Geography Reference
In-Depth Information
has been used in the
Immanent
project to compute the propagation of benzene
around a gasoline station from fictive measures (Cheaib et al.
2013
).
Having sensors, protocols and models, we can imagine an easy access to the
representation of phenomena at different levels of detail including the district level,
at any time. But if many web sites are mapping phenomena such as Air Quality,
most of them are spatially not very accurate (e.g. ESRI
2007
). In our research on the
representation of phenomena such as Air Quality or urban heat island effect, we
wish to work at the district level: we wish to represent phenomena in the street,
surrounding the buildings where people are working or living. In order to map real
time data, we need to set architecture to connect all the components (second
section) and we need dedicated models and methods to map the phenomena in a
meaningful way (third and fourth sections).
Improving the Data Flow to Take Advantage of the Existing
Models
Researchers who conceive models are mainly focusing on the quality of the
estimation and on the efficiency of the computation, not on their interfaces. As a
consequence, these models, even though they use geographical information, do not
read nor write OGC standard or even commercial standard such as shape. Figure
1a
illustrates difficulties we had to face in the Immanent project: the system could not
read 3D data in GIS format (we had to digitized existing data through a CAD
software!), and by default, the output is a snap shot capture on a physical viewer
software such as Paraview. These types of viewer are perfect to classify and map
the values is different ways. They are excellent tools for the expert but they are not
adapted for decision making. First of all, there is no geographical coordinate system
(the 3D field is in the image local coordinate system), and the symbolisation is very
far from being rich as a GIS can be. The graphical representation is very poor as
illustrated in Fig.
1a
, limited to the representation of the values of the phenomenon
but neither on its effect nor on its geographical context which is very simplified. We
could say that this kind of representation is not contextualised, there is nearly no
semantic. Nothing but the phenomenon is detailed, although the impact of the
phenomenon depends on its context.
In order to improve the representation as well as the access to the information, it
was necessary to modify the data flow. We introduced the concept of
pollution data
base
in order to store the results, we modified the output according to GIS format
and we developed web services to allow an easy access to the data (Fig.
2
).
The first task consisted in being able to project the output of the model (the
pollution field
) on top of geographical data. A simple solution was developed to
write a set of files that looked like DTMs. Even if the phenomenon is 3D, we
extracted a grid defined by a
grid threshold
and an altitude (
Z0
). The system
