise, numerical resolution of the PDEs entails temporal discretisation of the period being studied. Changes in concentrations of pollutants are calculated with a temporal resolution (defined by the time step) that depends on the spatial grid, the duration of the period in question and, obviously, on the properties of the pollutants, the computer resources and the input data.

Physical-Chemical Processes

In the deterministic chemistry-transport models used within the context of the PREV’AIR system, changes in pollutant concentrations over time are calculated by linking the variation of pollutant concentrations in the area over time to physical-chemical processes that increase (production processes) or reduce (loss processes) the concentration of a chemical compound in the atmosphere.

For example, the following physical-chemical processes are used in the CHIMERE-Continental model:

  • Horizontal dispersion of chemical compounds due to the horizontal component of the wind. This process is important for a compound like ozone, which can be transported over distances of several hundred kilometres.
  • Vertical dispersion of chemical compounds due to the vertical component of the wind and vertical convection, which is caused particularly by the effect of the sun heating the ground: not very noticeable at night, vertical convection takes place in the morning, creating a mixed layer, which "disappears" by the end of the day.
  • Dry deposition of chemical compounds on the ground. The importance of this deposition process depends on the chemical compound in question, the soil type and the atmospheric conditions. This is a process of loss of chemical compounds, which may be predominant on a large scale for compounds such as ozone.
  • Chemical reactions. The chemical compounds that exist in the atmosphere interact chemically with each other. A specific case of a chemical reaction is solar-energy-activated photolysis reactions. Therefore, the chemistry of ozone causes its precursors, which are nitrogen oxides (NOx) and Volatile Organic Compounds (VOCs) to come into play. The latter, however, include several thousand compounds that cannot be completely taken into account during operational implementation of an air quality model. Simplifying hypotheses are also applied, which aim to reduce the number of chemical compounds and chemical reactions taken into consideration and to optimise computation time.
  the following physical-chemical processes are used

signification symboles
  • For particles, other physical-chemical processes are involved: nucleation, condensation, aggregation... One talks more generally of aerosol microphysics et chemistry. Oxidising reactions in the atmosphere tend to form products with low saturation vapour pressure (therefore highly condensable). These oxidised compounds can be formed by nucleation of new particles or can condense directly from existing particles. Particles group together to form new particles (coagulation process). In addition, particles behave as true catalysts for chemical reactions involving certain gaseous compounds (NO2, N2O5, HO2). Complex aqueous chemistry inside clouds is responsible for the creation of certain sulphates. The particles settle due to gravity, by Brownian diffusion, and are also very efficiently lixiviated by precipitation.

Input data

In order to calculate the flows of production or loss of pollutants related to physical-chemical processes, a certain amount of input data must be provided for the chemistry-transport models, particularly pollutant emissions and meteorological data. These input data must be obtained from a number of bodies.

  • Chemistry-transport models do not produce their own meteorological data, which are necessary to assess the flows related to the processes of dispersion, mixing, deposition, chemical reactions, etc. It is therefore necessary to enter meteorological data as input for the area to be modelled and the period we wish to simulate. In PREV'AIR, models use data provided by meteorological models.
  • The area over which the pollutant concentrations are calculated by the models is not isolated from the space surrounding it: chemical compounds are transported from the "outside" towards the modelling area. The concentrations of pollutants on the borders of the model area are therefore specified in the form of model input data.
  • Precursor emissions, VOCs and NOx, must be entered into the models. These emissions come from localised, linear or surface sources and can be of manmade ("anthropogenic") or natural origin ("biogenic"). The emission data consist of information for the amounts emitted in each sector of the grid, at each moment, and of the chemical composition of such emissions.
  • The dry deposition process of a given chemical compound depends on the soil type. Likewise, biogenic VOC emissions depend on land cover. These inventories (soil type, vegetation species, etc.) are taken into account by the model in the area being studied.
  • Finally, the initial concentrations must be specified as model input data.


AEAT/ENV/R/0545 report, Speciation of UK emissions of NMVOC, N.R. Passant, February 2002

Derognat, C., 2002, Pollution photo-oxydante à l’échelle urbaine et interaction avec l’échelle régionale, thèse de doctorat, Université Paris 6

Hauglustaine, D.A., Brasseur, G.P., Walters, S., Rasch, P.J., Muller, J.-F., Emmons, L.K. and Carroll, M.A., 1998, MOZART : A global chemical transport model for ozone and related chemical tracers, 2. Model results and evaluation, Journal of Geophysical Research, 103, 28291-28336

Horowitz, L.W., Walters, S., Mauzerall, D.L., Emmons, L.K., Rasch, P.J., Granier, C., Tie, X., Lamarque, J.-F., Schultz, M.G. and G.P. Brasseur, 2003, A global simulation of tropospheric ozone and related tracers: Description and evaluation of MOZART, version 2, J. Geophys. Res., in press

Lattuati, M., 1997, Impact des émissions européennes sur le bilan de l’ozone troposphérique à l’interface de l’Europe et de l’Atlantique Nord: apport de la modélisation lagrangienne et des mesures en altitude, thèse de doctorat, Université Paris 6

Simpson, D., Winiwarter, W., Borjesson, G., Cinderby, S., Ferreiro, A., Guenther, A., Hewitt, C.N., Janson, R., Khalil, M.A.K., Owen, S., Pierce, T.E., Puxbaum, H., Shearer, M., Steinbrecher, S., Svennson, B.H., Tarrason, L., and M.G. Oquist, 1999, Inventorying emissions from nature in Europe, J. Geophys. Res., 104 (D7) 8113-8152