Richard J. Perkins 教授 特別講演会

The concentration of an atmospheric pollutant in a city street depends on numerous phenomena, with a wide range of time and length scales. Some pollutants (vehicle emissions, for example) are emitted and mixed within a street, before escaping into the urban boundary layer. Other pollutants (industrial emissions, for example) can be emitted outside a city, and advected over the city by the prevailing winds, before being entrained into streets. The associated length scales for these processes can therefore vary from the order of a metre or less to the order of several hundred kilometres. The corresponding timescales vary from a minute or less to several days.

For many practical reasons we need to be able to compute the distribution of pollutants within individual streets, and to do this we have to be able to compute the concentrations at all scales. Any practical useful model for doing this needs to be operational:

• It must be able to incorporate and exploit all available data.
• It must be rapid enough to permit the simulation of a wide variety of situations.
• It must be rapid enough to enable authorities to intervene in real time.

Although very detailed topographical data are increasingly available for real urban sites, no single model is capable of reproducing all the different scales, and it is therefore necessary to envisage a system of embedded models, at different physical scales, which reproduce the dominant processes at each scale.

The key to the development of these models is the correct identification of the dominant physical processes at each scale, and this depends strongly on the structure of the flow at the different scales.

We have developed a set of nested models for computing pollution concentrations in city streets, and we have applied it to the Greater Lyon agglomeration. The basic components of the system are:

• A street canyon model - CARMEN
• A street network model - SIRANE
• A city-scale model, based on the EDF code MERCURE
• A regional scale model based on the code SAIMM and UAMS-V

The street canyon model is based on an experimental and numerical study of flow and dispersion in a variety of street geometries; the results of those studies show that the mixing of pollutants within the street, and the exchange of pollutants with the overlying boundary layer, depend on a rather sensitive interaction between the large scale recirculating flow in the street and the turbulence in the mixing layer at the top of the street canyon. A thorough understanding of these processes has enabled us to develop a simple, rapid, yet surprisingly accurate model for the concentration distribution with in the street and the pollutant fluxes into and out of the street. The influence of thermal effects (differential heating of one side of the street, through solar radiation, for example) is not yet fully understood.

The street network model was also developed on the basis of extensive wind tunnel experiments and numerical simulations. These showed how the flow in a network of streets was driven by the external wind field, and how the flow in street intersections redistributed pollutants within the connecting streets. In particular, although the flow in the intersection is strongly three dimensional, the redistribution of pollutants can be modelled satisfactorily in a surprisingly simple way.

The entire modelling system has been used to compute pollutant concentrations at different locations in the agglomeration, for real conditions, and the results have been compared with measurements made by pollution monitors. The meteorological data was provided from measurements, and the traffic emissions were estimated from a traffic model for the agglomeration, coupled with a standard vehicle emissions model.

The comparisons between the computations and the measured concentrations show that:

• It is necessary to include all the different scales of transport and dispersion to reproduce the measurements
• Once the entire range of processes is included, the model results agree rather well with the measured concentrations.
• Local emissions, and therefore small-scale dispersive processes, influence strongly the concentrations throughout the city.