Image Processing Reference
data are collected in near-real time and used to compute an air quality index, the
ATMO index (Garcia and Colosio 2001 ). This index aims at informing local
authorities, as well as the public, about air quality in the city. In instances of high
levels of pollution, public authorities are able to implement restrictive measures on
car traffic and on activities of some polluting companies.
However, the actual exposure of persons to ambient pollution cannot be estimated
with the present networks. The costs of a measuring station and its maintenance limit
the generation of index values to specific points of towns. Given the few measuring
stations composing a standard air quality network, an accurate knowledge of the
spatial distribution of the atmospheric pollutants over a city is presently impossible.
Several tools based on numerical modeling of the airflow and chemical transporta-
tion, and conversion provide maps of pollutant concentrations. However, these are
produced at a regional scale with a grid cell of 1-10 km, which is insufficient.
A methodology based on a multisource approach for mapping pollutants concen-
trations over a city has been proposed to overcome this problem (Ung et al. 2002a, b ).
The notion of a “virtual station” is defined, and sources related to air pollution and
urban shapes and morphology are exploited to virtually increase the number of
measuring stations, thus increasing the quality of mapping by interpolation techniques.
The approach was applied to the city of Strasbourg (France) and a ground truth
campaign achieved in June 2003 confirms the validity of the proposed approach.
The proposed methodology used in the above-mentioned exercise has four steps:
(1) the creation of identity cards of each actual measuring station, (2) the evaluation
of the sites of the city in order to create pseudo-stations, (3) the selection of virtual
stations in this set of pseudo-stations, and (4) the creation of a map.
The main idea is to study and evaluate the urban environment as factors influencing
the air pollution in the city. Urban space is a complex domain composed of built-up
areas, roads and streets, bare soil, residential areas, industrial areas, wooden and
parks areas. Atmospheric pollution does not behave the same in each of these areas
(Derbez et al. 2001). Several studies demonstrate the heterogeneity of air pollution and
the influence of the building positions and heights and the street orientations according
to the dominant wind situation. Indeed, dramatic differences in pollutant concentrations
have been observed for two adjacent streets (Derbez et al. 2001; Croxford and Penn
1998 ; Croxford et al. 1996 ; Scaperdas and Colvile 1999 ). Hence, a characterization of
the city's morphology is necessary in order to model its influence on the behavior of
air pollutants. This can be obtained by jointly analyzing and processing images and
databases, and the organization of urban features. It allows the establishment of the
so-called “identity card” of each place of the city.
This ID card comprises a set of attributes, such as:
The geographical position of the area
Its land use