Geography Reference
In-Depth Information
Models and Methods to Represent
and Explore Phenomena on GIS
Anne Ruas
Mapping Phenomena at the District Level: Needs
and Diagnostics
The increasing cost of energy cumulated with the attractiveness of the cities incites
scientists and politicians to imagine smart cities for the future. These smart cities
would concentrate the population and manage, as well as possible, all the drawback
of this concentration such as the noise, the urban heat island effect or the various
chemical pollutions. In the following of the paper we name these effects the
phenomena. They are continuous, they fill the space and their quantity varies
according to space and time.
To compute a phenomenon we need at least some sensors to measure real data
and models to extrapolate measures spatially and temporally. Today many types of
sensors exist and research and technology is focusing on reducing their price and
their size and to encapsulate sensors in boxes that are able to trigger the measure
and to send the result to a server according to appropriate protocols. These boxes
contain a set of sensors like the IGN France GeoCube that can measure, store and
send to a sever the location, the accurate time (GPS module), the wind and any other
measured quantities. All of these kinds of components are fast improving.
To extrapolate measures, two classical types of method exist. Either the sensor
network is dense enough so values can be interpolated and extrapolated with
methods such as the Kriging (using real and virtual stations as in Ung
et al.
2001
), or the phenomenon is built thanks to few measures, a specific model
and wind information for propagation. Models are interesting because they allow
forecasting a phenomenon (see for example the importance of meteorology fore-
casting for everyday life). This last solution requires also fewer sensors, which is
